Attorney Docket No.4210.0527WO DIAGNOSIS OF PATIENT TUMOR TISSUE GRANT STATEMENT [0001] This invention was made with government support under Grant Number TR003715 awarded by the National Institutes of Health. The government has certain rights in the invention. CROSS-REFERENCE TO RELATED APPLICATIONS [0002] The present application claims priority to U.S. Provisional Patent Appl. No. 63/404,866 to Satterlee et al., entitled “A Normalized Ex Vivo Platform For Functional Precision Diagnosis Of Patient Tumor Tissue”, filed on September 8, 2022, and to U.S. Provisional Patent Appl. No. 63/464,729 to Satterlee et al., entitled “A Normalized Ex Vivo Platform For Functional Precision Diagnosis Of Patient Tumor Tissue”, and filed on May 8, 2023, the disclosures of which are incorporated herein by reference in their entireties. TECHNICAL FIELD [0003] The subject matter disclosed herein relates generally to a normalized ex vivo platform for functional precision diagnosis of patient tumor tissue. More particularly, the subject matter disclosed herein relates to methods of diagnosing a tumor and/or screening for a therapeutic for a tumor, including using a living tissue substrate and an algorithm for determining a drug sensitivity score. BACKGROUND AND INTRODUCTION [0004] Effective precision diagnosis to guide brain cancer treatment is a critical unmet need. Genomic tumor profiling often lacks actionable outputs, while many in vitro and in vivo models of patient disease lack the accuracy or speed to provide timely, relevant information to guide patient care. Between 2006 and 2020, a time period during which an astounding number of new cancer-directed drugs were developed, eligibility for those drugs only increased from around 5% to 13%, and response to those drugs only increased from around 3% to 7%. The reasons for this are myriad, and include factors such as co- occurring oncogenic alterations, tumor heterogeneity, epistatic interactions, and adaptive
Attorney Docket No.4210.0527WO cellular circuitry. It is becoming increasingly clear that static histopathological and molecular measurements of tumors are still insufficient to most accurately and usefully classify tumors and cancers, and that to identify precision medicines for the majority of cancer and tumor patients, functional diagnostic platforms are needed. [0005] Patient-derived models of cancer (PDMCs), including cell lines, patient-derived organoids (PDOs), patient-derived explants (PDEs), and patient-derived xenografts (PDXs), provide functional models of a patient’s individual tumor that can be screened with multiple drugs. The potential for PDMCs to guide personalized care has been demonstrated in several studies which show their ability to successfully predict antitumor response. Indeed, drug screening assays using PDMCs have successfully predicted antitumor responses in humans, demonstrating their potential. Unfortunately, initiation time, cost, efficiency scales, and similarity to the parent tumor limit applications of PDXs, while patient-derived cell lines often lose the genetic and phenotypic heterogeneity of the parent tumor via cell selection during clonal expansion. Furthermore, generating these models from heterogeneous tumors also limits reproducibility among intra-tumor replicates. A platform which supports and rapidly engrafts both low- and high-grade tumor tissue, maintains genetic heterogeneity and resemblance to the parent tumor, and allows functional testing of approved and experimental therapeutics is still desperately needed. SUMMARY [0006] This summary lists several embodiments of the presently disclosed subject matter, and in many cases lists variations and permutations of these embodiments. This summary is merely an example of the numerous and varied embodiments. Mention of one or more representative features of a given embodiment is likewise for purposes of example. Such an embodiment can typically exist with or without the feature(s) mentioned; likewise, those features can be applied to other embodiments of the presently disclosed subject matter, whether listed in this summary or not. To avoid excessive repetition, this summary does not list or suggest all possible combinations of such features. [0007] Provided in some embodiments is a functional precision medicine that improves pre-clinical drug testing and guides clinical decisions. More particularly, in some embodiments, provided is a living tissue substrate (LTS) platform, in some embodiments an organotypic brain slice culture (OBSC)-based platform, and multi-parametric algorithm which enables rapid engraftment, treatment, and analysis of uncultured patient brain tumor
Attorney Docket No.4210.0527WO tissue and patient-derived cell lines. The platform supports engraftment of any patient tumor, including for example, but not limited to, high- and low-grade adult and pediatric tumor tissue rapidly establish on OBSCs among endogenous astrocytes and microglia while maintaining the tumor’s original DNA profile. The disclosed algorithm calculates dose-response relationships of both tumor kill and LTS toxicity, generating summarized drug sensitivity scores based on therapeutic window and allowing for the normalization of response profiles across a panel of FDA-approved and exploratory agents. Furthermore, in some embodiments, summarized patient tumor scores after LTS treatment show positive associations to clinical outcomes, demonstrating that the LTS platform can provide rapid, accurate, functional testing to ultimately guide patient care. [0008] In some embodiments, provide are methods of diagnosing a tumor and/or screening for a therapeutic for a tumor, the methods comprising providing a living tissue substrate (LTS), engrafting one or more tumor tissue cells to the LTS, wherein the one or more tumor tissue cells comprise tumor tissue and/or tumor cells obtained from a subject, and analyzing a dose-response of a candidate therapeutic for both tumor tissue cell kill and LTS toxicity, whereby the tumor is diagnosed or a candidate therapeutic to treat the tumor is identified. In some embodiments, the LTS comprises a tissue selected from brain, mesentery, kidney, liver, lung, bone and spinal cord. In some embodiments, the LTS comprises brain tissue, optionally an organotypic brain slice culture. In some embodiments, the LTS comprises mesentery tissue, optionally an organotypic mesentery membrane culture. [0009] In some embodiments, the one or more tumor tissue cells are derived from a primary or metastatic tumor of a subject, optionally wherein the one or more tumor tissue cells are dissociated into small pieces, and transfected with a reporter, prior to seeding onto the LTS, optionally wherein the tumor is a brain tumor or ovarian tumor. In some embodiments, the one or more tumor tissue cells are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and labelled with a fluorescent reporter, prior to seeding onto the LTS. In some embodiments, the one or more tumor tissue cells are engrafted to the LTS and tested for drug sensitivities, with assay completion in less than 10 days, optionally less than 5 days, optionally less than 4 days, optionally less than 3 days, optionally less than 2 days. In some embodiments, the genetic drift of the one or more tumor tissue cells is minimized due to the rapid engraftment, optionally wherein the genetic drift is less than about 50%, optionally less than about 25%,
Attorney Docket No.4210.0527WO optionally less than about 10%, at the time of diagnosing and/or screening, optionally wherein the mutational profile of the one or more tumor tissue cells is substantially similar to the mutational profile in vivo. [0010] In some embodiments, such methods further comprise providing a patient in need of treatment and/or having a tumor, and collecting a biopsy from the patient as the source of the one or more tumor tissue cells. In some embodiments, the one or more tumor tissue cells are cryopreserved after biopsy and thawed prior to engraftment on the LTS, optionally wherein the cryopreserved tumor tissue cells are preserved for a plurality of sequential and/or simultaneous applications of the method. In some embodiments, the cryopreserved tumor tissue cells are not exposed to an enzyme during dissociation. [0011] In some embodiments, analyzing a dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity comprises calculating a drug sensitivity score (DSS), wherein the DSS is optionally calculated by comparing tumor cell survival, optionally measured via bioluminescence imaging, to health of the LTS, optionally measured via Propidium Iodide (PI) assay. In some embodiments, DSS is calculated based on a plurality of weighted parameters, optionally a combination of all weighted parameters, selected from the group consisting of: (1) killing at maximum dose (Max Kill), (2) dose required to kill 10% of the tumor (EC10), (3) dose required to kill 25% of the tumor (EC25), (4) dose required to kill 50% of the tumor (EC50), (5) dose required to kill 75% of the tumor (EC75), (6) dose required to kill 90% of the tumor (EC90), (7) slope through the EC50, (8) the area under the curve (AUC), (9) tumor growth acceleration, (10) biphasic killing (rapid killing at low doses and limited additional killing at higher doses), and (11) incomplete kill at the highest dose, optionally wherein a DSS from 0 to 100 signifies increasing efficacy in tumor kill relative to LTS toxicity, and wherein a DSS from 0 to - 100 signifies increasing LTS toxicity relative to tumor kill. In some embodiments, each of the parameters is weighted at about 1% to about 45% in the DSS calculation, optionally wherein each is weighted as follows: Maximum Kill (10%), EC10 (5%), EC25 (5%), EC50 (10%), EC75 (5%), EC90 (5%), Slope through IC50 (10%), Tumor Growth Acceleration (5%; not compared to LTS toxicity), Biphasic Killing Curve (5%; not compared to LTS toxicity), Incomplete Kill (5%; not compared to LTS toxicity), and Area Under the Curve (35%). In some embodiments, the analysis of a dose-response of a candidate therapeutic for both tumor tissue cell kill and LTS toxicity is done substantially simultaneously. In some embodiments, the identified candidate therapeutic to treat the tumor comprises a
Attorney Docket No.4210.0527WO pharmaceutically active agent, a chemotherapeutic composition, a small molecule, an immunotherapeutic agent, an inhibitor, a radiation therapy, and combinations thereof. [0012] In some embodiments, provided is a functional precision diagnostic method, the method comprising performing the disclosed methods for diagnosing a tumor or screening for a therapeutic for a tumor, and further comprising iteratively testing additional therapeutics on cryopreserved patient tumor cells before administration to a subject, whereby a treatment can be adapted based on a DSS output. In some embodiments, such methods further comprise testing combinatorial therapies using LTS and DSS. In some embodiments, the methods comprise performing, by a module implemented using a non- transitory computer readable medium, the simultaneous analyzing of a dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity, including calculating a Drug sensitivity score (DSS). In some embodiments, the LTS is cultured in a multi-well format. [0013] In some embodiments, the presently disclosed subject matter comprises a diagnostic and/or therapeutic screening system, comprising a living tissue substrate (LTS), optionally cultured in a multi-well format, one or more tumor tissue cells engrafted to the LTS, optionally wherein the tumor tissue cells are dissociated into small pieces from a tumor biopsy or tumor resection tissue, transfected with a reporter, and seeded onto the LTS, and a multi-parametric algorithm which simultaneously analyzes a dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity. In some embodiments, the LTS comprises a tissue selected from brain, mesentery, kidney, liver, lung, bone and spinal cord. In some embodiments, the LTS comprises brain tissue, optionally an organotypic brain slice culture. In some embodiments, the LTS comprises mesentery tissue, optionally an organotypic mesentery membrane culture. [0014] In some embodiments, in the disclosed systems the one or more tumor tissue cells are derived from a primary or metastatic tumor of a subject, optionally wherein the one or more tumor tissue cells are dissociated into small pieces, and transfected with a reporter, prior to seeding onto the LTS, optionally wherein the tumor is a brain tumor or ovarian tumor. In some embodiments, the one or more tumor tissue cells are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and labelled with a fluorescent reporter, prior to seeing onto the LTS. In some embodiments, the one or more tumor tissue cells are engrafted to the LTS and tested for drug sensitivities, with assay completion in less than 10 days, optionally less than 5 days, optionally less than 4
Attorney Docket No.4210.0527WO days, optionally less than 3 days, optionally less than 2 days. In some embodiments, the genetic drift of the one or more tumor tissue cells is minimized due to the rapid engraftment, optionally wherein the genetic drift is less than about 50%, optionally less than about 25%, optionally less than about 10%, at the time of diagnosing and/or screening, optionally wherein the mutational profile of the one or more tumor tissue cells is substantially similar to the mutational profile in vivo. [0015] In some embodiments, in the disclosed systems, simultaneously analyzing a dose- response of a candidate therapeutic for both tumor cell kill and LTS toxicity comprises calculating a Drug sensitivity score (DSS), wherein the DSS is optionally calculated by comparing tumor cell survival, measured via bioluminescence imaging, to health of the LTS, measured via Propidium Iodide (PI) assay, and wherein the system further comprises a computer program product comprising computer-executable instructions embodied in a computer-readable medium for performing the method steps of calculating a Drug sensitivity score (DSS). In some embodiments, DSS is calculated based on a plurality of weighted parameters, optionally a combination of all weighted parameters, selected from the group consisting of: (1) killing at maximum dose (Max Kill), (2) dose required to kill 10% of the tumor (EC10), (3) dose required to kill 25% of the tumor (EC25), (4) dose required to kill 50% of the tumor (EC50), (5) dose required to kill 75% of the tumor (EC75), (6) dose required to kill 90% of the tumor (EC90), (7) slope through the EC50, (8) the area under the curve (AUC), (9) tumor growth acceleration, (10) biphasic killing (rapid killing at low doses and limited additional killing at higher doses), and (11) incomplete kill at the highest dose, optionally wherein a DSS from 0 to 100 signifies increasing efficacy in tumor kill relative to LTS toxicity, and wherein a DSS from 0 to - 100 signifies increasing LTS toxicity relative to tumor kill. In some embodiments, the parameters is weighted at about 1% to about 25% in the DSS calculation, optionally wherein each is weighted as follows: Maximum Kill (10%), EC10 (10%), EC25 (10%), EC50 (15%), EC75 (10%), EC90 (10%), Slope through IC50 (10%), Tumor Growth Acceleration (5%; not compared to LTS toxicity), Biphasic Killing Curve (5%; not compared to LTS toxicity), Incomplete Kill (5%; not compared to LTS toxicity), and Area Under the Curve (10%). [0016] In some embodiments, provided herein are methods of treating a subject, the method comprising performing a method of diagnosing a tumor as disclosed herein, and administering to the subject a treatment based on the diagnosis. In some embodiments, the
Attorney Docket No.4210.0527WO subject is a mammal, optionally wherein the subject is a human. In some embodiments, the treatment comprises a combinatorial treatment. [0017] In some implementations, the current subject matter relates to a computer implemented method for diagnosing a patient tumor tissue. The method may include receiving, using at least one processing circuitry, a first image of a living tissue substrate (LTS) engrafted with one or more tumor tissue cells, identifying, using a computer vision (CV) algorithm, the one or more tumor tissue cells, receiving a second image of the LTS, the second image being subsequent to the first image and subsequent to an application of a first candidate therapeutic in a plurality of candidate therapeutics, determining, based on an analysis of the second image, a tumor tissue cell kill parameter of the first candidate therapeutic, receiving a third image of the LTS without an engrafted tumor, where the LTS has been treated with the first candidate therapeutic, determining, based on an analysis of the third image, a toxicity of the first candidate therapeutic against the LTS, and generating, using a machine learning (ML) model, a drug sensitivity score (DSS) for the first candidate therapeutic and a type of the tumor tissue cells based on the toxicity and the tumor tissue cell kill parameter. [0018] In some implementations, the current subject matter may include one or more of the following optional features. The identifying may include determining, using the CV algorithm, a region of interest associated with the one or more tumor tissue cells, and generating, using the CV algorithm, a mask for the one or more tumor tissue cells. [0019] In some implementations, the identifying may include identifying, using the CV algorithm, one or more tumor tissue cells based on a brightness of one or more pixels associated with the one or more tumor tissue cells and a brightness of one or more pixels associated with the LTS in the first image. The brightness may represent an amount of light emitted by the one or more tumor tissue cells. [0020] In some implementations, the identifying may include determining, using the CV algorithm, the region of interest and the mask based on a brightness of one or more pixels associated with the one or more tumor tissue cells and a brightness of one or more pixels associated with the LTS in the first image. The brightness may represent an amount of light emitted by the one or more tumor tissue cells or an amount of light emitted by the LTS. [0021] In some implementations, the first image may be received prior to the application of the first candidate therapeutic.
Attorney Docket No.4210.0527WO [0022] In some implementations, the DSS may be generated by the ML model based on one or more respective weights applied to a plurality of parameters, the one or more weights of the ML model are trained based on a training data, the training data including a plurality of images of at least one another LTS engrafted with other tumor tissue cells. The plurality of parameters may include at least one of the following: a killing at maximum dose (Max Kill) parameter, a dose required to kill 10% of the tumor (EC10), a dose required to kill 25% of the tumor (EC25), a dose required to kill 50% of the tumor (EC50), a dose required to kill 75% of the tumor (EC75), a dose required to kill 90% of the tumor (EC90), a slope through the EC50, an area under the curve (AUC), a tumor growth acceleration, an incomplete kill parameter at the highest dose, a biphasic killing parameter corresponding to a rapid killing at low doses and limited additional killing at higher doses, and any combination thereof. One or more initial weights of the parameters include at least one of the following: a 10% weight for the Max Kill parameter, a 5% weight for the EC10 parameter, a 5% weight for the EC25 parameter, a 10% weight for the EC50 parameter, a 5% weight for the EC75 parameter, a 5% weight for the EC90 parameter, a 10% weight for the slope through IC50 parameter, a 5% weight for a tumor growth acceleration parameter, a 5% weight for a biphasic killing curve parameter, a 5% weight for an incomplete kill parameter, a 35% weight for the AUC parameter, and any combination thereof. The ML model may be configured to be trained by modifying at least one of the initial weights of the parameters and by, optionally, removing at least one of the parameters. [0023] In some implementations, a DSS from 0 to 100 may correspond to increasing efficacy in tumor kill relative to LTS toxicity. A DSS from 0 to -100 may correspond to increasing LTS toxicity relative to tumor kill. A negative DSS score may correspond to near zero LTS toxicity and increased tumor growth. [0024] In some implementations, the method may further include generating, using the CV algorithm applied to the second image, one or more masks of the one or more tumor tissue cells depicted in the second image. The tumor tissue cell kill of the first candidate therapeutic may be based on the one or more tumor tissue cells depicted in the second image. The DSS may be generated based on at least the tumor tissue cell kill measured using the one or more masks of the one or more tumor tissue cells depicted in the second image. The CV algorithm may be configured to be trained to identify the LTS and overlay one or more bioluminescence values to determine a signal corresponding to the one or
Attorney Docket No.4210.0527WO more tumor tissue cells. One or more tumor tissue cells depicted in the second image may include a first tumor spot and a second tumor spot. In some implementations, the method may further include bisecting, using the CV algorithm, the mask of the one or more tumor tissue cells depicted in the second image into a first portion including the first tumor spot and a second portion including the second tumor spot; determining, based on the first portion of the mask, a radiance of the first tumor spot; and determining, based on the second portion of the mask, a radiance of the second tumor spot. The tumor tissue cell kill of the first candidate therapeutic may be determined based on the first portion of the mask for the first tumor spot and the second portion of the mask for the second tumor spot. The DSS may be generated, using the ML model, based on the first portion of the mask and the second portion of the mask. [0025] In some implementations, each of the plurality of candidate therapeutics may be applied to respective LTS engrafted with a respective tumor tissue cell, wherein a respective tumor tissue cell kill of the respective candidate therapeutic is determined based on respective first and second images of the LTS. Each of the plurality of candidate therapeutics may be configured to be applied to respective LTS without engrafted tumor tissue cells, wherein a respective LTS toxicity of the respective candidate therapeutic is determined based on the third images of the LTS. A respective DSS for each candidate therapeutic may be generated, using the ML model, based on the respective LTS toxicity and the respective tumor tissue cell kill of the respective candidate therapeutic, wherein the first candidate therapeutic may be selected based on the DSS scores for each candidate treatment. [0026] In some implementations, the mask may include a plurality of attributes of the one or more tumor tissue cells, wherein the plurality of attributes include at least one of the following: a size of the one or more tumor tissue cells, a location of the one or more tumor tissue cells, an intensity of light emitted by the one or more tumor tissue cells, and any combination thereof. [0027] In some implementations, the first image may be a tumor fluorescence image obtained at a first predetermined time. The second image may be a tumor bioluminescence image obtained a second predetermined time. The third image may be an organotypic culture (e.g., organotypic brain slice culture (OBSC)) fluorescence image obtained at a third predetermined time. At least one of the second and third predetermined times may occur after the first predetermined time. The DSS may be generated, using the ML model,
Attorney Docket No.4210.0527WO based on one or more measurements made across at least one of: the first image, the second image, the third image, and any combination thereof. [0028] In some implementations, the method may further include generating, using the CV algorithm applied to the third image, one or more masks of the one or more LTS depicted in the third image, wherein the LTS toxicity of the first candidate therapeutic may be determined using the one or more masks of the one or more LTS depicted in the third image. The DSS may be generated, using an ML model, based at least on the values of LTS toxicity found by using the mask(s) of the one or more LTS depicted in the third image, wherein the CV algorithm is configured to be trained to identify the LTS and overlay fluorescence values to determine a signal corresponding to the one or more LTS. [0029] Non-transitory computer program products (i.e., physically embodied computer program products) are also described that store instructions, which when executed by one or more data processors of one or more computing systems, causes at least one data processor to perform operations herein. Similarly, computer systems are also described that may include one or more data processors and memory coupled to the one or more data processors. The memory may temporarily or permanently store instructions that cause at least one processor to perform one or more of the operations described herein. In addition, methods can be implemented by one or more data processors either within a single computing system or distributed among two or more computing systems. Such computing systems can be connected and can exchange data and/or commands or other instructions or the like via one or more connections, including but not limited to a connection over a network (e.g., the Internet, a wireless wide area network, a local area network, a wide area network, a wired network, or the like), via a direct connection between one or more of the multiple computing systems, etc. [0030] The details of one or more variations of the subject matter described herein are set forth in the accompanying drawings and the description below. Other features and advantages of the subject matter described herein will be apparent from the description and drawings, and from the claims. [0031] Although some of the aspects of the subject matter disclosed herein have been stated hereinabove, and which are achieved in whole or in part by the presently disclosed subject matter, other aspects will become evident as the description proceeds when taken in connection with the accompanying drawings as best described herein below.
Attorney Docket No.4210.0527WO BRIEF DESCRIPTION OF THE DRAWINGS [0032] The features and advantages of the present subject matter will be more readily understood from the following detailed description which should be read in conjunction with the accompanying drawings that are given merely by way of explanatory and non- limiting example, and in which: [0033] Figure 1 is a schematic illustration of a living tissue substrate (LTS) culture system as disclosed herein, and used for testing therapeutics against a tumor or cancer of interest. [0034] Figure 2 is a schematic illustration of a procedure for LTS-organotypic brain slice culture (OBSC) engraftment. Top row depicts OBSC generation, where rat pup brains are dissected, sliced via vibratome, and plated. Bottom row depicts the patient tumor preparation procedure, where resected patient tumors are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and stained with far-red lipid-soluble dye, and added to OBSCs. [0035] Figure 3 is a schematic illustration of a procedure for LTS-mesentery (M) engraftment. Top row depicts mesenteric surgery and plating. Bottom row depicts the disclosed patient tumor preparation procedure, where resected patient tumors are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and stained with far-red lipid-soluble dye, and added to mesentery. [0036] Figures 4A-4G show results of experiments conducted for OBSC characterization, standardization, and quality control. Fig.4A) Left: Brightfield and fluorescence images of standard healthy and dead (positive control; frozen in EtOH overnight) OBSCs, via PI stain. Right: Average fluorescence from standard OBSCs (n=287 biological replicates) and positive control OBSCs (n=131 biological replicates) OBSCs. Fig. 4B) Effect of pup age on OBSC viability (n≥6 per day), analyzed using one-way ANOVA **** P<0.0001. Fig. 4C) Viability comparison of healthy OBSCs vs those damaged during dissection and/or slicing (n=60 biological replicates for healthy, n=36 biological replicates for damaged), analyzed using Welch’s T-Test P= 0.0172. Fig. 4D) Viability of OBSCs based on brain slice media (BSM) quality and promptness during slicing (n=12 biological replicates per group; slicing delay after dissection=1h), analyzed using one-way ANOVA ** P=0.0052, **** P<0.0001. Fig. 4E) Representative batch-to-batch viability from n≥6 biological replicates randomly sampled OBSCs from each batch (n ~ 150 OBSCs per batch). Fig. 4F) PI signal from OBSCs measured on different days after slicing (n=6
Attorney Docket No.4210.0527WO biological replicates). Fig. 4G) 10X immunofluorescent maximum intensity projection images of healthy OBSCs showing activity level of astrocytes (GFAP), neurons (NeuN), and microglia (CD11b) immediately after slicing and on D4. All data, except where otherwise noted, were collected 4 days after slicing. [0037] Figures 5A-5I show the results of tumor growth and interaction on OBSCs. Fig. 5A) In vitro cell growth in 96h. 1500 cells were seeded into 96 well plates and measured on day 0 and 4 via BLI, n = 6 technical replicates. Fig. 5B) In vivo tumor growth in mice following intracranial implantation of 250,000 tumor cells, n = 5 biological replicates. Fig. 5C) Representative fluorescence image of 4 MB231Br tumor foci seeded into the thalamic region of two OBSCs within one well. Fig. 5D) Representative bioluminescence image depicting tumor seeding of 24 tumor foci onto 12 OBSCs in one six-well plate. Fig. 5E) Cell line growth in 96h on OBSCs. Bioluminescence images were taken on day 0 and 4. Day 0 fluorescence images were used to normalize initial tumor size. n = 4 biological replicates. Fig. 5F) Migratory behavior from initial tumor engraftment site on OBSCs, t = 96h. Circles larger than the normalized tumor circumference indicate outward migratory behavior while smaller circles indicate inward retraction. Fig. 5G) 10X immunofluorescent maximum intensity projection images of astrocytes and tumor interaction: astrocytes stained by GFAP (green) without or in the presence of GBM8 tumor cells (red) 96h after engraftment. Fig. 5H) Growth of GBM8 on OBSCs seeded early (day of slicing) or seeded late (7 days post slicing). Fig. 5I) Diameter change of GBM8 on OBSCs seeded early or late. [0038] Figures 6A-6D show tumor killing on OBSCs. Fig. 6A) Top, IC50s calculated based on linear interpolation of dose-response data on OBSCs. Cells were seeded on day 0 and dosed with therapeutics on day 1; survival was measured via bioluminescence on day 4. Concentrations of small molecule drugs are given in µM; XRad dose is given in Gy. NR indicates the IC50 was not reached within the dose range. Bottom, graphical representation of IC50s of all drugs vs all tumor lines. Fig. 6B) Killing of MB231Br, LN229, U373WT, and U373KO by TR107 on OBSCs and in vitro. Fig. 6C) Combination therapy of radiation and subsequent temozolomide against GBM8 and MS21. Fig. 6D) Combination therapy of etoposide and carboplatin against U373WT, U373KO, MB231Br, and PDIPG. [0039] Figures 7A-7D show results of drug sensitivity score algorithm and array. Fig. 7A) Dose-response curves of GBM8, U373 KO, MS21, and LN229 on OBSCs t = 3 days
Attorney Docket No.4210.0527WO after administration of etoposide. n = 4 biological replicate tumor foci per dose, 6 doses per cell line. Fig. 7B) Therapeutic windows across all DSS parameters for each treated tumor line from Fig. 6A. 7C) Dose-response curves of U373WT (red), U373KO (blue), and OBSC (black) against ten therapeutics. Fig.7D) DSS array for all cell lines against all drugs. DSS from 0 to 100 signify increasing efficacy in tumor kill relative to OBSC toxicity, while scores from 0 to -100 describe scenarios in which tumors thrive more effectively than OBSCs for a given treatment. [0040] Figures 8A-8E show patient tumor tissue on OBSCs. Fig. 8A) Qualitative (left) and quantitative (right) BLI of three different patient brain tumor tissues cultured in vitro, on transwell insert and on OBSC 4 days after seeding; (t-test, **p<0.001,****p<0.0001, n=12 biological replicates). Fig. 8B) Bioluminescence of each patient tumor t = 4 days after seeding on OBSCs (n=4 biological replicates per tumor). Fig. 8C) 10X immunofluorescent maximum intensity projection images of astrocytes and patient tumor interaction: astrocytes stained by GFAP (green) without or in the presence of PGBM patient tumor (red via mCherry expression) 96h after engraftment. Fig. 8D) Schematic of experimental design for DNA sequencing of patient tumor tissue (MG-II). Fig. 8E) Heatmap of the top 250 most significantly mutated somatic genes showing maintenance of HT profile in BSHT but not in CL, (top left), top 25 most significantly mutated genes, (top right; starred row = NF2), and mutations within the NF2 gene on chromosome 22 in HT, CL and BSHT (bottom). BSHT, n=4 biological replicates; HT, n=3 technical replicates; CL, n=3 biological replicates. [0041] Figures 9A-9D show further data demonstrating patient tumor tissue on OBSCs. Fig. 9A) Reproducibility in PGBM-R patient tumor seeding via Cell Tracker fluorescence t = 1h after seeding on OBSCs. Fig. 9B) Reproducibility in persistence of unique tumor foci from MMG-II patient tumor 4, 6, and 8 days after seeding, n = 6 biological replicates. Fig. 9C) Reproducibility in persistence of unique tumor foci from fresh and cryopreserved/thawed (frozen) tissue from the same GG-I patient tumor sample 4 and 8 days after seeding on OBSCs, n=6 biological replicates. Fig. 9D) Reproducibility in survival of fresh and cryopreserved/thawed (frozen) tumor tissue from the same GG-I patient tumor sample t = 3 days after treatment with carboplatin, azacitidine and trametinib, n = 4 biological replicate foci per dose. [0042] Figures 10A-10C: Combined DSS results. Fig. 10A) Combined DSS array for all cell lines and patient tumor tissue against all drugs. Tumor line data from Figure 7 is
Attorney Docket No.4210.0527WO repeated here for comparison. DSS from 0 to 100 signify increasing efficacy in tumor kill relative to slice toxicity, while scores from 0 to -100 describe scenarios in which tumors thrive more effectively than OBSCs for a given treatment. Gray boxes indicate drugs that were not tested against a certain tumor. Fig.10B) Waterfall plot of all 145 DSS presented in (A). DSS from established tumor lines are represented as blue lines; DSS from patient tumor tissues are represented as red lines. Fig. 10C) Waterfall plots depicting relative sensitivities to individual therapeutics. All tumors treated with each therapeutic are included. Bottom right: Head-to-head comparison of DSS from drugs used to treat PGBM and PGBM-R. Yellow lines = DSS for PGBM; green lines = DSS for PGBM-R; blue lines = DSS for established tumor lines; red lines = DSS for patient tumor tissue. [0043] Figures 11A-11C show the results of the testing and analysis of patient ovarian cancer tumor on living tissue substrates (LTS). The testing and evaluation included tumor engraftment, drug treatment and DSS calculation. Fig.11A shows dose response curves of LTS toxicity following a 3-day exposure to carboplatin. Fig. 11B shows bioluminescence of ovarian patient tumor t= 4 days after engraftment on LTS. Fig.11C shows dose response of ovarian patient tumor engrafted on LTS t=3 days after exposure to carboplatin. [0044] Figures 12A-12K show the results of OMMCs generation. Fig.12A is a schematic illustration of OMMCs generation. Fig. 12B shows region of interest from above view of the isolated mesentery (green drawing) and display of its net of cells and extracellular components by light microscopy and H&E staining. Figs. 12C and 12D show survival of 8-week-old rat mesentery on OMMCs. Fig. 12C shows BLI tracking of the transduced mesentery over a 10-day period. Fig. 12D shows survival of mesentery over a 17-day period using the PI assay. Fig. 12E shows mesentery killing by gradual increase in DMSO concentrations. The top right shows PI fluorescence measured with the AMI optical system and the bottom left and right sides shows the PI fluorescence from dead cells when exposed to 0% and 100% DMSO respectively. Consistent mesentery survival is observed and confirmed by PI assay across experiments as shown in Fig. 12F. Fig. 12G shows similar down trend in mesentery survival became significant after DAY 11 for all ages evaluated, except for 3 and 4 weeks old where the PI fluorescence was not measurable from Day 5 on. Fig. 12H is a photograph from three different mesentery ages showed a shrinkage of the region of interest for 3- and 4-week-old mesenteries on Day 8. Fig. 12I shows two regions of the rat mesentery were selected to determine cell count and membrane thickness, Ileum and Jejunum using PI staining and confocal imaging. There was a
Attorney Docket No.4210.0527WO homogenous number of cells in both regions not showing a significant difference (Fig. 12J), however, a difference was noticed in thickness in between the regions (Fig. 12K). [0045] Figures 13A-13G show tumor spots on OMMCs. Fig. 13A shows tumor seeding process on OMMC. Fig. 13B shows light, fluorescent and BLI pictures from above view of tumor spots on a mesentery with a magnified display of a well-rounded tumor spot. Fig. 13C shows ES-2 and SKOV3 showed consistent tumor growth on OMMC in 10 days (about 1 and a half weeks). Fig. 13D shows reproducibility in tumor growth for ES- and SKOV3 leading to survival above 100% across separate experiments. Fig. 13E shows minimal inter-well variability (>600 multiple comparisons for 36 wells) after manual tumor seeding on Day 0. Fluorescent imaging confirmed a clear potential macrophage activation when tumor is present. [0046] Figures 14A-14E show tumor drug response, drug toxicity on the mesentery and drug sensitivity score (DSS). Fig. 14A shows drug exposure effect on tumorless OMMCs survival, using increasing concentrations of FDA approved single and combination chemotherapies (Olaparib, Gemcitabine, Carboplatin, Paclitaxel, Paclitaxel-Carboplatin 10 and 100) in a 3-day period. Fig. 14B is a visual of schematic (Top) and real (Bottom) of OMMC system with ES-2 tumor spots suggesting its potential functionality to assess toxicity and tumor drug response by BLI quantification. Fig. 14C shows tumor drug- response curves on OMMCs along with mesentery viability after 3-day exposure to the same group of chemotherapies. Fig. 14D is an example of calculated DSS for the two cell lines against Gemcitabine and DSSs array for both cell lines against all drugs from 0 to 100 suggest increasing efficacy in tumor kill relative to OMMC toxicity, while scores from 0 to −100 describe scenarios in which tumors thrive more effectively than OMMC for a given treatment. Fig. 14E is an example of therapeutic window across all DSS weighted parameters for Gemcitabine treated tumor cell line. Values ranged from-1 to +1, where values approaching+1 indicate better tumor kill relative to less toxicity on the tissue, and values approaching −1 suggest tumors remained viable while toxicity to the normal OMMCs tissue was elevated. [0047] Figures 15A-15D show OC biopsies on OMMCs. Fig. 15A includes mean values of tumor growth on OMMMs showing all patient tumors stay alive and even proliferate for some of them in a 6-day period. Fig. 15B is a comparison of patient OC tumor growth in different cultured systems where OMMCs suggest a better tumor substrate. Fig. 15C includes two examples of inter-well variability of the human OC tumor spots at the time
Attorney Docket No.4210.0527WO of placement on the mesentery membrane, showing there was a consistent tumor cell manipulation with no significant difference in BLI values inter-well. Fig. 15D shows significant tumor response on OMMCs to 500uM of chemotherapies. [0048] Figure 16 shows all biopsy tumor response curves on OMMCs per individual treatment and their corresponding therapeutic window across all DSS parameters were calculated. The DSSs array for both cell lines against all drugs from 0 to 100 suggest increasing efficacy in tumor kill relative to OMMC toxicity, while scores from 0 to −100 describe scenarios in which tumors thrive more effectively than OMMC for a given treatment. [0049] Figures 17A-17E show how various LTS originating from other organs such as kidney, liver, and lung have been developed to engraft, treat, and analyze treatment response of various tumor cell lines and uncultured patient tumor tissue samples. Fig.17A shows growth of tumor cell lines of various origin on LTS from brain, kidney, liver, and lung. Fig. 17B shows dose-response curves of Lomustine vs LN229 tumor cells growing on LTS from liver, brain, and kidney. Fig.17C shows off-target toxicity of Lomustine and Azacitidine against LTS from liver and kidney. Fig. 17D shows brightfield images of various tissue substrates in the disclosed LTS systems. Figure 17E displays images of engrafted patient tumor tissue surviving on various tissue substrates in the disclosed LTS systems at t = 4 days after engraftment. This data, alongside data in Fig 11, provides additional evidence that various LTS systems generated as described in this disclosure can be used to engraft, treat, and analyze uncultured patient tumor tissue. FIG. 18 illustrates an example system for diagnosing of patient tumor, according to some implementations of the current subject matter. [0050] FIG. 19 illustrates a system in accordance with some implementations of the current subject matter. [0051] FIG. 20 illustrates an apparatus in accordance with some implementations of the current subject matter. [0052] FIG. 21 illustrates an artificial intelligence architecture in accordance with some implementations of the current subject matter. [0053] FIG. 22 illustrates an artificial neural network in accordance some implementations of the current subject matter. [0054] FIG. 23 illustrates further details of the image processing engine, according to some implementations of the current subject matter.
