Classifications

• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
  • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
  • G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
  • G01N33/53—Immunoassay; Biospecific binding assay; Materials therefor
  • G01N33/574—Immunoassay; Biospecific binding assay; Materials therefor for cancer
  • G01N33/57484—Immunoassay; Biospecific binding assay; Materials therefor for cancer involving compounds serving as markers for tumor, cancer, neoplasia, e.g. cellular determinants, receptors, heat shock/stress proteins, A-protein, oligosaccharides, metabolites
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/0002—Inspection of images, e.g. flaw detection
  • G06T7/0012—Biomedical image inspection
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/60—Analysis of geometric attributes
  • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N2333/00—Assays involving biological materials from specific organisms or of a specific nature
  • G01N2333/435—Assays involving biological materials from specific organisms or of a specific nature from animals; from humans
  • G01N2333/46—Assays involving biological materials from specific organisms or of a specific nature from animals; from humans from vertebrates
  • G01N2333/47—Assays involving proteins of known structure or function as defined in the subgroups
  • G01N2333/4701—Details
  • G01N2333/4727—Calcium binding proteins, e.g. calmodulin
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/10—Image acquisition modality
  • G06T2207/10024—Color image
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/30—Subject of image; Context of image processing
  • G06T2207/30004—Biomedical image processing
  • G06T2207/30024—Cell structures in vitro; Tissue sections in vitro

