DESC:F O R M 2
THE PATENTS ACT, 1970 (39 of 1970)
&
THE PATENTS RULES, 2003
COMPLETE SPECIFICATION
[See section 10 and rule 13]

1. TITLE OF THE INVENTION: SYSTEM AND METHOD
COMPRISING EXPRESSION MAPS FOR PROGNOSTICATING EARLY-STAGE ORAL CANCER
2. APPLICANT (A) NAME: MAZUMDAR SHAW MEDICA
FOUNDATION
(B)ADDRESS: MAZUMDAR SHAW MEDICAL
FOUNDATION, A-BLOCK, 8TH FLOOR, MAZUMDAR SHAW MEDICAL CENTRE, #258/A, NARAYANA HEALTH CITY, BOMMASANDRA, BANGALORE, KARNATAKA, INDIA, 560099
3. NATIONALITY (C) : INDIA

THE FOLLOWING SPECIFICATION PARTICULARLY DESCRIBES THE NATURE OF THIS INVENTION AND THE MANNER IN WHICH IT IS TO BE PERFORMED

[001] CROSS-REFERENCE TO RELATED APPLICATIONS
[002] This application claims priority from prior provisional patent application titled, “SYSTEM AND METHOD COMPRISING EXPRESSION MAPS FOR PROGNOSTICATING EARLY-STAGE ORAL CANCER”, application number: 202241040282 filed on July 14, 2022. The entire collective teachings thereof being herein incorporated by reference.
[003] TECHNICAL FIELD
[004] The present disclosure is in the technical field of system and method comprising expression maps for prognosticating oral cancer, more particularly, disclosed embodiment involves identification of specific CSC markers in specific neoplastic regions in the tumor-adjacent normal and the area of field cancer.
[005] BACKGROUND
[006] At present, oral squamous cell carcinoma (OSCC) comprises more than 90% of all oral neoplasms. In India, OSCC accounts for 40% of total cancer deaths and is regarded as a major public health challenge by WHO (World Health Organization).
[007] Despite the advances of therapeutic approaches, morbidity and mortality in OSCC have not improved significantly during the last 30 years (Markopoulos 2012), due to the increasing incidences of loco regional recurrence, distant metastasis and second primary tumor (Bauer et al. 2010). A key issue in OSCC outcome is that it recurs at the primary site in about 10-30% of cases, even after a complete resection of primary tumor.
[008] Smoking, alcohol use and tobacco use are the major risk factors for oral cavity cancer. Continuous exposure to carcinogens and other environmental parameters such as UV light in the oral mucosa are the underlying reason for the development of precancerous modifications in the adjacent mucosa surrounding the tumor.
[009] This ‘field’ that develops around the tumor is considered the primary cause for local recurrence and the formation of second primary tumors; both of which are primary causative factors for the poor prognosis/survival in oral cancers.
[010] There are several cues that indicate a definitive role for cancer stem cells in oral cancer prognosis; several studies have addressed the same (Rodini et al. 2017). Nevertheless, a comprehensive investigation into the correlation of CSCs with multiple parameters of poor prognosis such as recurrence and field cancer has not been attempted in oral cancer. It is well known that the path for carcinogenesis begins long before the clinical/pathological detection of the cancerous lesion in the tissue; the most effective way of confronting the disease is prevention and early detection. When viewed in combination with the evidence that CSCs are probably the cell types that can initiate, migrate, and generate tumors, a comprehensive profiling of their molecular pattern and an understanding of their correlation with the poor prognosticators of the disease will find a solution.
[011] In summary, there is an urgent need in the art to develop a system and method that will enable prediction of transforming clonal events in surrounding normal epithelium and in tumor margins at the time of cancer resection. This will in turn enable timely detection of susceptible patients and the customization of treatment management regimes and thereby improve the prognosis. Identification of targetable molecules that can be explored as therapeutic option will be an additional asset.
[012] SUMMARY OF THE DISCLOSED EMBODIMENT
[013] The present disclosure relates to a system and method comprising expression maps for prognosticating early-stage oral cancer. More particularly, disclosed embodiment involves identification of specific CSC (Cancer Stem Cells) markers in specific neoplastic regions in the tumor-adjacent normal and the area of field of cancer formed.
[014] According to the first aspect of the disclosed embodiments, a method for identifying a resection margin in an oral cavity of a subject and predicting recurrence in the subject is provided. The method includes mapping an area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44. The biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the excised specimen of the subject. The combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject. The three zones comprise a first zone, a second zone, and a third zone. The first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity. The method includes validating the combined expression of the markers using frozen section staining and Immuno-histochemical profiling. The method includes obtaining histology images of the mapped area, images of the frozen section staining, images of the immunohistochemistry profiling. The method includes analyzing the histology images of the mapped area, frozen section staining images, immunohistochemistry profiling images using a machine learning model trained to predict the resection margin and the recurrence using the combined expression of the biomarkers in the adjacent tissues around the tumor region.
[015] According to an embodiment, the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.
[016] According to another embodiment, the cells comprises oral squamous carcinoma cells.
[017] According to yet another embodiment, the combined expression of biomarkers in the adjacent tissues is used to predict the transforming clonal events in the resection margin in the subject without nodal metastasis.
[018] According to yet another embodiment, the machine learning model comprising one or more of a random forest (RF) model, a Logistic Regression model, or PCA Logistic regression model.
[019] According to yet another embodiment, the machine learning model predicts the neoplastic outcome of the cells with combined expression of the biomarkers comprising the CD44 and SOX2 with a sensitivity of 83% and a specificity of 100%.
[020] According to yet another embodiment, the machine learning model is trained using labeled histology images annotated with cancer stem cell biomarkers comprising SOX2 and CD44, clinical and histological variables comprising tumor size, node size, margin status, perineural invasion (PNI), extracapsular spread (ECS).
[021] According to the second aspect of the disclosed embodiments, a system for identifying a resection margin in the oral cavity of a subject is provided. The system includes a processor comprising a memory that stores a set of instructions to perform the following steps,
the processor is configured to obtain (i) histology images of a mapped area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44, (ii) images of frozen section staining and immunohistochemistry profiling. The frozen section staining and immunohistochemistry profiling validate the combined expression of the cancer stem cell biomarkers comprising SOX2 and CD44. The combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject. The three zones comprise a first zone, a second zone, and a third zone. The first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity. The processor is configured to analyzing the histology images of the mapped area, the frozen section staining images, the immunohistochemistry profiling images using a trained machine learning model to predict cells with combined expression of the biomarkers in the adjacent tissues around the tumor region for the identification of the resection margin in the oral cavity of the subject.