Attorney Docket No.4210.0527WO [0055] FIG. 24 illustrates an example process for processing of images by the image processing engine, according to some implementations of the current subject matter. [0056] FIG. 25 illustrates an example process for diagnosing patient tumor tissue, according to some implementations of the current subject matter. [0057] FIG.26 illustrates an example process, according to some implementations of the current subject matter. [0058] FIG. 27 illustrates an example system, according to some implementations of the current subject matter. [0059] FIG. 28 illustrates an aspect of the subject matter in accordance with some implementations of the current subject matter. [0060] FIG. 29 illustrates an aspect of the subject matter in accordance with some implementations of the current subject matter. DETAILED DESCRIPTION [0061] The presently disclosed subject matter now will be described more fully hereinafter, in which some, but not all embodiments of the presently disclosed subject matter are described. Indeed, the presently disclosed subject matter can be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Definitions [0062] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the presently disclosed subject matter. [0063] While the following terms are believed to be well understood by one of ordinary skill in the art, the following definitions are set forth to facilitate explanation of the presently disclosed subject matter. [0064] All technical and scientific terms used herein, unless otherwise defined below, are intended to have the same meaning as commonly understood by one of ordinary skill in the art. References to techniques employed herein are intended to refer to the techniques as commonly understood in the art, including variations on those techniques or
Attorney Docket No.4210.0527WO substitutions of equivalent techniques that would be apparent to one of skill in the art. While the following terms are believed to be well understood by one of ordinary skill in the art, the following definitions are set forth to facilitate explanation of the presently disclosed subject matter. [0065] In describing the presently disclosed subject matter, it will be understood that a number of techniques and steps are disclosed. Each of these has individual benefit and each can also be used in conjunction with one or more, or in some cases all, of the other disclosed techniques. [0066] Accordingly, for the sake of clarity, this description will refrain from repeating every possible combination of the individual steps in an unnecessary fashion. Nevertheless, the specification and claims should be read with the understanding that such combinations are entirely within the scope of the current subject matter and the claims. [0067] Following long-standing patent law convention, the terms “a”, “an”, and “the” refer to “one or more” when used in this application, including the claims. Thus, for example, reference to "a cell" includes a plurality of such cells, and so forth. [0068] Unless otherwise indicated, all numbers expressing quantities of ingredients, reaction conditions, and so forth used in the specification and claims are to be understood as being modified in all instances by the term “about”. Accordingly, unless indicated to the contrary, the numerical parameters set forth in this specification and attached claims are approximations that can vary depending upon the desired properties sought to be obtained by the presently disclosed subject matter. [0069] As used herein, the term “about,” when referring to a value or to an amount of a composition, dose, sequence identity (e.g., when comparing two or more nucleotide or amino acid sequences), mass, weight, temperature, time, volume, concentration, percentage, etc., is meant to encompass variations of in some embodiments ±20%, in some embodiments ±10%, in some embodiments ±5%, in some embodiments ±1%, in some embodiments ±0.5%, and in some embodiments ±0.1% from the specified amount, as such variations are appropriate to perform the disclosed methods or employ the disclosed compositions. [0070] The term “comprising”, which is synonymous with “including” “containing” or “characterized by” is inclusive or open-ended and does not exclude additional, unrecited elements or method steps. “Comprising” is a term of art used in claim language which
Attorney Docket No.4210.0527WO means that the named elements are essential, but other elements can be added and still form a construct within the scope of the claim. [0071] As used herein, the phrase “consisting of” excludes any element, step, or ingredient not specified in the claim. When the phrase “consists of” appears in a clause of the body of a claim, rather than immediately following the preamble, it limits only the element set forth in that clause; other elements are not excluded from the claim as a whole. [0072] As used herein, the phrase “consisting essentially of” limits the scope of a claim to the specified materials or steps, plus those that do not materially affect the basic and novel characteristic(s) of the claimed subject matter. [0073] With respect to the terms “comprising”, “consisting of”, and “consisting essentially of”, where one of these three terms is used herein, the presently disclosed and claimed subject matter can include the use of either of the other two terms. [0074] As used herein, the term “and/or” when used in the context of a listing of entities, refers to the entities being present singly or in combination. Thus, for example, the phrase “A, B, C, and/or D” includes A, B, C, and D individually, but also includes any and all combinations and subcombinations of A, B, C, and D. [0075] As used herein, “living tissue substrates” or “LTSs” can refer to any tissue base living substrate, including organotypic brain slice cultures (OBSCs) and organotypic mesentery membrane cultures (OMMCs), or other tissue types, e.g. liver, kidney, bone, etc. Additionally, in some embodiments “LTS”, “OBSC” and “OMMC” can be use interchangeably and can generally refer to any LTS within the context of the present disclosure. A. Living Tissue Substrate I. Subjects [0076] The subject treated, screened, tested, or from which a sample is taken, is desirably a human subject, although it is to be understood that the principles of the disclosed subject matter indicate that the compositions and methods are effective with respect to invertebrate and to all vertebrate species, including mammals, which are intended to be included in the term “subject”. Moreover, a mammal is understood to include any mammalian species in which screening is desirable, particularly agricultural and domestic mammalian species.
Attorney Docket No.4210.0527WO [0077] The disclosed compositions, formulations, therapeutics and methods of using the same are particularly useful in the treatment of warm-blooded vertebrates. Thus, the presently disclosed subject matter concerns mammals and birds. [0078] More particularly, provided herein is the treatment of mammals such as humans, as well as those mammals of importance due to being endangered (such as Siberian tigers), of economic importance (animals raised on farms for consumption by humans) and/or social importance (animals kept as pets or in zoos) to humans, for instance, carnivores other than humans (such as cats and dogs), swine (pigs, hogs, and wild boars), ruminants (such as cattle, oxen, sheep, giraffes, deer, goats, bison, and camels), and horses. Also provided is the treatment of birds, including the treatment of those kinds of birds that are endangered, kept in zoos, as well as fowl, and more particularly domesticated fowl, i.e., poultry, such as turkeys, chickens, ducks, geese, guinea fowl, and the like, as they are also of economic importance to humans. Thus, provided herein is the treatment of livestock, including, but not limited to, domesticated swine (pigs and hogs), ruminants, horses, poultry, and the like. [0079] In some embodiments, the subject to be used in accordance with the presently disclosed subject matter is a subject in need of treatment and/or diagnosis. In some embodiments, a subject can have or be believed to be suffering from thrombosis or other related condition or disease, or any inflammation-associated disease, condition or phenotype. II. Living tissue substrates (LTS) and related methods and systems for drug screening and diagnostic prediction [0080] The ideal functional precision medicine platform would maintain a representative, heterogeneous fraction of tumor tissue within a living tissue microenvironment containing complex native architecture. This platform would enable high engraftment efficiency and rapid, functional, drug sensitivity testing to quantify not only the potency of each drug against the tumor, but also the toxicity to healthy tissue. The readout of such an assay would comprehensively evaluate multiple therapeutics based on their therapeutic window, enabling direct drug-to-drug comparisons, and provide quantitative recommendations to guide clinical decisions. Such objectives are achieved by the disclosed living tissue substrate (LTS) systems and methods.
Attorney Docket No.4210.0527WO [0081] Disclosed herein are LTSs, including for example, but not limited to, organotypic brain slice cultures (OBSCs), organotypic mesentery membrane culture (OMMCs), kidney, liver, lung, bone and spinal cord, which allow rapid and high-efficiency engraftment, treatment, and functional analysis of living brain tumor cells and patient tumor tissue. For each of over 150 drug-tumor-LTS interactions, a novel, multi-parametric algorithm was developed to combine over 100 data points and measure dose-response relationships of both tumor kill and healthy tissue toxicity to calculate an overall drug sensitivity score (DSS). These toxicity-normalized DSS were used in some embodiments to compare both approved and exploratory drugs within and among each tumor type. By comparing this data to clinical genetic profiles, treatment regimens, and tumor responses of matched patients, positive correlations were identified in drug sensitivities and areas in which patients may have benefitted from an alternative approach to treatment. [0082] The disclosed LTS systems and methods include functional analysis of patient brain tumor tissue via a novel multi-parametric algorithm, comparing algorithm outputs to patients’ own genomic profiles and immediate responses to treatment. Disclosed herein for the first time are methods, devices and systems for engrafting solid tumor tissue from patient resection surgeries onto living tissue substrates (LTS) and the quantification of survival. Disclosed herein are methods to prepare and engraft solid tumor tissue from patient resection surgeries onto live, thin sections of living tissue. Engraftment of tumor tissue onto these living tissue substrates, or LTSs, improves tumor survival over other methods and allows immediate initiation of downstream assays. The disclosed methods to prepare, engraft, interrogate, and analyze solid patient tumor tissue allow rapid drug screening and diagnostic prediction in a complex ex vivo environment. [0083] By way of example only, and not being bound by any particular theory or mechanism of action, the same novel patient tumor preparation protocol is used to engraft (a) patient primary or metastatic brain tumor tissue onto living slices of brain and (b) patient tumor tissue which grows or metastasizes into the peritoneum (e.g. ovarian tumor tissue) onto living mesentery. Other LTS examples were tested and verified, including kidney, liver and lung. Based on these discoveries and the supporting data herein, the engraftment and interrogation of patient tumor tissue on numerous types of LTS provides for generalizable methods, devices, systems and assays for any tumor type and any substrate type. The general workflow for this tumor/substrate-agnostic assay is illustrated in Figure 1. More particularly, Figure 1 is a schematic illustration of a living tissue
Attorney Docket No.4210.0527WO substrate (LTS) culture system 100 as disclosed herein, and used for testing therapeutics against a tumor or cancer of interest. Specially prepared patient tumor tissue 122 is engrafted onto LTS (or tissue substrate) 120 and treated with escalating doses of one or more therapeutics in media 108. Such can be carried out in any suitable vessel 102, including a multiwell plate or assay system, e.g. a 96-well plate, with a well 104 configured for holding the LTS 120. A porous floor 106 can be provided in well 104 that allows media 108 to pass through and contact LTS 120. Several aspects of dose-response tumor killing profiles are compared to LTS toxicity profiles using an algorithm, as disclosed herein, to calculate one number from -100 to 100 representing the overall efficacy of the drug against the tumor (referred to herein as a drug sensitivity score (DSS). [0084] In some embodiments, disclosed herein is the use of a living tissue layer to support uncultured patient tumor tissue. While organotypic brain slice cultures (OBSCs) have been used as substrates, the preparation, engraftment, and interrogation of patient tumor tissue onto LTSs, including OBSCs, requires a significantly different protocol. Significant challenges were overcome in developing the protocol and associated methods for engrafting patient tumor tissue on LTS, including OBSCs. An example procedure for LTS- OSBC engraftment is illustrated in Figure 2, where the top row depicts standard OBSC generation, where rat pup brains are dissected, sliced via vibratome, and plated. The bottom row of Figure 2 depicts a novel patient tumor preparation procedure, where resected patient tumors are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and added to OBSCs. In some embodiments, far-red lipid-soluble dye, or any other suitable dye or marker, can optionally be used to stain the patient tumors. [0085] Additionally, disclosed herein for the first time is the use of a living layer of mesentery tissue to support ex vivo tumor growth of any kind. Living mesentery (from rat or other subjects), in some embodiments referred to herein as LTS-M, or LTS-OMMC, has not previously been used to engraft tumor cells or patient tumor tissue. Mesentery is already very thin and does not need to be sliced before tumor tissue is seeded. All other downstream methods to prepare, seed, interrogate, and analyze tumor tissue on mesentery are generally conserved as compared to the other LTS methods disclosed herein. Figure 3 provides an illustration of these novel methods and systems, where the top row depicts mesenteric surgery and plating. The bottom row of Figure 3 depicts novel patient tumor preparation procedure where resected patient tumors are finely minced with no enzyme,
Attorney Docket No.4210.0527WO strained through a 100µm filter, infected with lentiviral luciferase and stained with far-red lipid-soluble dye, and added to mesentery. [0086] Importantly, capturing intratumoral heterogeneity is a feature of this approach, which, in some embodiments, allows for the analysis of the tumor as a whole and not just a portion. [0087] As disclosed herein, LTS can in some embodiments be a preferred or at least advantageous method for rapid testing of patient tumor tissue with high engraftment rate. As shown herein, patient tumor tissue reproducibly persists on LTS-OBSC and LTS-M in a transwell format, but does not survive on transwell membranes without a LTS. When cultured in polystyrene in vitro flasks, significant cell death/selection occurs. The remnant of tumor tissue able to survive in vitro takes significant time to expand and loses heterogeneity due to clonal expansion. Growing tumor tissue in mice or other animals is a lengthy process with a low engraftment rate. The experiments herein confirmed that patient tissue only survives on LTS. More particularly, there was no survival of minced patient brain tumor tissue on transwell insert only nor in standard in vitro culture, but there was reproducible survival on LTSs, including LTS-OBSCs, LTS-OMMCs, and LTSs from kidney, liver and lung. [0088] Provided herein in some embodiments, is the cryopreserving and thawing patient solid tumor tissue for use in functional diagnostic assays. Details of the disclosed cryopreservation and tissue banking protocol are as follows: Patient Tissue Banking (cryopreservation) Protocol Equipment Needed: • Scale • Biosafety Hood • 1ml pipette with large orifice tips • Tabletop centrifuge • Solvent-resistant pen • Cryovial Freezing Container • -80 oC Freezer • Styrofoam Box Reagents Needed: • 10cm culture dish
Attorney Docket No.4210.0527WO • 50ml conical tube • Disposable Scalpels • PBS • Disposable plastic spatula • 4oC CryoStor CS10 • Cryovials • Dry Ice Procedure: 1. Receive tissue in Hibernate-A media from hospital. 2. Remove media and drop tissue into a 10cm cell culture dish. 3. Separate and discard necrotic areas 4. Weigh remaining tissue (mg). 5. In a biosafety hood, use scalpel to chop tissue into very small pieces (set a timer for 10 minutes to ensure the tissue is finely chopped). Keep scraping tissue pieces into a pile and continue chopping. The tissue will turn into a slurry over time. As this happens, it will be tough to see how small the tissue pieces are. 6. Weigh an empty 50ml conical tube and fill with 45ml PBS. 7. Add tissue slurry to tube of PBS. This can be done by first scooping the tissue slurry with the side of the scalpel or a disposable plastic spatula, and then retrieve the remainder by pipetting PBS onto the culture dish and aspirating the diluted tissue pieces back into the large orifice pipette tip. 8. The 50ml conical tube should now be full of very small pieces of tissue. Close and invert the tube a few times to disperse and wash. 9. Centrifuge the tube at low RPM for 5 minutes to gently pellet. 10. During centrifugation, label cryovials with the Specimen, tumor type, and/or deidentified patient number using a solvent-resistant pen. 11. Discard supernatant. When close to the bottom of the tube, do not use vacuum aspiration. Instead, use a 1ml manual pipette to carefully remove the rest of the supernatant. Be careful not to disturb the pelleted cells. 12. Weigh the 50ml tube + pellet. The number of cryovials needed will depend on the total initial amount of tissue (see below). 13. Add 1ml 4oC CryoStor CS10 per 200mg tissue (minimum 1ml CryoStor if there is less than 200mg of tissue) and resuspend the cell pellet. Add cells in CryoStor to
Attorney Docket No.4210.0527WO each cryovial and label with tissue mass added. As you pipette, keep resuspending the cells to prevent settling. For exceptions, see examples below: a. Example 1 – 100mg tissue: use 1ml CryoStor to resuspend and add into one vial b. Example 2 – 250mg tissue: this is just barely over 200mg, so use 1 ml CryoStor to resuspend and add into one vial c. Example 3 – 500mg tissue: use 2.5ml cryostor to resuspend. Add 1 ml into each of two vials (200mg each) and 0.5ml into a third vial (100mg) 14. Cap cryovials and add to freezing container. 15. Close freezing container and put into -80 freezer. 16. Wait overnight for cryovials to freeze. [0089] Performing this disclosed cryopreservation procedure on patient tumor tissue between “Mince” and “Strain and Disperse” steps allows safe shipment of tumor tissue from other sites. Surprisingly, it was discovered that drug sensitivity profiles of tumor tissue engrafted on LTSs systems disclosed herein, including OBSC and OMMC, are not changed after cryopreservation and thawing compared to fresh tumor tissue engrafted onto LTS. Notably, no enzyme is used when dissociating solid tumor tissue in preparation for cryopreservation. Moreover, experiments were conducted to evaluate the killing profiles of fresh and cryopreserved/thawed patient brain tumors tissue. The data confirmed that cryopreservation does not affect the drug sensitivity of patient tumor tissue. [0090] Provided herein in some embodiments, and for the first time, is a drug (therapeutic, active agent, and/or candidate compound) efficacy scoring method which normalizes drug efficacy by comparing the drug’s tumor kill versus the drug’s toxicity to the LTS, referred to herein as a Drug Sensitivity Score (DSS). In some embodiments, the exact methods to quantify LTS viability can differ slightly between tissue types of LTS, including for example between OBSC and mesentery tissues. Using the multi-point dose- response curves generated using the disclosed assay, the disclosed multiparametric equation calculates the following parameters for both tumor killing and LTS toxicity, comparing both at each parameter and weighting (weighted value in parentheses) each parameter to calculate a single number between -100 and 100 which is weighted at about 1% to about 45% in the DSS calculation, optionally, and in one exemplary embodiment, weighted as follows: Maximum Kill (about 10%), EC10 (about 5%), EC25 (about 5%), EC50 (about 10%), EC75 (about 5%), EC90 (5%), Slope through IC50 (about 10%),
Attorney Docket No.4210.0527WO Tumor Growth Acceleration (about 5%; not compared to LTS toxicity), Biphasic Killing Curve (about 5%; not compared to LTS toxicity), Incomplete Kill (about 5%; not compared to LTS toxicity), and Area Under the Curve (about 35%). While in some embodiments all of the above parameters are calculated and weighted in calculating a number between -100 and 100, in some embodiments only select parameters of the above are used depending on the LTS and/or drug tested. [0091] To elaborate further, in some embodiments, the equations used in the algorithm(s) for the disclosed DSS include one or more, or all, of the following: MAX KILL Max Kill of Tumor:
Amount of tumor killed at highest dose 1
െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ Max Kill of Slice:
Amount of slice killed at highest dose 1
െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ Max Kill Window Ratio between the amount of tumor killed and slice killed If the amount of tumor killed is more than the amount of slice killed Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ No: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ Max Kill Score If the amount killed at the highest dose is more than 0 Yes: If the amount of tumor killed is more than the amount of slice killed: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ If the amount of slice killed is more than the amount of tumor killed:
Attorney Docket No.4210.0527WO ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ∗ െ1 No: Whichever is smaller: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ∗ െ1 O
r %
^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ∗ െ1 ECXX ECXX of Tumor (where ECXX is generic for EC10, EC25, EC50, etc.): ECXX by extrapolating between the point above and below the ECXX Low Dose = Lowest dose above ECXX High Dose = Highest dose below ECXX ^^1 െ ^^ ^^ ^^ ^^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ∗ ^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^
Corresponding Slice Health at ECXX of Tumor Determine the amount of slice that is alive at the EC XX of the tumor Low Dose = Lowest dose above tumor ECXX High Dose = Highest dose below tumor ECXX ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ∗
^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^
ECXX Window Ratio between where XX% of the tumor is killed and the health of the slice at that dose If the health of the slice if more than 1-XX%:
Attorney Docket No.4210.0527WO Yes: ^^ ^^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ െ ^1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ^
^ ^^ ^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ No: ^1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ െ ^^ ^^ ^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ^
1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^ ECXX Score If tumor growth is seen that results in XX% more than the max kill of the slice: Yes -1*% Weight in DSS No If tumor killing is seen but not more than XX% Yes 0 No If the amount of slice that is killed at the tumor ECXX is more than XX%: Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ No: െ1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ Slope Tumor Slope Slope through the EC50 or the highest dose if EC50 isn’t reached ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50
Slope through the EC50 or the highest dose if EC50 isn’t reached ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50
Attorney Docket No.4210.0527WO Ratio between slopes If the slice slope is larger than the tumor slope: Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ No: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ Slope Score If the max kill is between -.05 and .05 Yes 0 No If the tumor slope is positive Yes -1 * %Weight in DSS No: If the tumor slope is less than the slice slope Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ No: െ1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ Tumor Growth Acceleration Is there an increase in tumor growth? Highest survival is more than 150%: -1 * %Weight in DSS Highest survival is between 150% and 125%: 0 Highest survival is less than 125%:
Attorney Docket No.4210.0527WO 1 * %Weight in DSS Biphasic killing curve Fast initial kill, followed by much greater difficulty to kill remaining fraction Yes: -1 * %Weight in DSS No: 1 * %Weight in DSS Incomplete Kill Is there a tumor population remaining? If there is more than 25% remaining: -1 * %Weight in DSS If there is between 10% and 25% remaining: 0 If there is less than 10% remaining: 1 * %Weight in DSS Area Under the Curve Tumor AUC AUC as calculated using the trapezoidal rule ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 2 ^
∗ ^ ^^ ^^ ^^ ^^ 2 െ ^^ ^^ ^^ ^^ 1^
^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 2 ^
∗ ^ ^^ ^^ ^^ ^^ 2 െ ^^ ^^ ^^ ^^ 1^
AUC Window Ratio between slice AUC and tumor AUC If slice AUC is less than tumor AUC
Attorney Docket No.4210.0527WO Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ No: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^
^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ AUC Score If the slice AUC is less than the tumor AUC Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ No: െ1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ [0092] In some embodiments, each of the above equations making components of the DSS can be adjusted, modified and/or eliminated based on the LTS used/tested, and/or pertinent characteristics of the same. [0093] In some embodiments, disclosed are diagnostic and/or screening methods, assays and systems based on the LTS platform for diagnosing a tumor in a subject and/or screening potential drug and treatment options, including combinatorial treatments. In some embodiments, the present disclosure provides adaptive treatment methods where cryopreserved patient samples, e.g. brain tumor samples, can be serially thawed and analyzed using the disclosed methods and LTS platform, coupled with DSS analysis, to identify optimal patient specific treatment programs after testing and/or screening a plurality of options. Drug sensitivity information gathered from testing the first thawed vial of patient samples would inform the drugs tested on patient samples from the second thawed vial, etc. This adaptive, iterative, testing would maximize the information that can be gathered and more effectively pinpoint the most appropriate therapeutic or combination for a single patient before the patient begins treatment. This is possible in the disclosed system because (1) LTS supports engraftment of cryopreserved/thawed patient samples, and (2) LTS assays are rapid (4 days from engraftment to readout), allowing multiple
Attorney Docket No.4210.0527WO iterations of testing within a clinically relevant time frame. In some embodiments, this clinically relevant timeframe can vary depending on the type of tissue used for the LTS. [0094] Finally, while data provided herein directly supports the use of brain tissue (OBSCs) and mesentery based LTSs, other tissues suitable for LTSs, and that have been tested, include but are not limited to kidney, liver, lung, bone and spinal cord (see, e.g. Figures 17A-17E). Similar to mesentery and OBSC LTSs, these other tissues can serve as an LTS for growing patient samples and tumor cells for purposes of diagnosing and/or screening drug candidates. [0095] Taken together, the present disclosure offers several advantages not previously contemplated, and many of which were unexpected. For example, the LTS provides: o A tissue layer which supports tumor viability more effectively and reproducibly than culture methods without the substrate. o A natural microenvironment which communicates with tumor tissue. o A healthy tissue control with which to measure off-target drug toxicity. [0096] Moreover, compared to directly culturing slices of the patient tumor tissue itself for functional testing, using a separate living tissue substrate to engraft patient tumor: o Decreases the amount of tumor tissue required per replicate, allowing broader/deeper testing per unit of tumor pass received from the clinic. o Increases intra-sample reproducibility: the homogeneous cell suspension we make from the tumor tissue we receive allows each tumor grown on LTS to be very similar; in contrast, slices of patient tumor are each from a different tumor region and therefore unique, especially in more heterogeneous tumor tissue samples. [0097] Additionally, the disclosed methods, systems and protocols to cryopreserve and thaw patient tumor tissue just after resection allows for: o Ship tumor tissue from other sites without risking viability loss from sustained live tissue transport. o Strategically thaw tumor tissue samples at the right time(s) for optimal study impact. [0098] The disclosed methods also allow for quickly and gently dissociate, label, and seed patient tumor tissue onto LTS allows even low-grade tumors to more quickly and successfully engraft compared to other culture methods. [0099] Additionally, seeding patient tumor tissue onto living tissue substrates without any time in culture:
Attorney Docket No.4210.0527WO o More effectively maintains tumor heterogeneity. o Minimizes cell selection and genetic drift before assay initiation. [0100] Moreover, the disclosed methods, systems and assays require just four days from the time we receive tumor tissue to provide the final output. [0101] Surprisingly, 100% of the patient tumor tissue seeded onto slices have retained evaluable levels of engraftment, regardless of tumor grade or type. [0102] Importantly, the present disclosure confirms the successful testing of the following treatment types against tumors grown on living tissue substrates: o Small Molecule Drugs o Radiation Therapy o Protein Therapies o CAR-T and iNSC Cellular Therapies o Combination Therapies Once a tumor or cancer is diagnosed in a subject, an appropriate treatment, including any of these listed herein, or combinations thereof, can be administered to the subject. [0103] Each Drug Sensitivity Score (DSS) is generated from over 100 unique data points. Each drug is tested against an individual tumor at six dose levels at n = 4 tumors per dose, providing a dose-response profile with which to gauge efficacy. o Tumor response is compared to the toxicity of each drug to the LTS (six dose levels at n = 12 LTS per dose), allowing for the to quantification of a “therapeutic window” to ask: ^ Which tumors can be killed at doses which present minimal toxicity to healthy tissue? ^ Which drugs present the lowest toxicity to healthy tissue while still killing the tumor? ^ Can combination therapy maintain effective tumor kill while decreasing healthy tissue toxicity compared to single-agent therapy? [0104] The disclosed novel, multi-point algorithm combines several pieces of data from dose-response toxicity curves of tumor and LTS to output one normalized number (-100 to 100) that can be compared in the following ways: o Measure and rank efficacies of different drugs against a single tumor. o Measure and rank which tumors respond most and least to a single drug.
Attorney Docket No.4210.0527WO [0105] Thus, provided in some embodiments are methods of diagnosing a tumor and/or screening for a therapeutic for a tumor. Such methods can include providing a living tissue substrate (LTS), engrafting one or more tumor tissue cells (e.g. tumor tissue and/or tumor cells obtained from a subject) to the LTS, and analyzing a dose-response of a candidate therapeutic for both tumor tissue cell kill and LTS toxicity. The result of such methods can provide for the diagnosis of the tumor and/or identification of a candidate therapeutic to treat the tumor. The tumor tissue cell kill, or tumor toxicity, can be defined or quantified on a continuous scale and can be measured anywhere between about 1% and 99%, i.e. "dead". In some embodiments, the tumor tissue cell kill, or tumor toxicity, can be defined as the degree to which the tumor tissue is killed, degraded, or rendered unviable, and can range from about 10% to about 99% kill, from about 20% to about 95% kill, from about 30% to about 90% kill, from about 40% to about 90% kill, or about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 99% or more kill (in some aspects as compared to an untreated tumor tissue cell. The LTS toxicity can be defined or quantified on a continuous scale and can be measured anywhere between about 1% and 99%, i.e. "dead". The LTS toxicity can be defined as the degree to which the living tissue substrate is killed, degraded, or rendered unviable (i.e. an undesirable side effect of the treatment), and can range from about 1% to about 99% toxicity, from about 1% to about 90% toxicity, from about 5% to about 75% toxicity, from about 5% to about 50% toxicity, or about 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or less toxicity (in some aspects as compared to an untreated LTS). [0106] As described herein, the LTS of these methods and systems can comprise any tissue that forms solid tumors, and in some embodiments can include a tissue selected from brain, mesentery, kidney, liver, lung, bone and spinal cord. Brain tissue can in some embodiments be an organotypic brain slice culture (OBSC). Mesentery tissue can in some embodiments be an organotypic mesentery membrane culture (OMMC). [0107] In some embodiments, the one or more tumor tissue cells are derived from a primary or metastatic tumor of a subject. As described further herein, the one or more tumor tissue cells are dissociated into small pieces, and transfected with a reporter, prior to seeding onto the LTS, optionally wherein the tumor is a brain tumor or ovarian tumor. In some aspects, the one or more tumor tissue cells are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and labeled with a
Attorney Docket No.4210.0527WO fluorescent reporter (e.g. stained with far-red lipid-soluble dye), prior to seeding onto the LTS. [0108] Advantageously, in some aspects, the one or more tumor tissue cells are engrafted to the LTS and tested for drug sensitivities, with assay completion in less than 10 days, optionally less than 5 days, optionally less than 4 days, optionally less than 3 days, optionally less than 2 days, optionally less than about one day. [0109] Unexpectedly, and as demonstrated in the working Examples, the genetic drift of the one or more tumor tissue cells in these methods and systems is minimized due to the rapid engraftment. By way of example and not limitation, the genetic drift is less than about 50%, optionally less than about 25%, optionally less than about 10%, at the time of diagnosing and/or screening. In some cases, the mutational profile of the one or more tumor tissue cells is substantially similar to the mutational profile in vivo. [0110] In some embodiments, the source of the one or more tumor tissue cells used in these systems and methods is a patient in need of treatment and/or having a tumor, including for example a human subject, and collecting a biopsy from the patient as the source of the one or more tumor tissue cells. The one or more tumor tissue cells can be cryopreserved after biopsy and thawed prior to engraftment on the LTS, optionally wherein the cryopreserved tumor tissue cells are preserved for a plurality of sequential and/or simultaneous applications of the methods. Unlike prior methods, the cryopreserved tumor tissue cells are not exposed to an enzyme during dissociation. [0111] As discussed further herein, the analysis of the dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity can be assessed or characterized by calculating a drug sensitivity score (DSS), wherein the DSS is calculated by comparing tumor cell survival to health of the LTS. By way of example and not limitation, the DSS can be calculated based on a plurality of weighted parameters, optionally a combination of all weighted parameters, selected from the group consisting of: (1) killing at maximum dose (Max Kill), (2) dose required to kill 10% of the tumor (EC10), (3) dose required to kill 25% of the tumor (EC25), (4) dose required to kill 50% of the tumor (EC50), (5) dose required to kill 75% of the tumor (EC75), (6) dose required to kill 90% of the tumor (EC90), (7) slope through the EC50, (8) the area under the curve (AUC), (9) tumor growth acceleration, (10) biphasic killing (rapid killing at low doses and limited additional killing at higher doses), and (11) incomplete kill at the highest dose. As disclosed herein, the DSS from 0 to 100 signifies increasing efficacy in tumor kill relative to LTS toxicity, and
Attorney Docket No.4210.0527WO wherein a DSS from 0 to -100 signifies increasing LTS toxicity relative to tumor kill. For exemplary purposes only, each of the parameters can in some embodiments be weighted at about 1% to about 45% in the DSS calculation, optionally wherein each is weighted as follows: Maximum Kill (10%), EC10 (5%), EC25 (5%), EC50 (10%), EC75 (5%), EC90 (5%), Slope through IC50 (10%), Tumor Growth Acceleration (5%; not compared to LTS toxicity), Biphasic Killing Curve (5%; not compared to LTS toxicity), Incomplete Kill (5%; not compared to LTS toxicity), and Area Under the Curve (35%). [0112] These methods can advantageously identify a suitable therapeutic or drug compound for treating a particular tumor or cancer. Such candidate therapeutics or drugs to treat the tumor can in some embodiments comprise a pharmaceutically active agent, a chemotherapeutic composition, a small molecule, an immunotherapeutic agent, an inhibitor, a radiation therapy, and combinations thereof. [0113] In some embodiments, provided are functional precision diagnostic methods that include performing any of the above methods for diagnosing a tumor or screening for a therapeutic for a tumor, and further comprising iteratively testing additional therapeutics on cryopreserved patient tumor cells before administration to a subject, whereby a treatment can be adapted based on a DSS output. For example, such functional precision diagnostics can include testing combinatorial therapies using LTS and DSS. [0114] As disclosed further herein, in some embodiments provided are diagnostic and/or therapeutic screening systems. Such screening systems can include a LTS, one or more tumor tissue cells engrafted to the LTS, and a multi-parametric algorithm which simultaneously analyzes a dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity, i.e., the disclosed DSS. The LTSs are configured the same as in the above described methods. [0115] Finally, in some embodiments, methods of treating a subject are provided. Such treatment methods can begin with the diagnostic and/or screening methods and systems described herein, followed by administering to the subject a treatment or therapeutic based on the diagnosis. In some embodiments, the treatment can comprise a combinatorial treatment, i.e., one or more drugs or therapeutic compositions/modalities based on the DSS analysis. [0116] The subject matter disclosed herein can be implemented in software in combination with hardware and/or firmware. For example, the subject matter described herein can be implemented in software executed by a processor. In one exemplary
Attorney Docket No.4210.0527WO implementation, the subject matter described herein can be implemented using a computer readable medium having stored thereon computer executable instructions that when executed by a processor of a computer control the computer to perform steps. Exemplary computer readable mediums suitable for implementing the subject matter described herein include non-transitory devices, such as disk memory devices, chip memory devices, programmable logic devices, and application specific integrated circuits. In addition, a computer readable medium that implements the subject matter described herein can be located on a single device or computing platform or can be distributed across multiple devices or computing platforms. EXAMPLES [0117] The following examples are included to further illustrate various embodiments of the presently disclosed subject matter. However, those of ordinary skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the spirit and scope of the presently disclosed subject matter. Materials and Methods for Examples 1-10 Cell Lines: [0118] Cell lines used: LN229 (established GBM line), GBM8 (high-passage patient- derived GBM line
56), MS21 (low-passage patient-derived GBM line), U373WT (wild-type established GBM line), U373KO (RAD18 knockout of U373WT via CRISPR), MB231Br (breast cancer metastasis to brain), and PDIPG (low-passage pediatric patient-derived line). [0119] GBM8 cells were cultured in Neurobasal-A medium (Gibco) with 7.5 ml L- glutamine, 10ml B27 supplement, 2.5 ml N2 supplement, 1 mg heparin, 10 μg EGF, 10 μg FGF, and 2.5 ml anti-anti. LN229 cells were from American Type Culture Collection. MDA-MB231-Br cells were obtained through a material transfer agreement (MTA) (T. Yoneda). MS21 cells were derived in the Hingtgen Laboratory from a GBM patient biopsy. U373WT and U373KO cells were provided by C. Vaziri (University of North Carolina). LN229, MDA-MB231-Br, MS21, U373WT and U373KO cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (Gibco). IFF-BT105 PDIPG cells were obtained from
Attorney Docket No.4210.0527WO Ian’s Friends Foundation and cultured in Neurobasal Medium(-A):DMEM/F-12, GlutaMAX Medium (1:1) supplemented with 1x Antibiotic-Antimycotic, 1x Sodium Pyruvate, 1x MEM Non-Essential Amino Acids, 10mM HEPES buffer, 1x GlutaMAX-I, 1X B27 minus vitamin A, 20ng/ml EGF, 20ng/ml bFGF, 10ng/ml PDGF-AA, 10ng/ml PDGF-BB, 2ug/ml heparin. All cell lines were cultured at 37
oC, 5% CO2 and 95% humidity. Generating OBSCs: [0120] All OBSCs were generated from P8 Sprague-Dawley rat pups. Dissected brains were mounted on a vibratome (Leica VT1000S) platform submerged in ice-cold brain slice media (BSM). Coronal OBSCs were sliced at a thickness of 300 µm at ~15 OBSCs/animal. Visibly damaged brains or OBSCs were discarded. Acceptable OBSCs were transferred onto transwell inserts in a 6-well culture plate. 1ml of OBSC media (BSM
22) was added under each transwell. BSM comprised of Neurobasal-A medium supplemented with 10% heat-inactivated pig serum, 5% heat-inactivated rat serum, 1 mM L-glutamine, 10 mM KCl, 10 mM HEPES, 1 mM sodium pyruvate and 100 U/mL penicillin-streptomycin. The plates were then transferred to a 37°C incubator with 5% CO2 and 95% humidity. For QC purposes, six random OBSCs from every batch were selected to undergo the PI assay test for cell death. On D4, the PI signals from the QC group were compared to those from previous batches. In Vivo Tumor Studies: [0121] 6-8 week-old female athymic nude mice were used; n = 5 mice per cell line with no exclusion criteria, randomization, or blinding as all mice were used in one group. 250,000 tumor cells were stereotactically implanted into the brain parenchyma (1 mm, 1 mm, 2.5 mm) of mice anesthetized with isoflurane. All mice underwent serial bioluminescence imaging to measure tumor growth over time and were monitored for changes in weight or behavior to indicate the endpoint had been reached. Luciferin was injected i.p. into mice at 3mg/mouse in 250 mL PBS. Brains from each group, dissected at time of death, were coronally sectioned along the tumor implant site and imaged for tumor fluorescence.