Definitions

• Warnakulasuriya et al. 201 1
• Warnakulasuriya et al. 2015
• biomarkers have been proposed for association with oral lesion risk, including but not limited to hypermethylation of endothelin receptor type B (EDNRB) and kinesin family member 1 A (KIF1 A) (Pattani et al., 2010), loss of heterozygosity (Zhang et al., 2012), p16 methylation/HPV (Liu et al., 2015), DNA content (Xiao et al., 2015), and S100A7 (Kaur, R. et al. , 2014).
• EDNRB endothelin receptor type B
• KIF1 A kinesin family member 1 A
• US 2014/0235487 provides a quantitative method for determining the risk of developing oral cancers wherein certain morphological data of individual cells are used to calculate a risk score.
• a method, and system that can provide a reliable means of determining the risk of a subject developing cancer, such as head and neck or oral cancer.
• SUMMARY OF THE DESCRIPTION [0009] In general, described herein is a method of automatically quantifying a risk score representative of the risk that a subject, or individual, will develop a cancer. The method is based on a tissue sample obtained from the subject and prepared to visually identify at least one biological marker. The preparation may involve, for example, staining.
• a method of prognosing the risk of developing a cancer in a subject comprising: [0011] - preparing a tissue sample obtained from the subject for visually identifying at least one biological marker associated with the cancer; [0012] - digitally scanning the prepared tissue sample with a digital scanner to generate a scanned image of the sample; [0013] - analyzing the scanned image with an image analyzer to identify at least one region of interest and to quantify at least one parameter that characterizes the presence of the at least one biological marker; [0014] - transmitting the at least one quantified parameter to a processor, the processor being programmed to execute an algorithm for determining a risk score representative of the risk of the individual developing the cancer based on the at least one quantified parameter; and, [0015] - executing the algorithm to generate the risk score.
• a system for prognosing the risk of developing a cancer in a subject comprising: [0017] - a scanning module for digitally scanning a biological sample obtained from the subject and generating a scanned image of the sample, the sample being pre-treated to visually identify at least one biological marker associated with the cancer; [0018] - a visualization module for analyzing the scanned image to identify at least one region of interest and to quantify at least one parameter that characterizes the presence of the at least one biological marker; and, [0019] - a processor programmed to execute an algorithm for determining a risk score representative of the risk of the individual developing the cancer based on the at least quantified parameter.
• FIG. 1 illustrates the automated region of interest (ROI) selection.
• Figure 2 illustrates the automated labelling of the ROI.
• Figure 3 illustrates an example of overlapping ROIs.
• Figures 4a and 4b illustrate examples of input screens from the VisiopharmTM software used for the image analysis in the example.
• Figures 5a and 5b shows the Kaplan-Meier survival curves for the two groups of cases studied in the example.
• the present description relates to a method, in particular an automated method, of determining a subject's (i.e. patient's) risk of developing a cancer.
• the description involves conducting an automated image analysis of a tissue sample that has been stained for one or more biological markers associated with the cancer.
• markers may be cell morphologic properties (such as protein or nucleic acid markers for example.
• the one or more biological markers are one or more protein biomarkers.
• the cancer in question may be any cancer to which the biological marker has been correlated. The correlation may be a reduction or increase in the amount or concentration of the marker when the cancer is present.
• a preferred biological marker is one or more protein markers.
• the image analysis would involve treating a biological sample obtained from the subject with an agent for visually detecting the presence or absence of the protein marker.
• the means of visually detecting the presence of a given protein involves the use of 3,3'-diaminobenzidine (or "DAB").
• DAB 3,3'-diaminobenzidine
• the sample obtained from the subject is contacted with an antibody specific to at least one of the protein markers under investigation, the antibody being conjugated with a peroxidase enzyme.
• the sample is treated with hydrogen peroxide and DAB.
• the DAB is thereby oxidized, forming a brown precipitate.
• the brown precipitate can then be visually detected and the presence and intensity of such color enables the detection of the protein(s).
• the visualization system first identifies a region of interest (ROI) within the sample.
• ROI is preferably identified or delineated based on the detected concentration of the marker(s) in question in the tissue as well as the location of the marker(s) in the tissue.
• Such location information would preferably also include topologic location information concerning the location of the marker(s) within the tissue.
• the information may indicate the amount of the marker(s) in the epithelium
• This detection may involve a "heat map" methodology, wherein the intensity of a stain color (e.g. brown, in the case of DAB) indicates concentration of the marker.
• the ROI determination would preferably require a threshold intensity to be determined initially. Once the threshold intensity is set, the visualization system would automatically identify the ROI. Such ROI may be visually represented by one or more boundary lines to identify areas of marker concentration that exceed the concentration threshold. These regions, or "hot spots", superimposed on a digital image of the sample. [0032] After this, the visualization system then conducts a further data acquisition step on the ROIs that are identified.
• the system calculates the value of at least one parameter representative of the presence of the biological marker.
• the visualization system calculates the values of at least two parameters, namely: (1) a first parameter, P1 , comprising for example a value representing the percentage of the ROI area that is positive for the marker(s) in question; and (2) a second parameter, P2, comprising for example a value representing the average cell size of the cells within the ROI.
• a first parameter, P1 comprising for example a value representing the percentage of the ROI area that is positive for the marker(s) in question
• P2 comprising for example a value representing the average cell size of the cells within the ROI.
• this step can be conducted automatically using software associated with the visualization system. It will be understood that in other embodiments, further parameters may be determined either quantitatively or qualitatively.
• the presence of the marker(s) in question may be determined at the subcellular level, thus providing data on the sub-cellular localization of the marker(s) within the cells in the ROI. It will be understood that the description is not limited to the number of other parameters that may be incorporated into the analysis.
• the visualization system transmits such data to a processor for further processing.
• the processor may be part of or otherwise associated with the same hardware system used to conduct the visualization procedure or it may be associated with a separate hardware component such as a local or remote computer or server.
• the processor is programmed to receive the values of P1 and P2 to perform one or more further mathematical operations on same.
• the processor of the present description is encoded to execute a first algorithm to calculate a risk score, RS.
• the risk score is calculated by multiplying each of the parameters, P1 and P2, with a suitable weighting factor and then subtracting the weighted average cell size value from the weighted marker-positive area value.
• the weighting factors can vary based on the marker or markers in question and on the desired sensitivity of the analysis.