[022] According to an embodiment, the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.
[023] According to another embodiment, the cells comprise oral squamous carcinoma cells.
[024] According to yet another embodiment, the combined expression of biomarkers in the adjacent tissues is used to predict the transforming clonal events in the resection margin in the subject without nodal metastasis.
[025] According to yet another embodiment, the machine learning model comprises one or more of a random forest (RF) model, a Logistic Regression model, or PCA Logistic regression model.
[026] According to yet another embodiment, the machine learning model predicts the neoplastic outcome of the cells with combined expression of the biomarkers comprising the CD44 and SOX2 with a sensitivity of 83% and a specificity of 100%.
[027] According to yet another embodiment, the machine learning model is trained using labeled histology images annotated with cancer stem cell biomarkers comprising SOX2 and CD44, clinical and histological variables comprising tumor size, node size, margin status, perineural invasion (PNI), extra capsular spread (ECS).
[028] According to the third aspect of the disclosed embodiments, a kit for identifying a resection margin in an oral cavity of a subject, wherein the kit comprising reagent/s for identifying the cancer stem cell biomarkers comprising SOX2 and CD44 and a combination thereof. The reagent/s comprising anti-SOX2 polyclonal antibody and anti- CyclinD1 or anti- CD44 antibodies or a fragment or derivative thereof labeled with a detectable label. An area is mapped around a tumor region in the oral cavity of the subject with a combined expression of the SOX2 and CD44. The biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the subject.
[029] According to an embodiment, the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to on left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.
[030] Several aspects of the disclosed embodiment are described below with reference to examples for illustration. However, one skilled in the relevant art will recognize that the disclosed embodiment can be practiced without one or more of the specific details or with other methods, components, materials and so forth. In other instances, well-known structures, materials, or operations are not shown in detail to avoid obscuring the features of the disclosed embodiment. Furthermore, the features/aspects described can be practiced in various combinations, though only some of the combinations are described herein for conciseness.
[031] BRIEF DESCRIPTION OF THE DRAWINGS
[032] The foregoing and other objects and features of the disclosed embodiment will become more fully apparent from the following description and appended claims, taken in conjunction with the accompanying drawings. Understanding that these drawings depict only typical embodiments of the disclosed embodiment and are, therefore, not to be considered limiting of its scope, the disclosed embodiment will be described with additional specificity and detail through use of the accompanying drawings briefly described below:
[033] FIG. 1 illustrates the area of samples collection from around the tumor, according to the aspects of disclosed embodiment.
[034] FIG. 2 illustrates additional sites of sample collections to study field cancer, according to the aspects of disclosed embodiment.
[035] FIG. 3 illustrates univariate and bivariate analysis of features, according to the aspects of disclosed embodiment.
[036] FIG. 4(A) illustrates Heat map represents correlation between features, according to the aspects of disclosed embodiment.
[037] FIG. 4(B) illustrates the selection of optimal number of features. This graph represents number of features vs accuracy, according to the aspects of disclosed embodiment.
[038] FIG. 4(C) Screen plot represents variance, 2 PCs captured the 80% of variance in the data set, according to the aspects of disclosed embodiment.
[039] FIG. 5- FIG. 5(A), FIG. 5(B), FIG. 5(C), FIG. 5(D) illustrates the expression of markers in tumor samples, according to the aspects of disclosed embodiment.
[040] FIG. 6- FIG. 6(A), FIG. 6(B), FIG. 6(C), FIG. 6(D) illustrates the gene expression of markers in adjacent normal area of the tumor, according to the aspects of disclosed embodiment.
[041] FIG. 7 illustrates the distribution and correlation of markers, according to the aspects of disclosed embodiment.
[042] FIG. 8 illustrates the representation of PCs. (A, D) Distribution of PCs. (B, C) Graph depicts that two PCs formed from 4 markers can significantly delineate the recurrence and non- recurrence oral squamous cell carcinoma, according to the aspects of disclosed embodiment.
[043] FIG. 9(A) illustrates the representative image of IHC expression of the markers (SOX2, CD44) in the tumor-adjacent normal mucosa samples of the recurrent and non- recurrent cohort (400x), according to the aspects of disclosed embodiment.
[044] FIG. 9(B) illustrates the average and median protein expression of selected markers, according to the aspects of disclosed embodiment.
[045] FIG. 10 illustrates ROC analysis of the selected markers. The AUC, sensitivity and specificity are represented in the graph. CD44 (AUC: 0.83) (FIG. 10(A)) and SOX2 (AUC: 0.94) (FIG. 10(B)), according to the aspects of disclosed embodiment.
[046] FIG. 11 illustrates representation of histological status according to distance, according to the aspects of disclosed embodiment.
[047] FIG. 12(A) illustrates the representative images (CD44, SOX2) field of cancer samples, according to the aspects of disclosed embodiment.
[048] FIG. 12(B) illustrates the images represent expression of both the markers in recurrent and non- recurrent patients, according to the aspects of disclosed embodiment.
[049] FIG. 12(C) illustrates the representative images of the IHC staining, according to the aspects of disclosed embodiment.
[050] FIG. 12(D), FIG. 12(E) illustrates the co-relation of markers expression with dysplasia, according to the aspects of disclosed embodiment.
[051] FIG. 12(D) illustrates the median expression of CD44 (p<0.01) was highest in moderate dysplastic tissue, according to the aspects of disclosed embodiment.
[052] FIG. 12(E) illustrates the median expression of SOX2 was highest in moderate dysplastic tissues (*p<0.05, **p<0.001), according to the aspects of disclosed embodiment.
[053] FIG. 13 illustrates the representation of markers expression according to distance, in field of cancer samples, according to the aspects of disclosed embodiment.
[054] FIG. 14 illustrates the co-relation of markers, distance and dysplasia. Median expression of CD44 (A) and SOX2 (B), according to the aspects of disclosed embodiment.
[055] FIG. 15 illustrates the survival analysis of dysplasia. No co-relation was found among the different grades of dysplasia and disease-free survival (p<0.05), according to the aspects of disclosed embodiment.
[056] FIG. 16 illustrates the co-relation of markers, distance, dysplasia and prognosis of CD44, according to the aspects of disclosed embodiment.
[057] FIG. 17 illustrates the co-relation of markers, distance, dysplasia and prognosis of SOX2, according to the aspects of disclosed embodiment.
[058] FIG. 18(A) illustrates the representation of marker expression in the entire cohort, according to the aspects of disclosed embodiment.
[059] FIG. 18(B) illustrates the representation of marker expression in the node-positive (N+) cohort, according to the aspects of disclosed embodiment.