Attorney Docket No.4210.0527WO Chemicals: [0122] The following chemicals were purchased from Sigma-Aldrich: carboplatin, temozolomide, etoposide, azacitidine, vincristine. The following chemicals were purchased from EMD Millipore: cisplatin. The following chemicals were purchased from Selleck Chemicals: lomustine, gemcitabine, and trametinib. TR-107 was provided by Madera Therapeutics, LLC. Lentiviral Vectors: [0123] The following LVs were used in this study: eGFP fused to firefly luciferase (LV– eGFP-FL) and mCherry protein fused to firefly luciferase (LV–mCh-FL). Propidium Iodide Assay: [0124] PI fluorescence was used to quantify the health/quality of OBSCs as well as the toxicity of each treatment vs OBSCs at the end of each assay: t = 4 days after OBSC generation and t = 3 days after treatment initiation. 5 µg/mL PI was mixed with BSM and added under transwells 60 minutes before fluorescence measurement on an AMI optical imaging system (Spectral Instruments). Positive control was generated by killing the OBSC via incubation with 70% ethanol and freezing overnight between days 3 and 4 of the assay. For all DSS calculations, n = 12 OBSCs from N = 2 separate experiments. Immunohistochemistry: [0125] OBSCs dedicated for Day 0 IHC were fixed in 4% paraformaldehyde immediately following sectioning and stored at 4°C. After 48 hours of fixation, the sections were transferred to 30% sucrose and stored at 4°C until IHC was conducted. The sections dedicated for Day 4 IHC were transferred to 6 well plates with BSM and stored at 37°C. GBM8-mch-FLuc cells were engrafted atop select OBSCs as described below. On Day 1, the sections were transferred to new 6 well plates with fresh BSM. The sections were fixed in 4% paraformaldehyde on Day 4 and stored at 4°C. After 48 hours of fixation, the sections were transferred to 30% sucrose and stored at 4°C. When day 0 and day 4 sections were ready for IHC, they were first washed for 10 minutes in 0.1% triton X-100 in 1X Dulbecco’s phosphate-buffered saline (PBST) at room temperature. Sections were then blocked in 5% fetal bovine serum in PBST for 1h at room temperature. The sections were incubated in a primary antibody solution consisting of primary antibodies and blocking
Attorney Docket No.4210.0527WO buffer rotating overnight at room temperature. The primary antibodies used were glial fibrillary acidic protein (GFAP [Abcam, ab7260] at 1:1000), neuronal nuclear protein (NeuN [Abcam, ab177487] at 1:1000), and cluster of differentiation molecule 11B (CD11B [Abcam, ab133357], 1:500). The sections were washed 3 times in PBST for 10 minutes after 18-24 hours in primary antibodies. They were then incubated in a secondary antibody consisting of blocking buffer solution and Alexa Fluor 488 goat anti-rabbit IgG (Thermo Fisher Scientific, A-11008, 1:1000) for 1 hour in a darkroom. The sections were washed 3 times in PBST for 10 minutes and mounted on microscopic slides. Liquid mountant (ProLong Gold Antifade Mountant) was applied on the sections, coverslips were added, and slides were allowed to cure overnight. Z-stack images were acquired using a Zeiss 780 confocal microscope at UNC Neuroscience Microscopy Core. Z-stack images are processed by converting to maximum intensity projection (Max IP) images. The brightness of Max IP images was then further adjusted to accurately assess and present the morphological differences of astrocytes. In Vitro Studies: [0126] Cells were seeded at a density of 5000 cells/µL in 96 well plates. 24h after seeding, drugs were added and allowed to incubate for 72 hours. Cell viability was then assessed via bioluminescence, measured using an AMI optical imaging system. Tumor Growth on OBSCs: [0127] Tumor cells (.17 µL, 25,000 cells) were seeded onto OBSCs 2h after slicing, with one tumor foci seeded in the center of each hemisphere for a total of two tumor foci per OBSC. BSM was changed 24h after slicing. Fluorescence images were taken 0 hours, 24 hours, and 96 hours post-seeding for normalization of tumor size. Bioluminescence readouts were also taken at 0 hours, 24 hours, and 96 hours post seeding for assessment of cell viability. Luciferin was added underneath the transwell insert and allowed to incubate for 10 minutes before bioluminescence measurement on an AMI optical imaging system. Drug Screening on OBSCs: [0128] Tumor cells were seeded onto OBSCs 2h after slicing as described above. One day after tumor seeding, six concentrations of each drug were diluted in the media underneath each transwell, or XRad was applied using an X-Rad 320 Precision X-Ray
Attorney Docket No.4210.0527WO machine (n = 4 tumor foci per concentration; n ≥ 24 tumor foci per drug per cell line). Three days after treatment, the bioluminescence of live tumor cells was measured using an AMI optical imager (Spectral Instruments) and dose-response curves were generated. Tumor survival was normalized to day one tumor fluorescence and compared to an untreated control group. Combination Therapy: [0129] Carboplatin + etoposide: On Day 1 after slicing, both small molecules were administered in the BSM underneath the transwells at the desired concentrations. TMZ + radiation: On Day 1 after slicing, radiation was first administered by an X-Rad 320 Precision X-Ray machine. TMZ was then administered in the BSM underneath the transwells at the desired concentrations. Preparing Patient Tissue for Engraftment onto OBSCs: [0130] Fresh brain tumor tissue surgically resected at UNC hospitals was placed in sterile 4°C Neurobasal-A medium and immediately taken to the UNC Tissue Procurement Facility (TPF). The amount of brain tumor tissue received ranged from 0.05g to 2g. Either at TPF or in the Hingtgen Laboratory, the resected tumor tissue was minced into approximately 0.5 mm diameter pieces using a disposable scalpel and washed with PBS. Tumor pieces which were to be assayed at a later time were placed in a cryogenic vial and frozen in tissue freezing medium (CryoStor CS10) in a FreezeCell™ at - 80°C overnight before transfer into liquid nitrogen. [0131] To engraft onto OBSCs, the cryopreserved/thawed or fresh brain tumor tissue was filtered through a 100 µm cell strainer. Each 50 mg of tissue was transfected with 1ml mcherry-FLuc Lentivirus at 1.5e7 vg/ml with 1µl polybrene for 4h at 37°C. After incubation, the brain tumor tissue was washed with PBS three times to remove the residual virus and reconstituted in PBS with a final volume of 150 µl. Then the tissue solution was engrafted at OBSCs at ~1 mg tissue in 3 µl of PBS on each hemisphere of the slice. The OBSC engrafted with patient tumor tissue was incubated at 37°C with 5% CO2 and 95% humidity. The BSM was changed after 24h and subsequently every 3 days. Tumor treatment studies were executed as described above for tumor lines. The viability of the patient tissue was measured by an AMI optical imaging system with the addition of 1.5 mg/mL luciferin underneath the transwell.
Attorney Docket No.4210.0527WO [0132] Patient tumor drug screening results were not shown to clinicians with information that could identify the patients they reference. Preparing Cell Line from Patient Tissue: [0133] For the primary patient cell line generation, the fresh or frozen tissue were mechanically dissected to approximately single cell under sterile conditions. The resulting single-cell suspension was then washed with PBS, resuspended in growth medium chosen to allow growth of greatest number of tumor cells (DMEM with 20% FBS and 1% PS), and dispensed into 6-well plates. All cultures were initiated in a volume of 2 ml per well and incubated at 37°C and 5% CO2. The viable cells and tissue chunks were allowed to settle and attach to the bottom of the plates for 2-3 days. The floating cellular debris was then carefully aspirated, the attached cells were carefully washed with PBS, and 2 ml of fresh medium was added to each well. Culture medium was routinely changed every 3-5 days, and proliferating adherent cells were passaged after detachment with trypsin. DNASeq: [0134] Cryopreserved/thawed human brain tumor tissue was prepared as described above and either added onto OBSCs (BSHT group), cultured in vitro (CL group) or snap frozen (HT group). After 4 days, tumor tissue was carefully excised from OBSCs using forceps and snap frozen. After 30 days, when cell line had established, cells were collected and snap frozen. All replicates of all samples were shipped to Novogene, Inc. for DNA sequencing. Whole Exome Sequencing was run on the DNA samples following library preparation with an Agilent SureSe-lectXT2 All Exon V6 kit. Raw data was downloaded and analyzed at the UNC Bioinformatics core facility. [0135] The BBsplit algorithm from the BBtools suite was run on all samples to eliminate rat DNA contamination in the BSHT samples as well as to account for any biases that may result as a part of that process. Only the reads that were binned to the human reference were used for subsequent analysis. Reads were then mapped to the GRCh38 version of the human genome with BWA v0.7.17 and realigned together with ABRA2 v.23. Quality control was implemented using the GATK/Picard v4.1.7.0 toolkit. Somatic variants were called for each sample using the MuTect2 algorithm v4.1.7.0. Variants were merged into a single cohort variant call file and then converted to MAF via vcf2maf v1.6.21 tool. Variants were annotated using VEP v87. To identify mutations with potentially high
Attorney Docket No.4210.0527WO biological impact, multiple filtering steps were applied to somatic mutation calling. First, we selected only the somatic variants that passed all filters from the MuTect2 FilterMutectCalls algorithm and second, only high/moderate impact (change coding) variants as defined by the VEP annotation were further analyzed. Over 1900 single nucleotide variants (SNVs) were detected across all samples. Figures that summarized the results were generated using maftools. Calculating Drug Sensitivity Scores: [0136] Several weighted parameters were factored into an overall drug sensitivity score. DSS were calculated by comparing the tumor cell survival, measured via BLI, to the health of the OBSC, measured via PI assay (parameter 1-8), or by quantifying the behavior of the tumor dose-response curve (parameter 9-11). (1) Killing at maximum dose (Max Kill Window, 10% of DSS) (2) Dose required to kill 10% of the tumor (EC10 Window, 5% of DSS) (3) Dose required to kill 25% of the tumor (EC25 Window, 5% of DSS) (4) Dose required to kill 50% of the tumor (EC50 Window, 10% of DSS) (5) Dose required to kill 75% of the tumor (EC75 Window, 5% of DSS) (6) Dose required to kill 90% of the tumor (EC90 Window, 5% of DSS) (7) Slope through the EC50 (Slope Window, 10% of DSS) (8) The area under the dose-response curve (AUC Window, 35% of DSS) (9) Tumor growth acceleration (treated tumor grows faster than untreated tumor: for max growth up to 125%, window = +1; 125%-150% window = 0; over 150% window = -1; 5% of DSS) (10) Biphasic killing (rapid killing at low doses and limited additional killing at higher doses: no biphasic curve shape, window = +1; biphasic curve shape, window = -1; 5% of DSS) (11) Incomplete kill at the highest dose (some tumor remaining at highest dose: for <10% remaining, window = +1; 10%-25% remaining, window = 0; >25% remaining, window = -1; 5% of DSS). [0137] For DSS parameters 1-6, therapeutic windows were calculated by comparing OBSC toxicity and tumor response at the doses where tumor kill passed through the DSS parameter. For DSS parameter 7, the therapeutic window was calculated by comparing the
Attorney Docket No.4210.0527WO slopes through the tumor EC50 and the OBSC tox EC50. For DSS parameter 8, the therapeutic window was calculated by comparing the areas under tumor kill and OBSC tox curves. Normalized therapeutic window ratios for DSS parameters 1-8 within each drug-tumor-OBSC interaction were calculated in the following manner: within each window, values ranged from +1 to -1, where values approaching +1 signify increasing tumor kill relative to normal tissue toxicity, and values approaching -1 indicate agents where tumors remained highly viable while toxicity to the normal OBSC tissue was elevated. DSS parameters 9-11 were determined based on the behavior of the tumor in response to the drug, as defined above. [0138] All individually weighed parameters were added together to generate the DSS. Overall DSS from 0 to 100 signify increasing efficacy in tumor kill relative to OBSC toxicity, while scores from 0 to -100 describe scenarios in which tumors thrive more effectively than OBSCs for a given treatment. Dose-response values were calculated via linear interpolation of raw data, not from best-fit curve equations. Calculating ZIP Synergy Scores and Plots: [0139] All synergy scores and plots were calculated using SynergyFinder. Statistical Analysis: [0140] All statistical tests and sample sizes are included in the Figure Legends. All data are shown as mean ± SEM. In all cases, the p values are represented as follows: ∗∗∗p < 0.001, ∗∗ p < 0.01, ∗p < 0.05, and not statistically significant when p > 0.05. In all cases, the stated ‘‘n’’ value is either number of OBSCs, number of tumor spots placed on the OBSCs, or mice with multiple independent images used to obtain data points for each. Mean values between two groups were compared using t-tests with Welch’s correction when variances were deemed significant by F tests. Mean values between three or more groups were compared to the control by using one-way ANOVA followed by Dunnett’s multiple comparisons test. All statistical analyses were performed using GraphPad Prism (Version 9.1.0). All statistical analysis methods and resulting p values are included within the Figure Legends. For all quantifications of immunohistology, the samples being compared were processed in parallel and imaged using the same settings and laser power.
Attorney Docket No.4210.0527WO Example 1 Generation and Characterization of Living OBSCs as LTSs [0141] These studies use OBSCs generated from rat pups (RRID:MGI:5651135) as living tissue substrates to culture and treat tumor cell lines and uncultured patient brain tumor resection tissue (as discussed herein, other LTSs beyond OBSCs were also tested). To generate this rapid, high-fidelity ex vivo testing platform, first assessed was OBSC quality and reproducibility, optimizing the process using quantitative fluorescence-based imaging methods. After piloting several assays, an established nuclear permeability assay using propidium iodide23–30 (PI, Sigma Aldrich, CAT#P4170) was adapted to quantify cell death within OBSCs. This method provided a large dynamic range to detect nuanced differences between healthy and unhealthy OBSCs, with higher fluorescence values indicating more dead cells (Fig. 4A). OBSC quality was impacted by rat pup age at time of generation (Fig. 4B); therefore, OBSCs were generated from eight-day-old pups for every experiment described in this study. It was also found that optimal OBSCs were generated using improved methods for brain dissection and OBSC culture conditions (Fig. 4C-D), establishing the disclosed robust, standardized procedure for OBSC generation. Following this optimization period, over 6,000 living OBSCs were reproducibly generated across nearly 40 separate batches, including randomly sampling and testing 6 OBSCs per batch for QC analysis (Fig. 4E). All OBSCs with physical damage after slicing, such as a nick or a tear, were discarded, but this amounted to fewer than 2% of OBSCs over all experiments. [0142] Just as brain tissue suffers acute injury after brain tumor resection, OBSCs suffer acute injury after slicing. Serial measurements of OBSC health showed an initially high PI signal that resolves over time, reaching baseline levels by Day 7 (Fig 4F). In parallel, IHC analysis showed astrocytes (GFAP, RRID:AB_305808) and macrophages/microglia (CD11b, RRID:AB_2650514) displayed reactive morphologies
31 immediately post-slicing (Fig. 4G). This activation attenuates in astrocytes by Day 4 but persists in macrophages/microglia, suggesting that the OBSCs may still contain dead cells and debris that are phagocytosed by the myeloid cells
32,33. The morphology of neurons (NeuN, RRID:AB_2650514) remained unchanged between Day 0 and 4. Together, these results describe a high-quality, well controlled, reproducible method for generating and tracking the health of OBSCs. As such, further studies were conducted to test engrafting tumor cells onto these living tissue substrates (see further Examples herein).
Attorney Docket No.4210.0527WO Example 2 Tumor Modeling on OBSCs [0143] Before attempting to move into uncultured patient tissue samples, a diverse panel of human brain cancer lines (adult, pediatric, primary and metastatic; defined in Figure 5 legend) were selected to further identify the myriad capabilities of the disclosed LTS assay, particularly the OBSC assay, and show how LTSs, including OBSCs, provide a rapid, representative, and applicable niche for broad tumor engraftment and analysis. First, multiple tumor characteristics were compared across in vitro, in vivo, and OBSC culture. All tumor lines were transduced with lentiviral vectors encoding mCherry-Firefly Luciferase (LV-mCh-FLuc) optical reporters, then the lines were either seeded in 96-well culture plates, orthotopically xenografted into the brains of Nude mice, or seeded atop the living OBSCs. The bilateral symmetry of each coronally sectioned OBSC was leveraged to seed tumor cells within the anatomically homogenous thalamic region of each hemisphere. Serial imaging and quantitative analysis was performed to track growth, morphology, and rate of invasion. [0144] Four high-passage tumor lines commonly used within the field, LN229, U373, MB231Br, and GBM8, grew rapidly in vitro (Fig 5A), in vivo via orthotopic implantation (Fig 5B), and on OBSCs (Fig 5C-E). These tumor lines all expanded several-fold in just four days in vitro and on OBSCs, and over 100-fold in vivo within 40 days. The low- passage adult GBM line MS21 and pediatric DIPG line PDIPG displayed a different set of results. These two lines had been initially cultured via clonal expansion in vitro after human tumor biopsy, and both were able to be cultured/expanded in vitro between passage 3 and 20 during this work. Both lines grew several fold within four days in vitro, but neither reproducibly established in vivo, even after 90 days. On OBSCs, MS21 and PDIPG tumors both nearly doubled in size over four days, displaying how OBSCs can support growth of low-passage tumor cells not supported in vivo. [0145] Tumors seeded on OBSCs rapidly establish among the endogenous astrocytes and microglia. Observed were tumor-specific trends in outward invasion or inward retraction of tumor cells on the thalamic region of OBSCs. The invasive cell line GBM8, metastatic cell line MB231Br, and diffuse pediatric DIPG were all found to invade radially outward on OBSCs, while other established lines that grow more densely in vivo retract inward (Fig 5F). Interestingly, microglia within OBSCs also reciprocally respond to the presence
Attorney Docket No.4210.0527WO of tumor, maintaining activated morphologies and associating with nearby tumor cells, indicating a two-way interaction between the tumor and this ex vivo microenvironment (Fig 5G). Adding tumor cells to OBSCs on Day 0 after slicing (“Early Seeding”, the standard method used herein), compared to Day 7 (“Late Seeding”) when microglial activation and PI signaling have attenuated, also leads to faster tumor growth and invasion (Fig 5H-I), consistent with in vivo tumor response. Together, this suggests that LTSs, including OBSCs, provide a niche for tumor growth and interrogation that is more appropriate and accurate than in vitro culture systems and less harsh than in vivo models. Example 3 Tumor Treatment on LTSs, including OBSCs [0146] The panel of OBSC tumor models was next used to investigate treatment responses within the platform. To assess the molecular specificity of drug-induced tumor killing on OBSCs, a CRISPR-modified variant of the U373 tumor line harboring a single- gene knockout in RAD18 was used. Deletion of this DNA repair mediator was expected increase sensitivity and killing for agents that utilize this pathway. The deletion did not affect the growth of untreated wildtype or knockout cells, now termed U373WT and U373KO, both in vitro and on the OBSCs. [0147] All seven OBSC-engrafted tumors were subjected to treatment with a panel of approved CNS tumor therapeutics and external radiation therapy (XRad), as well as the experimental agent TR107 (Madera), a second-generation derivative of ONC201
35,36. Six concentrations of each agent were tested, assessing treatment response by quantitative bioluminescence imaging (BLI) to specifically measure tumor kill. Dose-response curves and half maximal inhibitory concentrations (IC50s) were then calculated for each agent (Fig 6A). If treatment elicited <50% death at the highest dose, a “Not Reached (NR)” was reported. [0148] By comparing IC50s (Fig 6A), it was found that 7 of 7 lines showed robust response to the proven agents Carboplatin (Sigma-Aldrich, Cat# C2538-100MG) Cisplatin (Sigma-Alrich, Cat# 232120-50MG), and Lomustine (Selleck Chemicals, Cat# S1840), with the PDX GBM cell line GBM8 showing markedly higher sensitivity. Several agents, including Temozolomide (Sigma-Aldrich, Cat# T2577-100MG), Vincristine (Sigma- Aldrich, Cat # V8388-1MG), XRad, and Gemcitabine (Sigma-Aldrich, Cat# S1149) failed to induce significant killing at even the highest tested doses in more than 5 of the 7 cell
Attorney Docket No.4210.0527WO lines, resulting in designations of NR. Azacitidine (Sigma-Aldrich, Cat# A2385-250MG) and Etoposide (Sigma-Aldrich, Cat# E1383-100MG) showed broad potency, inducing killing in 7 of 7 and 6 of 7 lines, respectively, at IC50s in the 200-400 µM range. [0149] When comparing U373WT vs U373KO, it was found that IC50s were significantly decreased and killing significantly increased in U373KO following treatment with TMZ, azacitidine, gemcitabine, and etoposide. Interestingly, U373KO also showed increased resistance to radiation, possibly due to an upregulation of the nonhomologous end joining pathway. [0150] The experimental agent TR107 was the most potent therapeutic in the panel, inducing complete killing in 7 of 7 cell lines with IC50s in the 0.04-0.1 µM range. Interestingly, potent but incomplete killing of tumor cells by TR107 is often observed in standard in vitro culture (Fig 6B). Because ClpP inhibitors such as TR107 and ONC201 can both directly act on tumor cells and indirectly increase tumor kill by acting on normal cells in the tumor microenvironment (TME), the complete killing observed here when tumors are grown on OBSCs highlights a difference in functional killing patterns in vitro and on OBSCs that may be due to drug-tumor-OBSC interactions. [0151] In the clinical setting, combination regimens are the mainstay of patient care. Following this investigation of single agent therapy, drug responses for clinically relevant therapeutic combinations on OBSCs was tested (Fig 6C-D). Treatment of GBM8 across low doses of TMZ and varying doses of radiation resulted in additive killing with synergistic kill at high radiation doses as indicated by the ZIP synergy score, while combining TMZ and radiation against MS21 revealed strong antagonism. When treatment response of etoposide and carboplatin combination therapy were compared against U373WT and U373KO, the addition of low-dose carboplatin to etoposide surprisingly produced an antagonistic effect in U373WT, which was reversed as higher doses of carboplatin led to enhanced killing by the combination regimen. This effect was not observed in U373KO, where the elevated sensitivity to etoposide monotherapy led to robust killing and synergy from the combination regimen. Synergistic killing of MB231Br and PDIPG is also observed when treating these tumor lines with carboplatin and etoposide. These data highlight the ability of the OBSC platform to measure synergistic effects of combination therapies.
Attorney Docket No.4210.0527WO Example 4 Advancing LTSs, including OBSCs, as a tool for standardized drug testing [0152] As therapeutics are developed for clinical use, assessment of (1) direct tumor killing and (2) toxicity to normal tissue, e.g. brain tissue, both play central roles in defining efficacy. To provide a normalized comparison among drugs with varying potencies, a drug scoring system based on LTS, e.g. OBSC, functional testing was created. Leveraging quantitative imaging methods to measure tumor volumes and health of OBSCs, multiple parameters of drug activity were assessed, methods for normalized assessment that established a “therapeutic window” were developed, and data were collapsed into a single ranked scoring system for multi-component comparison. This process was designed and optimized using the panel of brain tumor lines before moving into uncultured patient brain tumor tissue. [0153] Numerous parameters such as IC10, IC50, and AUC are commonly used to define drug activity. The disclosed analysis calculates 11 such parameters from tumor dose- response curves to determine multiple efficacy values for each agent against each cell line (Fig 7A-B). The ability to separately define values across tumor foci using BLI and normal brain tissue (OBSCs with no tumor added) using the disclosed predefined PI assay allowed for the measurement of both efficacy and off-target toxicity within many parameters and generation of normalized therapeutic window ratios for each drug-tumor-LTS (OBSC) interaction. Within these windows, values ranged from +1 to -1, where values approaching +1 signify increasing tumor kill relative to normal tissue toxicity, and values approaching -1 indicate agents where tumors remained highly viable while toxicity to the normal LTS, e.g. OBSC, tissue was elevated. Incorporating LTS or OBSC toxicity in this way allowed comparison among drugs with less bias toward more potent compounds. [0154] This approach was first applied to GBM8, U373KO, MS21, and LN229 treated with etoposide (Fig 7A). Multi-parameter analysis was driven by comparing each tumor’s dose-response curve, with survival normalized to untreated tumor BLI, to the OBSC toxicity curve (Fig 7B). MS21 displays a consistent increase in growth in response to treatment, producing therapeutic windows close to -1 for many parameters. LN229 is moderately killed by etoposide treatment but still less than the OBSC itself, producing slightly negative therapeutic window values. U373KO and GBM8 all display incrementally more sensitivity to etoposide relative to the OBSC and thus each produce slightly more positive therapeutic windows across all parameters.
Attorney Docket No.4210.0527WO [0155] Next, a multi-parametric algorithm was developed to collapse the normalized parameters into a single score that simplifies comparison of each drug-tumor-LTS/OBSC interaction. This formula integrates over 100 data points from tumor kill and LTS/OBSC toxicity dose-response curves across the 11 individually weighted parameters to derive an overall Drug Sensitivity Score (DSS, described more fully herein, including in the Methods). DSS were set to range from -100 to +100, where +100 describes maximal performing agents (greatest tumor kill, lowest toxicity) and -100 describes the lowest performing agents (poorest tumor kill, highest toxicity). Fig 7A shows how DSS increases as tumor kill increases. [0156] The value of this disclosed algorithm is strongest in relative comparisons, for example, when comparing tumor responses to several drugs or when comparing dose- response curves of varying shapes. The algorithm was first applied to define DSS for U373WT cells treated with different therapeutic agents (Fig 7C: U373WT = red lines with circles; OBSC = black lines with triangles). As shown in Figure 7C-D, the established agents cisplatin and lomustine resulted in DSS of 62 and 64, respectively, near the top of the range for this cell line. TR107 received the highest DSS at 78, while etoposide scored near the bottom, with a DSS of -13. Although TR107 was extremely potent, displaying robust tumor kill at concentrations orders of magnitude lower than most other drugs, it also presented some toxicity to OBSCs within that range. By basing the DSS on therapeutic window, TR107 could be more fairly compared alongside less potent drugs like temozolomide or therapeutics like XRad which use different dosage units. [0157] To investigate the ability of the algorithm to detect differences in single-gene alterations, drug response profiles of U373WT were compared to those of the RAD18- knockout line U373KO (Fig 7C: U373KO = blue lines with squares). As expected, a marked difference in DSS was detected for several therapies between the two cell lines, where RAD18 deletion led to increased sensitivity to TMZ, etoposide, and azacitidine, and a corresponding increase in DSS. In comparing each pair of dose-response profiles against each therapeutic, some showed the most significant differences at low drug doses, while others showed greater curve separation at high doses, and still others displayed distinct killing profiles altogether, underscoring the importance of calculating a unified DSS which incorporates many aspects of each curve. [0158] Lastly, the testing was expanded and DSS generated for 11 therapies across 8 different tumor types (Fig 7D), enabling broader comparisons and providing insight into
Attorney Docket No.4210.0527WO drug:tumor activity. This allowed for the comparison of (1) the relative efficacy of a single agent across multiple tumor types (horizontal rows) and (2) the relative sensitivity of a single tumor to multiple agents (vertical columns). It was found that GBM8 was the most sensitive tumor line to most therapeutics, with the majority of DSS scores near 90. TR107 was, on average, the most effective agent across tumor lines, but interestingly was least effective against GBM8. TMZ and radiation, agents commonly used in clinical care for GBM, showed wide variability among tumors, with DSS scores ranging from -62 to 96 (TMZ) and -39 to 77 (radiation). Example 5 Utilizing LTSs, including OBSCs, as a diagnostic platform for uncultured patient brain tumor tissue [0159] While the field can choose from many models and assays when testing tumor lines, generating a viable, representative tumor model from uncultured patient tumor tissue has historically been a difficult task, especially in the field of brain cancer. Significant initial cell death is often observed when culturing cells in vitro, and in vivo PDX models require extensive lead times while yielding low rates of establishment
41–43. Even brain tumor PDOs are most successfully established when modeling the most aggressive GBM subtypes
20. To fill this need, disclosed herein is a method to prepare and engraft a diverse panel of living, uncultured patient brain tumor tissues onto LTSs, e.g. OBSCs, for rapid, functional drug screening and diagnosis. To increase flexibility in timing and assay selection, also created and validated was a method to cryopreserve patient tumor tissue while preserving the tumor’s original genetic profile, persistence, and drug response on OBSCs. These methods (1) maximize tumor engraftment and viability independent of tumor grade or subtype, (2) limit cell loss/selection and genetic drift, (3) maintain a normal distribution of tumor per OBSC engraftment site to combat intratumoral heterogeneity, and (4) maximize the number of replicate tumor foci per mass of clinical tumor biopsy tissue. [0160] Fresh surgical biopsies were obtained from patients undergoing standard-of-care resection surgeries at University of North Carolina at Chapel Hill (UNC) hospitals following informed consent. Following cryopreservation and thaw according to the disclosed optimized protocol, the tissue was dissociated into a homogeneous near-single- cell suspension, rapidly transduced with LV-mCh-FLuc, and seeded as tumor foci each
Attorney Docket No.4210.0527WO containing a representative sample of ~1 mg tissue onto OBSCs. To measure the ability of OBSCs to engraft patient brain tumor tissue, tumor tissue survival four days after seeding was first compared on (1) OBSCs in a transwell setup (top left), (2) the transwell membrane without OBSCs (bottom left), and (3) standard in vitro culture (right) via BLI (Fig 8A). Uncultured tumor resection tissue from three different patients bearing three different tumor types (grade I ganglioglioma (GG-I), grade 1 meningioma (MG-1), and a heterogeneous glioblastoma/meningioma tumor (GBM-MG)) showed consistent survival on OBSCs, but not in other culture formats. GG-I, MG-I, and GBM-MG showed <1%, 29%, and <1% viability compared to OBSCs when cultured on the transwell insert alone without OBSC tissue, and <1%, 2%, and <1% viability, respectively, compared to OBSCs when cultured in standard in vitro culture plates. [0161] This tumor seeding method was then applied to eight additional types of adult, pediatric, primary, and metastatic patient brain tumor resection tissue, 11 tumors in all (Fig 8B). As before, quantitative optical imaging revealed that every patient tumor tested reproducibly persisted on OBSCs at t = 4 days after seeding, regardless of tumor type. While efforts were undertaken to understand why OBSCs could provide such an accommodating niche for patient tumor engraftment, OBSC astrocyte activation around tumor tissue was again measured via GFAP. It was found that the engrafted patient tumor tissue also induces activation of – and interaction with – OBSC-embedded astrocytes in a similar manner to the astrocyte activation observed after tumor line engraftment (Fig 8C). [0162] Uncultured tumor resection tissue from an NF2 mutation-driven grade II meningioma (MG-II) was then used to investigate whether tumors engrafted onto OBSCs maintained the genetic profile of the parent tumor. Three groups of samples were prepared for whole exome sequencing (WES) from the same pool of dissociated MG-II tissue (Fig 8D). Group HT is the original uncultured human tumor tissue (n = 3). Group CL is the parent tumor tissue expanded in standard in vitro cell culture until the minimum number of cells required for WES had grown. Because of initial cell loss and subsequent clonal expansion, this required six passages over the span of approximately one month (n = 3). Group BSHT is the uncultured tumor resection tissue engrafted onto OBSCs and subsequently dissected from the OBSCs at the conclusion of our standardized assay length (4 days; n = 4). [0163] WES analysis showed that tumor tissue engrafted onto OBSCs maintained a significant genetic resemblance to the parent tumor, while tumor tissue expanded in vitro
Attorney Docket No.4210.0527WO displayed a distinctly different profile (Fig 8E: left plot displays top 250 most significantly mutated somatic genes; right plot displays top 25, derived from left plot). Furthermore, the mutational profiles of all four BSHT biological replicates were markedly similar, indicating that each OBSC-engrafted tumor indeed contained a representative sample of the original patient tumor. A closer look at the hallmark NF2 mutations existing within all samples revealed that while all samples from the original tumor (HT1-3) and the OBSC- engrafted tumor (BSHT1-4) maintain the frame shift deletion at V24, this mutation was lost in all samples expanded in vitro (CL1-3) and replaced by mutations in other areas. Together, this data suggests that the rapid assay design and tumor-accommodating niche of the disclosed LTS platform, including OBSC platform, enables effective maintenance of the original patient tumor profile. [0164] Consistency in seeding and viability is essential for accurate and reproducible results when assessing therapeutic agents. To investigate consistency in initial seeding, patient tumor tissue was labelled with a fluorescent dye and compared the initial sizes of 24 tumor foci at 1 hour post-seeding (two foci per OBSC; two OBSCs per well). It was found that well-to-well reproducibility was high and yielded no statistically significant differences (Fig 9A). Similar consistency in longitudinal tumor persistence was observed, with no statistical differences in viability (via BLI) among unique samples on day 4, 6, and 8 after seeding (Fig 9B, n=6 foci). [0165] OBSCs can also support the persistence of cryopreserved and thawed patient tumor tissue. The disclosed method of cryo-storage, thaw, and engraftment onto OBSCs generates reproducible persistence (Fig 9C) and response to therapeutic agents (Fig 9D) when compared to fresh tumor tissue from the same patient tumor, significantly enhancing the versatility and flexibility of this platform. Example 6 Patient Tumor Sensitivities Assessed By DSS [0166] Viable brain tumor resection tissue is often limited, leading in part to the dearth of functional testing strategies for brain cancers. While functional diagnostic platforms for other solid tumors generally test many therapeutics and require a large amount of tissue, the studies disclosed herein purposefully limited the number of screened therapeutics to those already being considered by attending clinicians. The workflow has been set up to accept any amount of available tumor tissue and screen even a small number of drugs in
Attorney Docket No.4210.0527WO order to facilitate clinical decisions among top therapeutic options. For each tissue acquired, a clinical team helps curate the panel of relevant therapeutics and determine which drugs to test on each tumor based on tumor type, mutational status, and amount of available tissue. By working with clinicians in this way, a focus on comparisons can directly guide care for each patient in real time. [0167] Uncultured brain tumor resection tissue from ten patients was seeded onto OBSCs and treated with various therapeutics to generate DSS (Fig 10A). Fig 10A compares patient DSS alongside DSS for all established tumor lines (data repeated from Fig 7D). Trends across every DSS for all tumors (Fig 10B) show that our DSS algorithm can calculate scores throughout the entire scoring range: patient tumors are generally more sensitive to treatment than the cell-culture-selected established lines in our panel, save for the ubiquitously sensitive GBM8 tumor line, but there is still a wide distribution in patient tumor sensitivities across all drugs and tumors. [0168] Table 1 describes all tumor types and reported mutational statuses of patient tumor tissue used in this study were compared. Multitudes of co-occurring mutations appear frequently in this set of tumors, presenting difficulty in making treatment decisions based on mutational status alone. We can begin to correlate our results to this clinical data by comparing mutations, treatments, and outcomes in individual patients to DSS calculated from treating matched live patient tissue on OBSCs. Table 1: Patient Tumor Clinical Data. Data was derived from clinical records in accord with our IRB-approved protocol.
Attorney Docket No.4210.0527WO
Example 7 Correlating OBSC-Derived Patient Sensitivities to Clinical Data [0169] Next, the correlation between treatment responses predicted by the OBSC platform and the true clinical response using this small cohort of patient biopsies was investigated. To this end, clinical pathologic and genomic data on patient biopsies was collected under this open protocol. Here, correlative data is reported on specimens which received additional treatment after surgical resection.