• the values for the weighting factors can be calculated using any statistical modelling techniques or methods as would be known in the art.
• weighting factors may involve linear regression or Cox regression methods based on a given data set.
• weighting factors i.e. w 1 and w 2
• w 1 and w 2 weighting factors
• the processor is also programmed to calculate a probability value, PV, that the cancer will develop within a time period t.
• PV a probability value
• - PV is, as indicated above, the probability of developing cancer within a time period t.
• - S(t) is the probability of not developing cancer within the time period t.
• the system for performing the visualization of the sample need not necessarily be physically located together with the processor that conducts the aforementioned risk calculations.
• the visualization procedure can be performed at a lab located in one location.
• the results, i.e. quantified parameters P1 and P2 can then be transferred or transmitted to another location where the processor may be situated, which processor can the execute the subsequent calculations.
• the processor calculating the risk value(s) may be located a separate office, such as an office of a data analysis service provider.
• the service provider can then provide a clinician or physician or patient with the calculated risk value.
• the above description has involved two entities, the lab and the service provider. However, it will be understood that any number (i.e. one or more entities) may be involved in the data analysis/manipulation process. [0038] As described in the example below, the above automated method was used to determine the risk of developing oral cancer in patients, wherein the protein marker S100A7 was utilized. This marker has been described as a known biomarker for head and neck cancer (Ralhan et al., 2008; Tripathi et al., 2010) and later for oral mucosal dysplasia (Kaur et al. 2014).
• each of the steps of preparing the tissue sample, scanning the tissue sample, analyzing the scanned image and processing the quantified parameter(s) can be performed at discrete locations or in the same location as needed. Further any group or subset of the steps can be performed at the same location or at different locations.
• the data generated by each step can be transmitted in any manner as would be commonly known.
• the data from one step can be passed to the other step over any data communication network, or may be physically transported from one location to another by means of a memory device, such as a USB device, disk etc.
• a memory device such as a USB device, disk etc.
• the present description is not limited by the means by which data is transmitted.
• the description also encompasses a system for performing the aforementioned method steps.
• Such a system would include a digital scanning system or device (also referred to herein as a scanning module, which would be understood to encompass hardware and associated software), that is capable of scanning a biological sample that has been treated to visually identify one or more biological markers.
• the system would also include an image analysis system, or visualization system (also referred to herein as a visualization module, which would be understood to encompass hardware and associated software), for analyzing the digital image generated by the scanning system.
• the visualization system may, for example, identify the regions of interest in the scanned image and also generate one or more quantified parameters representative of the presence and concentration etc. of the marker.
• the system described herein would preferably also include a processor for receiving the quantified parameter(s) and for executing an algorithm that calculates a risk score based on the value(s) of the
• OPL Oral pre-malignant lesion
• the Aalen-Link-Tsiatis estimate used to estimate the variance of expected cancer-free survival probability, provided the 95% confidence interval (CI) of the cancer-free survival curve.
• CI 95% confidence interval
• histopathological dysplasia grading was found to outperform histopathological dysplasia grading in two clinical indices.
• the sensitivity between the low-risk vs. non-low-risk using the present method was 96% compared to the mild vs. non-mild dysplasia grading which was 75%, with a negative predictive value of 80% and 59%, respectively.
• the present method is believed to better categorize a patient's 5-year risk of OPLs progressing to cancer. The method can be easily incorporated into clinical practice as no additional tissue samples are needed for the assessment.
• Materials and methods [0049] Tissue biopsy slides [0050] 150 samples were used in this application.
• tissue biopsy slides, tissue microarrays and immunohistochemistry staining have been described previously (Kaur, Sawhney et al. 2013). The staining was performed at a commercial clinical lab accredited in the province of Ontario, Canada. The slides were then digitally scanned on a Hamamatsu Nanozoomer-XRTM slide scanner. The images of the slides were visualized using
• VisiopharmTM VIS software version 5.0.1 .1 122, Hoersholm, Denmark. Clinical information for each sample such as dysplasia grading, gender, age, etc. was provided by Mount Sinai Hospital. This project including the informed consent form was approved by Mount Sinai Hospital Research Ethics Board (project 13-0197-E).
• VisiopharmTM APPs for automated regions of interest identification, cell classification, and counting [0052] The ROI selection, cell classification, and counting were performed using Visiopharm VIS. Five independent VisiopharmTM APPs were used in tandem to carry out the process.
• APP1 Whole Tissue selection. The tissue on a slide was outlined for further analysis. Glass with no tissue or with staining debris was excluded.
• APP2 Whole Tissue to DAB Area conversion. DAB (3,3'-diaminobenzidine) positive regions (intensity below a user-defined threshold) on the tissue were selected.
• APP3 DAB Area to cell classification. Cell nuclei were labeled in the DAB positive regions.
• APP4 DAB Heat mapping. A heat map was generated throughout the DAB positive region based on the density of nuclei (density was defined as number of nuclei per 10 ⁇ diameter circle). Five hottest spots (with highest density of nuclei) were selected and five 500 ⁇ diameter circles were created the centers at each of the hot spot. These circles might overlap, giving irregular shapes ( Figures 1 and 3).
• APP5 Nuclei classification and Positive DAB% calculation. Only tissue found inside the regions of interest (ROIs) as determined in APP4 were included at this stage. Areas void of tissue (background) was designed to be absent from any analysis. The nuclei were re-labeled which not necessarily overlapped with the labels in APP3. Cytoplasm were classified and labeled as either positive or negative for DAB staining. Data on two parameters were recorded, the average size of cells and the percentage of DAB positive areas versus the total area of ROIs ( Figure 2). [0058] Algorithm [0059] The present algorithm has two major steps. The first step, "image analysis", is used to obtain measurements of S100A7 from slide images.
• image analysis is used to obtain measurements of S100A7 from slide images.
• the second step "risk calculation” is used to feed the measurements from the first step to a formula to produce cancer progression probability. These two steps are discussed further below.
• Table 1 summarises the results of the 150 cases that were reviewed in this study. [0075] Table 1
• the algorithm used in the present example was found to classify dysplasia more accurately than histopathological grading in relation to cancer progression.
• a patient can be classified into one of three risk groups.
• the Nelson-Aalen-Breslow estimate was used to calculate the baseline cancer free survival curve. Based on risk score and the baseline cancer free survival curve, the expected cancer free survival probability for a patient can be calculated.
• the Aalen-Link-Tsiatis estimate was used to estimate the variance of expected cancer free survival probability. Based on the variance and expected cancer free survival probability, the 95% confidence interval (CI) of the cancer free survival curve is generated.
• Dysplasia grading (mild, moderate, 0.67 0.67 severe)