[060] FIG. 18(C) illustrates the representation of marker expression in the node-negative (N0) cohort. NR stands for non-recurrent and R stands for recurrent, according to the aspects of disclosed embodiment.
[061] The median expression of CD44 and maximum expression of SOX2 was higher in recurrent patients in the entire cohort. There was not much difference between CD44 and SOX2 expression in recurrent/non-recurrent patient cohort in node positive cohort. Both CD44 and SOX2 expression were higher in recurrent patient than non-recurrent patient in node negative cohort (**p<0.01). Survival plot (DFS) of CD44 median and SOX2 maximum. Higher levels of SOX2 maximum FIG. 18(D) and CD44 median FIG. 18(E) significantly co-related with poor disease-free survival represented in months. The high and low cutoff was provided by ROC curve of individual markers. FIG. 18(F). 4-dimensional plot of CD44 and SOX2 expression, node status and recurrence status. The x and y axis represent the SOX2 and CD44 scores respectively, while the z axis represents the lymph node metastasis cases (0= negative and 1= positive cases). The floor of the plot represents the lymph node negative cases. The wall represents the lymph node positive cases. The color code represents the recurrence status (blue/ 0 represents non- recurrence; yellow/1 represents recurrence). Each dot represents each patient. A clear segregation of markers high group (blue dots) and markers low group (yellow dots) was observed in the lymph node negative cases but not in lymph node positive cases, according to the aspects of disclosed embodiment
[062] FIG. 19 Representation of all the samples collected in the N0 patients from different zones, according to the aspects of disclosed embodiment.
[063] FIG. 19(B) is a representation of all the samples collected in the N+ patient from different zones, according to the aspects of disclosed embodiment.
[064] FIG. 20 illustrates the correlation between features (patient-wise) which is used for developing the machine learning model in which CD44 features highly correlated to recurrence, according to the aspects of disclosed embodiment.
[065] FIG 21 illustrates the visualization of patient-wise features after PCA dimension reduction, 3 PC showed the two-cluster distribution of recurrence subjects according to lymph node metastasis, according to the aspects of disclosed embodiment.
[066] FIG 22 illustrates the correlation between features (site-wise) which is used for developing the machine learning model in which CD44 features highly correlated to recurrence, according to the aspects of disclosed embodiment.
[067] FIG 23 illustrates the visualization of site-wise features after PCA dimension reduction, explains non-linear separation of recurrence and non-recurrence patients, according to the aspects of disclosed embodiment.
[068] FIG. 24 illustrates Digital Processing System, according to the aspects of disclosed embodiment.
[069] In the drawings, like reference numbers generally indicate identical, functionally similar, and/or structurally similar elements. The drawing in which an element first appears is indicated by the leftmost digit(s) in the corresponding reference number.
[070] DETAILED DESCRIPTION OF THE DISCLOSED EMBODIMENT
[071] It is to be understood that the present disclosure is not limited in its application to the details of construction and the arrangement of components set forth in the following description or illustrated in the drawings. The present disclosure is capable of other embodiments and of being practiced or of being carried out in various ways. Also, it is to be understood that the phraseology and terminology used herein is for the purpose of description and should not be regarded as limiting.
[072] The use of “including”, “comprising” or “having” and variations thereof herein is meant to encompass the items listed thereafter and equivalents thereof as well as additional items. The terms “a” and “an” herein do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced item. Further, the use of terms “first”, “second”, and “third”, and the like, herein do not denote any order, quantity, or importance, but rather are used to distinguish one element from another.
[073] As used herein, the singular forms “a”, “an”, and “the” include both singular and plural referents unless the context clearly dictates otherwise. By way of example, “a dosage” refers to one or more than one dosage.
[074] The terms “comprising”, “comprises” and “comprised of” as used herein are synonymous with “including”, “includes” or “containing”, “contains”, and are inclusive or open-ended and do not exclude additional, non-recited members, elements or method steps.
[075] All documents cited in the present specification are hereby incorporated by reference in their totality. In particular, the teachings of all documents herein specifically referred to are incorporated by reference.
[076] Example embodiments of the disclosed embodiment are described with reference to the accompanying figures.
[077] In the drawings, like reference numbers generally indicate identical, functionally similar, and/or structurally similar elements. The drawing in which an element first appears is indicated by the leftmost digit(s) in the corresponding reference number.
[078] Definitions
[079] The following terms are used as defined below throughout this application unless otherwise indicated.
[080] The terms “tumour” or “tumour tissue” refer to an abnormal mass of tissue which results from uncontrolled cell division. A tumour or tumour tissue comprises “tumour cells” which are neoplastic cells with anomalous growth properties and no functional bodily function. Tumours, tumour cells and tumour tissue can be benign or malignant.
[081] The term "Marker" or "biomarker" are used interchangeably, and in the context of the disclosed embodiment refer to a polypeptide and/or changes in glycosylation, which is differentially present in a sample collected from patients having OPMD/HNSCC as compared to a comparable sample taken from control subjects.
[082] The phrase "differentially present" refers to differences in the quantity of the marker present in a sample taken from patients as compared to a control subject. A biomarker can be differentially present in terms of frequency, quantity or both.
[083] The term "Diagnostic" means identifying a pathologic condition.
[084] The terms "detection", "detecting" and the like, may be used in the context of detecting markers or biomarkers.
[085] A "test amount" of a marker refers to an amount of a marker present in a sample being tested. A test amount can be either in absolute amount (e.g., µg/ml) or a relative amount (e.g., relative intensity of signals).
[086] The terms "polypeptide," "peptide" and "protein" are used interchangeably herein to refer to a polymer of amino acid residues. "Polypeptide," "peptide" and "protein” can be modified, e.g., by the addition of carbohydrate residues to form glycoproteins.
[087] The term "Detectable moiety" or a "label" refers to spectroscopic, photochemical, biochemical, immunochemical, or chemical means of detection of a composition. For example, labels may include 32P, 35S, fluorescent dyes, biotin-streptavidin, dioxigenin, haptens, electron-dense reagents, and enzymes. The detectable moiety generates a measurable signal that can quantify the amount of bound detectable moiety in a sample. Quantitation of the signal is done by scintillation counting, densitometry, or flow cytometry.
[088] The term "Antibody” refers to a polypeptide ligand encoded by an immunoglobulin gene(s), which specifically binds and recognizes an epitope.
