CN111863119A

CN111863119A - Adjoint diagnosis model based on PDC/PDX drug sensitivity experiment and multigroup chemical detection analysis and application

Info

Publication number: CN111863119A
Application number: CN202010469410.6A
Authority: CN
Inventors: 戴文韬; 李园园; 刘伟
Original assignee: Shanghai Pudai Biotech Partnership LP
Current assignee: Shanghai Pudai Biotech Partnership LP
Priority date: 2020-05-28
Filing date: 2020-05-28
Publication date: 2020-10-30

Abstract

The invention provides a PDX or PDC model based drug susceptibility experiment and multi-group chemical detection analysis adjoint diagnosis model, a construction method and an application, comprising four modules of construction of a human tumor cell line model and an animal transplantation model, drug susceptibility experiment based on the human tumor cell model and the animal model, multi-group chemical data acquisition of a human tumor sample, drug susceptibility phenotype-multi-group chemical data mining analysis and the like. The invention provides an implementation thought and a key method of four modules and reasonable design of mutual correlation and matching of the four modules, forms a whole set of integration strategy of model construction, drug sensitivity experiments, omics data acquisition, data integration and data mining analysis, overcomes the limitation of the technical means of the conventional accompanying diagnosis scheme research and development strategy in links of sample planning, data acquisition, data analysis and the like, can develop accompanying diagnosis scheme research and development more flexibly and more widely, and simultaneously obtains mechanism explanatory clues related to markers.

Description

Adjoint diagnosis model based on PDC/PDX drug sensitivity experiment and multigroup chemical detection analysis and application

Technical Field

The invention belongs to the technical field of biomedicine, relates to a set of research and development strategies accompanied with a diagnosis scheme, and particularly integrates cell and animal model construction, drug sensitivity experiments, multigroup chemical data acquisition and related data integration analysis methods.

Background

Tumors are a serious public health problem worldwide. According to data published by the international tumor research institute, about 800 million people die of tumors every year worldwide, becoming the first cause of death of human beings. Practice proves that the conventional commonly used mouse spontaneous mutation tumor model, mouse compound mutation tumor model, tumor cell line model, cell line transplantation tumor model and the like can not well simulate the biological characteristics of tumor lesions in a patient body, so that the antitumor drug with good performance in a laboratory stage can not always produce obvious curative effect on clinical patients. That is, the lack of preclinical tumor models with good correspondence to tumor patients is an important cause of transformation failure. In response to this problem, human tumor xenograft models/cell line models (PDX/PDC), derived from human tumor tissue, have given the best solution to date, which exhibit very good biological relevance to clinical tumor patients and can be used to predict clinical outcome. Meanwhile, the existing serum-based biochemical immunoassay marker is limited by the technical platform, the detection target and other reasons, cannot find out a key combination from the complex genetic background and heterogeneity of the tumor, and is difficult to be used as a concomitant diagnosis scheme of the antitumor drug. In response to this problem, molecular markers provide the most reliable solution for companion diagnosis, which benefits from the rapid development of molecular omics detection and analysis technologies including genome, exome, transcriptome and proteome, represented by sequencing, and becomes an important cornerstone for precise medical treatment of tumors.

A human derived tumor cell model (PDC) is an important means for researching canceration mechanism and drug screening, mainly relates to the culture of primary tumor cells, namely, tumor cells are directly separated from a living tumor specimen and then cultured in an in-vitro environment, the biological characteristics and genetic characteristics of the cells are higher in consistency with clinical phenotype, the heterogeneity of original tumors can be maintained, and the method is particularly important in a plurality of oncology researches (such as individual tumor drug sensitivity experiments and individual diagnosis and treatment scheme establishment).

The human tumor transplantation model (abbreviated as PDX) refers to a human tumor allograft model established by inoculating an immunodeficient animal with a tumor excised from a patient in an operation. In recent years, human tumors are transplanted into immunodeficient mice, the shapes of tumor cells, the chromosome content and the isozyme level are the same as those of the human tumors, and the cell dynamics and the biochemical characteristics are well maintained, so the mouse xenograft human tumors become ideal models in immunology and oncology research. Liver cancer, colon cancer, breast cancer, lung cancer, ovarian cancer, melanoma, gastric cancer, lymphoma, leukemia, nephropathy, cervical cancer, soft tissue sarcoma, osteosarcoma and the like are successfully transplanted to immunodeficient mice at present, and a certain proportion of good growing tumor parts can be obtained and can be subjected to passage.

The human tumor transplantation model can simulate the internal environment of a human body and provides a model similar to clinical research on anti-tumor drugs. Studies have shown that the human tumor transplant model has the same histological features as the original patient tumor and maintains the heterogeneity of the patient tumor, and that the human tumor transplant model retains the basic genomic and expression profile characteristics of the patient tumor. Based on the characteristics, the human tumor transplantation model is mainly applied as a research tool for an anticancer drug sensitivity experiment in vivo to carry out a drug sensitivity experiment and detect a tumor drug effect related marker.

By combining PDX/PDC technology with multigroup chemical detection, the invention provides a system research and development strategy of 'model construction-drug sensitivity experiment-omics data acquisition-correlation analysis', can simultaneously collect pharmacodynamic data of an anti-tumor drug and multigroup chemical data related to a PDX/PDC model in a pharmacodynamic experiment, and performs integration and mining; the method can overcome the problems of tumor complex genetic background, tumor heterogeneity, model drug effect correspondence and the like, and effectively accelerate the research and development of corresponding concomitant diagnosis schemes.

Disclosure of Invention

The invention provides a system-integrated companion diagnosis research and development strategy, which comprises a companion diagnosis model based on PDX or PDC model drug sensitivity experiments and multigroup chemical detection analysis, a construction method and application, wherein the companion diagnosis model comprises the construction of a human tumor cell line model (PDC) and a human tumor transplantation animal model (PDX), drug sensitivity experiments based on the PDC and the PDX, multigroup chemical data acquisition of a human tumor sample, and drug sensitivity phenotype-multigroup chemical data mining analysis.