Attorney Docket No.4210.0527WO [0170] A pediatric patient underwent subtotal tumor resection for a rim-enhancing, centrally necrotic, left temporal mass. Tumor PGBM underwent extensive histopathological and genomic analysis for clinical purposes, and demonstrated diffuse and high-grade glioblastoma histology, IDH1-mutation (IDH1 c.395G>A (p.Arg132His), MGMT-promoter methylation profile, TERT-promoter-wild-type, and absence of 1p/19q codeletion. Immunohistochemistry (IHC) also demonstrated “patchy” BRAF V600E positivity, although DNA analysis of the tumor did not demonstrate any BRAF V600E alteration. DNA analysis for hotspot mutations also demonstrated CDK4 amplification, PAX5 V129M, PIK3CA H1047R, and TP53 R273. [0171] Despite the many identified histopathological and genomic events in this tumor, the optimal choice based on driver predictions and drug efficacy is simply not well- understood within the current paradigm of precision oncology. Several potential targets for precision medicine are theoretically conceivable, including TMZ to target MGMT methylation44, IDH inhibitor to target IDH1 mutation
45, CDK4/6 inhibitor to target CDK4 amplification
46, PIK3CA inhibitor to target PIK3CA mutation-induced AKT activation
47, and BRAF and MEK inhibitors to target the patchy BRAF V600E positivity on IHC
48, but it is difficult to predict which will be most effective. [0172] Bolstered by the clinical prediction that this MGMT-methylated tumor should respond well to TMZ, the patient underwent standard-of-care treatment with concurrent XRad and TMZ following subtotal resection. The patient exhibited minimal response to therapy and the tumor rapidly progressed. While this outcome was not predicted by MGMT methylation status, it was consistent with OBSC data: the DSS suggested that TMZ (DSS = 26) and XRad (DSS = 16) would show limited efficacy against this tumor (Fig 10A). [0173] The discrepant IHC and molecular results surrounding the BRAF V600E status of this tumor may have suggested a possible role for BRAF or MEK inhibition. While this patient never received a MEK inhibitor to target MAPK pathway upregulation associated with the possible BRAF V600E mutation, trametinib vs PGBM on OBSCs killed ~1/3 of the tumor at the highest dose (DSS = 16), correlating slightly better with the negative DNA mutation analysis of the tumor rather than the patchy positive IHC, although the term “patchy” is certainly qualitative. [0174] This patient underwent re-resection of the recurrent tumor, yielding PGBM-R. The patient resumed TMZ treatment but again showed minimal response to therapy. This
Attorney Docket No.4210.0527WO again aligns with OBSC results for PGBM-R, which suggested that TMZ (DSS = 17) would continue to be ineffective. Clinical genomic profiling reported that the MGMT- promotor was no longer methylated in the post-treatment tumor, suggesting development of a possible resistance mechanism
49. [0175] Genomic profiles are known to change upon tumor recurrence, and these differences can lead to changes in drug sensitivities as well. Fig 10C shows DSS profiles of PGBM (yellow) and PGBM-R (green) relative to all other tested tumor lines (blue) and patient tissues (red). While sensitivity to TMZ and XRad did not significantly change upon tumor recurrence, DSS for azacitidine and trametinib increased by 96 and 58 points, respectively, in the recurrent tumor. This functional data could be integral in developing updated treatment plans for recurrent tumors which fail first-line treatment. [0176] Analysis of other tumor samples revealed additional associations between DSS scoring, clinical outcomes, and genomic analysis. Patient tumor PIS received XRad, etoposide, and carboplatin/etoposide/Ifosfamide (ICE) treatment (DSS via OBSC for XRad = 37; etoposide = 80; carboplatin = 88) and has not experienced recurrence. Patient tumor MMG-II received XRad (DSS = 53; top among patient tumors) and continues to be in relapse-free remission. Example 8 Patient Ovarian Cancer Tumor on Living Tissue Substrates (LTS) [0177] Figures 11A-11C show the results of the testing and analysis of patient ovarian cancer tumor on various living tissue substrates (LTS). The testing and evaluation included tumor engraftment, drug treatment, and DSS calculation. Fig. 11A shows dose response curves of LTS toxicity following a 3-day exposure to carboplatin. Fig. 11B shows bioluminescence of ovarian patient tumor t= 4 days after engraftment on LTS, coupled with bioluminescence and fluorescence imaging and analysis (not shown). Bothconfirm the success of culturing uncultured patient tumor tissue on the disclosed LTS systems. Fig. 11C shows dose response of ovarian patient tumor engrafted on LTS t=3 days after exposure to carboplatin. DSS of ovarian patient tumor against carboplatin was calculated as follows: Brain, 80.1; Kidney, 92.8; and Mesentery, 90.9. Although the same tissue was engrafted on the different LTS, there was varied dose response, captured by the range of DSS scores, for tumor kill and similarly drug toxicity to each LTS.
Attorney Docket No.4210.0527WO Example 9 Development of LTSs using organotypic mesentery membrane culture (OMMCs) as a tool for standardized drug testing [0178] Figures 12A-12K show the results of OMMCs generation. Fig.12A is a schematic illustration of OMMCs generation. Fig. 12B shows region of interest from above view of the isolated mesentery (green drawing) and display of its net of cells and extracellular components by light microscopy and H&E staining. Figs. 12C and 12D show survival of 8-week-old rat mesentery on OMMCs. Fig. 12C shows BLI tracking of the transduced mesentery over a 10-day period. Fig. 12D shows survival of mesentery over a 17-day period using the PI assay. Fig. 12E shows mesentery killing by gradual increase in DMSO concentrations. Right top shows PI fluorescence measured with the AMI optical system and bottom left and right sides, shows the PI fluorescence from dead cells when expose to 0% and 100% DMSO respectively. Consistent mesentery survival is observed and confirmed by PI assay across experiments as shown in Fig. 12F. Fig. 12G shows similar down trend in mesentery survival became significant after DAY 11 for all ages evaluated, except for 3 and 4 weeks old where the PI fluorescence was not measurable from Day 5 on. Fig. 12H is a photograph from three different mesentery ages showed a shrinkage of the region of interest for 3- and 4-week-old mesenteries on Day 8. Fig. 12I shows two regions of the rat mesentery were selected to determine cell count and membrane thickness, Ileum and Jejunum using PI staining and confocal imaging. There was a homogenous number of cells in both regions not showing a significant difference (Fig. 12J), however, a difference was noticed in thickness in between the regions (Fig. 12K). [0179] After a meticulous optimization process during OMMC generation we were able to show the selected region of interest on the rat mesentery is consistently viable up to 10- 11-day period independently of animal age and across independent experiments. In addition, homogenous cell content and similar tissue structure in between Ileum and Jejunum suggest a potential LTS for a new ex vivo platform to study metastatic cancer tumors. [0180] Figures 13A-13G show tumor spots on OMMCs. Fig. 13A shows tumor seeding process on OMMC. Fig. 13B shows light, fluorescent and BLI pictures from above view
Attorney Docket No.4210.0527WO of tumor spots on a mesentery with a magnified display of a well-rounded tumor spot. Fig. 13C shows ES-2 and SKOV3 showed consistent tumor growth on OMMC in 10 days. Fig. 13D shows reproducibility in tumor growth for ES- and SKOV3 leading to survival above 100% across separate experiments. Fig. 13E shows minimal inter-well variability (>600 multiple comparisons for 36 wells) after manual tumor seeding on Day 0. Fluorescent imaging confirmed a clear potential macrophage activation when tumor is present. [0181] These results showed OC established cell lines, survived, and proliferated in OMMC during the time studied, while minimal variability in manual tumor spot placing was observed. Interestingly, a marked tumor spot-mesentery interaction observed by activation of immune cells in the present of tumor showed the existence of a dynamic and responsive living system. [0182] Figures 14A-14E show tumor drug response, drug toxicity on the mesentery and drug sensitivity score (DSS). Fig. 14A shows drug exposure effect on tumorless OMMCs survival, using increasing concentrations of FDA approved single and combination chemotherapies (Olaparib, Gemcitabine, Carboplatin, Paclitaxel, Paclitaxel-Carboplatin 10 and 100) in a 3-day period. Fig. 14B is a visual of schematic (Top) and real (Bottom) of OMMC system with ES-2 tumor spots suggesting its potential functionality to assess toxicity and tumor drug response by BLI quantification. Fig. 14C shows tumor drug- response curves on OMMCs along with mesentery viability after 3 day exposure to the same group of chemotherapies. Fig. 14D is an example of calculated DSS for the two cell lines against Gemcitabine and DSSs array for both cell lines against all drugs from 0 to 100 suggest increasing efficacy in tumor kill relative to OMMC toxicity, while scores from 0 to −100 describe scenarios in which tumors thrive more effectively than OMMC for a given treatment. Fig. 14E is an example of therapeutic window across all DSS weighted parameters for Gemcitabine treated tumor cell line. Values ranged from-1 to +1, where values approaching+1 indicate better tumor kill relative to less toxicity on the tissue, and values approaching −1 suggest tumors remained viable while toxicity to the normal OMMCs tissue was elevated. [0183] Figures 15A-15D show OC biopsies on OMMCs. Fig. 15A includes mean values of tumor growth on OMMMs showing all patient tumors stay alive and even proliferates for some in a 6-day period. Fig. 15B is a comparison of patient OC tumor growth in different cultured systems where OMMCs suggests a better tumor substrate. Fig. 15C includes two examples of inter-well variability of the human OC tumor spots at the time
Attorney Docket No.4210.0527WO of placement on the mesentery membrane, showing there was a consistent tumor cell manipulation with no significant difference in BLI values inter-well. Fig. 15D shows significant tumor response on OMMCs to 500uM of chemotherapies. [0184] Figure 16 shows all biopsy tumor response curves on OMMCs per individual treatment and their corresponded therapeutic window across all DSS parameters were calculated. The DSSs array for both cell lines against all drugs from 0 to 100 suggest increasing efficacy in tumor kill relative to OMMC toxicity, while scores from 0 to −100 describe scenarios in which tumors thrive more effectively than OMMC for a given treatment. [0185] Our OMMC system allowed us to test tumor responses of OC established cell lines and tumor tissues surgically dissected from patients when we exposed them to FDA approved standard of care treatments for OC patients. Simultaneously, we could also measure drug toxicity on the healthy mesentery. Human tumor tissues thrived significantly better in our OMMC than in vitro or ex vivo but with no mesentery. This suggests a microenvironment of a living substrate might be necessary to establish certain biological conditions for optimal tumor tissue engraftment or survival. We believe this will increase the study window for these difficult growing resected human tumor tissues in the laboratory. The calculated DSS reflects a more complete measure of the treatment efficacy and potentially facilitates a tool to guide precision medicine in the future. [0186] Example 10 Further testing of cell line culturing on various tissue substrates in the LTS models as a tool for standardized drug testing [0187] Figures 17A-17E show how various LTS originating from other organs such as kidney, liver, and lung have been developed to engraft, treat, and analyze treatment response of various tumor cell lines and uncultured patient tumor tissue samples. Fig.17A shows growth of tumor cell lines of various origin on LTS from brain, kidney, liver, and lung. on LTS. Fig. 17B shows dose-response curves of Lomustine vs LN229 tumor cells growing on LTS from liver, brain, and kidney. Fig. 17C shows off-target toxicity of Lomustine and Azacitidine against LTS from liver and kidney. Fig.17D shows brightfield
Attorney Docket No.4210.0527WO images of various tissue substrates in the disclosed LTS systems. Figure 17E displays images of engrafted patient tumor tissue surviving on various tissue substrates in the disclosed LTS systems at t = 4 days after engraftment. This data, alongside data in Fig 11, provides additional evidence that various LTS systems generated as described in this disclosure can be used to engraft, treat, and analyze uncultured patient tumor tissue. Example 11 Discussion of Examples 1-10 [0188] While significant advancement in CNS tumor diagnostics have been made in the past decades, with a meaningful move away from grouping tumors exclusively based on their appearance under light microscopy and towards integrated histopathological and molecular diagnoses, there remains a critical functional gap in the diagnostic schema: how will the tumor respond to therapy? Even within contemporary, well-defined, uniform tumor subtypes there exist profoundly stubborn non-uniform clinical outcomes
50,51. The data presented here describes LTS-based technology as a functional diagnostic tool that fills this critical gap. [0189] The data presented herein (1) further characterizes and standardizes the disclosed LTS platforms and assays, including but not limited to OBSC and OMMC platforms and assays, (2) describes how these LTSs can model tumor lines which are not readily established in vivo, (3) treats a diverse set of brain cancer cell lines with a panel of therapeutics, (4) introduces a multi-parametric algorithm to analyze dose-response data of both tumor kill and normal brain tissue toxicity, (5) translates the technology toward the clinic via testing on a wide variety of uncultured patient tumor specimens, and (6) validates this translational approach by identifying areas of association between patient tumor DSS and clinical outcomes/genomic data. [0190] The DSS array generated from established and low-passage tumor lines helps substantiate these findings in patient tumor tissue. This array acts as a reference dataset, defining therapeutic window and DSS values for responders and non-responders while probing responses to known mutational profiles and predicted sensitivities. MB231Br and GBM8 are both tumor cell lines which express upregulation or dependency on the MAPK pathway – MB231Br via the BRAF V600E mutation52 and GBM8 via the PTEN deletion53 – suggesting sensitivity to the MEK inhibitor Trametinib. Indeed, these demonstrated the two highest DSS vs trametinib among cell lines. Further, the invariably
Attorney Docket No.4210.0527WO high DSS in the GBM8 cell line correlates with other experiments using this cell line which have demonstrated a broad sensitivity to many drugs
53,54. Furthermore, the RAD18 knockout U373 cell line, a line with an inability to activate the DNA damage response after alkylator chemotherapy exposure, resulted in expected changes in therapeutic sensitivity to TMZ compared to the wild-type line
55. The increased sensitivity to Temozolomide exhibited by U373KO on OBSCs furthers the reliability of OBSCs to capture nuanced differences in sensitivity and reflect these differences in a DSS. Together, these data indicate the ability of LTSs, including OBSCs and OMMCs, to preserve targetable mutations in tumor cells while also affording the requisite DSS sensitivity to differentiate between those lines that are more and less sensitive. [0191] As evidenced by the presently disclosed data for TR107, this LTS assay and algorithm can also evaluate the efficacy of experimental therapeutics. TR107 averaged the highest DSS among all treated tumors and obtained the highest individual DSS in 5/12 tumors, suggesting that this drug may have potential for further preclinical development in these indications. Furthermore, the distinct tumor killing patterns of TR107 on LTSs compared to in vitro data suggests that LTSs may model drug interactions within the TME as well. Other experimental therapeutics may be screened in this way to determine relative efficacy vs standard-of-care, help interrogate resistance mechanisms to a specific therapeutic or class of drugs as they are being developed, or even use this platform to screen patients for eligibility on a clinical trial. [0192] In summary, the disclosed LTS platforms, including OBSC and OMMC platforms, fills an unmet gap in the diagnosis and treatment of CNS tumors: functional precision diagnosis. This assay and algorithm treat the patient’s own tumor cells with both approved and experimental therapeutics and provide a normalized and simply summarized DSS output, which will help guide treatment and evaluate new therapies. B. Drug Sensitivity Score Algorithm [0193] Functional precision medicine platforms are strategies to improve pre-clinical drug testing and guide clinical decisions. Effective precision diagnosis to guide cancer treatment is a critical unmet need. Genomic tumor profiling often lacks actionable outputs, while many in vitro and in vivo models of patient disease lack the accuracy or speed to provide timely, relevant information to guide patient care. In some example, non-limiting implementations, the current subject matter relates to an organotypic brain slice culture
Attorney Docket No.4210.0527WO (OBSC)-based platform and multi-parametric algorithm which may enable rapid engraftment, treatment, and analysis of uncultured patient brain tumor tissue and patient- derived cell lines. As can be understood, an OBSC is one example of a living tissue substrate (LTS) and is discussed herein by way of illustration only and is not intended to limit the current subject matter. In various experiments, the current subject matter platform supported engraftment of every patient tumor tested, including, but not limited to, high- and low-grade adult and pediatric tumor tissue, which may rapidly establish on OBSCs among endogenous astrocytes and microglia while maintaining the tumor’s original DNA profile. In that regard, the current subject matter processes may be configured to determine dose-response relationships of both tumor kill and normal brain tissue toxicity, generating summarized drug sensitivity scores based on therapeutic window and allowing normalized response profiles across a panel of FDA-approved and exploratory agents. [0194] The accurate classification of central nervous system (CNS) tumors has made significant progress in the past two decades, moving from an exclusive reliance on histopathological features toward more integrated diagnoses based on molecular insights. While the improved classification of CNS tumors is already enhancing the scientific rigor of clinical trials, the impact of these integrated diagnoses on precision oncology medicine – the molecular-guided pairing of tumors with drugs – has yet to demonstrate widespread clinical benefit. Between 2006 and 2020, a time period during which an astounding number of new cancer-directed drugs were developed, eligibility for those drugs only increased from around 5% to 13%, and response to those drugs only increased from around 3% to 7%. The reasons for this are myriad, and include factors such as co-occurring oncogenic alterations, tumor heterogeneity, epistatic interactions, and adaptive cellular circuitry. It is becoming increasingly clear that static histopathological and molecular measurements of tumors are still insufficient to most accurately and usefully classify CNS tumors, and that to identify precision medicines for the majority of CNS tumor patients, functional diagnostic platforms are needed. [0195] Patient-derived models of cancer (PDMCs), including cell lines, patient-derived organoids (PDOs), patient-derived explants (PDEs), and patient-derived xenografts (PDXs), provide functional models of a patient’s individual tumor that can be screened with multiple drugs. The potential for PDMCs to guide personalized care has been demonstrated in several studies which show their ability to successfully predict antitumor response. Indeed, drug screening assays using PDMCs have successfully predicted
Attorney Docket No.4210.0527WO antitumor responses in humans, demonstrating their potential. Unfortunately, initiation time, cost, efficiency scales, and similarity to the parent tumor limit applications of PDXs, while patient-derived cell lines often lose the genetic and phenotypic heterogeneity of the parent tumor via cell selection during clonal expansion. Some PDOs and PDEs can effectively capture the histological and mutational diversity of human cancers, but are most successfully generated from aggressive, high-grade tumors such as IDH-wildtype glioblastoma. Furthermore, generating these models from heterogeneous tumors also limits reproducibility among intra-tumor replicates. A platform which supports and rapidly engrafts both low- and high-grade brain tumor tissue, maintains genetic heterogeneity and resemblance to the parent tumor, and allows functional testing of approved and experimental therapeutics is still desperately needed. [0196] In some implementations, to address the above needs, the current subject matter relates to a process for diagnosing a patient tumor tissue. The diagnosing may be executed using a computer-implemented system that may be configured to receive a first image of a living tissue substrate (LTS) that may be engrafted with one or more tumor tissue cells. The first image may be obtained at a predetermined time and/or after a predetermined time period. For example, the first image may be a tumor fluorescence image obtained using an imaging apparatus (e.g., any type of known imaging apparatus may be used) on day one (and/or any other time, time period, etc.) using any known fluorescence imaging techniques. The first image may be used to identify one or more tissue cells. A computer vision (CV) algorithm may be used for identification of such tumor tissue cells. Another or second image of the LTS may then be obtained using the imaging apparatus (as can be understood, same and/or different imaging apparatus may be used). The second image may be a tumor bioluminescence image that may be obtained at another or second predetermined time or time period. The second image may be obtained subsequent to an application of a candidate therapeutic (e.g., a candidate cancer treatment drug, etc.) to the LTS. An analysis of the second image may be performed to determine a tumor tissue cell kill parameter of the candidate therapeutic. A further or a third image of the LTS may then be obtained. The image may be obtained without an engrafted tumor, where the LTS has been treated with the candidate therapeutic. This image may be an organotypic slice culture (e.g., organotypic brain slice culture organotypic (OBSC), but, as can be understood, can be any type of organotypic culture) fluorescence image obtained at a third predetermined time, which may be after the first predetermined time. Using such image,
Attorney Docket No.4210.0527WO the current subject matter may then determine toxicity of the candidate therapeutic against the LTS. The obtained information/data, e.g., toxicity and/or the tumor tissue cell kill parameter may be used to determine or generate a drug sensitivity score (DSS) of the candidate therapeutic and a type of tumor tissue cells. One or more machine learning models and/or artificial intelligence platforms may be used to determine or generate the DSS. [0197] For determining the DSS, the ML/AI platform(s) may be trained using various historical data (e.g., prior images of cells obtained at different times, prior DSS determinations, patient data, candidate therapeutic data, etc.). The trained platforms may then be used to determine one or more optimal treatments for patient tumors. For example and without limitation, the ML and/or AI platform(s) may determine, for each of a plurality of candidate drugs, dose-response relationships of both tumor kill and normal tissue toxicity, generating summarized DSSs based on therapeutic window and allowing to normalize response profiles across a panel of FDA-approved and exploratory agents. As such, as one of the benefits of the current subject matter, summarized patient tumor scores obtained after treatment may be used to reveal positive associations to clinical outcomes and thus, provide rapid, accurate, functional testing and, ultimately, guide patient care. [0198] The ML/AI platform may be configured to analyze images of patient tumors at various stages of growth. For example, a tissue slice may be engrafted with one or more tumor cells. One or more images of the engrafted tissue slice may be captured at one or more different time periods. In one example, images may be captured one day after the engraftment, followed by capturing images four days after engraftment. However, images may be captured at any timing intervals, e.g., 2 days and 5 days, 0 days and 6 days, etc. [0199] By way of a non-limiting example and for ease of illustration, the first set of images, e.g., the images captured one day after engraftment (referred to as day 1 tumor fluorescence (D1TF) images), may be prior to the application of a candidate drug to treat the tumor. The D1TF images may therefore depict the initial size of the tumor prior to the application of the candidate drug. The D1TF images may depict the fluorescence (or luminance or brightness) of the tumor. Tumor cells may express different fluorescent/luminescent protein markers which produce light. Therefore, the D1TF images may depict the fluorescence and/or luminescence of the tumor, which may express a fluorescent marker protein due to transfection or some similar process. In the D1TF
Attorney Docket No.4210.0527WO images, the tumor is the only tissue that fluoresces; the LTS behind the tumor is just a background signal. [0200] In some implementations, a mask of the tumor present in a D1TF image may be generated. The mask may then be used to determine one or more attributes of the tumor, e.g., an average fluorescence, a total fluorescence, an area of the tumor, and/or any other attributes, and/or any combinations thereof. The mask may be programmatically generated using one or more computer vision (CV) algorithms, which allows regions of interest (ROI) for tumors of any shape and/or type to be identified and masks to be generated for the tumors. Doing so may improve the detection of tumors, including irregularly shaped tumors, for which ROIs and/or masks cannot be easily determined using regular shapes (e.g., circles, etc.). One non-limiting example of such CV algorithm may include an existing Otsu’s method. Another non-limiting example may include measuring a background signal from the corners of the image and using the background signal as the threshold or a starting point for threshold calculation. A further non-limiting example may include using edge detection (e.g. the Canny edge detector) to identify the edges of tumor spots. As can be understood, any other suitable CV algorithms may be used. Generally, using CV algorithm may more accurately identify a tumor and its shape, which may allow for more accurate generation of ROIs and masks than existing methodologies. [0201] As discussed above, following obtaining of the first set of images, a second set of images, e.g., the images captured four days after tumor engraftment on the organotypic culture, may be obtained after application of the candidate therapeutic to the tumor co- culture. These images may be generated as part of an experiment that also produces a first set (D1TF) of images. The second set of images may capture tumors, which may also be represented in the D1TF images. These images may for example include one or more of the following components: one component of the image(s) may be captured using brightfield techniques (e.g., using the entire spectrum of visible, white light to capture the image), another component may be captured at a specific wavelength of light corresponding to the signal generated by a fluorescent and/or bioluminescent marker, and/or any combination thereof. For ease of discussion and illustration only, the first component of the image may be referred to as a brightfield component, and the second component may be referred to as a signal component. Since tumor cells may be imaged using this bioluminescent technique, these images may be referred to as day 4 tumor bioluminescence (D4TB) images. The size of the tumor may be determined from the
Attorney Docket No.4210.0527WO signal component of the image by measuring the total brightness of the pixels associated with the tumor depicted in the images. The D4TB image may reflect the tumor size after treatment using a respective candidate therapeutic, and the change in tumor size from the D1TF image to the D4TB image may be used to determine the efficacy of the respective candidate therapeutic at one dose and/or a range of doses against the engrafted tumor. [0202] A third set of images (which may be obtained after the first set of images), e.g., the images of the organotypic culture taken four days after creation of the organotypic culture, may be taken after the application of a candidate therapeutic to the organotypic culture. These images may be captured in an experiment that may be separate from (and/or concurrent to) the experiment which produces D1TF and D4TB images. Like the D4TB images above, these images may include one or more of the following components: a brightfield component, a signal component, and any combination thereof. Since the health of the organotypic culture may be measured using a fluorescent marker (e.g., propidium iodide, etc.), these images may be referred to as a day 4 organotypic culture fluorescence (D4OF) images. The fluorescent signal component of the image may quantify the presence of cell death. The health of a particular organotypic culture may be measured by finding an average brightness per area across the entire organotypic culture. The process may also include positive controls, which may correspond to complete organotypic culture(s) death, and/or negative controls, which may correspond to maximally healthy organotypic culture(s). The toxicity of a candidate therapeutic to an organotypic culture may be quantified by measuring the signals from D4OF images and comparing treatment groups to the negative and/or positive controls. The toxicity of a candidate therapeutic to organotypic culture, as measured by D4OF images, may be compared to the efficacy of the candidate therapeutic against one or more tumor types, as measured by D1TF and D4TB images. [0203] In some implementations, CV algorithms may be applied to the brightfield components of the D4TB and D4OF images to identify the organotypic culture (e.g., a brain slice, and/or any other slice) and generate a mask, which may denote which pixels in the brightfield image represent the organotypic culture. The CV algorithms may be trained to identify tissue (e.g., organotypic cultures, tumors, etc.) and use the organotypic culture and/or tumors as masks. In some examples, the mask generated from the brightfield component of a D4TB or D4OF image may be overlaid on the corresponding fluorescent and/or bioluminescent signal component of the D4TB or D4OF image to
Attorney Docket No.4210.0527WO determine the signal generated by a tumor or OBSC. For example, the bioluminescence associated with the tumor may be determined by generating masks (e.g., via an ML algorithm which identifies organotypic culture in the brightfield component of the image) and applying these masks to the corresponding pixels in the bioluminescent signal component of the image. Alternatively, or in addition, the fluorescence associated with an organotypic culture may be determined by generating a mask (e.g., via an ML algorithm which identifies organotypic culture in the brightfield component of the image) and applying the mask to the corresponding pixels in the fluorescent signal component of the image. In some implementations, the masks representing organotypic culture in the D4TB images may be programmatically bisected, so that each organotypic culture mask can be used to measure two separate tumor spots. The bisection of the organotypic culture masks may generate a separate mask for each organotypic culture portion, each portion including one of the tumor spots. The masks may then be used to extract radiance information describing the tumor spots in the D4TB images. In some implementations, the masks, as generated from the brightfield images, may be used without alteration, but additional algorithms, e.g., algorithms similar to the ones described for D1TF images, may be used to identify the tumor spot with even more accuracy. [0204] Generally, the day 4 images may be processed to determine the effectiveness of a given candidate treatment. The effectiveness may be based at least in part on the amount of tumor cells killed and the amount of toxicity to the LTS. The amount of tumor cells and the toxicity to the LTS may be measured by comparing the tissues in the day 4 images to that in the day 1 image, by comparing treatment groups to control groups, or a combination of both. As stated, the day 4 images may be measured based on fluorescence values and/or bioluminescence values extracted from the images to determine the tumor cells killed and/or the toxicity to the LTS, and the comparison may be made relative to the tumor cells and the LTS in the day 1 image. In some implementations, the bioluminescence of tumor cells in the day 4 images may indicate that the tumor cells may still be alive. In some implementations, the fluorescence of the LTS in the day 4 images may indicate cell death. In some implementations, the LTS fluorescence on day 4 may be compared between negative control groups, positive control groups, and/or treatment groups in order to quantify the effect of the treatment on the LTS. Thus, by determining which cells are alive and/or dead in the day 4 images relative to the day 1 image and/or relative to control
Attorney Docket No.4210.0527WO groups, the current subject matter may be configured to measure an amount of tumor cells killed and a amount of toxicity to the LTS. [0205] The output of the image analysis may be used to determine one or more scores for each of a plurality of candidate therapeutics to treat the tumor. The scores may include the drug sensitivity score (DSS) for each candidate therapeutic. The DSS for each candidate therapeutic may be used to select a treatment for the patient. The DSS may be computed according to any range of values, such as values from -100 to 100, where 100 is the most effective treatment (e.g., greatest tumor kill, lowest toxicity to non-tumor cells) and -100 is the least effective treatment (e.g., poorest tumor kill, highest toxicity to non-tumor cells). The DSS scores may be used to identify new treatments for a patient and/or new doses of treatments for a patient (e.g., a lower dose and/or higher dose for the patient that is different than a standard dose). In some implementations, a machine learning (ML) model may be used to determine the DSS scores. [0206] In some implementations, the DSS is determined from a dose-response curve fitting procedure with appropriate logic tests. Attempted curve fits may be made by testing different dose response curve equations and the least squares calculation. The fitted dose response curves may be ordered by Akaike information criterion (AIC) calculation from least to greatest. The model with the lowest AIC and is the first model to have a non-NaN and non-infinite log-likelihood score, and has less than 0.3 residual variance is chosen as the best fit model. No effective dose numbers are calculated for dose response models of the tumor where the AUC curve is greater than the AUC curve of the slice health dose response model. [0207] The DSS values may be determined using an ML model that may apply weights to one or more and/or each of a plurality of different features. The ML model may be trained to learn the weights of one or more or each feature. The input to the model may include a plurality of metadata attributes of a patient and the output of the image analysis described herein (e.g., the analysis of the D1TF, D4TB and D4OF images). For example, the features may include at least one of the following: a killing at maximum dose (Max Kill) parameter, a dose required to kill 10% of the tumor (EC10), a dose required to kill 25% of the tumor (EC25), a dose required to kill 50% of the tumor (EC50), a dose required to kill 75% of the tumor (EC75), a dose required to kill 90% of the tumor (EC90), a slope through the EC50, an area under the curve (AUC), a tumor growth acceleration, an incomplete kill parameter at the highest dose, a biphasic killing parameter corresponding
Attorney Docket No.4210.0527WO to a rapid killing at low doses and limited additional killing at higher doses, and any combination thereof. As can be understood, any other parameters may be used. Because the model is trained on attributes including tumor type, the weights of each parameter may be tuned according to the tumor type, such that effective treatments for a particular tumor type may be determined. For example, the weights of the parameters include at least one of the following: a 10% weight for the Max Kill parameter, a 5% weight for the EC10 parameter, a 5% weight for the EC25 parameter, a 10% weight for the EC50 parameter, a 5% weight for the EC75 parameter, a 5% weight for the EC90 parameter, a 10% weight for the slope through IC50 parameter, a 5% weight for a tumor growth acceleration parameter, a 5% weight for a biphasic killing curve parameter, a 5% weight for an incomplete kill parameter, a 35% weight for the AUC parameter, and any combination thereof. As can be understood, any other weights may be used. The flexibility of the ML model may allow the ML model to be applied to determine treatment plans for any number and types of tumors. As patient treatment plans and outcomes are monitored over time, the ML model may be retrained based on these patient treatment plans and outcomes, further improving the accuracy of the ML model in generating effective treatment plans. [0208] In some example, non-limiting implementations, feature weights could comprise anywhere from 0%-100% in a ML model as more information is added and in certain instances some parameters will be dropped (e.g., not included in the trained ML model). As an example, for a non-aggressive tumor line (patient or cell) that is susceptible to most drugs, and provides a high DSS the EC90 may not be as important and so the weight of how it affects the DSS is going to be limited. Another consideration may be if a cell line that responds well to any kind of treatment, if it always reaches complete response, then incomplete kill and biphasic both become obsolete and would therefore not be needed when comparing what treatment strategy should be used across that section to be returned to the provider. Thus, the DSS that may be presented to a provider might not constitute a reflection of all DSS scores ever generated, but may be weighted specifically to the patient in which case if a portion of the DSS may be preserved in a non-floating weight, it would make it more difficult to tease out the best treatment option if it was not actually important. In some implementations, features that do not change may be dropped (e.g., making their DSS weight set to zero), thereby making all other DSS components weights increase. [0209] FIG. 18 illustrates an example system 1800 for diagnosing of patient tumor, according to some implementations of the current subject matter. The system 1800 may
Attorney Docket No.4210.0527WO include an imaging apparatus 1802 capable of imaging a living tissue substrate 1804 and generating one or more images 1806, a drug sensitivity score engine 1808 that may receive images 1806, a storage location 1826 that may store images 1806, and a computing device 1822. The drug sensitivity score engine 1808 may be configured to include a tumor tissue cell identification engine 1816, a tumor tissue cell kill parameter engine 1818, a candidate therapeutic toxicity engine 1820, an image processing engine 1812, and one or more machine learning (ML) models 1810. The engine 1808 may be configured generate one or more DSS scores 1824 that may be presented on a graphical user interface of the computing device 1822. [0210] One or more components of the system 1800 may be communicatively coupled using one or more communications networks. The communications networks may include one or more of the following: a wired network, a wireless network, a metropolitan area network ("MAN"), a local area network ("LAN"), a wide area network ("WAN"), a virtual local area network ("VLAN"), an internet, an extranet, an intranet, and/or any other type of network and/or any combination thereof. [0211] Further, one or more components of the system 1800 may include any combination of hardware and/or software. In some implementations, one or more components of the system 100 may be disposed on one or more computing devices, such as, server(s), database(s), personal computer(s), laptop(s), cellular telephone(s), smartphone(s), tablet computer(s), virtual reality devices, and/or any other computing devices and/or any combination thereof. In some example implementations, one or more components of the system 1800 may be disposed on a single computing device and/or may be part of a single communications network. Alternatively, or in addition to, such devices may be separately located from one another. A device may be a computing processor, a memory, a software functionality, a routine, a procedure, a call, and/or any combination thereof that may be configured to execute a particular function associated with validation processes disclosed herein. [0212] In some implementations, the system 1800’s one or more components may include network-enabled computers. As referred to herein, a network-enabled computer may include, but is not limited to a computer device, or communications device including, e.g., a server, a network appliance, a personal computer, a workstation, a phone, a smartphone, a handheld PC, a personal digital assistant, a thin client, a fat client, an Internet browser, or other device. One or more components of the system 1800 also may
Attorney Docket No.4210.0527WO be mobile computing devices, for example, an iPhone, iPod, iPad from Apple® and/or any other suitable device running Apple’s iOS® operating system, any device running Microsoft's Windows®. Mobile operating system, any device running Google's Android® operating system, and/or any other suitable mobile computing device, such as a smartphone, a tablet, or like wearable mobile device. [0213] One or more components of the system 1800 may include a processor and a memory, and it is understood that the processing circuitry may contain additional components, including processors, memories, error and parity/CRC checkers, data encoders, anti-collision algorithms, controllers, command decoders, security primitives and tamper-proofing hardware, as necessary to perform the functions described herein. One or more components of the system 1800 may further include one or more displays and/or one or more input devices. The displays may be any type of devices for presenting visual information such as a computer monitor, a flat panel display, and a mobile device screen, including liquid crystal displays, light-emitting diode displays, plasma panels, and cathode ray tube displays. The input devices may include any device for entering information into the user's device that is available and supported by the user's device, such as a touch-screen, keyboard, mouse, cursor-control device, touch-screen, microphone, digital camera, video recorder or camcorder. These devices may be used to enter information and interact with the software and other devices described herein. [0214] In some example implementations, one or more components of the system 1800 may execute one or more applications, such as software applications, that enable, for example, network communications with one or more components of system 1800 and transmit and/or receive data. [0215] One or more components of the system 1800 may include and/or be in communication with one or more servers via one or more networks and may operate as a respective front-end to back-end pair with one or more servers. One or more components of the environment 1800 may transmit, for example from a mobile device application (e.g., executing on one or more user devices, components, etc.), one or more requests to one or more servers). The requests may be associated with retrieving data from servers. The servers may receive the requests from the components of the system 1800. Based on the requests, servers may be configured to retrieve the requested data from one or more databases (e.g., storage location 1826, as shown in FIG. 18). Based on receipt of the requested data from the databases, the servers may be configured to transmit the received
Attorney Docket No.4210.0527WO data to one or more components of the system 1800, where the received data may be responsive to one or more requests. [0216] The system 1800 may include one or more networks, such as, for example, networks that may be communicatively coupling one or more components of the system 1800, including the computing device 1822. In some implementations, networks may be one or more of a wireless network, a wired network or any combination of wireless network and wired network and may be configured to connect the components of the system 1800 and/or the components of the system 1800 to one or more servers. For example, the networks may include one or more of a fiber optics network, a passive optical network, a cable network, an Internet network, a satellite network, a wireless local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a virtual local area network (VLAN), an extranet, an intranet, a Global System for Mobile Communication, a Personal Communication Service, a Personal Area Network, Wireless Application Protocol, Multimedia Messaging Service, Enhanced Messaging Service, Short Message Service, Time Division Multiplexing based systems, Code Division Multiple Access based systems, D-AMPS, Wi-Fi, Fixed Wireless Data, IEEE 802.11b, 802.15.1, 802.11n and 802.11g, Bluetooth, NFC, Radio Frequency Identification (RFID), Wi-Fi, and/or any other type of network and/or any combination thereof. [0217] In addition, the networks may include, without limitation, telephone lines, fiber optics, IEEE Ethernet 802.3, a wide area network, a wireless personal area network, a LAN, or a global network such as the Internet. Further, the networks may support an Internet network, a wireless communication network, a cellular network, or the like, or any combination thereof. The networks may further include one network, or any number of the exemplary types of networks mentioned above, operating as a stand-alone network or in cooperation with each other. The networks may utilize one or more protocols of one or more network elements to which they are communicatively coupled. The networks may translate to or from other protocols to one or more protocols of network devices. The networks may include a plurality of interconnected networks, such as, for example, the Internet, a service provider's network, a cable television network, corporate networks, home networks, etc. [0218] The system 1800 may include one or more servers, which may include one or more processors that maybe coupled to memory. Servers may be configured as a central system, server or platform to control and call various data at different times to execute a