[089] The terms "subject", "patient" or "individual" generally refer to a human or mammals. "Sample" refers to a polynucleotide, antibodies fragments, polypeptides, peptides, genomic DNA, RNA, or cDNA, polypeptides, a cell, a tissue, and derivatives thereof may comprise a bodily fluid or a soluble cell preparation, or culture media, a chromosome, an organelle, or membrane isolated or extracted from a cell.The subject can have a pre-existing disease or condition, such as cancer. Furthermore, the subject may not have any known pre-existing condition. In addition, the subject may also be non- responsive to an existing or past treatment, such as a treatment for cancer.
[090] The term “oral cancer” refers to a group of malignant or neoplastic cancers originating in the oral cavity of an individual. Non-limiting examples of oral cancers include cancers of the buccal vestibule, hard or soft palate, tongue, gums (including gingival and alveolar carcinomas), lingual cancer, buccal mucosa carcinoma, and the like.
[091] The “oral cavity” includes the buccal mucosa, upper and lower alveolar ridges, floor of the mouth, retromolar trigone, hard palate, and anterior two thirds of the tongue.
[092] “Periodontal disease” refers diseases affecting the gums of an individual, including gingivitis, periodontitis, and the like.
[093] “Therapeutically effective amount or dose” refers to a dose that produces effects for which it is administered. The exact dose depends on the purpose of the treatment.
[094] “Prognosis” refers to prediction of the likelihood of metastasis, predictions of disease free and overall survival, the probable course and outcome of cancer targetting, or the likelihood of recovery from the cancer, in a subject.
[095] “Diagnosis” refers to identification of a disease state, such as cancer in a subject. The methods of diagnosis provided by the disclosed embodiment can be combined with other methods of detection well known in the art. Non-limiting examples of other methods of detection include, detection of known disease biomarkers in oral tissue samples, co-axial tomography (CAT) scans, positron emission tomography (PET), oral radiography, oral biopsy, radionuclide scanning, and the like.
[096] “Nucleic acid” refers to deoxyribonucleotides or ribonucleotides and polymers thereof in either single- or double-stranded form, and complements thereof.
[097] A particular nucleic acid sequence may also implicitly encompass conservatively modified variants thereof (e.g., degenerate codon substitutions) and complementary sequences, in addition to the sequence explicitly indicated. Furthermore, degenerate codon substitutions may be achieved by generating sequences in which the third position of one or more selected (or all) codons is substituted with mixed-base and/or deoxyinosine residues.
[098] The cancer characterized by the methods of the disclosed embodiment can comprise, without limitation, a carcinoma, a germ cell tumor, a blastoma, a sarcoma, a lymphoma or leukemia, or other cancers.
[099] EMBODIMENTS OF THE DISCLOSURE
[0100] The disclosed embodiment involves the system and method comprising expression maps for prognosticating oral cancer. More particularly it involves identification of specific CSC markers in specific neoplastic regions in the tumor-adjacent normal and the area of field cancer.
[0101] Cancer stem cells (CSCs) are known to have definitive role in oral cancer prognosis; several studies have addressed the same [4]. When viewed in combination with the evidence that CSCs are probably the cell types that can initiate, migrate and generate tumors, a comprehensive profiling of their molecular pattern and an understanding of their correlation with the poor prognosticators of the disease is a significant approach. This knowledge will enable prediction of transforming clonal events in surrounding normal epithelium and in tumor margins at the time of cancer resection. The study is based on the hypothesis that CSC signatures identified in the tumor/adjacent normal will signify poor prognosis in patients with oral cancer and that these markers can be used to map the field of cancer.
[0102] A literature review was carried out to identify a panel of CSC markers associated with tumor adjacent normal/surgical margins. The markers identified were validated in oral cancer cohort of TCGA and in surgical margins of patients. The best marker panel selected based on prognostic efficacy were then validated by immune histochemistry in two pipelines i) retrospectively in multiple surgical margins and ii) prospectively in samples collected at different distances (1cm, 2cm, additional sites; 4 quadrants) around the tumor as described in the FIG. 1.
[0103] Also, samples were collected from additional sites where the propensity of developing tumor is high. The significant markers were selected based on correlation with disease-free survival and clinical/pathological parameters (p<0.05). The selected markers were used to map the field of cancer around the tumor as illustrated in FIG. 2. The figure illustrates left upper buccal sulcus (202), right upper buccal sulcus (204), right lower buccal sulcus (206), left upper buccal sulcus (208).
[0104] Literature review identified CSC markers (N=14) associated with surgical margin/adjacent normal and prognosis. The markers were selected based on the following criteria i) detected in the tumor adjacent normal mucosa ii) associated with field of cancer of oral/head and neck squamous cell carcinoma, iii) role in field of cancer of epithelial tumors. Initial validation in TCGA OSCC cohort (N=313) identified a panel of 4 genes (CD44, SOX2, NOTCH1, Ki67). Analysis in surgical margins identified CD44 and SOX2 as associated with poor disease-free survival (p<0.05). Prospective validation in patients (n=40) with samples collected from around the tumor (n=232) indicated that the two markers increased the specificity of existing clinical/histological features in predicting outcome from 56% to 78% (p<0.05). Additionally, the markers predicted recurrence (specificity: 100%; sensitivity: 83.5% sensitivity) in patients without lymph-node metastasis (p<0.05).
[0105] Important attributes of the marker panel
[0106] Cancer stem cells (CSC) are involved in many aspects of tumor like- tumor initiation, progression and metastasis. They are considered the driving force for the therapeutic resistance in the tumor and in maintaining tumor aggressiveness. They are undifferentiated, divide asymmetrically and have unlimited self-renewal capacity. However, most of the markers used to identify a CSC, fall in some pathways that describe their properties, for example stemness markers, hypoxia markers, EMT markers, drug-resistant markers and cell adhesion molecules. CD44 is a known stem cell marker, CD44+ CSCs are known to attribute tumorigenicity, drug resistance and metastasis in head and neck cancer and thereby contribute to poor prognosis. However, reports also attribute CSC properties to CD44- population, emphasizing the use of multiple markers to specify CSCs populations. A panel of markers is hence used to identify CSCs.
[0107] The CSCs and their markers (n=27) (Table-1) identified from the literature were scored based on number of evidences (>2 articles), function (common epithelial markers/enzymes excluded) and detection in adjacent normal tissue. The top 14 markers were selected for experimental validation (Table-2) at multiple levels i) correlation with recurrence in TCGA OSCC patient cohort ii) expression in the surgical margins and iii) prospective profiling in the field of cancer samples and correlation with prognosis.
[0108] Table 1: Markers of CSCs associated with field of cancer