The invention provides a concomitant diagnosis model based on PDX or PDC model drug sensitivity experiment and multigroup chemical detection analysis, which specifically comprises the following steps:

step 1) according to the research and development target of the accompanying diagnosis scheme, determining the detection and analysis mode of drug sensitivity experiment and multigroup chemical data: p A N, wherein P represents the number of tumor patients, P is an integer, and P is more than or equal to 3; a represents the number of experimental arms, A is an integer and is more than or equal to 1; n represents the number of mice carrying tumor tissues of the same patient divided into each arm, N is an integer and is more than or equal to 1; p <10, denoted as P0; p is more than or equal to 10 and is marked as P1;

wherein, the step 1) comprises the following steps 1.1) to 1.3)

Step 1.1) if a is 1 and P is <10, determining that the drug sensitivity test and multi-group chemical data detection analysis mode is P0 × 1 × N; otherwise, executing step 1.2);

step 1.2) if A is 1 and P is more than or equal to 10, determining that the detection and analysis mode of the drug sensitive experiment and the multiple groups of chemical data is P1 × 1 × N;

otherwise, executing step 1.3);

step 1.3) if A is more than 1 and P is less than 10, determining that the drug sensitivity test and the multi-group chemical data detection analysis mode is P0A N; otherwise, determining the detection and analysis mode of the drug sensitivity experiment and the multi-group chemical data as P1A N;

step 2) according to the drug sensitivity and the multi-group chemical data detection and analysis mode determined in the step 1), PDX and/or PDC model construction and/or resuscitation are carried out, and the method comprises the following steps;

Step 2.1) PDX model construction or recovery, which comprises the following specific steps:

step 2.1.1) tumor sample acquisition: when the PDX model is constructed, the tumor sample is a fresh tumor tissue collected in the operation process; when the PDX model is recovered, the tumor sample is a tumor tissue preserved by freezing;

step 2.1.2) tumor sample pretreatment: the tumor tissue sample is placed in a proper amount of sterile RPMI 1640 or DMEM cell culture solution containing penicillin 400U/mL, streptomycin 400ug/mL and 10% BSA, so as to ensure the sterility of the obtained material, and the tumor envelope, connective tissue, necrotic tissue, adipose tissue and the like are carefully stripped and removed.

Step 2.1.3) cutting the tissue into 1-3 mm³The small blocks are put into 500-1000 mu l matrigel (CoringMatrigel Matrix) for pretreatment for 1-5 minutes for standby.

Step 2.1.4) tumor tissue inoculation can be divided into subcutaneous inoculation, kidney envelope inoculation, peritoneal omentum majus inoculation and the like, and the specific inoculation steps are as follows:

step 2.1.4.1) subcutaneous inoculation: the tissue mass was removed 1 piece with forceps and placed in an inoculating needle/trocar and the tissue mass injected into the subcutaneous tissue at the junction of the forelimb or hind limb and torso of the immunodeficient mouse using the trocar. Each mouse was inoculated with 1-2 sites, and 4-5 small tissue blocks were injected at each site. The inoculation should be carried out in a clean bench.

Step 2.1.4.2) inoculating kidney capsule and omentum magnum: the immunodeficient mice are anesthetized, then the abdominal cavity of the mice is opened, the kidney capsule or the omentum majus of the abdominal cavity is cut off by surgical scissors, 1 tissue block is clamped by forceps and is placed under the membrane, and then the kidney capsule or the omentum majus of the abdominal cavity is sutured. The inoculation should be carried out in a clean bench.

Step 2.1.5) tumor growth observation: for a subcutaneous inoculated PDX model, after an immunodeficiency mouse is inoculated with tumor tissues, the growth condition of the tumor is observed at least 1 time per week by means of naked eyes or touch and the like, and when the tumor formation is observed by means of naked eyes or touch and the like, the weight and the volume of the tumor of the animal are regularly measured and recorded; for the PDX model inoculated on the kidney envelope and the abdominal omentum majus, the growth condition of the tumor is observed at least 1 time per week by touching or small animal living body imaging after the immunodeficiency mice are inoculated with the tumor tissue, and when the tumor formation is observed by means of touching or small animal living body imaging and the like, the weight and the volume of the tumor of the animals are regularly measured and recorded.

Step 2.1.6) tumor tissue harvest: when the volume of the tumor body reaches 1000-1500 mm³When the tumor-bearing mouse has a plurality of tumor bodies, the principle of the largest tumor body is taken, the tumor tissue can be harvested for subsequent expanding inoculation.

Step 2.1.7) evaluation of the consistency of the PDX model tumor tissue and the sample source tissue: and evaluating the PDX model to ensure that the model keeps high consistency with a sample source. The method specifically comprises the following steps: and (3) pathological and immunohistochemical staining evaluation is carried out on tumor tissues in the constructed PDX model, high-throughput sequencing/proteome detection can be carried out on the constructed PDX model, the constructed PDX model is compared with corresponding detection results of sample source tissues, and consistency evaluation is completed through the technical means.

Step 2.2) PDC model construction or recovery, which comprises the following specific steps:

step 2.2.1) tumor cell acquisition: when the PDC model is constructed, the tumor cells are obtained by digesting fresh tumor tissues collected in the operation process; when the PDC model is recovered, the tumor cells are the frozen tumor cells. The specific steps of digesting the tumor tissue to obtain the tumor cells are as follows:

step 2.2.2.1) tumor sample pretreatment: harvested fresh human tumor tissue was bubbled in 70% ethanol for 5 min. This step can ensure sterility of the material. Under aseptic conditions, the tissue was placed in an appropriate amount of sterile RPMI 1640 cell culture medium containing penicillin 400U/mL, streptomycin 400ug/mL, 10% BSA. BSA works well to protect cells after tissues are transferred to the outside of the living body.

Step 2.2.2.2) tumor tissue enzymolysis: the tissue was rapidly cut into pieces of 1mm3 and collagenase IV solution was added at a final concentration of 100Mg/ml, using HBSS (Mg2+, Ca2+) solution at a final concentration of 1 Mg/ml. According to the enzymolysis requirement of the tissue of the separation part, 0.25% trypsin can be properly added for cooperation. Cleavage at 37 ℃ for 10 min.

Step 2.2.2.3) tumor cell filtration and purification: the digested cell suspension was filtered through a 100 mesh cell strainer to a 50ml centrifuge tube and purified using percoll reagent. Percoll reagent is prepared by preparing mother liquor by using 9 parts of Percoll and 1 part of 8.5% NaCl, and then diluting the mother liquor to 70% by using 0.85% NaCl to prepare solution with the specific gravity of 1.090 g/ml. The prepared percoll separating medium is slowly added into the filtered cell suspension, and the cell suspension is centrifuged for 15min at 1500 rmp/min.