Attorney Docket No.4210.0527WO plurality of workflow actions. Servers may be configured to connect to the one or more databases. Servers may be incorporated into and/or communicatively coupled to at least one of the components of the system 1800. [0219] One or more components of the system 100 may be configured to execute one or more transactions using one or more containers. In some implementations, each transaction may be executed using its own container. A container may refer to a standard unit of software that may be configured to include the code that may be needed to execute the action along with all its dependencies. This may allow execution of actions to run quickly and reliably. [0220] As shown in FIG. 18, the imaging apparatus 1802 may be configured to obtain one or more images 1806 of the living tissue substrate (LTS). The imaging apparatus 1802 may be any known imaging apparatus that may be configured to obtain different types of images of the LTS 1804 (e.g., using different techniques, such as, for instance, brightfield or signal techniques, as discussed herein). The imaging apparatus 1802 may be configured to obtain such images at different times and/or during different periods, such as for example, at an initial imaging of the LTS, immediately after application of a candidate therapeutic (e.g., a cancer-treating drug, an illness treatment drug, etc.) to the LTS, some time (e.g., minute(s), hour(s), day(s), week(s), month(s), etc.) after application the candidate therapeutic to the LTS, etc., and/or at any other desired time. The images may be single images, collection of images, continuous images, etc. The images may include still images, videos, and/or any other types of images. [0221] The obtained images 1806 may be stored in one or more storage locations 1826. The storage location 1826 may be a database (e.g., a column-store, a row-store, etc.) and/or any other type of storage location, which may be accessed to store data (e.g., images, information, etc.) and/or to query and/or retrieve data for processing. Once obtained, the imaging apparatus 1802 may be configured to provide the images 1806 to the drug sensitivity score engine 1808 for determination of a drug sensitivity score (DSS) 1824 for presentation of a graphical user interface 1822. [0222] To determine the DSS score 1824, the engine 1808 may be configured to receive a first image or images 1806 of the LTS from the imaging apparatus 1802. The LTS may be engrafted with one or more tumor tissue cells. Such image may be obtained by the imaging apparatus 1802 at a first predetermined time (and/or period of time), e.g., prior to application of a candidate therapeutic. In some example, non-limiting implementations,
Attorney Docket No.4210.0527WO the first images may be a tumor fluorescence image. As can be understood, the first image may be any other type of image. [0223] Upon receiving the first image(s), the tumor tissue cell identification engine 1816 of the DSS engine 1808 may be configured to identify the tumor tissue cells in the LTS that may be present in the image. To do so, the tumor tissue cell identification engine 1816 may be configured to access and execute a computer vision (CV) algorithm 1814 of the image processing engine 1812. The CV algorithm 1814 may be configured to be any type of computer vision algorithm (e.g., an Otsu's method and/or any other type of method). The tumor tissue cell identification engine 1816, using the CV algorithm 1814, may determine and/or, otherwise, ascertain a region of interest associated with the identified tumor tissue cells and generate a mask for such identified cells. In some implementations, the tumor tissue cell identification engine 1816 may be configured to identify the tumor tissue cells based on a brightness (e.g., representing an amount of light emitted by the tumor tissue cells) of one or more pixels associated with the one or more tumor tissue cells and a brightness of one or more pixels associated with the LTS in the received first image. Further, the region of interest and the mask may be determined by the tumor tissue cell identification engine 1816 based on a brightness of one or more pixels associated with the tumor tissue cells and a brightness of one or more pixels associated with the LTS in the received first image. In this case, the brightness may represent an amount of light emitted by the tumor tissue cells and/or an amount of light emitted by the LTS. The mask may be characterized and/or may include one or more attributes of the tumor tissue cell(s). The attribute(s) may include at least one of the following: a size of one or more tumor tissue cells, a location of one or more tumor tissue cells, an intensity of light emitted by one or more tumor tissue cells, and any combination thereof. [0224] The determined region of interest and the mask, along with the first image and/or any other data, information, etc., may be stored by the engine 1808 in the storage location 1826. The stored information may then be queried and/or retrieved by the engine 1808 for further processing, e.g., determination of the DSS score 1824. As can be understood, the engine 1808 may include an internal temporary (and/or permanent) storage capability to store this information to be used in determination of the DSS score 1824. [0225] In some implementations, the imaging apparatus 1802 may be configured to obtain further images of the LTS. Such images may be obtained subsequent to application of one or more candidate therapeutics to the LTS. The images may also be obtained at
Attorney Docket No.4210.0527WO another or second predetermined period of time (or time period) and after the initial set of images that has been obtained by the imaging apparatus 1802. For example, the first set of images may be obtained by the imaging apparatus 1802 on day 1 (or at any other time) of the diagnosis of tumor cells in the LTS, and the second set of images may be obtained by the imaging apparatus 1802 on day 4 (or at any other time) after the initial set of images. The timing of the imaging of the LTS may be configured to be predetermined by the drug sensitivity score engine 1808 and/or any other parameters and/or factors. [0226] The second image(s) may include tumor bioluminescence image(s) and/or any other images. The images may be analyzed by the tumor tissue cell kill parameter engine 1818 of the engine 1808 to determine tumor tissue cell kill parameter(s) associated with the candidate therapeutic applied to the LTS. To determine the tumor tissue cell kill parameter(s), the tumor tissue cell kill parameter engine 1818 may access one or more CV algorithms 1814 of the image processing engine 1812. The CV algorithm 1814 used by the tumor tissue cell kill parameter engine 1818 may be the same and/or different than the CV algorithm 1814 used by the tumor tissue cell identification engine 1816. The CV algorithm 1814 may be trained to identify the LTS and overlay one or more bioluminescence values to determine a signal corresponding to one or more tumor tissue cells. The tumor tissue cell kill parameter engine 1818 may use the trained CV algorithm to generate one or more masks of one or more tumor tissue cells shown in the obtained second image(s). The tumor tissue cell kill parameter(s) of the candidate therapeutic applied to the LTS may then be determined by tumor tissue cell kill parameter engine 1818 using such tumor tissue cells shown in the second image. This information (e.g., masks, and/or tumor tissue cells shown in the second image(s)) may eventually be used by the drug sensitivity score engine 1808 to determine and/or generate the DSS score 1824. [0227] In some implementations, the tumor tissue cells shown in the second image(s) may include one or more tumor spots (e.g., a first tumor spot, a second tumor spot, etc.). The tumor spot(s) may be used by the tumor tissue cell kill parameter engine 1818 to determine radiance of each spot. To do so, the tumor tissue cell kill parameter engine 1818 may be configured to use the trained CV algorithm to bisect the mask(s) of one or more tumor tissue cells shown in the second image into one or more portions, where each portion may include a corresponding tumor spot (e.g., a first portion may include a first tumor spot, a second portion may include a second tumor spot, etc.). The radiance of each spot may be determined based on each portion. The radiance of the tumor spot(s) in each
Attorney Docket No.4210.0527WO portion of the mask may then be used by the tumor tissue cell kill parameter engine 1818 to determine the tumor tissue cell kill of the candidate therapeutic. This data may then be used during determination of the DSS score 1824. The determined tumor tissue cell kill parameter(s) may be stored in the storage location 1826 and/or in any other storage location and/or may be used for training/re-training of CV algorithm(s) 1824. [0228] The imaging apparatus 1802 of the system 1800 may be configured to obtain further or third image(s) of the LTS 1804. Such image(s) may be of the LTS 1804, but without an engrafted tumor. These images may be obtained by the imaging apparatus 1802 subsequent to being treated by the candidate therapeutic. The third images may be sent to the candidate therapeutic toxicity engine 1820 of the drug sensitivity score engine 1808 for analysis/processing, and in particular, determination of toxicity of the candidate therapeutic that had on the LTS. The third image(s) may include an organotypic culture fluorescence image and may be obtained at a third predetermined time (or time period). [0229] In some implementations, the third image(s) may be obtained by the imaging apparatus 1802 after the first image(s) that has been obtained. As discussed above, the first image(s) may have been obtained by the imaging apparatus 1802 on day 1 (or at any other time) of the diagnosis of tumor cells in the LTS, and the third image(s) may be obtained by the imaging apparatus 1802 on day 4 (or at any other time) after the first image(s). The third of image(s) may be obtained prior to, at the same time, and/or after the second image(s). Again, the timing of the imaging of the LTS may be determined by the drug sensitivity score engine 1808 and/or any other parameters and/or factors. [0230] The candidate therapeutic toxicity engine 1820 may likewise access one or more trained CV algorithm(s) 1814 from image processing engine 1812 and apply it to third image(s). The trained CV algorithm(s) 1814 may be trained to identify the LTS and overlay fluorescence values to determine a signal corresponding to the LTS. Application of the CV algorithm(s) 1814 to the third image(s) may result in generation of one or more masks of the LTS shown in the third image(s), which may then be used to determine toxicity of the candidate therapeutic. The toxicity data may be stored in the storage location 1826 and/or any other storage location, and may be used in the determination of the DSS score 1824 and/or may be used for training/re-training of CV algorithm(s) 1824. [0231] Once the data related to the identification of tumor tissue cell(s) (as determined from the first image(s)), the tumor tissue cell kill parameter(s) (as determined from the second image(s)), and/or toxicity of the candidate (as determined from the third image(s))
Attorney Docket No.4210.0527WO is obtained, the drug sensitivity score engine 1808 may be configured to invoke one or more ML models 1810 to generate a drug sensitivity score (DSS) for the candidate therapeutic and a type of the tumor tissue cell(s). The DSS may be generated by one or more ML models 1810, as discussed herein below, based on one or more respective weights applied to a plurality of parameters. The ML models 1810 may be trained using a plurality of images of of other LTSs that may be engrafted with other tumor tissue cells. In some example, non-limiting implementations, the parameters may include at least one of the following: a killing at maximum dose (Max Kill) parameter, a dose required to kill 10% of the tumor (EC10), a dose required to kill 25% of the tumor (EC25), a dose required to kill 50% of the tumor (EC50), a dose required to kill 75% of the tumor (EC75), a dose required to kill 90% of the tumor (EC90), a slope through the EC50, an area under the curve (AUC), a tumor growth acceleration, an incomplete kill parameter at the highest dose, a biphasic killing parameter corresponding to a rapid killing at low doses and limited additional killing at higher doses, and any combination thereof. Further, the example, non- limiting weights that may be applied to one or more such parameters may include at least one of the following: a 10% weight for the Max Kill parameter, a 5% weight for the EC10 parameter, a 5% weight for the EC25 parameter, a 10% weight for the EC50 parameter, a 5% weight for the EC75 parameter, a 5% weight for the EC90 parameter, a 10% weight for the slope through IC50 parameter, a 5% weight for a tumor growth acceleration parameter, a 5% weight for a biphasic killing curve parameter, a 5% weight for an incomplete kill parameter, a 35% weight for the AUC parameter, and any combination thereof. The ML models 1810 may be trained by modifying one or more weights of the parameters and/or removing parameters (thereby assigning more weight to other parameters). [0232] In some example, non-limiting, implementations, DSS score 1824 determined by the drug sensitivity score engine 1808 may range from -100 to 100. A DSS score 1824 having values from 0 to 100 may correspond to an increasing efficacy in tumor kill relative to LTS toxicity. Alternatively, or in addition, a DSS score 1824 having values from 0 to - 100 may correspond to an increasing LTS toxicity relative to the tumor kill. Further, a negative value of the DSS score 1824 may correspond to a near-zero LTS toxicity and an increased tumor growth. As can be understood, other values of the DSS score 1824 may be used to characterize toxicity of the candidate therapeutic being applied to the LTS, tumor kill, tumor growth, and/or any other parameters.
Attorney Docket No.4210.0527WO [0233] FIG.19 illustrates an embodiment of a system 1900. The system 1900 is suitable for implementing one or more embodiments as described herein. In one embodiment, for example, the system 1900 is an AI/ML system suitable for determination of the DSS score 1824 by the system 1800. [0234] The system 1900 comprises a set of M devices, where M is any positive integer. FIG.19 depicts three devices (M=3), including a client device 1902, an inferencing device 1904, and a client device 1906. The inferencing device 1904 communicates information with the client device 1902 and the client device 1906 over a network 1908 and a network 1910, respectively. The information may include input 1912 from the client device 1902 and output 1914 to the client device 1906, or vice-versa. In one alternative, the input 1912 and the output 1914 are communicated between the same client device 1902 or client device 1906. In another alternative, the input 1912 and the output 1914 are stored in a data repository 1916. In yet another alternative, the input 1912 and the output 1914 are communicated via a platform component 1926 of the inferencing device 1904, such as an input/output (I/O) device (e.g., a touchscreen, a microphone, a speaker, etc.). [0235] As depicted in FIG. 19, the inferencing device 1904 includes processing circuitry 1918, a memory 1920, a storage medium 1922, an interface 1924, a platform component 1926, ML logic 1928, and an ML model 1930. In some implementations, the inferencing device 1904 includes other components or devices as well. [0236] The inferencing device 1904 is generally arranged to receive an input 1912, process the input 1912 via one or more AI/ML techniques, and send an output 1914. The inferencing device 1904 receives the input 1912 from the client device 1902 via the network 1908, the client device 1906 via the network 1910, the platform component 1926 (e.g., a touchscreen as a text command or microphone as a voice command), the memory 1920, the storage medium 1922 or the data repository 1916. The inferencing device 1904 sends the output 1914 to the client device 1902 via the network 1908, the client device 1906 via the network 1910, the platform component 1926 (e.g., a touchscreen to present text, graphic or video information or speaker to reproduce audio information), the memory 1920, the storage medium 1922 or the data repository 1916. [0237] The inferencing device 1904 includes ML logic 1928 and an ML model 1930 to implement various AI/ML techniques for various AI/ML tasks. The ML logic 1928 receives the input 1912, and processes the input 1912 using the ML model 1930. The ML model 1930 performs inferencing operations to generate an inference for a specific task
Attorney Docket No.4210.0527WO from the input 1912. In some cases, the inference is part of the output 1914. The output 1914 is used by the client device 1902, the inferencing device 1904, or the client device 1906 to perform subsequent actions in response to the output 1914. [0238] In some implementations, the ML model 1930 is a trained ML model 1930 using a set of training operations. An example of training operations to train the ML model 1930 is described with reference to FIG. 20. [0239] FIG. 20 illustrates an apparatus 2000. The apparatus 2000 depicts a training device 2014 suitable to generate a trained ML model 1810 for the inferencing device 1904 of the system 1900. As depicted in FIG.20, the training device 2014 includes a processing circuitry 2016 and a set of ML components 2010 to support various AI/ML techniques, such as a data collector 2002, a model trainer 2004, a model evaluator 2006 and a model inferencer 2008. [0240] In general, the data collector 2002 collects data 2012 from one or more data sources, which may include images 1806 and/or any other data (e.g., LTS data, toxicity of therapeutics, masks, etc.), to use as training data for the ML model 1810. The data collector 2002 collects different types of data 2012, such as text information, audio information, image information, video information, graphic information, and so forth. The model trainer 2004 receives as input the collected data and uses a portion of the collected data as test data for an AI/ML algorithm to train the ML model 1810. The model evaluator 2006 evaluates and improves the trained ML model 1810 using a portion of the collected data as test data to test the ML model 1810. The model evaluator 2006 also uses feedback information from the deployed ML model 1810. The model inferencer 2008 implements the trained ML model 1810 to receive as input new unseen data, generate one or more inferences on the new data, and output a result such as an alert, a recommendation or other post-solution activity. [0241] An exemplary AI/ML architecture for the ML components 2010 is described in more detail with reference to FIG. 21. [0242] FIG. 21 illustrates an artificial intelligence architecture 2100 suitable for use by the training device 2014 to generate the ML model 1810 for deployment by the inferencing device 1904. The artificial intelligence architecture 2100 is an example of a system suitable for implementing various AI techniques and/or ML techniques to perform various inferencing tasks on behalf of the various devices of the system 1800.
Attorney Docket No.4210.0527WO [0243] AI is a science and technology based on principles of cognitive science, computer science and other related disciplines, which deals with the creation of intelligent machines that work and react like humans. AI is used to develop systems that can perform tasks that require human intelligence such as recognizing speech, vision and making decisions. AI can be seen as the ability for a machine or computer to think and learn, rather than just following instructions. ML is a subset of AI that uses algorithms to enable machines to learn from existing data and generate insights or predictions from that data. ML algorithms are used to optimize machine performance in various tasks such as classifying, clustering and forecasting. ML algorithms are used to create ML models that can accurately predict outcomes. [0244] In general, the artificial intelligence architecture 2100 includes various machine or computer components (e.g., circuit, processor circuit, memory, network interfaces, compute platforms, input/output (I/O) devices, etc.) for an AI/ML system that are designed to work together to create a pipeline that can take in raw data, process it, train an ML model 1930, evaluate performance of the trained ML model 1930, and deploy the tested ML model 1930 as the trained ML model 1930 in a production environment, and continuously monitor and maintain it. [0245] The ML model 1930 is a mathematical construct used to predict outcomes based on a set of input data. The ML model 1930 is trained using large volumes of training data 2126, and it can recognize patterns and trends in the training data 2126 to make accurate predictions. The ML model 1930 is derived from an ML algorithm 2124 (e.g., a neural network, decision tree, support vector machine, etc.). A data set is fed into the ML algorithm 2124 which trains an ML model 1930 to "learn" a function that produces mappings between a set of inputs and a set of outputs with a reasonably high accuracy. Given a sufficiently large enough set of inputs and outputs, the ML algorithm 2124 finds the function for a given task. This function may even be able to produce the correct output for input that it has not seen during training. A data scientist prepares the mappings, selects and tunes the ML algorithm 2124, and evaluates the resulting model performance. Once the ML logic 1928 is sufficiently accurate on test data, it can be deployed for production use. [0246] The ML algorithm 2124 may comprise any ML algorithm suitable for a given AI task. Examples of ML algorithms may include supervised algorithms, unsupervised algorithms, or semi-supervised algorithms.
Attorney Docket No.4210.0527WO [0247] A supervised algorithm is a type of machine learning algorithm that uses labeled data to train a machine learning model. In supervised learning, the machine learning algorithm is given a set of input data and corresponding output data, which are used to train the model to make predictions or classifications. The input data is also known as the features, and the output data is known as the target or label. The goal of a supervised algorithm is to learn the relationship between the input features and the target labels, so that it can make accurate predictions or classifications for new, unseen data. Examples of supervised learning algorithms include: (1) linear regression which is a regression algorithm used to predict continuous numeric values, such as stock prices or temperature; (2) logistic regression which is a classification algorithm used to predict binary outcomes, such as whether a customer will purchase or not purchase a product; (3) decision tree which is a classification algorithm used to predict categorical outcomes by creating a decision tree based on the input features; or (4) random forest which is an ensemble algorithm that combines multiple decision trees to make more accurate predictions. [0248] An unsupervised algorithm is a type of machine learning algorithm that is used to find patterns and relationships in a dataset without the need for labeled data. Unlike supervised learning, where the algorithm is provided with labeled training data and learns to make predictions based on that data, unsupervised learning works with unlabeled data and seeks to identify underlying structures or patterns. Unsupervised learning algorithms use a variety of techniques to discover patterns in the data, such as clustering, anomaly detection, and dimensionality reduction. Clustering algorithms group similar data points together, while anomaly detection algorithms identify unusual or unexpected data points. Dimensionality reduction algorithms are used to reduce the number of features in a dataset, making it easier to analyze and visualize. Unsupervised learning has many applications, such as in data mining, pattern recognition, and recommendation systems. It is particularly useful for tasks where labeled data is scarce or difficult to obtain, and where the goal is to gain insights and understanding from the data itself rather than to make predictions based on it. [0249] Semi-supervised learning is a type of machine learning algorithm that combines both labeled and unlabeled data to improve the accuracy of predictions or classifications. In this approach, the algorithm is trained on a small amount of labeled data and a much larger amount of unlabeled data. The main idea behind semi-supervised learning is that labeled data is often scarce and expensive to obtain, whereas unlabeled data is abundant
Attorney Docket No.4210.0527WO and easy to collect. By leveraging both types of data, semi-supervised learning can achieve higher accuracy and better generalization than either supervised or unsupervised learning alone. In semi-supervised learning, the algorithm first uses the labeled data to learn the underlying structure of the problem. It then uses this knowledge to identify patterns and relationships in the unlabeled data, and to make predictions or classifications based on these patterns. Semi-supervised learning has many applications, such as in speech recognition, natural language processing, and computer vision. It is particularly useful for tasks where labeled data is expensive or time-consuming to obtain, and where the goal is to improve the accuracy of predictions or classifications by leveraging large amounts of unlabeled data. [0250] The ML algorithm 2124 of the artificial intelligence architecture 2100 is implemented using various types of ML algorithms including supervised algorithms, unsupervised algorithms, semi-supervised algorithms, or a combination thereof. A few examples of ML algorithms include support vector machine (SVM), random forests, naïve Bayes, K-means clustering, neural networks, and so forth. A SVM is an algorithm that can be used for both classification and regression problems. It works by finding an optimal hyperplane that maximizes the margin between the two classes. Random forests is a type of decision tree algorithm that is used to make predictions based on a set of randomly selected features. Naive Bayes is a probabilistic classifier that makes predictions based on the probability of certain events occurring. K-Means Clustering is an unsupervised learning algorithm that groups data points into clusters. Neural networks is a type of machine learning algorithm that is designed to mimic the behavior of neurons in the human brain. Other examples of ML algorithms include a support vector machine (SVM) algorithm, a random forest algorithm, a naive Bayes algorithm, a K-means clustering algorithm, a neural network algorithm, an artificial neural network (ANN) algorithm, a convolutional neural network (CNN) algorithm, a recurrent neural network (RNN) algorithm, a long short-term memory (LSTM) algorithm, a deep learning algorithm, a decision tree learning algorithm, a regression analysis algorithm, a Bayesian network algorithm, a genetic algorithm, a federated learning algorithm, a distributed artificial intelligence algorithm, and so forth. Embodiments are not limited in this context. [0251] As depicted in FIG. 21, the artificial intelligence architecture 2100 includes a set of data sources 2102 to source data 2104 for the artificial intelligence architecture 2100. Data sources 2102 may comprise any device capable generating, processing, storing or
Attorney Docket No.4210.0527WO managing data 2104 suitable for a ML system. Examples of data sources 2102 include without limitation databases, web scraping, sensors and Internet of Things (IoT) devices, image and video cameras, audio devices, text generators, publicly available databases, private databases, and many other data sources 2102. The data sources 2102 may be remote from the artificial intelligence architecture 2100 and accessed via a network, local to the artificial intelligence architecture 2100 an accessed via a network interface, or may be a combination of local and remote data sources 2102. As stated above, the data sources may include, but are not limited to images 1806 and/or any other data (e.g., LTS data, toxicity of therapeutics, masks, etc.), [0252] The data sources 2102 source difference types of data 2104. By way of example and not limitation, the data 2104 includes image data from medical images, audio data from speech recognition, text data from emails, chat logs, customer feedback, news articles or social media posts, etc. The data 2104 includes publicly available datasets such as those from government agencies, academic institutions, or research organizations. These are just a few examples of the many sources of data that can be used for ML systems. It is important to note that the quality and quantity of the data is critical for the success of a machine learning project. [0253] The data 2104 is typically in different formats such as structured, unstructured or semi-structured data. Structured data refers to data that is organized in a specific format or schema, such as tables or spreadsheets. Structured data has a well-defined set of rules that dictate how the data should be organized and represented, including the data types and relationships between data elements. Unstructured data refers to any data that does not have a predefined or organized format or schema. Unlike structured data, which is organized in a specific way, unstructured data can take various forms, such as text, images, audio, or video. Unstructured data can come from a variety of sources, including social media, emails, sensor data, and website content. Semi-structured data is a type of data that does not fit neatly into the traditional categories of structured and unstructured data. It has some structure but does not conform to the rigid structure of a traditional relational database. Semi-structured data is characterized by the presence of tags or metadata that provide some structure and context for the data. [0254] The data sources 2102 are communicatively coupled to a data collector 2002. The data collector 2002 gathers relevant data 2104 from the data sources 2102. Once collected, the data collector 2002 may use a pre-processor 2106 to make the data 2104 suitable for
Attorney Docket No.4210.0527WO analysis. This involves data cleaning, transformation, and feature engineering. Data preprocessing is a critical step in ML as it directly impacts the accuracy and effectiveness of the ML model 1930. The pre-processor 2106 receives the data 2104 as input, processes the data 2104, and outputs pre-processed data 2116 for storage in a database 2108. Examples for the database 2108 includes a hard drive, solid state storage, and/or random access memory (RAM). [0255] The data collector 2002 is communicatively coupled to a model trainer 2004. The model trainer 2004 performs AI/ML model training, validation, and testing which may generate model performance metrics as part of the model testing procedure. The model trainer 2004 receives the pre-processed data 2116 as input 2110 or via the database 2108. The model trainer 2004 implements a suitable ML algorithm 2124 to train an ML model 1930 on a set of training data 2126 from the pre-processed data 2116. The training process involves feeding the pre-processed data 2116 into the ML algorithm 2124 to produce or optimize an ML model 1930. The training process adjusts its parameters until it achieves an initial level of satisfactory performance. [0256] The model trainer 2004 is communicatively coupled to a model evaluator 2006. After an ML model 1930 is trained, the ML model 1930 needs to be evaluated to assess its performance. This is done using various metrics such as accuracy, precision, recall, and F1 score. The model trainer 2004 outputs the ML model 1930, which is received as input 2110 or from the database 2108. The model evaluator 2006 receives the ML model 1930 as input 2112, and it initiates an evaluation process to measure performance of the ML model 1930. The evaluation process includes providing feedback 2118 to the model trainer 2004. The model trainer 2004 re-trains the ML model 1930 to improve performance in an iterative manner. [0257] The model evaluator 2006 is communicatively coupled to a model inferencer 2008. The model inferencer 2008 provides AI/ML model inference output (e.g., inferences, predictions or decisions). Once the ML model 1930 is trained and evaluated, it is deployed in a production environment where it is used to make predictions on new data. The model inferencer 2008 receives the evaluated ML model 1930 as input 2114. The model inferencer 2008 uses the evaluated ML model 1930 to produce insights or predictions on real data, which is deployed as a final production ML model 1930. The inference output of the ML model 1930 is use case specific. The model inferencer 2008 also performs model monitoring and maintenance, which involves continuously
Attorney Docket No.4210.0527WO monitoring performance of the ML model 1930 in the production environment and making any necessary updates or modifications to maintain its accuracy and effectiveness. The model inferencer 2008 provides feedback 2118 to the data collector 2002 to train or re- train the ML model 1930. The feedback 2118 includes model performance feedback information, which is used for monitoring and improving performance of the ML model 1930. [0258] Some or all of the model inferencer 2008 is implemented by various actors 2122 in the artificial intelligence architecture 2100, including the ML model 1930 of the inferencing device 1904, for example. The actors 2122 use the deployed ML model 1930 on new data to make inferences or predictions for a given task, and output an insight 2132. The actors 2122 implement the model inferencer 2008 locally, or remotely receives outputs from the model inferencer 2008 in a distributed computing manner. The actors 2122 trigger actions directed to other entities or to itself. The actors 2122 provide feedback 2120 to the data collector 2002 via the model inferencer 2008. The feedback 2120 comprise data needed to derive training data, inference data or to monitor the performance of the ML model 1930 and its impact to the network through updating of key performance indicators (KPIs) and performance counters. [0259] As previously described with reference to FIGS. 19, 20, the systems 1900, 2000 implement some or all of the artificial intelligence architecture 2100 to support various use cases and solutions for various AI/ML tasks. In various embodiments, the training device 2014 of the apparatus 2000 uses the artificial intelligence architecture 2100 to generate and train the ML model 1930 for use by the inferencing device 1904 for the system 1900. In one embodiment, for example, the training device 201414 may train the ML model 1930 as a neural network, as described in more detail with reference to FIG. 22. Other use cases and solutions for AI/ML are possible as well, and embodiments are not limited in this context. [0260] FIG. 22 illustrates an embodiment of an artificial neural network 2200. Neural networks, also known as artificial neural networks (ANNs) or simulated neural networks (SNNs), are a subset of machine learning and are at the core of deep learning algorithms. Their name and structure are inspired by the human brain, mimicking the way that biological neurons signal to one another. [0261] Artificial neural network 2200 comprises multiple node layers, containing an input layer 2226, one or more hidden layers 2228, and an output layer 2230. Each layer
Attorney Docket No.4210.0527WO comprises one or more nodes, such as nodes 2202 to 2224. As depicted in FIG. 22, for example, the input layer 2226 has nodes 2202, 2204. The artificial neural network 2200 has two hidden layers 2228, with a first hidden layer having nodes 2206, 2208, 2210 and 2212, and a second hidden layer having nodes 2214, 2216, 2218 and 2220. The artificial neural network 2200 has an output layer 2230 with nodes 2222, 2224. Each node 2202 to 2224 comprises a processing element (PE), or artificial neuron, that connects to another and has an associated weight and threshold. If the output of any individual node is above the specified threshold value, that node is activated, sending data to the next layer of the network. Otherwise, no data is passed along to the next layer of the network. [0262] In general, artificial neural network 2200 relies on training data 2126 to learn and improve accuracy over time. However, once the artificial neural network 2200 is fine- tuned for accuracy, and tested on testing data 2128, the artificial neural network 2200 is ready to classify and cluster new data 2130 at a high velocity. Tasks in speech recognition or image recognition can take minutes versus hours when compared to the manual identification by human experts. [0263] Each individual node 2202 to 424 is a linear regression model, composed of input data, weights, a bias (or threshold), and an output. [0264] Once an input layer 2226 is determined, a set of weights 2232 are assigned. The weights 2232 help determine the importance of any given variable, with larger ones contributing more significantly to the output compared to other inputs. All inputs are then multiplied by their respective weights and then summed. Afterward, the output is passed through an activation function, which determines the output. If that output exceeds a given threshold, it “fires” (or activates) the node, passing data to the next layer in the network. This results in the output of one node becoming in the input of the next node. The process of passing data from one layer to the next layer defines the artificial neural network 2200 as a feedforward network. [0265] In one embodiment, the artificial neural network 2200 leverages sigmoid neurons, which are distinguished by having values between 0 and 1. Since the artificial neural network 2200 behaves similarly to a decision tree, cascading data from one node to another, having x values between 0 and 1 will reduce the impact of any given change of a single variable on the output of any given node, and subsequently, the output of the artificial neural network 2200.
Attorney Docket No.4210.0527WO [0266] The artificial neural network 2200 has many practical use cases, like image recognition, speech recognition, text recognition or classification. The artificial neural network 2200 leverages supervised learning, or labeled datasets, to train the algorithm. As the model is trained, its accuracy is measured using a cost (or loss) function. This is also commonly referred to as the mean squared error (MSE). [0267] Ultimately, the goal is to minimize the cost function to ensure correctness of fit for any given observation. As the model adjusts its weights and bias, it uses the cost function and reinforcement learning to reach the point of convergence, or the local minimum. The process in which the algorithm adjusts its weights is through gradient descent, allowing the model to determine the direction to take to reduce errors (or minimize the cost function). With each training example, the parameters 2234 of the model adjust to gradually converge at the minimum. [0268] In one embodiment, the artificial neural network 2200 is feedforward, meaning it flows in one direction only, from input to output. In one embodiment, the artificial neural network 2200 uses backpropagation. Backpropagation is when the artificial neural network 2200 moves in the opposite direction from output to input. Backpropagation allows calculation and attribution of errors associated with each neuron 2202 to 2224, thereby allowing adjustment to fit the parameters 2234 of the ML model 1930 appropriately. [0269] The artificial neural network 2200 is implemented as different neural networks depending on a given task. Neural networks are classified into different types, which are used for different purposes. In one embodiment, the artificial neural network 2200 is implemented as a feedforward neural network, or multi-layer perceptrons (MLPs), comprised of an input layer 2226, hidden layers 2228, and an output layer 2230. While these neural networks are also commonly referred to as MLPs, they are actually comprised of sigmoid neurons, not perceptrons, as most real-world problems are nonlinear. Trained data 2104 usually is fed into these models to train them, and they are the foundation for computer vision, natural language processing, and other neural networks. In one embodiment, the artificial neural network 2200 is implemented as a convolutional neural network (CNN). A CNN is similar to feedforward networks, but usually utilized for image recognition, pattern recognition, and/or computer vision. These networks harness principles from linear algebra, particularly matrix multiplication, to identify patterns within an image. In one embodiment, the artificial neural network 2200 is implemented
Attorney Docket No.4210.0527WO as a recurrent neural network (RNN). A RNN is identified by feedback loops. The RNN learning algorithms are primarily leveraged when using time-series data to make predictions about future outcomes, such as stock market predictions or sales forecasting. The artificial neural network 2200 is implemented as any type of neural network suitable for a given operational task of system 1900, and the MLP, CNN, and RNN are merely a few examples. Embodiments are not limited in this context. [0270] The artificial neural network 2200 includes a set of associated parameters 2234. There are a number of different parameters that must be decided upon when designing a neural network. Among these parameters are the number of layers, the number of neurons per layer, the number of training iterations, and so forth. Some of the more important parameters in terms of training and network capacity are a number of hidden neurons parameter, a learning rate parameter, a momentum parameter, a training type parameter, an Epoch parameter, a minimum error parameter, and so forth. [0271] In some cases, the artificial neural network 2200 is implemented as a deep learning neural network. The term deep learning neural network refers to a depth of layers in a given neural network. A neural network that has more than three layers—which would be inclusive of the inputs and the output—can be considered a deep learning algorithm. A neural network that only has two or three layers, however, may be referred to as a basic neural network. A deep learning neural network may tune and optimize one or more hyperparameters 2236. A hyperparameter is a parameter whose values are set before starting the model training process. Deep learning models, including convolutional neural network (CNN) and recurrent neural network (RNN) models can have anywhere from a few hyperparameters to a few hundred hyperparameters. The values specified for these hyperparameters impacts the model learning rate and other regulations during the training process as well as final model performance. A deep learning neural network uses hyperparameter optimization algorithms to automatically optimize models. The algorithms used include Random Search, Tree-structured Parzen Estimator (TPE) and Bayesian optimization based on the Gaussian process. These algorithms are combined with a distributed training engine for quick parallel searching of the optimal hyperparameter values. [0272] FIG. 23 illustrates further details of the image processing engine 1812, according to some implementations of the current subject matter. As shown in FIG. 23, the image processing engine 1812 may be configured to receive one or more images A, B, … C 2302,
Attorney Docket No.4210.0527WO 2304, 2306. The images may be received from imaging apparatus 1802 (as shown in FIG. 18). For example, image(s) A 2302 may be images of the LTS obtained by the imaging apparatus 1802 at first predetermined period of time (and/or time period) and may be used by the image processing engine 1812 to identify one or more tumor tissue cells in the LTS. Image(s) B 2304 may likewise be obtained by the imaging apparatus 1802. However, image(s) 2304 may be obtained at a second predetermined period of time (and/or time period) and may be representative of LTS after application of one or more candidate therapeutics. The second period of time may be different from the first period of time. As discussed herein, the imaging apparatus 1802 may also obtain image(s) C 2306. These images may be obtained at a third predetermined period of time and may be indicative of a toxicity of the candidate therapeutic on the LTS. [0273] Upon obtaining the images, the images 2302, 2304, 2306 may be stored in the storage location 1826 along with any other data associated with the images (e.g., tumor information, candidate therapeutic information, patient information (e.g., anonymized, de- identified, etc.), and/or any other information/data). The images 2302-2306 may also be provided to the image processing engine 1812 for analysis. [0274] The image processing engine 1812 may identify and/or select one or more computer vision algorithm(s) 1814 for analyzing and/or processing of each of the images 2302-2306. For example, one CV algorithm 1814 that may be trained to identify tumor tissue cells in the LTS may be selected for processing of images A 2302. Another CV algorithm 1814 trained on data related to tumor tissue cell kill may be selected for processing of images B 2304. Further, yet another CV algorithm 1814 trained on data related to toxicity of a therapeutic may be selected for processing of images C 2306. The CV algorithms 1814 that may be selected for processing of the images 2302-2306 may be the same and/or different. The CV algorithms may include, but are not limited to, an Otsu’s method, an algorithm that measures a background signal from corners of an image and uses the background signal as a threshold and/or a starting point for threshold calculation, an edge detection algorithm (e.g., Canny edge detector) that identifies edges of tumor spots, etc., and/or any combination of algorithms. As can be understood, any suitable computer vision algorithm and/or combination of algorithms may be used. Computer vision algorithms may be used to more accurately identify a tumor and/or its shape, which may allow for more accurate generation of regions of interest (ROIs) and/or masks.