Types of Markers

Marker
Cancer Stem Cell Relation

Role in Field of cancer Detection in
Adjacent mucosa References

Pluripotent Markers

Oct4 Cancer stem cell marker in oral cancer; Associated with prognosis

Dedifferentiation of tumor/mature cells

? (Chiou et al. 2008, Sun et al. 2010)

Sox2 SOX2 has role in regulating cancer stem cell properties of pancreatic cancer cells.
Dedifferentiation of tumor/mature cells; Tumor Initiation

? (McCaughan et al. 2010, Yuan et al. 2010)

Nanog Moon et al has reported that Nanog has a role in genesis of cancer stem cells in GBM.

Dedifferentiation of tumor/mature cells ?

(Chiou et al. 2008, Herreros-Villanueva et al. 2013)

Aldehyde Dehydrogena se

ALDH1A1 ALDH1+/CD44+ cells show increased migration and tumor initiation

Intra-epithelial migration, tumor initiation

? (Biddle et al. 2011, Bhaijee et al. 2012)

Drug Transporter

ABCG2 Stem cell marker imparting drug resistance in HNSCC; ABCG2+ cells
increased tumor
initiation

Tumor initiation/drug resistance

? (Sun et al. 2010)

Adhesion Molecule

CD44
CSC marker in HNSCC

Tumor initiation
? (Bhaijee et al. 2012, Gonzalez-Moles et al. 2012)

CD133 Putative CSC marker in brain, prostrate and head and neck cancer.

Tumor initiation
? (Sun et al. 2010, Gallmeier et al. 2011)

CD147 Involved in chemo- resistance, invasiveness and
Tumorigenecity of CSCs
Promote invasion

? (Guo et al. 2000, Landras et al. 2019)

EMT markers

E-Cadherin

Marker of EMT and CSCs

Epithelial Migration

? (Dittmer et al. 2009, Gonzalez-Moles et al. 2012, Hu et al. 2012,
Shaw et al. 2013)

N Cadherin Markers of EMT and CSCs (high expression found in CD133+ pancreatic cancer
cells)

Epithelial Migration

? (Hoca et al. 2020)

S100A4 Putative CSC marker in HNSCC
Epithelial Migration ? (Lo et al. 2011, Trujillo et al. 2011)

MMP2 and MMP9
Implicated in the invasive behavior of CSCs in colorectal cancer and OSCC

Epithelial Migration

? (Bredin et al. 2003, Trujillo et al. 2011, Mane et al. 2013, Tseng et al. 2013)

SNAI1 EMT marker that maintains self-renewal properties of CSCs
Tumor Initiation/Migration
? (Trujillo et al. 2011, Heiden et al. 2014)

S100A8 Progression of disease in colorectal carcinoma and migration of cancer stem cells.

Epithelial Migration

? (Roesch-Ely et al. 2010, Duan et al. 2013)
Tumor Supressor genes/Oncog enes/ Cell Cycle regulatory
Gene

Cyclin D1

Induces EMT in CSCs in ovarian cancer

Epithelial Migration

? (Jiao et al. 2013, Wang et al. 2013)

K Ras Mutations in K-RAS activate CSCs contributing towards tumorigenesis as well as metastasis in the cells.

Tumor Initiation

? (Keohavong et al. 2001, Moon et al. 2014)

Differentiatio n Antigen
CK8, CK18
and CK19 CK 8/18 is expressed CSCs of papillary carcinoma; CK 19 in cutaneous epithelial lesions

Proliferation/Initiation

? (Abbas et al. 2011, Kale et al. 2012)

Telomerase Telomerase enzymatic blockers such as, Imetelstat has been shown to decrease CSC populations

Tumorigenesis

? (Joseph et al. 2010, Marian et al. 2010)
Retinoic Acid Receptor Expression correlates with CSC expression in pancreatic cancer
Tumorigenesis
? (Angadi et al. 2012, Bleul et al. 2015)

Proliferation Marker
Ki67 Ki67 is a marker of cancer stem cell of
Glioblastoma
Proliferation
? (Gonzalez-Moles et al. 2012, Li et al. 2012)

Drug Resistent genes

ATR Inhibition of ATR abrogates tumorigenicity of colon cancer cells through depletion of CD133 positive cancer stem cell population.