Step 2.2.2.4) cell layer aspiration and resuspension: the turbid cell layer between the cell separation and the digestion solution was slowly aspirated with a 20ml syringe. The cells were counted, the cell density was adjusted to 5X 106/L, and the cells were resuspended and mixed with the cell culture medium. The tumor cell culture solution is prepared by adding 20% fetal calf serum, penicillin 100U/mL, streptomycin 100ug/mL, insulin 5ug/mL, human recombinant epidermal growth factor 5ng/mL, hydrocortisone 2ug/mL, and dexamethasone 0.02ug/mL into RPMI 1640 basic culture medium.

Step 2.2.3) preparation before cell plating: before cell plating, cells were plated in advance with collagen i. Collagen I (3mg/ml) was diluted to 50ug/ml with 20mM acetic acid. The required collagen volume was converted to surface area of the well plate and unit concentration of 5ug/cm 2. After incubation for 1h at room temperature, PBS was washed three times and dried for use.

Step 2.2.4) cell plating and culture: the cell suspension is put into a pore plate paved with collagen I, cultured for 48 hours at 37 ℃ in a 5% CO2 incubator, and then the solution is changed. The growth state of the cells was observed. If there are many contaminating cells such as fibroblasts, the contaminating cells are removed by using the difference in digestion time.

Step 3) data acquisition and analysis of drug sensitivity experiments, quantification of drug effects, and grouping of tumor patients, comprising the following steps; step 3.1) drug sensitivity experiment data acquisition, including step 3.1.1) acquisition of PDX model drug sensitivity data; step 3.1.2) acquiring PDC model drug sensitivity data;

step 3.1.1) PDX model drug sensitivity data acquisition, which comprises the following specific steps:

step 3.1.1.1) confirming the number of the needed experimental mice according to the drug sensitivity and the multigroup data detection and analysis mode determined in the step 1), and completing the PDX model subculture according to the step 2.1.2-2.1.4, wherein the inoculation mode is subcutaneous inoculation.

Step 3.1.1.2) tumor growth observation: after the immunodeficiency mice are inoculated with tumor tissues, the growth condition of the tumor is observed at least 1 time per week by means of naked eyes or touch and the like, and when the tumor formation is observed by means of naked eyes or touch and the like, the weight and the volume of the tumor of the animals are regularly measured and recorded;

step 3.1.1.3) the volume of the tumor tissue reaches 100-300 mm³The mice are listed as drug sensitive experiment candidate mice, and when the number of the candidate mice meets the grouping requirement of the drug sensitive experiment, the candidate mice are randomly grouped according to the drug sensitive and multigroup chemical data detection and analysis mode determined in the step 1).

Step 3.1.1.4) according to the research and development target of the accompanying diagnosis scheme, determining a drug sensitivity experimental scheme and carrying out a drug sensitivity experiment. Tumor volume and mouse body weight were measured and recorded 3 times per week during the experiment.

3.1.2) collecting PDC model drug sensitivity data, which comprises the following steps:

step 3.1.2.1) determining the number of required experimental cells according to the drug sensitivity and multigroup chemical data detection and analysis mode determined in step 1), and completing the enlarged culture and grouping of the PDC model according to step 2.2).

Step 3.1.2.2) developing a target according to the companion diagnostic protocol, determining a drug sensitive protocol, and adding drugs or control reagents at corresponding concentrations to different experimental groups.

Step 3.1.2.3) according to the drug sensitive test method, detecting the survival amount of the cells at the corresponding time, and respectively marking as V (test group) and V (control group) according to the test groups.

Step 3.2) pretreatment of drug sensitive experimental data, which comprises the following specific steps:

step 3.2.1) preprocessing PDX drug sensitivity experimental data, comprising the following steps: for tumor volumes in excess of 2000mm³The experimental animal drug sensitivity data of (1) is that the tumor volume needs to be removed is more than 2000mm³Tumor volume data and mouse weight data at and after the time point; for drug sensitivity data of experimental animals with the relative change of the body weight of the mouse of more than 20%, removing tumor volume data and body weight data of the time point and later time points with the relative change of the body weight of the mouse of more than 20%; if the two are not in accordance, the data does not need other placesAnd (6) processing.

Step 3.3) quantification of the drug effect, the specific steps are as follows:

step 3.3.1) quantification of drug effects of the PDX model, comprising the steps of:

step 3.3.1.1) calculating the relative change of the body weight of the experimental mouse, wherein the specific calculation mode is as follows:

wherein the time t is 0 based on the time of starting administration; the unit of time t is: and (5) day.

Step 3.3.1.2) calculating the average growth rate of the tumor volume of the experimental mouse, wherein the specific calculation mode is as follows:

Wherein, the time t refers to the maximum time point of the data retained after the processing of the step 2), and the time for starting the administration is taken as a reference 0; the unit of time t is: and (5) day.

Step 3.3.1.3), calculating the drug sensitivity reaction grade of each experimental mouse according to the general rules in the reference field in the step 3.3.1.1-3.3.1.2) respectively as follows: complete response CR, partial response PR, stable disease SD, progressive disease PD; calculating the drug effect evaluation index disease control rate DCR of each experimental arm; the DCR is the proportion of CR, PR and SD, and is obtained by calculating according to the following formula:

the above DCR calculation formula can also adjust PR and/or SD data in the numerator denominator according to the situation, for example, it can be transformed into:

step 3.3.1.4) if the drug effect cannot be distinguished according to the step 3.3.1.3)And if the experiment arms are small, namely the experiment arms with the same pharmacodynamic index DCR exist, calculating the tumor inhibition rate TGI of each experiment arm, and calculating the TGI according to the following formula:

wherein the content of the first and second substances,

AU_CT，icumulative area/T under tumor volume growth curve for experimental group sample i_t，i

AUC_C，iCumulative area/C under tumor volume growth curve for control group sample i_t，i

T_t，iRepresenting the time corresponding to the accumulated area under the tumor volume growth curve of the sample i in the experimental group;

C_t，irepresenting the time corresponding to the accumulated area under the tumor volume growth curve of the control group sample i;

Step 3.3.2) quantification of the drug effect of the PDC model, comprising the steps of:

step 3.3.2.1) the cell Inhibition rate (TGI) is calculated. The specific calculation formula is as follows:

TGI ═ 1-V (control)/V (experimental)) × 100%

Step 3.3.2.2) calculating the half-lethality (IC 50) of the cells.