Attorney Docket No.4210.0527WO [0275] As discussed above, images A 2302, may for example, be captured at a first predetermined period of time (and/or during a first predetermined time period), e.g., one day, after engraftment (D1TF) and prior to application of a candidate therapeutic to treat a tumor. Images A 2302 may depict an initial size of the tumor prior to application of the candidate therapeutic. The D1TF images may depict a fluorescence and/or luminance and/or brightness of the tumor. Tumor cells may express different fluorescent/luminescent protein markers which produce light. Hence, the images A 2302 may express a fluorescent marker protein due to transfection and/or any other process. In the images A 2302, the tumor is the only tissue that fluoresces, where the LTS behind the tumor may be a background signal. An average fluorescence of the tumor may be determined based on images A 2302 to determine a size of the tumor using brightness of one or more pixels associated with the tumor. [0276] Images B 2304 may for example, be captured at a second predetermined period of time (and/or during a second predetermined time period), e.g., four days, after tumor engraftment on the organotypic culture (D4TB images). These images may be obtained after application of the candidate therapeutic to the organotypic culture-tumor. Images B 2304 may be generated as part of an experiment that also produces images A 2302 and/or separately. The images B 2304 may capture tumors which are also represented in images A 2302. Images B 2304 may include a brightfield component that may be captured using brightfield techniques (e.g., using an entire spectrum of visible, white light to capture the image), and a signal component using a specific wavelength of light corresponding to the signal generated by a fluorescent and/or bioluminescent marker. The size of the tumor may be determined from the signal component of the image by measuring the total brightness of the pixels associated with the tumor depicted in the image. The images B 2304 may reflect the tumor size after treatment using a respective candidate drug, and the change in tumor size from the images A 2302 to the images B 2304 may be used to determine the efficacy of the respective candidate therapeutic at one dose or a range of doses against the engrafted tumor. [0277] Images C may be obtained at a third predetermined period of time (and/or during a third predetermined time period), e.g., four days after creation of the organotypic culture (D4OF). These images may be taken after application of the candidate therapeutic to the organotypic culture. Images C 2306 may be captured in an experiment that is separate from the experiment which produces images A 2302 and images B 2304. Alternatively,
Attorney Docket No.4210.0527WO or in addition, images C may be obtained simultaneously and/or at any other time than images A and B. Images C 2306 may include a brightfield component and a signal component. These images may be used to measure health of the organotypic culture using a fluorescent marker, e.g., propidium iodide (PI). The fluorescent signal component of the image may quantify the presence of cell death. The health of a particular organotypic culture may be measured by determining an average brightness per area across the entire organotypic culture. The determination may include positive controls corresponding to complete organotypic culture death, and negative controls corresponding to maximally healthy organotypic culture. The toxicity of the candidate therapeutic to an organotypic culture may be determined by measuring signals from images C 2306 and comparing treatment groups to the negative and positive controls. The toxicity of a candidate therapeutic to organotypic culture, as measured by the images C 2306, may be compared to the efficacy of the candidate therapeutic against one or more tumor types, as measured by images A and B. [0278] Upon processing of images A-C by computer vision algorithm(s) 1814, the image processing engine 1812 may be configured to generated one or more masks 2314. The masks 2314 may denote which pixels in the brightfield image represent the organotypic culture. The algorithms 1814 may be trained to identify tissue (e.g., organotypic culture and/or tumors) and may use the organotypic culture and/or tumors to generate masks. For instance, the mask 1814 generated from the brightfield component of images B or C may be overlaid on the corresponding fluorescent or bioluminescent signal component of the images B or C to determine the signal generated by a tumor and/or organotypic culture. For example, the bioluminescence associated with the tumor may be determined by generating masks 2314 (e.g., using an ML model 1810 which may identify organotypic culture in the brightfield component of the image) and applying these masks to the corresponding pixels in the bioluminescent “signal” component of the image. Alternatively, or in addition, the fluorescence associated with an organotypic culture may be determined by generating masks 2314 (e.g., using an ML model 1810 which may identify organotypic culture in the “brightfield” component of the image) and applying the masks 2314 to the corresponding pixels in the fluorescent signal component of the image. Further, the masks 2314 representing organotypic culture in the images B 2304 may be programmatically bisected, so that each organotypic culture mask may be used to measure separate tumor spots. The bisection of the organotypic culture masks 2314 may generate
Attorney Docket No.4210.0527WO a separate mask for each organotypic culture, each including one or more tumor spots. The masks 2314 may then be used to extract radiance information describing the tumor spots in the images B 2304. The masks as generated from the brightfield images may be used without alteration. Alternatively, or in addition, further algorithms may be used to identify the tumor spot(s) with increased accuracy. [0279] In some implementations, upon generation of the masks 2314, the information/data retrieved from images A-C may be in the further processing pipeline to eventually determine the DSS score 1824. For example, the masks 2314 generated as a result of processing images A 2302 may be used by the tumor tissue cell identification engine 1816 to generate information/data related to tumor cell(s) 2324, including, but not limited, to a number of tumor cells, tumor shapes, tumor sizes, tumor densities, etc. The masks 2314 generated as a result of processing images B 2304 may be used by the tumor tissue cell kill parameter engine 1818 to determine tumor tissue cell kill parameter 2326. Last, but not least, masks produced as a result of analyzing images C 2306 may be used by the candidate therapeutic toxicity engine 1820 to ascertain toxicity of a therapeutic 2328. [0280] The data/information produced in 2324-2328 may be used by the image processing engine 1812 to generate the DSS score 1824. One or more ML models 1810 may be used for the purposes of generating the DSS score 1824. The ML model(s) 1810 may be trained using historical data/information associated with prior analysis of organotypic cultures, tumors, candidate therapeutics, and/or any other information. The model(s) 1810 may be re-trained, refresh-trained, and/or updated based on feedback that may be received from the user and/or any other data sources (e.g., including storage location 1826). Once the DSS score 1824 is generated, it along with data/information 2324-2328 may be stored in the storage location 1826. [0281] In some implementations, to determine the DSS score 1824, the image processing engine 1812 may be configured to use one or more of parameters, one or more of which may be assigned various weights. The image processing engine 1812 may be configured to determine the DSS score 1824 by comparing tumor cell survival, measured using bioluminescence imaging, to the health of the organotypic culture, measured via PI assay (as measured for parameters 1-8 below), and/or by quantifying the behavior of the tumor dose-response curve (as determined for parameters 9-11 below). However, as can be understood, any weighted combination of the below example parameters 1-11 (and/or any
Attorney Docket No.4210.0527WO other parameters) may be used to compute drug sensitivity scores. The example parameters and weights may include at least one of the following and/or any combinations thereof: 1. Killing at maximum dose (Max Kill Window, 10% of DSS) 2. Dose required to kill 10% of the tumor (EC10 Window, 5% of DSS) 3. Dose required to kill 25% of the tumor (EC25 Window, 5% of DSS) 4. Dose required to kill 50% of the tumor (EC50 Window, 10% of DSS) 5. Dose required to kill 75% of the tumor (EC75 Window, 5% of DSS) 6. Dose required to kill 90% of the tumor (EC90 Window, 5% of DSS) 7. Slope through the EC50 (Slope Window, 10% of DSS) 8. The area under the dose-response curve (AUC Window, 35% of DSS) 9. Tumor growth acceleration (e.g., treated tumor growths faster than untreated tumor: for max growth up to 125%, window = +1; 125%-150% window = 0; over 150% window = -1; 5% of DSS) 10. Biphasic killing (rapid killing at low doses and limited additional killing at higher doses: no biphasic curve shape, window = +1; biphasic curve shape, window = -1; 5% of DSS) 11. Incomplete kill at the highest dose (some tumor remaining at highest dose: for <10% remaining, window = +1; 10%-25% remaining, window = 0; >25% remaining, window = -1; 5% of DSS). [0282] By way of a non-limiting example, in determining the DSS score 1824, for parameters 1-6 above, therapeutic windows may be determined by comparing organotypic culture's toxicity and tumor response at the doses where tumor kill passed through a DSS parameter. For DSS parameter 7, the therapeutic window may be determined by comparing slopes through the tumor EC50 and the organotypic culture toxicity EC50. For DSS parameter 8, the therapeutic window may be determined by comparing areas under tumor kill and organotypic culture toxicity curves. Normalized therapeutic window ratios for DSS parameters 1-8 within each drug-tumor-organotypic culture interaction may be determined as follows: within each window, values ranged from +1 to -1, where values approaching +1 signify increasing tumor kill relative to normal tissue toxicity, and values approaching -1 indicate agents where tumors remained highly viable while toxicity to the normal OBSC tissue was elevated. DSS parameters 9-11 may be determined based on a behavior of the tumor in response to the candidate therapeutic.
Attorney Docket No.4210.0527WO [0283] In some example, non-limiting implementations, one or more of the above individually weighed parameters may be added together to generate the DSS score 1824. Overall DSS score 1824 values ranging from 0 to 100 may signify increasing efficacy in tumor kill relative to OBSC toxicity. DSS score 1824 values ranging from 0 to -100 may indicate scenarios in which tumors thrive more effectively than organotypic culture for a given treatment. Dose-response values may be determined using linear interpolation of raw data, best-fit curve equations, and/or any other methodologies. [0284] FIG.24 illustrates an example process 2400 for processing of images (e.g., images A-C 2302-2306) by the image processing engine 1812, according to some implementations of the current subject matter. At 2402, various metadata may be provided to the engine 1812. The metadata may include information about the patient (e.g., de-identified, etc.), the tumor, the candidate therapeutic, the organotypic culture, and/or any other information. This information may be sent to the image processing engine 1812 using a web-fillable form, and/or in any other fashion. [0285] At 2404, after images are obtained by the imaging apparatus 1802, one or more images (e.g., images A-C 2302-2306) may be selected for analysis. In some implementations, the image processing engine 1812 may be configured to select images automatically (and/or based on any desired criteria). Alternatively, or in addition, the images may be manually selected by the user. [0286] At 2406, the image processing engine 1812 may be configured to generate one or more masks 2314. The masks 2314 may be generated for one or more of the selected images A-C. The masks may be used for identification of one or more tumor cells, tumor tissue cell kills 2326, toxicity of a therapeutic 2328, etc. [0287] At 2408, optionally, the user may be presented with the generated masks 2314. The user may review the masks 2314 and provide feedback. The feedback may include approval of the masks, rejection of masks, etc. The feedback may be used by the image processing engine 1812 in training the ML models 1810 that may be used for generation of DSS score 1824 and/or for any other purposes. [0288] At 2410, the masks 2314 may be used to measure a signal that may be generated by one or more pixels associated with the imaged cells, e.g., tumor cells, organotypic culture, and/or any combination thereof. The signals may be reflective of tumor information (e.g., size, number, shape, density, etc.), tumor tissue cell kills (e.g., after
Attorney Docket No.4210.0527WO application of candidate therapeutic), toxicity of candidate therapeutic (e.g., healthy cells being killed by the candidate therapeutic), and/or any other data. [0289] At 2412, the image processing engine 1812 may be configured to determine tumor dose response and find best-fit curve, at 2414. This information may be determined based on the data/metadata that may have been obtained during operations 2402-2410. The image processing engine 1812 may use this information to determine the DSS score 1824, which may be presented to the user on a graphical user interface of the computing device 1822. As can be understood, any other information, along with DSS score 1824, may be presented to the user on the computing device 1822. [0290] FIG. 25 illustrates an example process 2500 for diagnosing patient tumor tissue, according to some implementations of the current subject matter. The process 2500 may be executed using one or more components of the system 1800 shown in FIG. 1800. In particular, the process 2500 may be used to generate a DSS score 1824, which may be determined by the image processing engine 1812 based on various gathered data. The data may include images A-C 2302-2306. [0291] At 2502, a first image of a living tissue substrate (LTS) engrafted with one or more tumor tissue cells may be received and/or obtained. The image may be obtained using imaging apparatus 1802 shown in FIG. 18 and received by the image processing engine 1812. The first image may be similar to image(s) A 2302 as shown in FIG. 23. [0292] At 2504, the image processing engine 1812 may identify, using a computer vision (CV) algorithm (e.g., computer vision algorithm(s) 1814), one or more tumor tissue cells. Identification of the tumor tissue cells may be performed based on analysis of brightness, luminescence, etc. and/or any other graphical parameters of the images A 2302. [0293] At 2506, the imaging apparatus 1802 may obtain and/or the image processing engine 1812 may receive a second image of the LTS. The second image may be obtained/received subsequent to the first image and subsequent to an application of a candidate therapeutic. As can be understood, one or more candidate therapeutics may be applied. The process 2500 may be executed in connection with a single candidate therapeutic and/or multiple candidate therapeutics. The second image may be similar to image(s) B 2304, as shown in FIG. 23. [0294] At 2508, the image processing engine 1812 may determine, based on an analysis of the second image, a tumor tissue cell kill parameter 2326 of the first candidate
Attorney Docket No.4210.0527WO therapeutic. Again, the determination of the tumor tissue cell kill parameter 2326 may be performed based on graphical analysis of the images obtained at 2506, as discussed herein. [0295] At 2510, the imaging apparatus 1802 may obtain and/or the image processing engine 1812 may receive a third image of the LTS. The third image may be similar to image(s) C 2306 and may be without an engrafted tumor. The LTS shown in this image has been treated with the candidate therapeutic. [0296] At 2512, the image processing engine 1812 may determine, based on an analysis of the third image, a toxicity 2328 of the candidate therapeutic against the LTS. [0297] At 2514, the image processing engine 1812 may generate, using one or more machine learning (ML) models 1810, a drug sensitivity score (DSS) 1824 for the candidate therapeutic and a type of the tumor tissue cells based on the toxicity and the tumor tissue cell kill parameter. [0298] In some implementations, the images may be used to generate one or more masks 2314, as discussed herein, and as shown by process 2600 in FIG.26. The masks 2314 may be used to identify one or more tumor spots, at 2602. At 2604, the image processing engine 1812 may bisect, using the CV algorithm 1814, the mask of the one or more tumor tissue cells that may be shown in the second image (as obtained at 2506) into a first portion including one or first tumor spot and second portion including another or a second tumor spot. [0299] At 2606, the image processing engine 1812 may determine, based on the first portion of the mask, a radiance of the first tumor spot, and, based on the second portion of the mask, a radiance of the second tumor spot, at 2608. At 2610, the image processing engine 1812 may determine the tumor tissue cell kill of the candidate therapeutic based on the first portion of the mask for the first tumor spot and the second portion of the mask for the second tumor spot. At 2612, the image processing engine 1812 may generate a DSS score 1824, using an ML model(s) 1810, based on the first portion of the mask and the second portion of the mask. [0300] In some implementations, the current subject matter can be configured to be implemented in a system 2700, as shown in FIG. 27. The system 2700 can include a processor 2702, a memory 2704, a storage device 2706, and an input/output device 2708. Each of the components 2702-2708 can be interconnected using a system bus 2710. The processor 2702 can be configured to process instructions for execution within the system 2700. In some implementations, the processor 2702 can be a single-threaded processor. In
Attorney Docket No.4210.0527WO alternate implementations, the processor 2702 can be a multi-threaded processor. The processor 2702 can be further configured to process instructions stored in the memory 2704 or on the storage device 2706, including receiving or sending information through the input/output device 2708. The memory 2704 can store information within the system 2700. In some implementations, the memory 2704 can be a computer-readable medium. In alternate implementations, the memory 2704 can be a volatile memory unit. In yet some implementations, the memory 2704 can be a non-volatile memory unit. The storage device 2706 can be capable of providing mass storage for the system 2700. In some implementations, the storage device 2706 can be a computer-readable medium. In alternate implementations, the storage device 2706 can be a floppy disk device, a hard disk device, an optical disk device, a tape device, non-volatile solid state memory, or any other type of storage device. The I/O device 2708 can be configured to provide input/output operations for the system 2700. In some implementations, the I/O device 2708 can include a keyboard and/or pointing device. In alternate implementations, the I/O device 2708 can include a display unit for displaying graphical user interfaces. EXAMPLE EXPERIMENTAL IMPLEMENTATION [0301] The following is a discussion of an example experimental implementation of the current subject matter system, as discussed herein. It is provided herein for illustrative, non-limiting purposes only. The system 1800 shown in FIG. 18 may be used for implementation of the system, but, the system 1800 and/or this experimental implementation are not limited thereto. Introduction [0302] The image analysis process may begin with three types of images needed to compute a DSS from an organotypic brain slice cultures (OBSC). As shown in FIG. 25, users may fill out a form which collects metadata relevant to the experiment. After users upload and/or select the experiment images which they want to analyze, the current subject matter may begin by identifying objects in the images and generating binary masks. A binary mask records which pixels in an image should be included in a measurement. The three different types of images may use different methods of mask generation. For the day one tumor fluorescence (D1TF) images, a simple thresholding algorithm may be used. A thresholding algorithm may determine whether a particular pixel is “background” or “signal” by comparing it to a threshold value. This threshold value may be predetermined,
Attorney Docket No.4210.0527WO derived from the image, and/or determined on a pixel-by-pixel basis. One example of a thresholding algorithm that may be used may include determining the highest pixel brightness value in an area that is known to be background (e.g., the corner of the image) and using that as the threshold. Other suitable algorithms (e.g., Otsu’s Method) may be used. Day 4 images may be analyzed using ML computer vision algorithms which may be trained on large numbers of images. One example of a platform that uses machine learning (ML) computer vision services tailored for biological applications may include, for example, Biodock AI, as available from Biodock, Inc., Austin, TX, USA (hereinafter, “Biodock”). Biodock includes an API which may allow for automated analysis using their platform. After masks are generated for all images, users may have the ability to review, edit, and confirm the results. Once all of the masks are confirmed, each image may be measured by utilizing its respective masks. The current subject matter may then organize all of the raw data collected from images (with its related metadata) and save it. The data may be added to a storage location (e.g., a database) and analyzed in order to produce final DSS scores, as described herein. Methods [0303] The OBSC platform may use data from three categories of images 2802 (Day 1 tumor fluorescence), 2804 (Day 4 tumor bioluminescence) and 2806 (Day 4 OBSC fluorescence), as shown in FIG. 28. [0304] Two of these image types may be generated in a tumor kill assay, and the third image type may be generated in a separate OBSC assay. The Day 1 Tumor Fluorescence (D1TF) images 2802, the first set of images produced in the tumor kill assay, may determine the initial size of tumor spots before treatment is administered. The D4 Tumor Bioluminescence (D4TB) images 2804 may measure the final tumor sizes after ~72h of treatment. Each tumor spot may be imaged twice: once on day 1 (D1), and once on day 4 (D4). The final tumor sizes on D4 may be normalized using the D1 data to control for variations in final tumor size that may be due to differences in initial tumor size. For example, a final D4TB image may find that tumor spot A is 10% larger than tumor spot B, but without comparing these results to the D1TF image, it may be difficult to know whether this difference is due to a difference in initial size or a difference in growth over the course of the tumor kill experiment. In both image types, the total brightness of the tumor spot may be used to estimate the tumor size. The total brightness may be determined by adding up all of the individual brightness values from each pixel within the
Attorney Docket No.4210.0527WO region of interest. The brightness value of an individual pixel may indicate how much fluorescence or bioluminescence was generated at that location in the image. It is worth mentioning that the D1TF image brightness may be dependent on the imaging settings used (e.g., exposure time, focus, etc.), and may only be useful for comparison within a given experiment, whereas the D4TB images, taken on an AMI imaging system (Spectral Instruments) using consistent settings, produce precise radiance data in units of photons/second/cm
2/steradian which can be compared between trials. [0305] The OBSC health assay may produce one image type, the D4 OBSC Fluorescence (D4OF) image 2806. These images may be acquired using an AMI (Spectral Instruments). Propidium Iodide (PI) is a fluorescent marker of apoptotic cell death, and soaking an OBSC in PI causes fluorescence which can be quantified to indicate OBSC health. Inclusion of negative and positive control groups generates a window within which we can determine whether a treatment causes no additional killing, complete killing, or something in between. All the OBSCs used in this experimental implementation are 300 µm thick, so it may be assumed that on average they have the same cell density per unit surface area. Thus, the average brightness (as opposed to the total) of pixels may provide a measure of proportional health that controls for variations in OBSC size. [0306] It should be noted that the Aura AMI (used to generate the D4TB and D4OF images) actually generates a composite image, which may include of a brightfield image and a signal image. The brightfield image may look like a regular black-and-white photograph taken across the entire visible spectrum of light. The signal image may include the fluorescence and/or luminescence values. The brightfield and signal components may be seen in FIG. 28: the black-and-white part of the D4 images represents the brightfield image, and the rainbow-colored regions represent the bioluminescent or fluorescent signal overlayed on the brightfield. Overlaying these two components may allow users to determine where signal is localized in the subject being photographed. [0307] Each of these image types may be processed in a slightly different way, as discussed herein, using existing packages (e.g., OpenCV (a Python package)), Biodock, etc. OpenCV is a Python package which provides many functions necessary for importing images and performing Computer Vision (CV) processes on them. OpenCV functions are used for each image type, although the applications will vary somewhat between image types. As can be understood, other packages that provide similar functionality may be used. The web-based platform Biodock may be used to generate masks for D4 images.
Attorney Docket No.4210.0527WO Biodock allows users to upload images, label them, and fine-tune a machine learning (ML) algorithm which has been pre-trained on biological images. The ML system can generate precise masks for each object in an image, even if the objects are touching. This is perfect for the D4 images, since they capture an entire 6 well plate with as many as 12 OBSCs. An ML model was trained to recognize OBSCs based on brightfield images so that the masks may be applied to signal images in order to measure either luminescent or fluorescent data. However, other services that provide similar functionality may be used. It is also feasible to train a ML model using basic programming packages rather than relying on an external, third-party platform. Amazon Web Services (AWS) may be used for various computing processes, data storage, and/or other computing functions. [0308] An object-oriented programming (OOP) may be used to execute the entire analysis process discussed herein. OOP allows programmers to create unique data classes for specific applications. In Python, classes can have attributes, which store data, and methods, which accomplish tasks. By carefully determining what data should be stored and how that data will need to be transformed, it is possible to create classes that ensure all the right data is present and is processed the right way. OOP may be implemented in numerous ways to enable automated image analysis. The image analysis program currently features a hierarchy of four classes: ^ Experiment o Organizes all of the data from a given experiment and features high-level analysis functions. ^ Image o Stores all of the data for a given image and provides functions for analyzing the image based on image type. ^ Well o Represents the wells in D4 images. Allows for organization which is necessary to connect results to relevant experimental metadata. ^ Slices o Represent the OBSCs and allow for measurement of each OBSC. Image Analysis 1. Day 1 Tumor Fluorescence (D1TF) Images [0309] After the D1TF images are obtained (as shown by images 2902-2906 in FIG.29), a binary mask for one or more tumors in the image may be generated by finding the
Attorney Docket No.4210.0527WO brightest pixel value present near the edge of the image (since our researchers always put the spot in the center of the image) and using this value as the threshold for mask generation. For example, the algorithm looks at a 100x100 pixel square in the corner of each image to find the maximum brightness, but any other area known to be background (e.g. just one corner, or a strip along each edge of the image) may be used. Since this algorithm finds the maximum brightness present in an area of known background and sets the threshold to this value, it will be referred to as maximum background thresholding (MBT). As can be understood, a number of other approaches may be used to generate masks for these images, including, for example, finding a threshold via Otsu’s method or performing edge detection via the Canny edge detector. [0310] Once each image has a mask (as shown by masks images 2908-2912 in FIG. 29), the mask may be used to gather data from the tumor spot. The mask may be binary, meaning it only contains values of 0 or 255 (the maximum signal in an 8-bit pixel). The OpenCV function bitwise_and() may apply the mask by making a copy of the original image in which all pixels not included in the mask are set to 0. Numpy functions sum() and count_nonzero() may be used to quickly find the total brightness of the remaining signal and the area of the mask in pixels, respectively. Each image name has a number which represents the order in which the images were taken. This information may be stored in order to connect the spot’s data to the appropriate spot in the D4TB images. [0311] A software program (e.g., ImageJ) used to analyze these images previously, measures the average brightness and the “area” based on some arbitrary conversion from pixels to in
2. Currently, this program multiplies the results by a factor of (20/1144)
2 to make the results comparable to ImageJ, but this factor may be removed once comparison to previous ImageJ measurements is no longer necessary. 2. Day 4 Tumor Bioluminescence (D4TB) Images [0312] The D4TB images may be analyzed using a more complex machine learning algorithm. A ML algorithm may be trained to identify OBSCs in the brightfield images, and the resultant masks may be used to measure bioluminescence from the corresponding signal image. The D4TB brightfield images may be uploaded via an API to Biodock in order to generate masks for one or more tumors in the images. When Biodock is done analyzing the images, users may be asked to confirm the Biodock results and even edit masks if necessary. Once users confirm the masks, JSON files containing the mask data may be automatically downloaded for use. The pycocotools package may parse this data,
Attorney Docket No.4210.0527WO storing masks as arrays and gathering information about which photo each mask belongs to. This mask may be generated based on the ML model’s identification of OBSCs in the brightfield component of the D4TB image. The ML model may be “taught” what an OBSC looks like by being trained on human-labelled images. After the training process, it may recognize OBSCs within a brightfield image and generate a mask for each OBSC. The masks may be used to measure bioluminescence from the signal component of the D4TB image. [0313] Implementation of OOP may be helpful to organize the masks, images, and resultant data. In some implementations, each mask may be stored in a slice object, each slice may be stored in a well object, and each well object may be stored in an image object. This hierarchy of classes may ensure that each mask may be associated with the correct image, treatment, and dose. The masks may be used to collect data from the bioluminescent signal images, which may contain exact information about the radiance of the tumor spots in units of photons/second/cm
2/steradian. [0314] The masks may be generated from the brightfield component of D4TB images, with a size of 1144x776 pixels, but the signal component of the D4TB images, which contains bioluminescence data, have a size of 572x388 pixels. Thus, the masks and their bounding box data may be resized accordingly. The masks may be resized using the OpenCV function cv.resize, which can scale the x and y coordinates by any factor (in this case, the factor is 0.5 for both). For example, an inter-area interpolation method may be used. The bounding box coordinates may undergo integer division. [0315] Additionally, while the masks may cover entire OBSCs, each D4TB image may have one spot per hemisphere. This means one mask may be used to capture two separate tumor spots. Since the OBSCs may be oriented horizontally (so that their transverse/major axis runs left to right), it is possible to split the bounding box into two even halves and measure the two halves of the OBSC separately. Using this technique, looping through the relevant pixels and storing the total radiance may yield two data points per OBSC, one for the left hemisphere and one for the right. In some implementations, the entire OBSC hemisphere may be measured, but thresholding algorithms (similar to the approach for D1TF images) in conjunction with the OBSC masks may be used in order to pinpoint the exact location of each tumor. 3. Day 4 OBSC Fluorescence (D4OF) Images
Attorney Docket No.4210.0527WO [0316] Analysis of these images may be similar to the D4TB images. The images may be uploaded to Biodock, which may compute masks for one or more tumors in the images. As before, the masks are based on the ML model’s identification of OBSCs in the brightfield component of the D4OF image. Since this process may be similar to that for the D4TB images, the same ML model may be used to identify OBSCs in both image types. The resultant masks may be downloaded, parsed, and applied to the images. Since the relevant measurement for OBSC PI is average radiance for the whole OBSC, it is not necessary to split the mask in half. However, the number of pixels measured may be recorded so that the total can be divided by the area. [0317] Once the data is gathered via these separate pipelines, it may be organized into one or more files which store the metadata for each tumor spot. Users may generate properly formatted files/forms by filling out a custom form designed to expedite metadata collection and minimize errors. The files/forms may be uploaded to a database where they may be read by the DSS determination program. [0318] The systems and methods disclosed herein can be embodied in various forms including, for example, a data processor, such as a computer that also includes a database, digital electronic circuitry, firmware, software, or in combinations of them. Moreover, the above-noted features and other aspects and principles of the present disclosed implementations can be implemented in various environments. Such environments and related applications can be specially constructed for performing the various processes and operations according to the disclosed implementations or they can include a general- purpose computer or computing platform selectively activated or reconfigured by code to provide the necessary functionality. The processes disclosed herein are not inherently related to any particular computer, network, architecture, environment, or other apparatus, and can be implemented by a suitable combination of hardware, software, and/or firmware. For example, various general-purpose machines can be used with programs written in accordance with teachings of the disclosed implementations, or it can be more convenient to construct a specialized apparatus or system to perform the required methods and techniques. [0319] The systems and methods disclosed herein can be implemented as a computer program product, i.e., a computer program tangibly embodied in an information carrier, e.g., in a machine readable storage device or in a propagated signal, for execution by, or to control the operation of, data processing apparatus, e.g., a programmable processor, a
Attorney Docket No.4210.0527WO computer, or multiple computers. A computer program can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program can be deployed to be executed on one computer or on multiple computers at one site or distributed across multiple sites and interconnected by a communication network. [0320] As used herein, the term “user” can refer to any entity including a person or a computer. [0321] Although ordinal numbers such as first, second and the like can, in some situations, relate to an order; as used in this document ordinal numbers do not necessarily imply an order. For example, ordinal numbers can be merely used to distinguish one item from another. For example, to distinguish a first event from a second event, but need not imply any chronological ordering or a fixed reference system (such that a first event in one paragraph of the description can be different from a first event in another paragraph of the description). [0322] The foregoing description is intended to illustrate but not to limit the scope of the current subject matter, which is defined by the scope of the appended claims. Other implementations are within the scope of the following claims. [0323] These computer programs, which can also be referred to programs, software, software applications, applications, components, or code, include machine instructions for a programmable processor, and can be implemented in a high-level procedural and/or object-oriented programming language, and/or in assembly/machine language. As used herein, the term “machine-readable medium” refers to any computer program product, apparatus and/or device, such as for example magnetic discs, optical disks, memory, and Programmable Logic Devices (PLDs), used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term “machine-readable signal” refers to any signal used to provide machine instructions and/or data to a programmable processor. The machine-readable medium can store such machine instructions non-transitorily, such as for example as would a non-transient solid state memory or a magnetic hard drive or any equivalent storage medium. The machine-readable medium can alternatively or additionally store such machine instructions in a transient manner, such as for example as
Attorney Docket No.4210.0527WO would a processor cache or other random access memory associated with one or more physical processor cores. [0324] To provide for interaction with a user, the subject matter described herein can be implemented on a computer having a display device, such as for example a cathode ray tube (CRT) or a liquid crystal display (LCD) monitor for displaying information to the user and a keyboard and a pointing device, such as for example a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well. For example, feedback provided to the user can be any form of sensory feedback, such as for example visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including, but not limited to, acoustic, speech, or tactile input. [0325] The subject matter described herein can be implemented in a computing system that includes a back-end component, such as for example one or more data servers, or that includes a middleware component, such as for example one or more application servers, or that includes a front-end component, such as for example one or more client computers having a graphical user interface or a Web browser through which a user can interact with an implementation of the subject matter described herein, or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication, such as for example a communication network. Examples of communication networks include, but are not limited to, a local area network (“LAN”), a wide area network (“WAN”), and the Internet. [0326] The computing system can include clients and servers. A client and server are generally, but not exclusively, remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. [0327] The implementations set forth in the foregoing description do not represent all implementations consistent with the subject matter described herein. Instead, they are merely some examples consistent with aspects related to the described subject matter. Although a few variations have been described in detail above, other modifications or additions are possible. In particular, further features and/or variations can be provided in addition to those set forth herein. For example, the implementations described above can be directed to various combinations and sub-combinations of the disclosed features and/or
Attorney Docket No.4210.0527WO combinations and sub-combinations of several further features disclosed above. In addition, the logic flows depicted in the accompanying figures and/or described herein do not necessarily require the particular order shown, or sequential order, to achieve desirable results. Other implementations can be within the scope of the following claims. [0328] The following examples pertain to further embodiments, from which numerous permutations and configurations will be apparent. [0329] In one aspect, a computer-implemented method may include receiving, using at least one processing circuitry, a first image of a living tissue substrate (LTS) engrafted with one or more tumor tissue cells; identifying, using the at least one processing circuitry, using a computer vision (CV) algorithm, the one or more tumor tissue cells; receiving, using the at least one processing circuitry, a second image of the LTS, the second image being subsequent to the first image and subsequent to an application of a first candidate therapeutic in a plurality of candidate therapeutics; determining, using the at least one processing circuitry, based on an analysis of the second image, a tumor tissue cell kill parameter of the first candidate therapeutic; receiving, using the at least one processing circuitry, a third image of the LTS without an engrafted tumor, where the LTS has been treated with the first candidate therapeutic; determining, using the at least one processing circuitry, based on an analysis of the third image, a toxicity of the first candidate therapeutic against the LTS; generating, using the at least one processing circuitry, using a machine learning (ML) model, a drug sensitivity score (DSS) for the first candidate therapeutic and a type of the tumor tissue cells based on the toxicity and the tumor tissue cell kill parameter. [0330] The method may also include wherein the identifying includes determining, using the CV algorithm, a region of interest associated with the one or more tumor tissue cells; and generating, using the CV algorithm, a mask for the one or more tumor tissue cells. [0331] The method may also include wherein the identifying includes identifying, using the CV algorithm, the one or more tumor tissue cells based on a brightness of one or more pixels associated with the one or more tumor tissue cells and a brightness of one or more pixels associated with the LTS in the first image; wherein the brightness represents an amount of light emitted by the one or more tumor tissue cells. [0332] The method may also include wherein the identifying includes determining, using the CV algorithm, the region of interest and the mask based on a brightness of one or more pixels associated with the one or more tumor tissue cells and a brightness of one or more
Attorney Docket No.4210.0527WO pixels associated with the LTS in the first image; wherein the brightness represents an amount of light emitted by the one or more tumor tissue cells or an amount of light emitted by the LTS. [0333] The method may also include wherein the first image is received prior to the application of the first candidate therapeutic. [0334] The method may also include wherein the DSS is generated by the ML model based on one or more respective weights applied to a plurality of parameters, the one or more weights of the ML model are trained based on a training data, the training data including a plurality of images of at least one another LTS engrafted with other tumor tissue cells. [0335] The method may also include wherein the plurality of parameters include at least one of the following: a killing at maximum dose (Max Kill) parameter, a dose required to kill 10% of the tumor (EC10), a dose required to kill 25% of the tumor (EC25), a dose required to kill 50% of the tumor (EC50), a dose required to kill 75% of the tumor (EC75), a dose required to kill 90% of the tumor (EC90), a slope through the EC50, an area under the curve (AUC), a tumor growth acceleration, an incomplete kill parameter at the highest dose, a biphasic killing parameter corresponding to a rapid killing at low doses and limited additional killing at higher doses, and any combination thereof. [0336] The method may also include wherein one or more initial weights of the parameters include at least one of the following: a 10% weight for the Max Kill parameter, a 5% weight for the EC10 parameter, a 5% weight for the EC25 parameter, a 10% weight for the EC50 parameter, a 5% weight for the EC75 parameter, a 5% weight for the EC90 parameter, a 10% weight for the slope through IC50 parameter, a 5% weight for a tumor growth acceleration parameter, a 5% weight for a biphasic killing curve parameter, a 5% weight for an incomplete kill parameter, a 35% weight for the AUC parameter, and any combination thereof; wherein the ML model is configured to be trained by modifying at least one of the initial weights of the parameters and by, optionally, removing at least one of the parameters. [0337] The method may also include wherein a DSS from 0 to 100 corresponds to increasing efficacy in tumor kill relative to LTS toxicity; a DSS from 0 to -100 corresponds to increasing LTS toxicity relative to tumor kill; and a negative DSS score corresponds to near zero LTS toxicity and increased tumor growth.
Attorney Docket No.4210.0527WO [0338] The method may also include generating, using the at least one processing circuitry, using the CV algorithm applied to the second image, one or more masks of the one or more tumor tissue cells depicted in the second image; wherein the tumor tissue cell kill of the first candidate therapeutic is based on the one or more tumor tissue cells depicted in the second image; wherein the DSS is generated based on at least the tumor tissue cell kill measured using the one or more masks of the one or more tumor tissue cells depicted in the second image; wherein the CV algorithm is configured to be trained to identify the LTS and overlay one or more bioluminescence values to determine a signal corresponding to the one or more tumor tissue cells. [0339] The method may also include wherein the one or more tumor tissue cells depicted in the second image include a first tumor spot and a second tumor spot. [0340] The method may also include bisecting, using the at least one processing circuitry, using the CV algorithm, the mask of the one or more tumor tissue cells depicted in the second image into a first portion including the first tumor spot and a second portion including the second tumor spot; determining, using the at least one processing circuitry, based on the first portion of the mask, a radiance of the first tumor spot; and determining, using the at least one processing circuitry, based on the second portion of the mask, a radiance of the second tumor spot, wherein the tumor tissue cell kill of the first candidate therapeutic is determined based on the first portion of the mask for the first tumor spot and the second portion of the mask for the second tumor spot; wherein the DSS is generated, using the ML model, based on the first portion of the mask and the second portion of the mask. [0341] The method may also include wherein each of the plurality of candidate therapeutics are applied to respective LTS engrafted with a respective tumor tissue cell, wherein a respective tumor tissue cell kill of the respective candidate therapeutic is determined based on respective first and second images of the LTS; each of the plurality of candidate therapeutics is configured to be applied to respective LTS without engrafted tumor tissue cells, wherein a respective LTS toxicity of the respective candidate therapeutic is determined based on the third images of the LTS; a respective DSS for each candidate therapeutic is generated, using the ML model, based on the respective LTS toxicity and the respective tumor tissue cell kill of the respective candidate therapeutic, wherein the first candidate therapeutic is selected based on the DSS scores for each candidate treatment.