Drug Resistance

? (Gallmeier et al. 2011, Suresh et al. 2012)

Hypoxia marker

HIF2a Micro-environmental factor, involved in maintaining CSCs self- renewal by enhancing the expression of
OCT4 and Nanog

Tumor propagation and amplification

? (Li and Rich 2010)

NOTCH1
Regulate the self- renewal and survival of CSCs Decreased NOTCH1 expression was found in the field of cancer of Skin cancer

? (Hu et al. 2012)

B Catenin
It is involved in the Wnt signalling of CSCs self-renewal B Catenin activation leads to large field of cells involved in tumorigenesis of digestive tract
(Coste et al. 2007)

? (Pandit et al. 2018)
[0109] Table 2: Selection of markers based on literature search

Sl No

Markers No of articles as CSCs
No of articles in detection of adj normal area No of articles in relation to field of cancer of OSCC No of articles in relation to field of cancer of HNSCC No of articles in relation to field of cancer of epithelial cancer
1 CD44 113 5 0 0 0
2 CD147 4 1 0 0 0
3 ATR 1 0 0 0 1
4 OCT4 33 1 0 0 0
5 SOX2 35 1 0 0 2
6 MMP2 2 5 0 0 2
7 MMP9 2 4 0 1 3
8 SNAI 20 0 0 0 1
9 NOTCH1 7 1 0 0 1
10 E Cadherin 5 6 2 0 6
11 N Cadherin 22 6 1 0 3
12 Cyclin D1 9 2 1 2 6
13 Ki67 9 3 1 0 3
14 B Catenin 19 6 0 1 1
[0110] Association of the CSC panel with recurrence in OSCC (TCGA cohort)
[0111] Patient Details:
[0112] In the TCGA OSCC cohort (n=313), 67% (211/313) of the cases were male with 52% (162/313) patients having a smoking habit. Tongue cancers (42%; 132/313) formed the major site in this cohort, a majority of the cases were of the T2-T4 stage (25–33%). Analysis of the markers (n=14) in the patients, wherein follow up was available (n=198; recurrent: 82; non-recurrent:116) after outlier removal (<10%) (FIG.3) indicated that SOX2 showed the most significant association (p<0.000) with recurrence (FIG. 4(A)). Additionally, logistic regression model (K-cross-validation) with all the markers indicated the minimum number of markers (4-7) that could provide similar accuracy (FIG.4B). Validation in test data showed a high sensitivity (86%) and specificity (100%) in predicting recurrence (Table-3). The best markers (n=7; SOX2, CD44, NOTCH1, Ki67, CyclinD1, E Cadherin and MMP9) were selected by recursive feature elimination (Table 4). Evaluation of multi-collinearity of selected features using Variation Inflation Factor (VIF) indicated an absence of collinearity (VIF=5) (Table 5). Among these markers, four that significantly associated with recurrence (SOX2, CD44, NOTCH1 and Ki67) were selected for further analysis. FIG. 4(C) illustrates Screen plot represents variance; 2 PCs had captured the 80% of variance in the data set.
[0113] Table-3: Sensitivity and specificity of training and test set
Features Training set (n=158) Test Set (n=40)
Accuracy 90 93
Sensitivity 89 86
Specificity 91 100
[0114] Table-4: Logistic regression with the best 7 markers
Coeffici ent Std error Z P value 0.025 0.975
Constant -4.4579 1.103 -4.041 0.000 -6.620 -2.296
SOX2 -9.1701 1.916 -4.785 0.000 -12.926 -5.414
CD44 -0.5939 0.277 -2.142 0.032 -1.137 -0.050
NOTCH1 -0.5915 0.290 -2.039 0.041 -1.160 -0.023
E
Cadherin 0.3695 0.327 1.131 0.258 -0.271 1.010
CyclinD1 -0.4260 0.247 -1.726 0.084 -0.910 0.058
Ki67 0.7424 0.313 2.373 0.018 0.129 1.356
MMP9 -0.2190 0.305 -0.719 0.472 -0.816 0.378
[0115] Table-5: Variation inflation factor (VIF) of the markers
Features VIF
Ki67 1.26
CD44 1.20
E Cadherin 1.17
NOTCH1 1.14
CyclinD1 1.07
MMP9 1.07
SOX2 1.06
[0116] Table 6: List of retrospective samples
Variables Outcome Numbers (%) Variables Outcome Numbers (%)
Age (n=23) Range 34-75 Recurrence Status (n=23) No Recurrence 16 (70)
Gender (n=23) Male 18 (78) Recurrence 7 (30)
Female 5 (22)

T Stage (n=23) T1 2 (9)

Site (n=23) Buccal Mucosa 10 (43) T2 3 (13)
Tongue 8 (35) T3 8 (34)
Other Oral cavity site
5 (22)
T4
10 (43)
Tumor Differentiatio n
Details
(n=23) WDSCC 7 (30) N Stage (n=23) N0 7 (31)
MDSCC 11 (48) N+ 16 (69)