The specific calculation formula is as follows:

IC50 ═ V (control)/V (experimental)) × 100%

And 3.4) sequencing the tumor patients according to the drug effect quantified in the step 3.3) and the effect of receiving drug treatment by the experimental arm, and dividing the tumor patients into a drug sensitive group Gp and a drug resistant group Gn according to the median or quantile. The specific calculation criteria are as follows:

the drug sensitive group Gp is composed of tumor patients with higher DCR proportion or larger TGI value in the step 3) under the treatment of the experimental arm, namely the DCR or TGI of the PDX/PDC model drug sensitive experiment of the patients is at least in the first 50% in the descending order of the experimental arm, and when the P value of the number of cases is larger, the ranking proportion can be gradually reduced to the first 10%;

the drug-resistant group Gn is composed of tumor patients with lower DCR proportion or smaller TGI value in the step 3) under the treatment of the experimental arm, namely DCR or TGI of PDX/PDC model drug sensitivity experiment of the patients are arranged in descending order of the experimental arm at least in the last 50%, and when the P value of the number of cases is larger, the ranking proportion can be gradually reduced until the last 5%;

And 4) carrying out multigroup chemical detection on tumor patients of a drug sensitive group Gp and a drug resistant group Gn in a related drug treatment experimental arm developed along with a diagnosis scheme according to scheme design, recording and sorting data of the tumor patients, wherein the data comprises at least one group or all combinations of genome, exome and transcriptome, and on the basis, the proteome detection information of the tumor patients can be included as supplement.

Step 5) selecting a plurality of groups of chemical accompanying diagnostic marker mining modes, and mining and analyzing the plurality of groups of chemical data in the step 4) according to the traditional Chinese medicine sensitivity experiment and the plurality of groups of chemical data detection and analysis modes P A N in the step 1), wherein the specific steps are as follows:

step 5.1) if the pattern is P0 x 1 x N, or the pattern is P1 x 1 x N but the number of tumor patients in the drug sensitive group Gp <5 or the drug resistant group Gn <5, adopting step 6.1) to mine a concomitant diagnosis scheme of the drug therapy;

step 5.2) if the pattern is P1 x 1 x N and the number of tumor patients in the drug sensitive group Gp and the drug resistant group Gn is more than 5, adopting the step 6.2) to excavate a concomitant diagnosis scheme of the drug therapy;

step 5.3) if the pattern is P0 a N, for each experimental arm (pharmacotherapy approach), respectively adopting step 6.1) to mine a concomitant diagnosis protocol of the pharmacotherapy;

Step 5.4) if the pattern is P1 a N, mining after classifying the experimental arms, and the specific analysis steps are as follows:

if the Gp of the drug sensitive group in the experimental arm is less than 5 or the number of tumor patients in the drug resistant group Gn is less than 5, adopting the step 6.1) to mine the accompanying diagnosis scheme of the drug therapy for the experimental arm (drug therapy method);

if the number of tumor patients in the drug sensitive group Gp and the drug resistant group Gn in the experimental arm is more than 5, adopting the step 6.2) to excavate a concomitant diagnosis scheme of the drug treatment;

step 6) a excavation scheme of multiple chemical markers matched with PDX or PDC model drug sensitivity experiments, which specifically comprises the following steps:

step 6.1) a multigenomic integration mining scheme featuring transcriptome and/or proteome differential expression, comprising the following steps:

step 6.1.1) if the tumor patients of the drug sensitive group Gp and the drug-resistant group Gn in the step 4) only have transcriptome and/or proteome data, screening the differential expression genes and/or proteins between the two groups of tumor patients, and sequencing according to the difference multiple and the P value for the combination optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3);

step 6.1.2) if the tumor patients of the drug sensitive group Gp and the drug-resistant group Gn in the step 4) only have genome and/or exome data, calculating and sequencing the somatic gene variation frequency of the tumor patients, identifying high-frequency variation genes, namely genes with the gene variation frequency of more than or equal to 5 percent, further screening important variation genes with difference between the two groups of tumor patients, sequencing according to the occurrence ratio and frequency in a sample, and optimizing the gene marker combination related to the tumor concomitant diagnosis in the step 6.3);

Step 6.1.3) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 4) simultaneously have genome and/or exome, transcriptome and/or proteome data, respectively executing the steps 6.1.1 and 6.1.2 for the combination optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3);

step 6.2) a multigenomic integration mining scheme featuring transcriptome and/or proteome differential gene regulation, comprising the specific steps of:

step 6.2.1) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 4) only have transcriptome and/or proteome data, screening the difference regulating genes and/or proteins between the two groups of tumor patients for the combination optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3), which comprises the following specific steps:

step 6.2.1.1) constructing a reference gene regulation network;

step 6.2.1.2) adopts a feature selection algorithm based on machine learning, including Boruta,

Bayes, NMF and univariate linear regression, and realizes acceleration by an isomeric calculation or parallelization method, and TFs which significantly contribute to TF-target relationship in a drug sensitive group Gp and a drug resistant group Gn are screened to form a drug effect associated specific gene regulation network;

step 6.2.1.3) quantifying the gene regulation strength in the condition-specific gene regulation network by adopting a multiple linear regression model;

Performing regression by a De-biased LASSO method, solving to obtain the regulation and control strength and the confidence interval of each gene regulation and control relationship, and judging whether the regulation and control difference is obvious or not by comparing whether the confidence intervals of the same regulation and control relationship in the gene regulation and control networks with different condition specificities are overlapped or not; or the intensity mean value change of the same regulation relation in the gene regulation and control network with different specific conditions is compared, the confidence interval does not need to be calculated, and the regulation and control difference is directly quantified.