Attorney Docket No.4210.0527WO [0342] The method may also include wherein the mask includes a plurality of attributes of the one or more tumor tissue cells, wherein the plurality of attributes include at least one of the following: a size of the one or more tumor tissue cells, a location of the one or more tumor tissue cells, an intensity of light emitted by the one or more tumor tissue cells, and any combination thereof. [0343] The method may also include wherein the first image is a tumor fluorescence image obtained at a first predetermined time; the second image is a tumor bioluminescence image obtained a second predetermined time; and the third image is an organotypic brain slice culture organotypic brain slice culture (OBSC) fluorescence image obtained at a third predetermined time; wherein at least one of the second and third predetermined times occur after the first predetermined time; wherein the DSS is generated, using the ML model, based on one or more measurements made across at least one of: the first image, the second image, the third image, and any combination thereof. [0344] The method may also include generating, using the at least one processing circuitry, using the CV algorithm applied to the third image, one or more masks of the one or more LTS depicted in the third image, wherein the LTS toxicity of the first candidate therapeutic may be determined using the one or more masks of the one or more LTS depicted in the third image; wherein the DSS is generated, using an ML model, based at least on the values of LTS toxicity found by using the mask(s) of the one or more LTS depicted in the third image, the CV algorithm is configured to be trained to identify the LTS and overlay fluorescence values to determine a signal corresponding to the one or more LTS. [0345] In one aspect a system may include at least one processing circuitry; and at least one non-transitory storage media storing instructions, that when executed by the at least one processing circuitry, cause the at least one processing circuitry to perform any of the above operations. [0346] In one aspect a computer program product comprising a non-transitory machine- readable medium storing instructions that, when executed by at least one programmable processing circuitry, cause the at least one programmable processing circuitry to perform any of the above operations.
Attorney Docket No.4210.0527WO REFERENCES All references listed herein including but not limited to all patents, patent applications and publications thereof, scientific journal articles, and database entries (e.g., GENBANK® database entries and all annotations available therein) are incorporated herein by reference in their entireties to the extent that they supplement, explain, provide a background for, or teach methodology, techniques, and/or compositions employed herein. 1. Louis, D.N., Wesseling, P., Aldape, K., Brat, D.J., Capper, D., Cree, I.A., Eberhart, C., Figarella-Branger, D., Fouladi, M., Fuller, G.N., et al. (2020). cIMPACT-NOW update 6: new entity and diagnostic principle recommendations of the cIMPACT-Utrecht meeting on future CNS tumor classification and grading. Brain Pathology 30, 844–856. 10.1111/bpa.12832. 2. Haslam, A., Kim, M.S., and Prasad, V. (2021). Updated estimates of eligibility for and response to genome-targeted oncology drugs among US cancer patients, 2006-2020. Annals of Oncology 32, 926–932.10.1016/j.annonc.2021.04.003. 3. Blakely, C.M., Watkins, T.B.K., Wu, W., Gini, B., Chabon, J.J., McCoach, C.E., McGranahan, N., Wilson, G.A., Birkbak, N.J., Olivas, V.R., et al. (2017). Evolution and clinical impact of co-occurring genetic alterations in advanced-stage EGFR-mutant lung cancers. Nat Genet 49, 1693–1704.10.1038/ng.3990. 4. de Bruin, E.C., McGranahan, N., Mitter, R., Salm, M., Wedge, D.C., Yates, L., Jamal- Hanjani, M., Shafi, S., Murugaesu, N., Rowan, A.J., et al. (2014). Spatial and temporal diversity in genomic instability processes defines lung cancer evolution. Science (1979) 346, 251–256.10.1126/science.1253462. 5. Sulkowski, P.L., Corso, C.D., Robinson, N.D., Scanlon, S.E., Purshouse, K.R., Bai, H., Liu, Y., Sundaram, R.K., Hegan, D.C., Fons, N.R., et al. (2017).2-Hydroxyglutarate produced by neomorphic IDH mutations suppresses homologous recombination and induces PARP inhibitor sensitivity. Sci Transl Med 9.10.1126/scitranslmed.aal2463. 6. Chandarlapaty, S., Sawai, A., Scaltriti, M., Rodrik-Outmezguine, V., Grbovic-Huezo, O., Serra, V., Majumder, P.K., Baselga, J., and Rosen, N. (2011). AKT inhibition relieves feedback suppression of receptor tyrosine kinase expression and activity. Cancer Cell 19, 58– 71.10.1016/j.ccr.2010.10.031. 7. Manchado, E., Weissmueller, S., Morris, J.P., Chen, C.-C., Wullenkord, R., Lujambio, A., de Stanchina, E., Poirier, J.T., Gainor, J.F., Corcoran, R.B., et al. (2016). A combinatorial strategy for treating KRAS-mutant lung cancer. Nature 534, 647–651. 10.1038/nature18600. 8. Poulikakos, P.I., Zhang, C., Bollag, G., Shokat, K.M., and Rosen, N. (2010). RAF inhibitors transactivate RAF dimers and ERK signalling in cells with wild-type BRAF. Nature 464, 427–430.10.1038/nature08902.
Attorney Docket No.4210.0527WO 9. Barretina, J., Caponigro, G., Stransky, N., Venkatesan, K., Margolin, A.A., Kim, S., Wilson, C.J., Lehár, J., Kryukov, G. v, Sonkin, D., et al. (2012). The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483, 603– 607.10.1038/nature11003. 10. Gao, H., Korn, J.M., Ferretti, S., Monahan, J.E., Wang, Y., Singh, M., Zhang, C., Schnell, C., Yang, G., Zhang, Y., et al. (2015). High-throughput screening using patient- derived tumor xenografts to predict clinical trial drug response. Nat Med 21, 1318–1325. 10.1038/nm.3954. 11. Lu, M., Zessin, A.S., Glover, W., and Hsu, D.S. (2017). Activation of the mTOR Pathway by Oxaliplatin in the Treatment of Colorectal Cancer Liver Metastasis. PLoS One 12, e0169439.10.1371/journal.pone.0169439. 12. Majumder, B., Baraneedharan, U., Thiyagarajan, S., Radhakrishnan, P., Narasimhan, H., Dhandapani, M., Brijwani, N., Pinto, D.D., Prasath, A., Shanthappa, B.U., et al. (2015). Predicting clinical response to anticancer drugs using an ex vivo platform that captures tumour heterogeneity. Nat Commun 6, 6169.10.1038/ncomms7169. 13. Vlachogiannis, G., Hedayat, S., Vatsiou, A., Jamin, Y., Fernández-Mateos, J., Khan, K., Lampis, A., Eason, K., Huntingford, I., Burke, R., et al. (2018). Patient-derived organoids model treatment response of metastatic gastrointestinal cancers. Science (1979) 359, 920– 926.10.1126/science.aao2774. 14. Drost, J., and Clevers, H. (2018). Organoids in cancer research. Nat Rev Cancer 18, 407–418.10.1038/s41568-018-0007-6. 15. Namekawa, T., Ikeda, K., Horie-Inoue, K., and Inoue, S. (2019). Application of Prostate Cancer Models for Preclinical Study: Advantages and Limitations of Cell Lines, Patient-Derived Xenografts, and Three-Dimensional Culture of Patient-Derived Cells. Cells 8.10.3390/cells8010074. 16. Jenkins, R.W., Aref, A.R., Lizotte, P.H., Ivanova, E., Stinson, S., Zhou, C.W., Bowden, M., Deng, J., Liu, H., Miao, D., et al. (2018). Ex Vivo Profiling of PD-1 Blockade Using Organotypic Tumor Spheroids. Cancer Discov 8, 196–215.10.1158/2159-8290.CD- 17-0833. 17. Neal, J.T., Li, X., Zhu, J., Giangarra, V., Grzeskowiak, C.L., Ju, J., Liu, I.H., Chiou, S.-H., Salahudeen, A.A., Smith, A.R., et al. (2018). Organoid modeling of the tumor immune microenvironment. Cell 175, 1972-1988.e16.10.1016/j.cell.2018.11.021. 18. Yuki, K., Cheng, N., Nakano, M., and Kuo, C.J. (2020). Organoid models of tumor immunology. Trends Immunol 41, 652–664.10.1016/j.it.2020.06.010. 19. Nickl, V., Schulz, E., Salvador, E., Trautmann, L., Diener, L., Kessler, A.F., Monoranu, C.M., Dehghani, F., Ernestus, R.-I., Löhr, M., et al. (2022). Glioblastoma-Derived Three-Dimensional Ex Vivo Models to Evaluate Effects and Efficacy of Tumor Treating Fields (TTFields). Cancers (Basel) 14, 5177.10.3390/cancers14215177. 20. Jacob, F., Salinas, R.D., Zhang, D.Y., Nguyen, P.T.T., Schnoll, J.G., Wong, S.Z.H., Thokala, R., Sheikh, S., Saxena, D., Prokop, S., et al. (2020). A Patient-Derived
Attorney Docket No.4210.0527WO Glioblastoma Organoid Model and Biobank Recapitulates Inter- and Intra-tumoral Heterogeneity. Cell 180, 188-204.e22.10.1016/j.cell.2019.11.036. 21. Satterlee, A.B., Dunn, D.E., Lo, D.C., Khagi, S., and Hingtgen, S. (2019). Tumoricidal stem cell therapy enables killing in novel hybrid models of heterogeneous glioblastoma. Neuro Oncol 21, 1552–1564.10.1093/neuonc/noz138. 22. Satterlee, A.B., Dunn, D.E., Valdivia, A., Malawsky, D., Buckley, A., Gershon, T., Floyd, S., and Hingtgen, S. (2022). Spatiotemporal analysis of induced neural stem cell therapy to overcome advanced glioblastoma recurrence. Mol Ther Oncolytics 26, 49–62. 10.1016/j.omto.2022.06.004. 23. Lossi, L., Alasia, S., Salio, C., and Merighi, A. (2009). Cell death and proliferation in acute slices and organotypic cultures of mammalian CNS. Prog Neurobiol 88, 221–245. 10.1016/j.pneurobio.2009.01.002. 24. Hezel, M., Ebrahimi, F., Koch, M., and Dehghani, F. (2012). Propidium iodide staining: A new application in fluorescence microscopy for analysis of cytoarchitecture in adult and developing rodent brain. Micron 43, 1031–1038.10.1016/j.micron.2012.04.006. 25. Zimmer, J., Kristensen, B.W., Jakobsen, B., and Noraberg, J. (2000). Excitatory amino acid neurotoxicity and modulation of glutamate receptor expression in organotypic brain slice cultures. Amino Acids 19, 7–21.10.1007/s007260070029. 26. Romero-Leguizamón, C.R., Elnagar, M.R., Kristiansen, U., and Kohlmeier, K.A. (2019). Increasing cellular lifespan with a flow system in organotypic culture of the Laterodorsal Tegmentum (LDT). Sci Rep 9, 1486.10.1038/s41598-018-37606-3. 27. Buskila, Y., Breen, P.P., Tapson, J., van Schaik, A., Barton, M., and Morley, J.W. (2014). Extending the viability of acute brain slices. Sci Rep 4, 5309.10.1038/srep05309. 28. Happ, D.F., and Tasker, R.A. (2016). A method for objectively quantifying propidium iodide exclusion in organotypic hippocampal slice cultures. J Neurosci Methods 269, 1–5. 10.1016/j.jneumeth.2016.05.006. 29. Noraberg, J. (2004). Organotypic Brain Slice Cultures: An Efficient and Reliable Method for Neurotoxicological Screening and Mechanistic Studies. Alternatives to Laboratory Animals 32, 329–337.10.1177/026119290403200403. 30. Suh, E.C., Jung, Y.J., Kim, Y.A., Park, E.M., and Lee, K.E. (2008). Aβ25–35 induces presynaptic changes in organotypic hippocampal slice cultures. Neurotoxicology 29, 691– 699.10.1016/j.neuro.2008.04.001. 31. Escartin, C., Galea, E., Lakatos, A., O’Callaghan, J.P., Petzold, G.C., Serrano-Pozo, A., Steinhäuser, C., Volterra, A., Carmignoto, G., Agarwal, A., et al. (2021). Reactive astrocyte nomenclature, definitions, and future directions. Nat Neurosci 24, 312–325. 10.1038/s41593-020-00783-4. 32. Fu, R., Shen, Q., Xu, P., Luo, J.J., and Tang, Y. (2014). Phagocytosis of microglia in the central nervous system diseases. Mol Neurobiol 49, 1422–1434.10.1007/s12035-013- 8620-6.
Attorney Docket No.4210.0527WO 33. Katayama, T., Kobayashi, H., Okamura, T., Yamasaki-Katayama, Y., Kibayashi, T., Kimura, H., Ohsawa, K., Kohsaka, S., and Minami, M. (2012). Accumulating microglia phagocytose injured neurons in hippocampal slice cultures: involvement of p38 MAP kinase. PLoS One 7, e40813.10.1371/journal.pone.0040813. 34. Wu, S.-Y. (2017). The roles of microglia macrophages in tumor progression of brain cancer and metastatic disease. Frontiers in Bioscience 22, 4573.10.2741/4573. 35. Bonner, E.R., Waszak, S.M., Grotzer, M.A., Mueller, S., and Nazarian, J. (2021). Mechanisms of imipridones in targeting mitochondrial metabolism in cancer cells. Neuro Oncol 23, 542–556.10.1093/neuonc/noaa283. 36. Fennell, E.M.J., Aponte-Collazo, L.J., Wynn, J.D., Drizyte-Miller, K., Leung, E., Greer, Y.E., Graves, P.R., Iwanowicz, A.A., Ashamalla, H., Holmuhamedov, E., et al. (2022). Characterization of TR-107, a novel chemical activator of the human mitochondrial protease ClpP. Pharmacol Res Perspect 10, e00993.10.1002/prp2.993. 37. Huang, J., Huen, M.S.Y., Kim, H., Leung, C.C.Y., Glover, J.N.M., Yu, X., and Chen, J. (2009). RAD18 transmits DNA damage signalling to elicit homologous recombination repair. Nat Cell Biol 11, 592–603.10.1038/ncb1865. 38. Prabhu, V.V., Morrow, S., Rahman Kawakibi, A., Zhou, L., Ralff, M., Ray, J., Jhaveri, A., Ferrarini, I., Lee, Y., Parker, C., et al. (2020). ONC201 and imipridones: Anti- cancer compounds with clinical efficacy. Neoplasia 22, 725–744.10.1016/j.neo.2020.09.005. 39. Yadav, B., Wennerberg, K., Aittokallio, T., and Tang, J. (2015). Searching for Drug Synergy in Complex Dose-Response Landscapes Using an Interaction Potency Model. Comput Struct Biotechnol J 13, 504–513.10.1016/j.csbj.2015.09.001. 40. Zheng, S., Wang, W., Aldahdooh, J., Malyutina, A., Shadbahr, T., Tanoli, Z., Pessia, A., and Tang, J. (2022). SynergyFinder Plus: Toward Better Interpretation and Annotation of Drug Combination Screening Datasets. Genomics Proteomics Bioinformatics 20, 587–596. 10.1016/j.gpb.2022.01.004. 41. Chokshi, C.R., Savage, N., Venugopal, C., and Singh, S.K. (2020). A Patient-Derived Xenograft Model of Glioblastoma. STAR Protoc 1, 100179.10.1016/j.xpro.2020.100179. 42. Kerstetter-Fogle, A.E., Harris, P.L.R., Brady-Kalnay, S.M., and Sloan, A.E. (2020). Generation of Glioblastoma Patient-Derived Intracranial Xenografts for Preclinical Studies. Int J Mol Sci 21.10.3390/ijms21145113. 43. William, D., Mullins, C.S., Schneider, B., Orthmann, A., Lamp, N., Krohn, M., Hoffmann, A., Classen, C.-F., and Linnebacher, M. (2017). Optimized creation of glioblastoma patient derived xenografts for use in preclinical studies. J Transl Med 15, 27. 10.1186/s12967-017-1128-5. 44. Alnahhas, I., Alsawas, M., Rayi, A., Palmer, J.D., Raval, R., Ong, S., Giglio, P., Murad, M.H., and Puduvalli, V. (2020). Characterizing benefit from temozolomide in MGMT promoter unmethylated and methylated glioblastoma: a systematic review and meta- analysis. Neurooncol Adv 2, vdaa082.10.1093/noajnl/vdaa082.
Attorney Docket No.4210.0527WO 45. DiNardo, C.D., Stein, E.M., de Botton, S., Roboz, G.J., Altman, J.K., Mims, A.S., Swords, R., Collins, R.H., Mannis, G.N., Pollyea, D.A., et al. (2018). Durable Remissions with Ivosidenib in IDH1-Mutated Relapsed or Refractory AML. N Engl J Med 378, 2386– 2398.10.1056/NEJMoa1716984. 46. Tien, A.-C., Li, J., Bao, X., Derogatis, A., Kim, S., Mehta, S., and Sanai, N. (2019). A Phase 0 Trial of Ribociclib in Recurrent Glioblastoma Patients Incorporating a Tumor Pharmacodynamic- and Pharmacokinetic-Guided Expansion Cohort. Clinical Cancer Research 25, 5777–5786.10.1158/1078-0432.CCR-19-0133. 47. Janku, F., Wheler, J.J., Naing, A., Falchook, G.S., Hong, D.S., Stepanek, V.M., Fu, S., Piha-Paul, S.A., Lee, J.J., Luthra, R., et al. (2013). PIK3CA mutation H1047R is associated with response to PI3K/AKT/mTOR signaling pathway inhibitors in early-phase clinical trials. Cancer Res 73, 276–284.10.1158/0008-5472.CAN-12-1726. 48. Fusco, M.J., Piña, Y., Macaulay, R.J., Sahebjam, S., Forsyth, P.A., Peguero, E., and Walko, C.M. (2021). Durable Progression-Free Survival With the Use of BRAF and MEK Inhibitors in Four Cases With BRAF V600E-Mutated Gliomas. Cancer Control 28, 10732748211040012.10.1177/10732748211040013. 49. Feldheim, J., Kessler, A.F., Feldheim, J.J., Schulz, E., Wend, D., Lazaridis, L., Kleinschnitz, C., Glas, M., Ernestus, R.-I., Brandner, S., et al. (2022). Effects of Long-Term Temozolomide Treatment on Glioblastoma and Astrocytoma WHO Grade 4 Stem-like Cells. Int J Mol Sci 23.10.3390/ijms23095238. 50. Hegi, M.E., Diserens, A.-C., Gorlia, T., Hamou, M.-F., de Tribolet, N., Weller, M., Kros, J.M., Hainfellner, J.A., Mason, W., Mariani, L., et al. (2005). MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352, 997–1003. 10.1056/NEJMoa043331. 51. Nobre, L., Zapotocky, M., Ramaswamy, V., Ryall, S., Bennett, J., Alderete, D., Balaguer Guill, J., Baroni, L., Bartels, U., Bavle, A., et al. (2020). Outcomes of BRAF V600E pediatric gliomas treated with targeted BRAF inhibition. JCO Precis Oncol 4. 10.1200/PO.19.00298. 52. Hollestelle, A., Nagel, J.H.A., Smid, M., Lam, S., Elstrodt, F., Wasielewski, M., Ng, S.S., French, P.J., Peeters, J.K., Rozendaal, M.J., et al. (2010). Distinct gene mutation profiles among luminal-type and basal-type breast cancer cell lines. Breast Cancer Res Treat 121, 53– 64.10.1007/s10549-009-0460-8. 53. Piccirillo, S.G.M., Colman, S., Potter, N.E., van Delft, F.W., Lillis, S., Carnicer, M.- J., Kearney, L., Watts, C., and Greaves, M. (2015). Genetic and functional diversity of propagating cells in glioblastoma. Stem Cell Reports 4, 7–15.10.1016/j.stemcr.2014.11.003. 54. Pavel, A.B., and Korolev, K.S. (2017). Genetic load makes cancer cells more sensitive to common drugs: evidence from Cancer Cell Line Encyclopedia. Sci Rep 7, 1938. 10.1038/s41598-017-02178-1. 55. Hanisch, D., Krumm, A., Diehl, T., Stork, C.M., Dejung, M., Butter, F., Kim, E., Brenner, W., Fritz, G., Hofmann, T.G., et al. (2022). Class I HDAC overexpression promotes
Attorney Docket No.4210.0527WO temozolomide resistance in glioma cells by regulating RAD18 expression. Cell Death Dis 13, 293.10.1038/s41419-022-04751-7. 56. Wakimoto, H., Kesari, S., Farrell, C.J., Curry, W.T., Zaupa, C., Aghi, M., Kuroda, T., Stemmer-Rachamimov, A., Shah, K., Liu, T.-C., et al. (2009). Human glioblastoma-derived cancer stem cells: establishment of invasive glioma models and treatment with oncolytic herpes simplex virus vectors. Cancer Res 69, 3472–3481.10.1158/0008-5472.CAN-08-3886. It will be understood that various details of the presently disclosed subject matter may be changed without departing from the scope of the presently disclosed subject matter. Furthermore, the foregoing description is for the purpose of illustration only, and not for the purpose of limitation.
Corresponding Slice Health at ECXX of Tumor Determine the amount of slice that is alive at the EC XX of the tumor Low Dose = Lowest dose above tumor ECXX High Dose = Highest dose below tumor ECXX ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^
ECXX Window Ratio between where XX% of the tumor is killed and the health of the slice at that dose If the health of the slice if more than 1-XX%:
Attorney Docket No.4210.0527WO Yes: ^^ ^^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ െ ^1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ^^ ^^ ^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ No: ^1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ െ ^^ ^^ ^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^^ ^1 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^ ECXX Score If tumor growth is seen that results in XX% more than the max kill of the slice: Yes -1*% Weight in DSS No If tumor killing is seen but not more than XX% Yes 0 No If the amount of slice that is killed at the tumor ECXX is more than XX%: Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ No: െ1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ Slope Tumor Slope Slope through the EC50 or the highest dose if EC50 isn’t reached ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ℎ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50 െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^50
^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ 2 ^ ∗ ^ ^^ ^^ ^^ ^^ 2 െ ^^ ^^ ^^ ^^ 1^
AUC Window Ratio between slice AUC and tumor AUC If slice AUC is less than tumor AUC
Attorney Docket No.4210.0527WO Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ No: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ െ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ AUC Score If the slice AUC is less than the tumor AUC Yes: ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ No: െ1 ∗ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ^^ ∗ % ^^ ^^ ^^ ^^ℎ ^^ ^^ ^^ ^^ ^^ ^^ [0092] In some embodiments, each of the above equations making components of the DSS can be adjusted, modified and/or eliminated based on the LTS used/tested, and/or pertinent characteristics of the same. [0093] In some embodiments, disclosed are diagnostic and/or screening methods, assays and systems based on the LTS platform for diagnosing a tumor in a subject and/or screening potential drug and treatment options, including combinatorial treatments. In some embodiments, the present disclosure provides adaptive treatment methods where cryopreserved patient samples, e.g. brain tumor samples, can be serially thawed and analyzed using the disclosed methods and LTS platform, coupled with DSS analysis, to identify optimal patient specific treatment programs after testing and/or screening a plurality of options. Drug sensitivity information gathered from testing the first thawed vial of patient samples would inform the drugs tested on patient samples from the second thawed vial, etc. This adaptive, iterative, testing would maximize the information that can be gathered and more effectively pinpoint the most appropriate therapeutic or combination for a single patient before the patient begins treatment. This is possible in the disclosed system because (1) LTS supports engraftment of cryopreserved/thawed patient samples, and (2) LTS assays are rapid (4 days from engraftment to readout), allowing multiple
Attorney Docket No.4210.0527WO iterations of testing within a clinically relevant time frame. In some embodiments, this clinically relevant timeframe can vary depending on the type of tissue used for the LTS. [0094] Finally, while data provided herein directly supports the use of brain tissue (OBSCs) and mesentery based LTSs, other tissues suitable for LTSs, and that have been tested, include but are not limited to kidney, liver, lung, bone and spinal cord (see, e.g. Figures 17A-17E). Similar to mesentery and OBSC LTSs, these other tissues can serve as an LTS for growing patient samples and tumor cells for purposes of diagnosing and/or screening drug candidates. [0095] Taken together, the present disclosure offers several advantages not previously contemplated, and many of which were unexpected. For example, the LTS provides: o A tissue layer which supports tumor viability more effectively and reproducibly than culture methods without the substrate. o A natural microenvironment which communicates with tumor tissue. o A healthy tissue control with which to measure off-target drug toxicity. [0096] Moreover, compared to directly culturing slices of the patient tumor tissue itself for functional testing, using a separate living tissue substrate to engraft patient tumor: o Decreases the amount of tumor tissue required per replicate, allowing broader/deeper testing per unit of tumor pass received from the clinic. o Increases intra-sample reproducibility: the homogeneous cell suspension we make from the tumor tissue we receive allows each tumor grown on LTS to be very similar; in contrast, slices of patient tumor are each from a different tumor region and therefore unique, especially in more heterogeneous tumor tissue samples. [0097] Additionally, the disclosed methods, systems and protocols to cryopreserve and thaw patient tumor tissue just after resection allows for: o Ship tumor tissue from other sites without risking viability loss from sustained live tissue transport. o Strategically thaw tumor tissue samples at the right time(s) for optimal study impact. [0098] The disclosed methods also allow for quickly and gently dissociate, label, and seed patient tumor tissue onto LTS allows even low-grade tumors to more quickly and successfully engraft compared to other culture methods. [0099] Additionally, seeding patient tumor tissue onto living tissue substrates without any time in culture:
Attorney Docket No.4210.0527WO o More effectively maintains tumor heterogeneity. o Minimizes cell selection and genetic drift before assay initiation. [0100] Moreover, the disclosed methods, systems and assays require just four days from the time we receive tumor tissue to provide the final output. [0101] Surprisingly, 100% of the patient tumor tissue seeded onto slices have retained evaluable levels of engraftment, regardless of tumor grade or type. [0102] Importantly, the present disclosure confirms the successful testing of the following treatment types against tumors grown on living tissue substrates: o Small Molecule Drugs o Radiation Therapy o Protein Therapies o CAR-T and iNSC Cellular Therapies o Combination Therapies Once a tumor or cancer is diagnosed in a subject, an appropriate treatment, including any of these listed herein, or combinations thereof, can be administered to the subject. [0103] Each Drug Sensitivity Score (DSS) is generated from over 100 unique data points. Each drug is tested against an individual tumor at six dose levels at n = 4 tumors per dose, providing a dose-response profile with which to gauge efficacy. o Tumor response is compared to the toxicity of each drug to the LTS (six dose levels at n = 12 LTS per dose), allowing for the to quantification of a “therapeutic window” to ask: ^ Which tumors can be killed at doses which present minimal toxicity to healthy tissue? ^ Which drugs present the lowest toxicity to healthy tissue while still killing the tumor? ^ Can combination therapy maintain effective tumor kill while decreasing healthy tissue toxicity compared to single-agent therapy? [0104] The disclosed novel, multi-point algorithm combines several pieces of data from dose-response toxicity curves of tumor and LTS to output one normalized number (-100 to 100) that can be compared in the following ways: o Measure and rank efficacies of different drugs against a single tumor. o Measure and rank which tumors respond most and least to a single drug.
Attorney Docket No.4210.0527WO [0105] Thus, provided in some embodiments are methods of diagnosing a tumor and/or screening for a therapeutic for a tumor. Such methods can include providing a living tissue substrate (LTS), engrafting one or more tumor tissue cells (e.g. tumor tissue and/or tumor cells obtained from a subject) to the LTS, and analyzing a dose-response of a candidate therapeutic for both tumor tissue cell kill and LTS toxicity. The result of such methods can provide for the diagnosis of the tumor and/or identification of a candidate therapeutic to treat the tumor. The tumor tissue cell kill, or tumor toxicity, can be defined or quantified on a continuous scale and can be measured anywhere between about 1% and 99%, i.e. "dead". In some embodiments, the tumor tissue cell kill, or tumor toxicity, can be defined as the degree to which the tumor tissue is killed, degraded, or rendered unviable, and can range from about 10% to about 99% kill, from about 20% to about 95% kill, from about 30% to about 90% kill, from about 40% to about 90% kill, or about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 99% or more kill (in some aspects as compared to an untreated tumor tissue cell. The LTS toxicity can be defined or quantified on a continuous scale and can be measured anywhere between about 1% and 99%, i.e. "dead". The LTS toxicity can be defined as the degree to which the living tissue substrate is killed, degraded, or rendered unviable (i.e. an undesirable side effect of the treatment), and can range from about 1% to about 99% toxicity, from about 1% to about 90% toxicity, from about 5% to about 75% toxicity, from about 5% to about 50% toxicity, or about 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or less toxicity (in some aspects as compared to an untreated LTS). [0106] As described herein, the LTS of these methods and systems can comprise any tissue that forms solid tumors, and in some embodiments can include a tissue selected from brain, mesentery, kidney, liver, lung, bone and spinal cord. Brain tissue can in some embodiments be an organotypic brain slice culture (OBSC). Mesentery tissue can in some embodiments be an organotypic mesentery membrane culture (OMMC). [0107] In some embodiments, the one or more tumor tissue cells are derived from a primary or metastatic tumor of a subject. As described further herein, the one or more tumor tissue cells are dissociated into small pieces, and transfected with a reporter, prior to seeding onto the LTS, optionally wherein the tumor is a brain tumor or ovarian tumor. In some aspects, the one or more tumor tissue cells are finely minced with no enzyme, strained through a 100µm filter, infected with lentiviral luciferase and labeled with a
Attorney Docket No.4210.0527WO fluorescent reporter (e.g. stained with far-red lipid-soluble dye), prior to seeding onto the LTS. [0108] Advantageously, in some aspects, the one or more tumor tissue cells are engrafted to the LTS and tested for drug sensitivities, with assay completion in less than 10 days, optionally less than 5 days, optionally less than 4 days, optionally less than 3 days, optionally less than 2 days, optionally less than about one day. [0109] Unexpectedly, and as demonstrated in the working Examples, the genetic drift of the one or more tumor tissue cells in these methods and systems is minimized due to the rapid engraftment. By way of example and not limitation, the genetic drift is less than about 50%, optionally less than about 25%, optionally less than about 10%, at the time of diagnosing and/or screening. In some cases, the mutational profile of the one or more tumor tissue cells is substantially similar to the mutational profile in vivo. [0110] In some embodiments, the source of the one or more tumor tissue cells used in these systems and methods is a patient in need of treatment and/or having a tumor, including for example a human subject, and collecting a biopsy from the patient as the source of the one or more tumor tissue cells. The one or more tumor tissue cells can be cryopreserved after biopsy and thawed prior to engraftment on the LTS, optionally wherein the cryopreserved tumor tissue cells are preserved for a plurality of sequential and/or simultaneous applications of the methods. Unlike prior methods, the cryopreserved tumor tissue cells are not exposed to an enzyme during dissociation. [0111] As discussed further herein, the analysis of the dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity can be assessed or characterized by calculating a drug sensitivity score (DSS), wherein the DSS is calculated by comparing tumor cell survival to health of the LTS. By way of example and not limitation, the DSS can be calculated based on a plurality of weighted parameters, optionally a combination of all weighted parameters, selected from the group consisting of: (1) killing at maximum dose (Max Kill), (2) dose required to kill 10% of the tumor (EC10), (3) dose required to kill 25% of the tumor (EC25), (4) dose required to kill 50% of the tumor (EC50), (5) dose required to kill 75% of the tumor (EC75), (6) dose required to kill 90% of the tumor (EC90), (7) slope through the EC50, (8) the area under the curve (AUC), (9) tumor growth acceleration, (10) biphasic killing (rapid killing at low doses and limited additional killing at higher doses), and (11) incomplete kill at the highest dose. As disclosed herein, the DSS from 0 to 100 signifies increasing efficacy in tumor kill relative to LTS toxicity, and
Attorney Docket No.4210.0527WO wherein a DSS from 0 to -100 signifies increasing LTS toxicity relative to tumor kill. For exemplary purposes only, each of the parameters can in some embodiments be weighted at about 1% to about 45% in the DSS calculation, optionally wherein each is weighted as follows: Maximum Kill (10%), EC10 (5%), EC25 (5%), EC50 (10%), EC75 (5%), EC90 (5%), Slope through IC50 (10%), Tumor Growth Acceleration (5%; not compared to LTS toxicity), Biphasic Killing Curve (5%; not compared to LTS toxicity), Incomplete Kill (5%; not compared to LTS toxicity), and Area Under the Curve (35%). [0112] These methods can advantageously identify a suitable therapeutic or drug compound for treating a particular tumor or cancer. Such candidate therapeutics or drugs to treat the tumor can in some embodiments comprise a pharmaceutically active agent, a chemotherapeutic composition, a small molecule, an immunotherapeutic agent, an inhibitor, a radiation therapy, and combinations thereof. [0113] In some embodiments, provided are functional precision diagnostic methods that include performing any of the above methods for diagnosing a tumor or screening for a therapeutic for a tumor, and further comprising iteratively testing additional therapeutics on cryopreserved patient tumor cells before administration to a subject, whereby a treatment can be adapted based on a DSS output. For example, such functional precision diagnostics can include testing combinatorial therapies using LTS and DSS. [0114] As disclosed further herein, in some embodiments provided are diagnostic and/or therapeutic screening systems. Such screening systems can include a LTS, one or more tumor tissue cells engrafted to the LTS, and a multi-parametric algorithm which simultaneously analyzes a dose-response of a candidate therapeutic for both tumor cell kill and LTS toxicity, i.e., the disclosed DSS. The LTSs are configured the same as in the above described methods. [0115] Finally, in some embodiments, methods of treating a subject are provided. Such treatment methods can begin with the diagnostic and/or screening methods and systems described herein, followed by administering to the subject a treatment or therapeutic based on the diagnosis. In some embodiments, the treatment can comprise a combinatorial treatment, i.e., one or more drugs or therapeutic compositions/modalities based on the DSS analysis. [0116] The subject matter disclosed herein can be implemented in software in combination with hardware and/or firmware. For example, the subject matter described herein can be implemented in software executed by a processor. In one exemplary
Attorney Docket No.4210.0527WO implementation, the subject matter described herein can be implemented using a computer readable medium having stored thereon computer executable instructions that when executed by a processor of a computer control the computer to perform steps. Exemplary computer readable mediums suitable for implementing the subject matter described herein include non-transitory devices, such as disk memory devices, chip memory devices, programmable logic devices, and application specific integrated circuits. In addition, a computer readable medium that implements the subject matter described herein can be located on a single device or computing platform or can be distributed across multiple devices or computing platforms. EXAMPLES [0117] The following examples are included to further illustrate various embodiments of the presently disclosed subject matter. However, those of ordinary skill in the art should, in light of the present disclosure, appreciate that many changes can be made in the specific embodiments which are disclosed and still obtain a like or similar result without departing from the spirit and scope of the presently disclosed subject matter. Materials and Methods for Examples 1-10 Cell Lines: [0118] Cell lines used: LN229 (established GBM line), GBM8 (high-passage patient- derived GBM line56), MS21 (low-passage patient-derived GBM line), U373WT (wild-type established GBM line), U373KO (RAD18 knockout of U373WT via CRISPR), MB231Br (breast cancer metastasis to brain), and PDIPG (low-passage pediatric patient-derived line). [0119] GBM8 cells were cultured in Neurobasal-A medium (Gibco) with 7.5 ml L- glutamine, 10ml B27 supplement, 2.5 ml N2 supplement, 1 mg heparin, 10 μg EGF, 10 μg FGF, and 2.5 ml anti-anti. LN229 cells were from American Type Culture Collection. MDA-MB231-Br cells were obtained through a material transfer agreement (MTA) (T. Yoneda). MS21 cells were derived in the Hingtgen Laboratory from a GBM patient biopsy. U373WT and U373KO cells were provided by C. Vaziri (University of North Carolina). LN229, MDA-MB231-Br, MS21, U373WT and U373KO cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco) supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin (Gibco). IFF-BT105 PDIPG cells were obtained from
Attorney Docket No.4210.0527WO Ian’s Friends Foundation and cultured in Neurobasal Medium(-A):DMEM/F-12, GlutaMAX Medium (1:1) supplemented with 1x Antibiotic-Antimycotic, 1x Sodium Pyruvate, 1x MEM Non-Essential Amino Acids, 10mM HEPES buffer, 1x GlutaMAX-I, 1X B27 minus vitamin A, 20ng/ml EGF, 20ng/ml bFGF, 10ng/ml PDGF-AA, 10ng/ml PDGF-BB, 2ug/ml heparin. All cell lines were cultured at 37oC, 5% CO2 and 95% humidity. Generating OBSCs: [0120] All OBSCs were generated from P8 Sprague-Dawley rat pups. Dissected brains were mounted on a vibratome (Leica VT1000S) platform submerged in ice-cold brain slice media (BSM). Coronal OBSCs were sliced at a thickness of 300 µm at ~15 OBSCs/animal. Visibly damaged brains or OBSCs were discarded. Acceptable OBSCs were transferred onto transwell inserts in a 6-well culture plate. 1ml of OBSC media (BSM22) was added under each transwell. BSM comprised of Neurobasal-A medium supplemented with 10% heat-inactivated pig serum, 5% heat-inactivated rat serum, 1 mM L-glutamine, 10 mM KCl, 10 mM HEPES, 1 mM sodium pyruvate and 100 U/mL penicillin-streptomycin. The plates were then transferred to a 37°C incubator with 5% CO2 and 95% humidity. For QC purposes, six random OBSCs from every batch were selected to undergo the PI assay test for cell death. On D4, the PI signals from the QC group were compared to those from previous batches. In Vivo Tumor Studies: [0121] 6-8 week-old female athymic nude mice were used; n = 5 mice per cell line with no exclusion criteria, randomization, or blinding as all mice were used in one group. 250,000 tumor cells were stereotactically implanted into the brain parenchyma (1 mm, 1 mm, 2.5 mm) of mice anesthetized with isoflurane. All mice underwent serial bioluminescence imaging to measure tumor growth over time and were monitored for changes in weight or behavior to indicate the endpoint had been reached. Luciferin was injected i.p. into mice at 3mg/mouse in 250 mL PBS. Brains from each group, dissected at time of death, were coronally sectioned along the tumor implant site and imaged for tumor fluorescence.