PDSCC
5 (22) M
Stage(n=23) M0 20 (87)
M+ 3 (13)
Habits (n=23) Yes 14 (61)
No 9 (39)
[0117] The disclosed embodiment will be further described in the following examples. It should be understood that these examples are for illustrative purposes only and are not to be construed as limiting this disclosed embodiment in any manner.
[0118] EXAMPLE. 1: CORRELATION OF MARKER PROFILE IN SURGICAL MARGINS OF PATIENTS WITH TREATMENT OUTCOME
[0119] 1. Patient samples
[0120] Retrospective surgical samples were collected from the patients (n=23; 2010-2014) with OSCC. Buccal mucosa (43%; 10/23) and tongue cancers (35%; 8/23) constituted the major sites. Majority of the cases were of T3 (34%) & T4 stage (43%), with 69% (16/23) cases being node positive. Based on follow-up, 70% (16/23) cases were non-recurrent. (Table-6) The validation was carried out sequentially by qPCR-based profiling (n=23) and by IHC analysis (n=17; surgical margins: 63).
[0121] 2. Correlation of the gene expression profile with recurrence
[0122] All the primers used in the study showed an efficiency ranging from 1.9 to 2.1 with the percentage efficiency being 93% to 110%. qPCR profiling indicated that all the 4 markers were upregulated in the recurrent tumor samples (reference genes: RPLPO, 18SrRNA) compared to the non-recurrent tumor samples (Table-7) (FIG.5). When evaluated in the tumor adjacent normal, while NOTCH1 and Ki67 were downregulated, SOX2 (1.66 fold; 1.87 vs 1.12; AUC: 0.71) (FIG.6) showed the highest fold difference in recurrent cohort (Table-8, 9). The dimension reduction was performed by PCA (FIG.7) and two principal components (PC) were selected (80% variance; for generating the logistic regression model. The model could significantly (p<0.02) delineate recurrent patients (Sensitivity: 80%; Specificity: 72%) (FIG.8). The important features contributing to PC1 and PC2, CD44 and SOX2, were carried forward for protein level validation in the retrospective cohort of patient.
[0123] Table 7: Median expression difference between recurrent and non-recurrent patient cohort of tumor tissues.
Markers Median expression fold difference
CD44 1.168
NOTCH1 1.1861
Ki67 2.3504
SOX2 3.5543
[0124] Table 8: Median expression fold difference between recurrent and non-recurrent patient cohort of adjacent normal area
Markers Sensitivity Specificity AUC
SOX2 75 73 0.71
NOTCH1 100 53 0.69
CD44 100 46.7 0.71
Ki67 75 86 0.85
[0125] Table 9 Details of sensitivity and specificity of the markers in gene expression in adjacent normal area of tumor

Markers Sensitivity Specificity AUC P Value
CD44 Average 83 70 0.78 0.05
CD44 Median 83 80 0.83 0.006
SOX2 Average 86 100 0.94 <0.0001
SOX2 Median 86 100 0.94 <0.0001
[0126] 3. Correlation of CD44 and SOX2 profile in surgical margins with prognosis
[0127] CD44 and SOX2 were evaluated in the surgical margins from each patient (average of 3 sites per patient; n=63) (FIG.9A). Comparison of the IHC scores (average, median and maximum score) indicated that the average and median expression of SOX2 and CD44 was significantly higher in the adjacent normal tissues of the recurrent cohort (n=17) (Table-10) (FIG.9B, 9C). Since the distribution of the samples were not normal, the median values of SOX2 and CD44 were considered. Correlation with prognosis indicated that SOX2 (AUC 0.94) showed high sensitivity (86%; 6/7) and specificity (100%; 10/10). CD44 (AUC 0.83) showed a sensitivity of 83% (5/6) and specificity of 80% (8/10) (FIG. 10(A), FIG. 10(B)), (Table-11). Univariate analysis (logistic regression) as well as hazard-ratio analysis also indicated that SOX2, CD44 were significantly (p<0.05) associated with tumor recurrence (Table 12-13). The combined multivariate logistic regression analysis of both the markers showed sensitivity 86% (6/7) and specificity 100% (10/10), p=0.002.
[0128] Table 10: Median expression of markers in recurrent and non-recurrent samples
Markers Median expression in recurrent sample Median expression in non-recurrent samples P Value
SOX2 Median 252 114 0.00009
SOX2 Average 263 125 0.0001
CD44 Median 152 106 0.03
CD44 Average 153 98 0.03
[0129] Table 11: Details of sensitivity and specificity of the markers in the protein expression

Markers
Sensitivity
Specificity
AUC % of patient Correctly classified
P Value
CD44 Average 50 90 0.77 75 0.04
CD44 Median 67 90 0.83 1 0.03
SOX2 Average 85 100 0.94 94 0.0002
SOX2 Median 86 100 0.94 94 0.0002
[0130] Table 12: Logistic regression analysis of the individual markers
Markers Sensitivity Specificity AUC P Value
CD44 Average 83 70 0.78 0.05
CD44 Median 83 80 0.83 0.006
SOX2 Average 86 100 0.94 <0.0001
SOX2 Median 86 100 0.94 <0.0001
[0131] Table 13: Hazard ratio of the selected markers

Markers Hazard Ratio 95% Confidence Interval
P-Value
CD44 median 9.4612 1.7591 to 50.8862 0.01
SOX2 median 19.5430 3.4047 to 112.1770 0.001
[0132] EXAMPLE. 2: PROSPECTIVE VALIDATION OF CD44 AND SOX2 IN THE FIELD AND CORRELATION WITH OUTCOME
[0133] 1. Patient cohort
[0134] Patients (n=40) diagnosed with OSCC (2014 to 2016) were recruited based on specific inclusion/exclusion criteria. Sixty percent (24/40) of patients were male, and majority of cases were from buccal mucosa (47%; 19/40) with 87% percent of the patients (n=35) with risk habits. Fifty three percent cases (21/40) were positive for peri-neural invasion and 60% had nodal metastasis (24/40) with 67% (16/24) cases being positive for extra capsular spread. The patients were followed up for minimum 36 months (Table-14). Among 40 patients, 13 developed recurrence, second primary tumor or leukoplakia at the same site (follow up: 36 months). One patient was lost to follow-up.
[0135] Table 14: Details of Prospective patient cohort
Variables Outcome Number (Percentage) Variables Outcome Percentage
Age (n=40) Median (Range) 54 (33-76)
Follow up status (n=39) No recurrence 26 (67)
Gender (n=40) Male 24 (60) Recurrence 13 (33)
Female 16 (40)