Step 6.2.1.4) integrating the three factors related to gene regulation, and screening the gene abnormal regulation relation between the condition-specific gene regulation networks in the drug sensitive and drug resistant states, comprising: the gene regulation intensity is obviously changed, the expression level of the regulation target gene is obviously changed, and the regulation intensity change direction of TF to target is consistent with the change direction of the expression level of target; meanwhile, sorting the screened gene abnormality regulation and control relations according to the difference degree of the regulation and control intensity between the drug sensitive group and the drug resistant group.

Step 6.2.2) if the tumor patients of the drug sensitive group Gp and the drug-resistant group Gn in the step 4) only have genome and/or exome data, calculating and sequencing the somatic gene variation frequency of the tumor patients, identifying high-frequency variation genes, namely genes with the gene variation frequency of more than or equal to 5 percent, further screening important variation genes with difference between the two groups of tumor patients, sequencing according to the occurrence ratio and frequency in a sample, and optimizing the gene marker combination related to the tumor concomitant diagnosis in the step 6.3);

Step 6.2.3) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 4) simultaneously have genome and/or exome, transcriptome and/or proteome data, respectively executing the steps 6.2.1 and 6.2.2 for the combination optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3);

step 6.3) tumor associated diagnosis related gene marker combination optimization is carried out based on successive increase iteration of a greedy algorithm or evolution iteration of a genetic algorithm, and the specific steps are as follows:

if the total number of the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 4) does not exceed 30, the step 6.3.1) is recommended; otherwise, if the total number of tumor patients in the drug sensitive group Gp and the drug resistant group Gn in step 4) exceeds 30, steps 6.3.1 and/or 6.3.2 can be adopted.

Step 6.3.1) successive addition iterative optimization based on greedy algorithm

Step 6.3.1.1), using the genes related to tumor diagnosis discovered in step 6.2), and taking the expression value and/or variation condition (0/1, wild type/mutant type) as quantitative input, respectively constructing models, predicting the classification of tumor patients obtained by drug sensitive experiments, evaluating by using an operator curve ROC, and selecting the model with the largest area under the curve AUC as an initial model.

Step 6.3.1.2), adding new genes related to tumor diagnosis to form a new model, randomly distributing a data set, and calculating the median AUC of K-fold cross validation by using a K-fold cross validation method to evaluate the accuracy of the new model.

And 6.3.1.3) repeating the iteration step 6.3.1.2) of the screening process, selecting the model with the largest AUC by adopting a greedy algorithm, and using the model as an initial model of the next iteration, wherein the model is considered to be converged and stable until the variation difference value of the AUC of the updated model is less than 0.001.

Finally, the gene combination in the convergence stable model in the step 6.3.1.3) is used as the gene marker combination related to tumor associated diagnosis after optimization.

Step 6.3.2) evolutionary iterative optimization based on genetic algorithm

Step 6.3.2.1) using the variant genes and difference regulation and control relations related to the tumor accompanying diagnosis mined in step 6.2), regarding the variant genes and difference regulation and control relations as "genes" on a "chromosome", wherein the length of the chromosome "is the number of the variant genes and difference regulation and control relations (here, based on the sample size, set according to experience and statistical principles, and is recommended to be 1/10 of the total number of tumor patients of the drug sensitive group Gp and the drug resistant group Gn in step 4), each" chromosome "is encoded in binary, and an initial population consisting of not less than 100 individuals (chromosomes, namely a set of variant genes and difference regulation and control relations) is generated;

Step 6.3.2.2), selecting an expression value related to a gene and/or a variation condition (0/1, wild type/mutant type) of a variation gene in a difference regulation and control relationship as quantitative input, constructing a model through K-fold cross validation, predicting the classification of tumor patients obtained by a drug sensitivity experiment, evaluating an operator curve ROC (taking AUC as an evaluation index), performing not less than 10 iterations, and sequencing all individuals in a population based on the prediction evaluation index; removing 20% of individuals with the lowest prediction evaluation index in each round, and calculating the average value of the prediction evaluation indexes of the other individuals to serve as fitness; taking 20% of individuals in the population before the prediction evaluation index as a preferred group, and randomly selecting 10% of the total amount of the rest individuals to form a selected population;

step 6.3.2.3) in the selected population of step 6.3.2.2), two individuals are randomly picked at a time and "chromosomes" are crossed, with crossover sites randomly selected between 0.1 and 0.9, resulting in two new next generation individuals. Then, 10% of the "genes" are randomly selected from the new individuals and mutated, i.e., the "gene locus" coding for 1 becomes 0, and the "gene locus" coding for 0 becomes 1. Offspring with "genes" greater than 0 remain as viable offspring. The round of cross-mutation ends when the number of surviving offspring individuals equals the number of individuals in the initial population.

And finally, carrying out evolution processes such as population generation, selection, crossing, variation and the like, and finishing a plurality of iterations to obtain individuals (variation genes and difference regulation and control relation sets) with the highest indexes. At this time, the chromosome length can be reduced according to the requirement, a new round of genetic evolution screening is carried out, and finally, individuals (variant genes and difference regulation and control relation sets) with the shortest chromosome length and stable indexes are obtained and are used as gene marker combinations related to tumor concomitant diagnosis after optimization.

The method has the advantages that a set of system research and development strategy of model construction, drug sensitivity experiment, omics data acquisition and correlation analysis is provided, and a PDC/PDX technology and a plurality of groups of chemical detection and analysis are integrated and used for mining the companion diagnosis marker; moreover, a set of comprehensive PDC/PDX model and original tumor sample consistency evaluation thinking is provided; and the accompanying diagnosis classification research and development mode of the construction system comprises the number of patients, the number of experimental arms and animals and the drug effect result which are all brought into the mode division basis; and a multigroup chemical integration mining strategy corresponding to the modes is provided, and the multigroup chemical integration mining strategy comprises different schemes such as differential expression and differential gene regulation, and a successive increase iteration based on a greedy algorithm or an evolution iteration optimization combination method based on a genetic algorithm. The models described in the present invention can be used for diagnostic purposes or non-diagnostic purposes for concomitant diagnostics based on PDX or PDC model drug sensitivity assays and multigroup chemical detection assays.

Drawings

FIG. 1 is a schematic flow diagram of a companion diagnostic protocol based on PDX or PDC model drug susceptibility testing and multi-set chemical detection analysis.

FIG. 2 is a tumor body display of the PDX drug sensitivity test of the antitumor drug in example 1.