Attorney Docket No.4210.0527WO Chemicals: [0122] The following chemicals were purchased from Sigma-Aldrich: carboplatin, temozolomide, etoposide, azacitidine, vincristine. The following chemicals were purchased from EMD Millipore: cisplatin. The following chemicals were purchased from Selleck Chemicals: lomustine, gemcitabine, and trametinib. TR-107 was provided by Madera Therapeutics, LLC. Lentiviral Vectors: [0123] The following LVs were used in this study: eGFP fused to firefly luciferase (LV– eGFP-FL) and mCherry protein fused to firefly luciferase (LV–mCh-FL). Propidium Iodide Assay: [0124] PI fluorescence was used to quantify the health/quality of OBSCs as well as the toxicity of each treatment vs OBSCs at the end of each assay: t = 4 days after OBSC generation and t = 3 days after treatment initiation. 5 µg/mL PI was mixed with BSM and added under transwells 60 minutes before fluorescence measurement on an AMI optical imaging system (Spectral Instruments). Positive control was generated by killing the OBSC via incubation with 70% ethanol and freezing overnight between days 3 and 4 of the assay. For all DSS calculations, n = 12 OBSCs from N = 2 separate experiments. Immunohistochemistry: [0125] OBSCs dedicated for Day 0 IHC were fixed in 4% paraformaldehyde immediately following sectioning and stored at 4°C. After 48 hours of fixation, the sections were transferred to 30% sucrose and stored at 4°C until IHC was conducted. The sections dedicated for Day 4 IHC were transferred to 6 well plates with BSM and stored at 37°C. GBM8-mch-FLuc cells were engrafted atop select OBSCs as described below. On Day 1, the sections were transferred to new 6 well plates with fresh BSM. The sections were fixed in 4% paraformaldehyde on Day 4 and stored at 4°C. After 48 hours of fixation, the sections were transferred to 30% sucrose and stored at 4°C. When day 0 and day 4 sections were ready for IHC, they were first washed for 10 minutes in 0.1% triton X-100 in 1X Dulbecco’s phosphate-buffered saline (PBST) at room temperature. Sections were then blocked in 5% fetal bovine serum in PBST for 1h at room temperature. The sections were incubated in a primary antibody solution consisting of primary antibodies and blocking
Attorney Docket No.4210.0527WO buffer rotating overnight at room temperature. The primary antibodies used were glial fibrillary acidic protein (GFAP [Abcam, ab7260] at 1:1000), neuronal nuclear protein (NeuN [Abcam, ab177487] at 1:1000), and cluster of differentiation molecule 11B (CD11B [Abcam, ab133357], 1:500). The sections were washed 3 times in PBST for 10 minutes after 18-24 hours in primary antibodies. They were then incubated in a secondary antibody consisting of blocking buffer solution and Alexa Fluor 488 goat anti-rabbit IgG (Thermo Fisher Scientific, A-11008, 1:1000) for 1 hour in a darkroom. The sections were washed 3 times in PBST for 10 minutes and mounted on microscopic slides. Liquid mountant (ProLong Gold Antifade Mountant) was applied on the sections, coverslips were added, and slides were allowed to cure overnight. Z-stack images were acquired using a Zeiss 780 confocal microscope at UNC Neuroscience Microscopy Core. Z-stack images are processed by converting to maximum intensity projection (Max IP) images. The brightness of Max IP images was then further adjusted to accurately assess and present the morphological differences of astrocytes. In Vitro Studies: [0126] Cells were seeded at a density of 5000 cells/µL in 96 well plates. 24h after seeding, drugs were added and allowed to incubate for 72 hours. Cell viability was then assessed via bioluminescence, measured using an AMI optical imaging system. Tumor Growth on OBSCs: [0127] Tumor cells (.17 µL, 25,000 cells) were seeded onto OBSCs 2h after slicing, with one tumor foci seeded in the center of each hemisphere for a total of two tumor foci per OBSC. BSM was changed 24h after slicing. Fluorescence images were taken 0 hours, 24 hours, and 96 hours post-seeding for normalization of tumor size. Bioluminescence readouts were also taken at 0 hours, 24 hours, and 96 hours post seeding for assessment of cell viability. Luciferin was added underneath the transwell insert and allowed to incubate for 10 minutes before bioluminescence measurement on an AMI optical imaging system. Drug Screening on OBSCs: [0128] Tumor cells were seeded onto OBSCs 2h after slicing as described above. One day after tumor seeding, six concentrations of each drug were diluted in the media underneath each transwell, or XRad was applied using an X-Rad 320 Precision X-Ray
Attorney Docket No.4210.0527WO machine (n = 4 tumor foci per concentration; n ≥ 24 tumor foci per drug per cell line). Three days after treatment, the bioluminescence of live tumor cells was measured using an AMI optical imager (Spectral Instruments) and dose-response curves were generated. Tumor survival was normalized to day one tumor fluorescence and compared to an untreated control group. Combination Therapy: [0129] Carboplatin + etoposide: On Day 1 after slicing, both small molecules were administered in the BSM underneath the transwells at the desired concentrations. TMZ + radiation: On Day 1 after slicing, radiation was first administered by an X-Rad 320 Precision X-Ray machine. TMZ was then administered in the BSM underneath the transwells at the desired concentrations. Preparing Patient Tissue for Engraftment onto OBSCs: [0130] Fresh brain tumor tissue surgically resected at UNC hospitals was placed in sterile 4°C Neurobasal-A medium and immediately taken to the UNC Tissue Procurement Facility (TPF). The amount of brain tumor tissue received ranged from 0.05g to 2g. Either at TPF or in the Hingtgen Laboratory, the resected tumor tissue was minced into approximately 0.5 mm diameter pieces using a disposable scalpel and washed with PBS. Tumor pieces which were to be assayed at a later time were placed in a cryogenic vial and frozen in tissue freezing medium (CryoStor CS10) in a FreezeCell™ at - 80°C overnight before transfer into liquid nitrogen. [0131] To engraft onto OBSCs, the cryopreserved/thawed or fresh brain tumor tissue was filtered through a 100 µm cell strainer. Each 50 mg of tissue was transfected with 1ml mcherry-FLuc Lentivirus at 1.5e7 vg/ml with 1µl polybrene for 4h at 37°C. After incubation, the brain tumor tissue was washed with PBS three times to remove the residual virus and reconstituted in PBS with a final volume of 150 µl. Then the tissue solution was engrafted at OBSCs at ~1 mg tissue in 3 µl of PBS on each hemisphere of the slice. The OBSC engrafted with patient tumor tissue was incubated at 37°C with 5% CO2 and 95% humidity. The BSM was changed after 24h and subsequently every 3 days. Tumor treatment studies were executed as described above for tumor lines. The viability of the patient tissue was measured by an AMI optical imaging system with the addition of 1.5 mg/mL luciferin underneath the transwell.
Attorney Docket No.4210.0527WO [0132] Patient tumor drug screening results were not shown to clinicians with information that could identify the patients they reference. Preparing Cell Line from Patient Tissue: [0133] For the primary patient cell line generation, the fresh or frozen tissue were mechanically dissected to approximately single cell under sterile conditions. The resulting single-cell suspension was then washed with PBS, resuspended in growth medium chosen to allow growth of greatest number of tumor cells (DMEM with 20% FBS and 1% PS), and dispensed into 6-well plates. All cultures were initiated in a volume of 2 ml per well and incubated at 37°C and 5% CO2. The viable cells and tissue chunks were allowed to settle and attach to the bottom of the plates for 2-3 days. The floating cellular debris was then carefully aspirated, the attached cells were carefully washed with PBS, and 2 ml of fresh medium was added to each well. Culture medium was routinely changed every 3-5 days, and proliferating adherent cells were passaged after detachment with trypsin. DNASeq: [0134] Cryopreserved/thawed human brain tumor tissue was prepared as described above and either added onto OBSCs (BSHT group), cultured in vitro (CL group) or snap frozen (HT group). After 4 days, tumor tissue was carefully excised from OBSCs using forceps and snap frozen. After 30 days, when cell line had established, cells were collected and snap frozen. All replicates of all samples were shipped to Novogene, Inc. for DNA sequencing. Whole Exome Sequencing was run on the DNA samples following library preparation with an Agilent SureSe-lectXT2 All Exon V6 kit. Raw data was downloaded and analyzed at the UNC Bioinformatics core facility. [0135] The BBsplit algorithm from the BBtools suite was run on all samples to eliminate rat DNA contamination in the BSHT samples as well as to account for any biases that may result as a part of that process. Only the reads that were binned to the human reference were used for subsequent analysis. Reads were then mapped to the GRCh38 version of the human genome with BWA v0.7.17 and realigned together with ABRA2 v.23. Quality control was implemented using the GATK/Picard v4.1.7.0 toolkit. Somatic variants were called for each sample using the MuTect2 algorithm v4.1.7.0. Variants were merged into a single cohort variant call file and then converted to MAF via vcf2maf v1.6.21 tool. Variants were annotated using VEP v87. To identify mutations with potentially high
Attorney Docket No.4210.0527WO biological impact, multiple filtering steps were applied to somatic mutation calling. First, we selected only the somatic variants that passed all filters from the MuTect2 FilterMutectCalls algorithm and second, only high/moderate impact (change coding) variants as defined by the VEP annotation were further analyzed. Over 1900 single nucleotide variants (SNVs) were detected across all samples. Figures that summarized the results were generated using maftools. Calculating Drug Sensitivity Scores: [0136] Several weighted parameters were factored into an overall drug sensitivity score. DSS were calculated by comparing the tumor cell survival, measured via BLI, to the health of the OBSC, measured via PI assay (parameter 1-8), or by quantifying the behavior of the tumor dose-response curve (parameter 9-11). (1) Killing at maximum dose (Max Kill Window, 10% of DSS) (2) Dose required to kill 10% of the tumor (EC10 Window, 5% of DSS) (3) Dose required to kill 25% of the tumor (EC25 Window, 5% of DSS) (4) Dose required to kill 50% of the tumor (EC50 Window, 10% of DSS) (5) Dose required to kill 75% of the tumor (EC75 Window, 5% of DSS) (6) Dose required to kill 90% of the tumor (EC90 Window, 5% of DSS) (7) Slope through the EC50 (Slope Window, 10% of DSS) (8) The area under the dose-response curve (AUC Window, 35% of DSS) (9) Tumor growth acceleration (treated tumor grows faster than untreated tumor: for max growth up to 125%, window = +1; 125%-150% window = 0; over 150% window = -1; 5% of DSS) (10) Biphasic killing (rapid killing at low doses and limited additional killing at higher doses: no biphasic curve shape, window = +1; biphasic curve shape, window = -1; 5% of DSS) (11) Incomplete kill at the highest dose (some tumor remaining at highest dose: for <10% remaining, window = +1; 10%-25% remaining, window = 0; >25% remaining, window = -1; 5% of DSS). [0137] For DSS parameters 1-6, therapeutic windows were calculated by comparing OBSC toxicity and tumor response at the doses where tumor kill passed through the DSS parameter. For DSS parameter 7, the therapeutic window was calculated by comparing the
Attorney Docket No.4210.0527WO slopes through the tumor EC50 and the OBSC tox EC50. For DSS parameter 8, the therapeutic window was calculated by comparing the areas under tumor kill and OBSC tox curves. Normalized therapeutic window ratios for DSS parameters 1-8 within each drug-tumor-OBSC interaction were calculated in the following manner: within each window, values ranged from +1 to -1, where values approaching +1 signify increasing tumor kill relative to normal tissue toxicity, and values approaching -1 indicate agents where tumors remained highly viable while toxicity to the normal OBSC tissue was elevated. DSS parameters 9-11 were determined based on the behavior of the tumor in response to the drug, as defined above. [0138] All individually weighed parameters were added together to generate the DSS. Overall DSS from 0 to 100 signify increasing efficacy in tumor kill relative to OBSC toxicity, while scores from 0 to -100 describe scenarios in which tumors thrive more effectively than OBSCs for a given treatment. Dose-response values were calculated via linear interpolation of raw data, not from best-fit curve equations. Calculating ZIP Synergy Scores and Plots: [0139] All synergy scores and plots were calculated using SynergyFinder. Statistical Analysis: [0140] All statistical tests and sample sizes are included in the Figure Legends. All data are shown as mean ± SEM. In all cases, the p values are represented as follows: ∗∗∗p < 0.001, ∗∗ p < 0.01, ∗p < 0.05, and not statistically significant when p > 0.05. In all cases, the stated ‘‘n’’ value is either number of OBSCs, number of tumor spots placed on the OBSCs, or mice with multiple independent images used to obtain data points for each. Mean values between two groups were compared using t-tests with Welch’s correction when variances were deemed significant by F tests. Mean values between three or more groups were compared to the control by using one-way ANOVA followed by Dunnett’s multiple comparisons test. All statistical analyses were performed using GraphPad Prism (Version 9.1.0). All statistical analysis methods and resulting p values are included within the Figure Legends. For all quantifications of immunohistology, the samples being compared were processed in parallel and imaged using the same settings and laser power.
Attorney Docket No.4210.0527WO Example 1 Generation and Characterization of Living OBSCs as LTSs [0141] These studies use OBSCs generated from rat pups (RRID:MGI:5651135) as living tissue substrates to culture and treat tumor cell lines and uncultured patient brain tumor resection tissue (as discussed herein, other LTSs beyond OBSCs were also tested). To generate this rapid, high-fidelity ex vivo testing platform, first assessed was OBSC quality and reproducibility, optimizing the process using quantitative fluorescence-based imaging methods. After piloting several assays, an established nuclear permeability assay using propidium iodide23–30 (PI, Sigma Aldrich, CAT#P4170) was adapted to quantify cell death within OBSCs. This method provided a large dynamic range to detect nuanced differences between healthy and unhealthy OBSCs, with higher fluorescence values indicating more dead cells (Fig. 4A). OBSC quality was impacted by rat pup age at time of generation (Fig. 4B); therefore, OBSCs were generated from eight-day-old pups for every experiment described in this study. It was also found that optimal OBSCs were generated using improved methods for brain dissection and OBSC culture conditions (Fig. 4C-D), establishing the disclosed robust, standardized procedure for OBSC generation. Following this optimization period, over 6,000 living OBSCs were reproducibly generated across nearly 40 separate batches, including randomly sampling and testing 6 OBSCs per batch for QC analysis (Fig. 4E). All OBSCs with physical damage after slicing, such as a nick or a tear, were discarded, but this amounted to fewer than 2% of OBSCs over all experiments. [0142] Just as brain tissue suffers acute injury after brain tumor resection, OBSCs suffer acute injury after slicing. Serial measurements of OBSC health showed an initially high PI signal that resolves over time, reaching baseline levels by Day 7 (Fig 4F). In parallel, IHC analysis showed astrocytes (GFAP, RRID:AB_305808) and macrophages/microglia (CD11b, RRID:AB_2650514) displayed reactive morphologies31 immediately post-slicing (Fig. 4G). This activation attenuates in astrocytes by Day 4 but persists in macrophages/microglia, suggesting that the OBSCs may still contain dead cells and debris that are phagocytosed by the myeloid cells32,33. The morphology of neurons (NeuN, RRID:AB_2650514) remained unchanged between Day 0 and 4. Together, these results describe a high-quality, well controlled, reproducible method for generating and tracking the health of OBSCs. As such, further studies were conducted to test engrafting tumor cells onto these living tissue substrates (see further Examples herein).
Attorney Docket No.4210.0527WO Example 2 Tumor Modeling on OBSCs [0143] Before attempting to move into uncultured patient tissue samples, a diverse panel of human brain cancer lines (adult, pediatric, primary and metastatic; defined in Figure 5 legend) were selected to further identify the myriad capabilities of the disclosed LTS assay, particularly the OBSC assay, and show how LTSs, including OBSCs, provide a rapid, representative, and applicable niche for broad tumor engraftment and analysis. First, multiple tumor characteristics were compared across in vitro, in vivo, and OBSC culture. All tumor lines were transduced with lentiviral vectors encoding mCherry-Firefly Luciferase (LV-mCh-FLuc) optical reporters, then the lines were either seeded in 96-well culture plates, orthotopically xenografted into the brains of Nude mice, or seeded atop the living OBSCs. The bilateral symmetry of each coronally sectioned OBSC was leveraged to seed tumor cells within the anatomically homogenous thalamic region of each hemisphere. Serial imaging and quantitative analysis was performed to track growth, morphology, and rate of invasion. [0144] Four high-passage tumor lines commonly used within the field, LN229, U373, MB231Br, and GBM8, grew rapidly in vitro (Fig 5A), in vivo via orthotopic implantation (Fig 5B), and on OBSCs (Fig 5C-E). These tumor lines all expanded several-fold in just four days in vitro and on OBSCs, and over 100-fold in vivo within 40 days. The low- passage adult GBM line MS21 and pediatric DIPG line PDIPG displayed a different set of results. These two lines had been initially cultured via clonal expansion in vitro after human tumor biopsy, and both were able to be cultured/expanded in vitro between passage 3 and 20 during this work. Both lines grew several fold within four days in vitro, but neither reproducibly established in vivo, even after 90 days. On OBSCs, MS21 and PDIPG tumors both nearly doubled in size over four days, displaying how OBSCs can support growth of low-passage tumor cells not supported in vivo. [0145] Tumors seeded on OBSCs rapidly establish among the endogenous astrocytes and microglia. Observed were tumor-specific trends in outward invasion or inward retraction of tumor cells on the thalamic region of OBSCs. The invasive cell line GBM8, metastatic cell line MB231Br, and diffuse pediatric DIPG were all found to invade radially outward on OBSCs, while other established lines that grow more densely in vivo retract inward (Fig 5F). Interestingly, microglia within OBSCs also reciprocally respond to the presence
Attorney Docket No.4210.0527WO of tumor, maintaining activated morphologies and associating with nearby tumor cells, indicating a two-way interaction between the tumor and this ex vivo microenvironment (Fig 5G). Adding tumor cells to OBSCs on Day 0 after slicing (“Early Seeding”, the standard method used herein), compared to Day 7 (“Late Seeding”) when microglial activation and PI signaling have attenuated, also leads to faster tumor growth and invasion (Fig 5H-I), consistent with in vivo tumor response. Together, this suggests that LTSs, including OBSCs, provide a niche for tumor growth and interrogation that is more appropriate and accurate than in vitro culture systems and less harsh than in vivo models. Example 3 Tumor Treatment on LTSs, including OBSCs [0146] The panel of OBSC tumor models was next used to investigate treatment responses within the platform. To assess the molecular specificity of drug-induced tumor killing on OBSCs, a CRISPR-modified variant of the U373 tumor line harboring a single- gene knockout in RAD18 was used. Deletion of this DNA repair mediator was expected increase sensitivity and killing for agents that utilize this pathway. The deletion did not affect the growth of untreated wildtype or knockout cells, now termed U373WT and U373KO, both in vitro and on the OBSCs. [0147] All seven OBSC-engrafted tumors were subjected to treatment with a panel of approved CNS tumor therapeutics and external radiation therapy (XRad), as well as the experimental agent TR107 (Madera), a second-generation derivative of ONC20135,36. Six concentrations of each agent were tested, assessing treatment response by quantitative bioluminescence imaging (BLI) to specifically measure tumor kill. Dose-response curves and half maximal inhibitory concentrations (IC50s) were then calculated for each agent (Fig 6A). If treatment elicited <50% death at the highest dose, a “Not Reached (NR)” was reported. [0148] By comparing IC50s (Fig 6A), it was found that 7 of 7 lines showed robust response to the proven agents Carboplatin (Sigma-Aldrich, Cat# C2538-100MG) Cisplatin (Sigma-Alrich, Cat# 232120-50MG), and Lomustine (Selleck Chemicals, Cat# S1840), with the PDX GBM cell line GBM8 showing markedly higher sensitivity. Several agents, including Temozolomide (Sigma-Aldrich, Cat# T2577-100MG), Vincristine (Sigma- Aldrich, Cat # V8388-1MG), XRad, and Gemcitabine (Sigma-Aldrich, Cat# S1149) failed to induce significant killing at even the highest tested doses in more than 5 of the 7 cell
Attorney Docket No.4210.0527WO lines, resulting in designations of NR. Azacitidine (Sigma-Aldrich, Cat# A2385-250MG) and Etoposide (Sigma-Aldrich, Cat# E1383-100MG) showed broad potency, inducing killing in 7 of 7 and 6 of 7 lines, respectively, at IC50s in the 200-400 µM range. [0149] When comparing U373WT vs U373KO, it was found that IC50s were significantly decreased and killing significantly increased in U373KO following treatment with TMZ, azacitidine, gemcitabine, and etoposide. Interestingly, U373KO also showed increased resistance to radiation, possibly due to an upregulation of the nonhomologous end joining pathway. [0150] The experimental agent TR107 was the most potent therapeutic in the panel, inducing complete killing in 7 of 7 cell lines with IC50s in the 0.04-0.1 µM range. Interestingly, potent but incomplete killing of tumor cells by TR107 is often observed in standard in vitro culture (Fig 6B). Because ClpP inhibitors such as TR107 and ONC201 can both directly act on tumor cells and indirectly increase tumor kill by acting on normal cells in the tumor microenvironment (TME), the complete killing observed here when tumors are grown on OBSCs highlights a difference in functional killing patterns in vitro and on OBSCs that may be due to drug-tumor-OBSC interactions. [0151] In the clinical setting, combination regimens are the mainstay of patient care. Following this investigation of single agent therapy, drug responses for clinically relevant therapeutic combinations on OBSCs was tested (Fig 6C-D). Treatment of GBM8 across low doses of TMZ and varying doses of radiation resulted in additive killing with synergistic kill at high radiation doses as indicated by the ZIP synergy score, while combining TMZ and radiation against MS21 revealed strong antagonism. When treatment response of etoposide and carboplatin combination therapy were compared against U373WT and U373KO, the addition of low-dose carboplatin to etoposide surprisingly produced an antagonistic effect in U373WT, which was reversed as higher doses of carboplatin led to enhanced killing by the combination regimen. This effect was not observed in U373KO, where the elevated sensitivity to etoposide monotherapy led to robust killing and synergy from the combination regimen. Synergistic killing of MB231Br and PDIPG is also observed when treating these tumor lines with carboplatin and etoposide. These data highlight the ability of the OBSC platform to measure synergistic effects of combination therapies.
Attorney Docket No.4210.0527WO Example 4 Advancing LTSs, including OBSCs, as a tool for standardized drug testing [0152] As therapeutics are developed for clinical use, assessment of (1) direct tumor killing and (2) toxicity to normal tissue, e.g. brain tissue, both play central roles in defining efficacy. To provide a normalized comparison among drugs with varying potencies, a drug scoring system based on LTS, e.g. OBSC, functional testing was created. Leveraging quantitative imaging methods to measure tumor volumes and health of OBSCs, multiple parameters of drug activity were assessed, methods for normalized assessment that established a “therapeutic window” were developed, and data were collapsed into a single ranked scoring system for multi-component comparison. This process was designed and optimized using the panel of brain tumor lines before moving into uncultured patient brain tumor tissue. [0153] Numerous parameters such as IC10, IC50, and AUC are commonly used to define drug activity. The disclosed analysis calculates 11 such parameters from tumor dose- response curves to determine multiple efficacy values for each agent against each cell line (Fig 7A-B). The ability to separately define values across tumor foci using BLI and normal brain tissue (OBSCs with no tumor added) using the disclosed predefined PI assay allowed for the measurement of both efficacy and off-target toxicity within many parameters and generation of normalized therapeutic window ratios for each drug-tumor-LTS (OBSC) interaction. Within these windows, values ranged from +1 to -1, where values approaching +1 signify increasing tumor kill relative to normal tissue toxicity, and values approaching -1 indicate agents where tumors remained highly viable while toxicity to the normal LTS, e.g. OBSC, tissue was elevated. Incorporating LTS or OBSC toxicity in this way allowed comparison among drugs with less bias toward more potent compounds. [0154] This approach was first applied to GBM8, U373KO, MS21, and LN229 treated with etoposide (Fig 7A). Multi-parameter analysis was driven by comparing each tumor’s dose-response curve, with survival normalized to untreated tumor BLI, to the OBSC toxicity curve (Fig 7B). MS21 displays a consistent increase in growth in response to treatment, producing therapeutic windows close to -1 for many parameters. LN229 is moderately killed by etoposide treatment but still less than the OBSC itself, producing slightly negative therapeutic window values. U373KO and GBM8 all display incrementally more sensitivity to etoposide relative to the OBSC and thus each produce slightly more positive therapeutic windows across all parameters.
Attorney Docket No.4210.0527WO [0155] Next, a multi-parametric algorithm was developed to collapse the normalized parameters into a single score that simplifies comparison of each drug-tumor-LTS/OBSC interaction. This formula integrates over 100 data points from tumor kill and LTS/OBSC toxicity dose-response curves across the 11 individually weighted parameters to derive an overall Drug Sensitivity Score (DSS, described more fully herein, including in the Methods). DSS were set to range from -100 to +100, where +100 describes maximal performing agents (greatest tumor kill, lowest toxicity) and -100 describes the lowest performing agents (poorest tumor kill, highest toxicity). Fig 7A shows how DSS increases as tumor kill increases. [0156] The value of this disclosed algorithm is strongest in relative comparisons, for example, when comparing tumor responses to several drugs or when comparing dose- response curves of varying shapes. The algorithm was first applied to define DSS for U373WT cells treated with different therapeutic agents (Fig 7C: U373WT = red lines with circles; OBSC = black lines with triangles). As shown in Figure 7C-D, the established agents cisplatin and lomustine resulted in DSS of 62 and 64, respectively, near the top of the range for this cell line. TR107 received the highest DSS at 78, while etoposide scored near the bottom, with a DSS of -13. Although TR107 was extremely potent, displaying robust tumor kill at concentrations orders of magnitude lower than most other drugs, it also presented some toxicity to OBSCs within that range. By basing the DSS on therapeutic window, TR107 could be more fairly compared alongside less potent drugs like temozolomide or therapeutics like XRad which use different dosage units. [0157] To investigate the ability of the algorithm to detect differences in single-gene alterations, drug response profiles of U373WT were compared to those of the RAD18- knockout line U373KO (Fig 7C: U373KO = blue lines with squares). As expected, a marked difference in DSS was detected for several therapies between the two cell lines, where RAD18 deletion led to increased sensitivity to TMZ, etoposide, and azacitidine, and a corresponding increase in DSS. In comparing each pair of dose-response profiles against each therapeutic, some showed the most significant differences at low drug doses, while others showed greater curve separation at high doses, and still others displayed distinct killing profiles altogether, underscoring the importance of calculating a unified DSS which incorporates many aspects of each curve. [0158] Lastly, the testing was expanded and DSS generated for 11 therapies across 8 different tumor types (Fig 7D), enabling broader comparisons and providing insight into
Attorney Docket No.4210.0527WO drug:tumor activity. This allowed for the comparison of (1) the relative efficacy of a single agent across multiple tumor types (horizontal rows) and (2) the relative sensitivity of a single tumor to multiple agents (vertical columns). It was found that GBM8 was the most sensitive tumor line to most therapeutics, with the majority of DSS scores near 90. TR107 was, on average, the most effective agent across tumor lines, but interestingly was least effective against GBM8. TMZ and radiation, agents commonly used in clinical care for GBM, showed wide variability among tumors, with DSS scores ranging from -62 to 96 (TMZ) and -39 to 77 (radiation). Example 5 Utilizing LTSs, including OBSCs, as a diagnostic platform for uncultured patient brain tumor tissue [0159] While the field can choose from many models and assays when testing tumor lines, generating a viable, representative tumor model from uncultured patient tumor tissue has historically been a difficult task, especially in the field of brain cancer. Significant initial cell death is often observed when culturing cells in vitro, and in vivo PDX models require extensive lead times while yielding low rates of establishment41–43. Even brain tumor PDOs are most successfully established when modeling the most aggressive GBM subtypes20. To fill this need, disclosed herein is a method to prepare and engraft a diverse panel of living, uncultured patient brain tumor tissues onto LTSs, e.g. OBSCs, for rapid, functional drug screening and diagnosis. To increase flexibility in timing and assay selection, also created and validated was a method to cryopreserve patient tumor tissue while preserving the tumor’s original genetic profile, persistence, and drug response on OBSCs. These methods (1) maximize tumor engraftment and viability independent of tumor grade or subtype, (2) limit cell loss/selection and genetic drift, (3) maintain a normal distribution of tumor per OBSC engraftment site to combat intratumoral heterogeneity, and (4) maximize the number of replicate tumor foci per mass of clinical tumor biopsy tissue. [0160] Fresh surgical biopsies were obtained from patients undergoing standard-of-care resection surgeries at University of North Carolina at Chapel Hill (UNC) hospitals following informed consent. Following cryopreservation and thaw according to the disclosed optimized protocol, the tissue was dissociated into a homogeneous near-single- cell suspension, rapidly transduced with LV-mCh-FLuc, and seeded as tumor foci each
Attorney Docket No.4210.0527WO containing a representative sample of ~1 mg tissue onto OBSCs. To measure the ability of OBSCs to engraft patient brain tumor tissue, tumor tissue survival four days after seeding was first compared on (1) OBSCs in a transwell setup (top left), (2) the transwell membrane without OBSCs (bottom left), and (3) standard in vitro culture (right) via BLI (Fig 8A). Uncultured tumor resection tissue from three different patients bearing three different tumor types (grade I ganglioglioma (GG-I), grade 1 meningioma (MG-1), and a heterogeneous glioblastoma/meningioma tumor (GBM-MG)) showed consistent survival on OBSCs, but not in other culture formats. GG-I, MG-I, and GBM-MG showed <1%, 29%, and <1% viability compared to OBSCs when cultured on the transwell insert alone without OBSC tissue, and <1%, 2%, and <1% viability, respectively, compared to OBSCs when cultured in standard in vitro culture plates. [0161] This tumor seeding method was then applied to eight additional types of adult, pediatric, primary, and metastatic patient brain tumor resection tissue, 11 tumors in all (Fig 8B). As before, quantitative optical imaging revealed that every patient tumor tested reproducibly persisted on OBSCs at t = 4 days after seeding, regardless of tumor type. While efforts were undertaken to understand why OBSCs could provide such an accommodating niche for patient tumor engraftment, OBSC astrocyte activation around tumor tissue was again measured via GFAP. It was found that the engrafted patient tumor tissue also induces activation of – and interaction with – OBSC-embedded astrocytes in a similar manner to the astrocyte activation observed after tumor line engraftment (Fig 8C). [0162] Uncultured tumor resection tissue from an NF2 mutation-driven grade II meningioma (MG-II) was then used to investigate whether tumors engrafted onto OBSCs maintained the genetic profile of the parent tumor. Three groups of samples were prepared for whole exome sequencing (WES) from the same pool of dissociated MG-II tissue (Fig 8D). Group HT is the original uncultured human tumor tissue (n = 3). Group CL is the parent tumor tissue expanded in standard in vitro cell culture until the minimum number of cells required for WES had grown. Because of initial cell loss and subsequent clonal expansion, this required six passages over the span of approximately one month (n = 3). Group BSHT is the uncultured tumor resection tissue engrafted onto OBSCs and subsequently dissected from the OBSCs at the conclusion of our standardized assay length (4 days; n = 4). [0163] WES analysis showed that tumor tissue engrafted onto OBSCs maintained a significant genetic resemblance to the parent tumor, while tumor tissue expanded in vitro
Attorney Docket No.4210.0527WO displayed a distinctly different profile (Fig 8E: left plot displays top 250 most significantly mutated somatic genes; right plot displays top 25, derived from left plot). Furthermore, the mutational profiles of all four BSHT biological replicates were markedly similar, indicating that each OBSC-engrafted tumor indeed contained a representative sample of the original patient tumor. A closer look at the hallmark NF2 mutations existing within all samples revealed that while all samples from the original tumor (HT1-3) and the OBSC- engrafted tumor (BSHT1-4) maintain the frame shift deletion at V24, this mutation was lost in all samples expanded in vitro (CL1-3) and replaced by mutations in other areas. Together, this data suggests that the rapid assay design and tumor-accommodating niche of the disclosed LTS platform, including OBSC platform, enables effective maintenance of the original patient tumor profile. [0164] Consistency in seeding and viability is essential for accurate and reproducible results when assessing therapeutic agents. To investigate consistency in initial seeding, patient tumor tissue was labelled with a fluorescent dye and compared the initial sizes of 24 tumor foci at 1 hour post-seeding (two foci per OBSC; two OBSCs per well). It was found that well-to-well reproducibility was high and yielded no statistically significant differences (Fig 9A). Similar consistency in longitudinal tumor persistence was observed, with no statistical differences in viability (via BLI) among unique samples on day 4, 6, and 8 after seeding (Fig 9B, n=6 foci). [0165] OBSCs can also support the persistence of cryopreserved and thawed patient tumor tissue. The disclosed method of cryo-storage, thaw, and engraftment onto OBSCs generates reproducible persistence (Fig 9C) and response to therapeutic agents (Fig 9D) when compared to fresh tumor tissue from the same patient tumor, significantly enhancing the versatility and flexibility of this platform. Example 6 Patient Tumor Sensitivities Assessed By DSS [0166] Viable brain tumor resection tissue is often limited, leading in part to the dearth of functional testing strategies for brain cancers. While functional diagnostic platforms for other solid tumors generally test many therapeutics and require a large amount of tissue, the studies disclosed herein purposefully limited the number of screened therapeutics to those already being considered by attending clinicians. The workflow has been set up to accept any amount of available tumor tissue and screen even a small number of drugs in
Attorney Docket No.4210.0527WO order to facilitate clinical decisions among top therapeutic options. For each tissue acquired, a clinical team helps curate the panel of relevant therapeutics and determine which drugs to test on each tumor based on tumor type, mutational status, and amount of available tissue. By working with clinicians in this way, a focus on comparisons can directly guide care for each patient in real time. [0167] Uncultured brain tumor resection tissue from ten patients was seeded onto OBSCs and treated with various therapeutics to generate DSS (Fig 10A). Fig 10A compares patient DSS alongside DSS for all established tumor lines (data repeated from Fig 7D). Trends across every DSS for all tumors (Fig 10B) show that our DSS algorithm can calculate scores throughout the entire scoring range: patient tumors are generally more sensitive to treatment than the cell-culture-selected established lines in our panel, save for the ubiquitously sensitive GBM8 tumor line, but there is still a wide distribution in patient tumor sensitivities across all drugs and tumors. [0168] Table 1 describes all tumor types and reported mutational statuses of patient tumor tissue used in this study were compared. Multitudes of co-occurring mutations appear frequently in this set of tumors, presenting difficulty in making treatment decisions based on mutational status alone. We can begin to correlate our results to this clinical data by comparing mutations, treatments, and outcomes in individual patients to DSS calculated from treating matched live patient tissue on OBSCs. Table 1: Patient Tumor Clinical Data. Data was derived from clinical records in accord with our IRB-approved protocol.