Landscapes

• Engineering & Computer Science (AREA)
• Health & Medical Sciences (AREA)
• Life Sciences & Earth Sciences (AREA)
• Physics & Mathematics (AREA)
• Immunology (AREA)
• General Physics & Mathematics (AREA)
• Molecular Biology (AREA)
• Biomedical Technology (AREA)
• Chemical & Material Sciences (AREA)
• Urology & Nephrology (AREA)
• Hematology (AREA)
• General Health & Medical Sciences (AREA)
• Theoretical Computer Science (AREA)
• Cell Biology (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Geometry (AREA)
• Medical Informatics (AREA)
• Quality & Reliability (AREA)
• Biotechnology (AREA)
• Radiology & Medical Imaging (AREA)
• Microbiology (AREA)
• Hospice & Palliative Care (AREA)
• Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
• Oncology (AREA)
• Food Science & Technology (AREA)
• Medicinal Chemistry (AREA)
• Analytical Chemistry (AREA)
• Biochemistry (AREA)
• Pathology (AREA)
• Investigating Or Analysing Biological Materials (AREA)

Priority Applications (2)

Applications Claiming Priority (2)

Publications (1)

Family

ID=59851352

Family Applications (1)

Country Status (2)

Citations (5)

Family Cites Families (1)

• 2017
  • 2017-03-14 EP EP17765612.1A patent/EP3430384A4/de active Pending
  • 2017-03-14 WO PCT/CA2017/050333 patent/WO2017156627A1/en active Application Filing

Patent Citations (5)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events