FIG. 3 disease control rate DCR and tumor inhibition rate TGI of the antitumor drug in colorectal cancer in example 1.

Detailed Description

The invention is further illustrated below with reference to examples and figures. It should be understood that these examples are only for illustrating the present invention, and are not to be construed as limiting the scope of the present invention. Variations and advantages that may occur to those skilled in the art may be incorporated into the invention without departing from the spirit and scope of the inventive concept, which is set forth in the following claims and their equivalents.

Example 1

This example was performed using the P0 x 1 x N model as a study, which was suitable for the development of a companion diagnostic protocol for 1 specific anti-tumor drug regimen (single or combination). According to the contents of the claims and the specification of the invention, a concomitant diagnosis scheme of a colorectal cancer medicament is mined by combining a PDX model, and the method is implemented as follows:

s1.1, selecting postoperative tumor tissue samples of 6 colorectal cancer patients at stage II, determining 1 anti-tumor drug treatment scheme, arranging 1 drug administration experimental arm, determining drug administration mode and concentration according to a method commonly used in the field, and distributing 3 PDX mice from each patient by the experimental arm; meanwhile, a total of 6 PDX mice were set up as a negative control group (physiological saline), consisting of 1 PDX mouse from each patient. Thus, the final pattern was 6 x 1 x 3, i.e. 6 patients, 18 mice dosed with PDX model mice and 6 mice dosed with control PDX model mice.

S1.2 preparation of 24 PDX model mice from 6 patients was accomplished by construction or resuscitation using subcutaneous inoculation according to protocol step 2, based on the experimental model identified in S1.1.

S1.3, developing a PDX drug sensitivity experiment according to the step 3 of the specification, adopting intraperitoneal injection for administration, recording and collecting tumor volume and weight data of a mouse, and carrying out quantitative analysis on drug effect, wherein partial experiment results are shown in a figure 2; disease control rate DCR and tumor inhibition rate TGI (see fig. 3) were calculated according to step 3.3.1.3), and 6 patients were divided into drug sensitive group (4 cases) and drug resistant group (2 cases) after ranking (DCR first, compare TGI when DCR is the same).

S1.4 according to the instruction, step 4, detecting exome and transcriptome based on the case grouping determined in step S1.3, and obtaining related data; and performing the integrated analysis of the multiomics by adopting the step 5.1 according to the step 5 of the specification.

S1.5 based on the grouped high-throughput sequencing data of S1.4, adopting a specification step 6.1) of a multigroup integrated mining scheme featuring transcriptome and/or proteome differential expression, determining the differentially expressed genes most relevant to the drug treatment effect, and determining potential transcriptome markers for drug treatment by calculating | Fold Change | and corrected P values of the gene expression values between tumor tissues and tissues beside cancer of 6 colorectal cancer patients, and ranking genes with corrected P values of <0.05 according to | Fold Change | descending order (the P values after correction are the same, and the P values after correction are smaller and higher), wherein the potential transcriptome markers comprise RUNX3, P2RY8, ATOH1, GLSIEC 1 and TLR 7; meanwhile, high-frequency variation genes are identified by comparing exon sequencing data between tumor tissues and cancer adjacent tissues of 6 colorectal cancer patients, and descending sequence arrangement (with consistent ratio and descending sequence according to gene variation frequency) is carried out according to the occurrence ratio of the variation genes in a sample to determine potential important variation gene markers for drug therapy, including KRAS, NRAS, BRAF, HER2, KIT and PDGFRA.

S1.6 adopts the instruction book step 6.3.1) to iteratively optimize the companion diagnostic marker combination based on successive increase of a greedy algorithm, a single gene marker is used as an initial model, potential genes are gradually increased, iterative excavation is carried out, the companion diagnostic marker combination consisting of RUNX3, P2RY8, SIGLEC1, KRAS, NRAS, BRAF, HER2 and KIT is finally obtained, and the response prediction AUC of a drug treatment scheme taking the disease control rate DCR as a standard is stabilized at 0.74.

Example 2:

this example was performed using the P1 a N model, which was suitable for use in a companion diagnostic protocol that was developed for 3 specific anti-tumor drug regimens (including single or combination). According to the contents of the claims and the specification of the invention, a concomitant diagnosis scheme of a colorectal cancer medicament is mined by combining a PDX/PDC model, and the concrete implementation is as follows:

s2.1, selecting an NIBR PDXE data set of Nowa biomedical research institute of public channel, extracting 59 cases of associated colorectal tumor cases of three administration schemes of Cetuximab, BYL719 and BYL719+ Cetuximab from the data set, and obtaining case transcriptome data and PDX drug sensitivity data.

S2.2 according to the step 3 of the specification, calculating the disease control rate DCR of each experimental arm, sorting PDX drug sensitivity experimental data of 59 patients of each experimental arm respectively, and dividing the 59 patients into a drug sensitivity group, a middle group and a drug resistance group, wherein the drug sensitivity group, the middle group and the drug resistance group are shown in a table 1.

TABLE 1 pharmacodynamic grouping of the three dosing regimens Cetuximab, BYL719+ Cetuximab in example 2

S2.3, grouping 3 experimental arm cases determined in the step S2.2 according to the step 4 of the specification to obtain transcriptome data related to the cases of the drug sensitive group and the drug resistant group; and performing the integrated analysis of the multiomics by adopting the step 5.2 according to the step 5 of the specification.

S2.4 take the Cetuximab experimental arm as an example, and the other two experimental arms (BYL719 and BYL719+ Cetuximab) can be referenced to the mining analysis protocol of the Cetuximab experimental arm.

Adopting an instruction step 6.2) of a multigenomics integration mining scheme featuring transcriptome and/or proteome differential gene regulation, Cetuximab drug-sensitive gene regulation network (transcriptome based on 16 drug-sensitive optimal response cases) and Cetuximab drug-resistant gene regulation network (transcriptome based on 15 drug-resistant progression cases) were constructed respectively, Cetuximab drug-efficacy-related differential regulation genes were identified based on instruction steps 6.2.1.3 and 6.2.1.4, and ranked according to their differential regulation strengths, and the following gene combinations were taken into subsequent analysis, including AXIN1, JUNB, MYC, SMAD5, SMAD4, TGIF2, UBB, ATF3, BMPR2, JUND, KLF10, NR2C2, PPP1CB, SKIL, SMURF1, SP 2, PTP 8, PITX2, E2F 2, SMDP 638, SMF 356, SMF 6, SMTP 3, etc.

S2.5 based on the Cetuximab drug effect related difference regulation and control gene combination in the step S2.4 of the embodiment, adopting the step 6.3.2) of the specification and based on evolutionary iterative optimization of genetic algorithm to obtain a Cetuximab colorectal cancer accompanying diagnosis scheme, wherein the following genes comprise BMPR2, MYC, TFDP2 and TGIF 2.

The protection of the present invention is not limited to the above embodiments. Variations and advantages that may occur to those skilled in the art may be incorporated into the invention without departing from the spirit and scope of the inventive concept, and the scope of the appended claims is intended to be protected.

Claims

1. The application of the companion diagnosis model based on PDX or PDC model drug sensitivity experiment and multigroup chemical detection analysis is characterized by comprising the following steps:

step 1) according to the research and development target of the accompanying diagnosis scheme, determining the detection and analysis mode of drug sensitivity experiment and multigroup chemical data: p is A is N; wherein, P represents the number of tumor patients, P is an integer, and P is more than or equal to 3; a represents the number of experimental arms, A is an integer and is more than or equal to 1; n represents the number of the tested individuals carrying the tumor tissues of the same patient divided into each arm, N is an integer and is more than or equal to 1; p <10, denoted as P0; p is more than or equal to 10 and is marked as P1; wherein, the step 1) comprises the following steps 1.1) to 1.3)

otherwise, executing step 1.3);

step 2) constructing and/or recovering a PDX and/or PDC model according to the drug sensitivity and the multi-group chemical data detection and analysis mode determined in the step 1);

step 3) acquiring and analyzing drug sensitivity experiment data, quantifying drug effect, and grouping tumor patients;

step 4) performing multigroup detection on grouped tumor patients according to scheme design and recording and sorting data of the grouped tumor patients;

step 5) selecting a mining mode of the multigroup diagnosis accompanying marker;

and 6) forming a multi-group chemical marker mining scheme matched with a PDX or PDC model drug sensitivity experiment, and obtaining a gene marker combination related to tumor concomitant diagnosis.

2. The use according to claim 1, wherein after PDX and/or PDC model construction and/or resuscitation in step 2), the model sample and the tissue sample of the original tumor patient from which the model was derived are evaluated for consistency, including any one or a combination of pathology, immunohistochemical staining, high-throughput sequencing, proteomic testing; among them, the consistency evaluation of high-throughput sequencing and proteome detection can be performed based on a full spectrum, and only target genes or proteins involved in COSMIC Cancer Gene Census can be evaluated.

3. The use according to claim 1, wherein step 3) comprises in particular the sub-steps of:

step 3.1) acquiring drug sensitivity experiment data;

step 3.2) preprocessing drug sensitive experiment data;

step 3.3) quantifying the drug effect, including drug sensitive reaction level disease control rate DCR and tumor inhibition rate TGI;

and 3.4) sequencing the tumor patients according to the drug effect quantified in the step 3.3) and the effect of receiving drug treatment by the experimental arm, and dividing the tumor patients into a drug sensitive group Gp and a drug resistant group Gn according to the median or quantile.

4. The use of claim 3, wherein the specific calculation criteria of step 3.4) is that the drug sensitive group Gp is treated by the experimental arm, and the tumor patients with higher DCR ratio of drug sensitive response grade disease control rate or larger TGI value in step 3) are composed of the tumor patients with higher DCR ratio or larger TGI value, i.e. the DCR or TGI of the PDX/PDC model drug sensitive experiment of the patients is at least in the first 50% in the descending order of the experimental arm, and the ranking ratio can be gradually reduced to the first 10% when the P value of the number of cases is larger; the drug resistance medicine Gn is composed of tumor patients with lower DCR ratio of drug sensitivity reaction level disease control rate or lower TGI value of tumor inhibition rate in the step 3) under the treatment of the experimental arm, namely DCR or TGI of the PDX/PDC model drug sensitivity experiment of the patients are arranged in descending order of the experimental arm at least in the last 50%, and when the P value of the number of cases is larger, the ranking proportion can be gradually reduced until the last 5%.

5. The use according to claim 1, wherein step 6) comprises in particular the sub-steps of:

step 6.1) a multigenomic integration mining scheme featuring transcriptome and/or proteome differential expression;

step 6.2) a multigroup chemical integration mining scheme featuring transcriptome and/or proteome differential gene regulation;

and 6.3) carrying out combined optimization on the gene markers related to tumor companion diagnosis, and carrying out successive increase iteration based on a greedy algorithm or evolution iteration based on a genetic algorithm.

6. The use according to claim 1, wherein step 5) comprises in particular the sub-steps of:

step 5.1) if the pattern is P0 x 1 x N, or the pattern is P1 x 1 x N but the number of tumor patients in the drug sensitive group Gp <5 or the drug resistant group Gn <5, adopting the steps 6.1) and 6.3) to excavate a concomitant diagnosis scheme of the drug therapy;

step 5.2) if the pattern is P1 x 1 x N and the number of tumor patients in the drug sensitive group Gp and the drug resistant group Gn is more than 5, adopting steps 6.2) and 6.3) to excavate a concomitant diagnosis scheme of the drug therapy;

step 5.3) if the pattern is P0 a N, for each experimental arm (pharmacotherapy approach), mining the concomitant diagnostic protocol for the pharmacotherapy using steps 6.1) and 6.3), respectively;

Step 5.4) if the pattern is P1 a N, mining after classifying the experimental arms, and the specific analysis steps are as follows: if the Gp of the drug sensitive group in the experimental arm is less than 5 or the number of tumor patients in the drug resistant group Gn is less than 5, adopting the steps 6.1) and 6.3) to excavate a concomitant diagnosis scheme of the drug therapy for the experimental arm (drug therapy method); if the number of tumor patients in the drug sensitive group Gp and the drug resistant group Gn in the experimental arm is more than 5, adopting the steps 6.2) and 6.3) to excavate a concomitant diagnosis scheme of the drug therapy.

7. The use according to claim 5, wherein step 6.1) a multigenomic integration mining scheme featuring differential expression of transcriptomes and/or proteomes, comprises the following sub-steps:

step 6.1.1) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 3) and the step 4) only have transcriptome and/or proteome data, screening the differential expression genes and/or proteins between the two groups of tumor patients, sorting the differential expression genes and/or proteins according to the difference multiple and the P value, and optimizing the combination of the gene markers related to the tumor accompanying diagnosis in the step 6.3);

step 6.1.2) if the tumor patients of the drug sensitive group Gp and the drug-resistant group Gn in the steps 3) and 4) only have genome and/or exome data, calculating and sequencing somatic gene variation frequency of the tumor patients, identifying high-frequency variation genes, namely genes with the gene variation frequency of more than or equal to 5%, further screening important variation genes with difference between the two groups of tumor patients, sequencing according to the appearance proportion and frequency in a sample, and optimizing the gene marker combination related to tumor concomitant diagnosis in the step 6.3);

Step 6.1.3) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in step 3) and step 4) have genomic and/or exome, transcriptome and/or proteome simultaneously, steps 6.1.1) and 6.1.2) are performed, respectively, for the optimization of the gene marker combination related to the tumor concomitant diagnosis of step 6.3).

8. The use according to claim 5, wherein step 6.2) a multigenomic integration mining scheme featuring transcriptome and/or proteome-differential gene regulation, comprises the following sub-steps:

step 6.2.1) if the tumor patients of the drug sensitive group Gp and the drug-resistant group Gn in the step 4) only have transcriptome and/or proteome data, screening the difference regulation genes and/or proteins between the two groups of tumor patients for the combination optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3);

Step 6.2.3) if the tumor patients of the drug sensitive group Gp and the drug resistant group Gn in the step 4) have genome and/or exome, transcriptome and/or proteome simultaneously, the steps 6.2.1 and 6.2.2 are respectively executed for the combined optimization of the gene markers related to the tumor concomitant diagnosis in the step 6.3).

9. Use according to claim 8, characterised in that step 6.2.1), in particular comprises the following sub-steps:

step 6.2.1.1) constructing a reference gene regulation network;

performing regression by a De-biased LASSO method, solving to obtain the regulation and control strength and the confidence interval of each gene regulation and control relationship, and judging whether the regulation and control difference is obvious or not by comparing whether the confidence intervals of the same regulation and control relationship in the gene regulation and control networks with different condition specificities are overlapped or not; or the intensity mean value change of the same regulation relation in the gene regulation and control network with different specific conditions is compared, the confidence interval is not required to be calculated, and the regulation and control difference is directly quantified;

10. The use according to claim 5, wherein step 6.3) tumor-associated diagnosis-related gene marker combination optimization is performed based on successive increase iteration of a greedy algorithm and/or evolution iteration of a genetic algorithm, scheme selection is performed according to the total number of tumor patients of the drug-sensitive group Gp and the drug-resistant group Gn in step 4), and if the total number of the two groups of patients does not exceed 30, successive increase iteration based on the greedy algorithm is employed; if the total number of the two groups of patients exceeds 30, adopting successive increase iteration based on a greedy algorithm and/or evolution iteration based on a genetic algorithm; and (3) predicting the classification of tumor patients obtained by the drug sensitivity experiment, evaluating by using an operator curve ROC (rock characteristic curve), and determining the gene marker combination related to the tumor concomitant diagnosis with the highest and most stable AUC (AUC) by taking the AUC as an evaluation index.

11. The use according to claim 10,

the successive increase iterative optimization based on the greedy algorithm comprises the following steps:

step 6.3.1.1) using the genes related to tumor concomitant diagnosis excavated in step 6.2), and taking the expression values and/or variation conditions (0/1, wild type/mutant type) as quantitative input, respectively constructing models, predicting the classification of tumor patients obtained by drug sensitivity experiments, evaluating by using an operator curve ROC, and selecting the model with the largest area under the curve AUC as an initial model;

step 6.3.1.2) adding new genes related to tumor diagnosis in sequence to form a new model, randomly distributing a data set, and calculating the median AUC of K-fold cross validation by using a K-fold cross validation method to evaluate the accuracy of the new model;

step 6.3.1.3) repeating the iteration step 6.3.1.2) of the screening process, selecting a model with the largest AUC by a greedy algorithm to be used as an initial model of the next iteration, and considering that the model is converged and stable until the variation difference value of the AUC of the updated model is less than 0.001;

finally, the gene combination in the convergence stable model in the step 6.3.1.3) is used as a gene marker combination related to tumor concomitant diagnosis after optimization;

The evolutionary iterative optimization based on the genetic algorithm comprises the following steps:

step 6.3.2.1) using the tumor associated diagnosis related variant genes and difference regulation and control relations mined in step 6.2), regarding the variant genes and difference regulation and control relations as a "gene" on a "chromosome", wherein the length of the "chromosome" is the number of variant genes and difference regulation and control relations, and each "chromosome" is encoded in binary, which generates an initial population consisting of not less than 100 individuals;

step 6.3.2.2), selecting an expression value related to a gene and/or a variation condition (0/1, wild type/mutant type) of a variation gene in a difference regulation and control relationship as quantitative input, constructing a model through K-fold cross validation, predicting the classification of tumor patients obtained by a drug sensitivity experiment, evaluating by using an operator curve ROC (rock characteristic) for not less than 10 rounds of iteration, and sequencing all individuals in a population based on prediction evaluation indexes; removing 20% of individuals with the lowest prediction evaluation index in each round, and calculating the average value of the prediction evaluation indexes of the other individuals to serve as fitness; taking 20% of individuals in the population before the prediction evaluation index as a preferred group, and randomly selecting 10% of the total amount of the rest individuals to form a selected population;

Step 6.3.2.3) randomly picking out two individuals each time in the selected population of step 6.3.2.2), hybridizing the two individuals, and randomly selecting the cross-over sites between 0.1 and 0.9 to generate two new next generation individuals; then randomly selecting 10% of 'genes' on a new individual for mutation, namely changing the 'gene locus' with the code of 1 into 0 and changing the 'gene locus' with the code of 0 into 1; progeny with a "gene" number greater than 0 are retained as viable progeny; the round of cross-mutation ends when the number of surviving offspring individuals equals the number of individuals in the initial population.