Site (n=40) Buccal Mucosa 19 (47)
Differentiation (n=40) PDSCC 5 (13)
Tongue 5 (13) MDSCC 18 (45)
Other sites 16 (40) WDSCC 17 (42)

Habits (n=40) Yes 35 (87)
(20 chewers)

Pathological features (n=40) Angio-lymphatic invasion +ve 9 (23)
No 5 (13) Peri nuclear invasion +ve 21(53)
Tumor Volume (n=40) < 3cm 16 (40)

Extra- capsular spread +ve

16 (40)
> 3cm 24 (60)
N Stage (n=40) Positive 24 (60)
Negative 16 (40)
[0136] 2. Histological analysis of the field of cancer samples
[0137] A total of 260 samples were collected, out of which, histological assessment of 232 samples (excluded: 28; tissues without epithelium) was carried out. Among the samples (1cm: 91; 2cm: 63, additional sites: 78), majority were mild (n=126) in comparison with moderately dysplastic (n=10) and hyperplastic (n=12), 79 tissues were non dysplastic. Three were atrophic, while 2 were suspected with tumor (one each at 1/2 cm distance). Correlation with distance indicated that mild (1cm=46; 51%; 2cm=27; 43%) and moderate dysplasia (1cm=8; 9%; 2cm=1; 1.6%) were higher at 1cm. Majority were mild dysplastic in the additional sites. (n=53; 68%) (FIG.11).
[0138] EXAMPLE. 3: METHOD OF IDENTIFYING A RESECTION MARGIN
A method for identifying a resection margin in an oral cavity of a subject and predicting recurrence in the subject, wherein the method comprises: mapping an area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44, wherein the biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the excised specimen of the subject, wherein the combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject, wherein the three zones comprise a first zone, a second zone, and a third zone, wherein the first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity, validating the combined expression of the markers using frozen section staining and immuno-histochemical profiling; obtaining histology images of the mapped area, images of the immunohistochemistry profiling; and analyzing the histology images of the mapped area, frozen section staining images, immunohistochemistry profiling images using a machine learning model trained to predict the resection margin and the recurrence using the combined expression of the biomarkers in the adjacent tissues around the tumor region.
[0139] 1. Immunohistochemical profiling of CD44 and SOX2
[0140] Representative image of sites of samples collection shown in FIG. 12(A). Representative image of sites of additional sites of samples collection is shown in FIG. 12(B). The figure illustrates left upper buccal sulcus (1202), right upper buccal sulcus (1204), right lower buccal sulcus (1206), left upper buccal sulcus (1208).
[0141] CD44 and SOX2 expression (FIG.12C) in the different samples was assessed based on i) distance from the tumor ii) histology and iii) recurrence. Analysis as per distance indicated that although the marker profiles showed a trend, the differences were not statistically significant (Table 15). Average expression of CD44 and SOX2 were highest in moderately (CD44: 189±26, SOX2: 136±31.1) as compared to mild dysplastic (CD44: 108.6±6.3, p<0.01, SOX2: 101±6.1) and non-dysplastic tissues (CD44: 86±7.3, p<0.01, SOX2:75.6±7.2, p<0.05) (FIG.12D, 12E). Similarly, the median expression of CD44 and maximum expression SOX2 showed a significant difference between the groups (p<0.01) and were considered for subsequent analysis. A three-way analysis, comparing expression, histology and distance, showed that both CD44 and SOX2 had highest expression in moderate dysplastic tissues as compared to mild and non-dysplastic tissues at 1cm and the additional areas (p<0.05) (Table 16).
[0142] Table 15: Expression of CD44 and SOX2 according to distance
1cm 2cm Additional
CD44 SOX2 CD44 SOX2 CD44 SOX2
Average expression 109±8.4 101±8 89±9.1 63±8.3 101±7.6 102±7.2
Median expression 100 80 80 45 95 100
[0143] Representation of markers expression according to distance, in field cancerization samples. The median CD44 expression was highest in tissues from 1cm distance from surgical margin. SOX2 expression was highest in tissues from additional sites. The difference in expression was not significant is shown in FIG. 13.
[0144] Co-relation of markers, distance and dysplasia. Median expression of CD44 (FIG 14(A)) and SOX2 (FIG. 14(B)) were highest in moderate dysplastic tissues at 1cm distance. (Dys: Dysplasia). CD44 showed a significant difference between the groups at 1cm distance.
[0145] Table 16: Expression of CD44 and SOX2 according to distance and dysplasia
1cm 2cm Additional

No Dys
Mild Dys
Moderate Dys
No Dys
Mild Dys Mode
rate Dys
No Dys
Mild Dys Mode
rate Dys
SOX2-
Median
70
95
150
35
70
0
86
100
200
CD44-
Median
60
100
210
88
100
10
85
100
210
[0146] Assessment of the marker parameters with prognosis indicated that expression of CD44 showed highest association with the recurrence (p<0.05) with a high hazard ratio (HR: 4.98; 95% CI: 1.2453 to 19.9270). ROC curve analysis, however, indicated a comparatively low accuracy for CD44 (Sensitivity: 69%; Specificity: 77%) and SOX2 (Sensitivity: 54%; Specificity: 77%). (Table 17, 18 and 19).
[0147] Table 17: Co-relation of marker expression with prognosis univariate Logistic regression analysis

Markers p value AUC Sensit ivity Specific ity 95%
Confidenc e Interval Odd's ratio
SOX2-Average 0.7 0.54 0 100 0.9910 to
1.0127 1.0018
SOX2 Maximum 0.4 0.615 0 100 0.9932 to
1.0160 1.0045
SOX2 Median 0.8 0.52 0 100 0.9916 to
1.0109 1.0012
CD44Average 0.03 0.68 38.46 96.15 1.0001 to
1.0275 1.0137
CD44 Maximum 0.1 0.666 7.69 92.31 0.9972 to
1.0180 1.0075
CD44 Median 0.03 0.679 38.46 92.31 1.0002 to 1.0119
[0148] Table 18: Co-relation of marker expression with prognosis in univariate Cox regression analysis
Markers P Value Sensitivity Specificity Area under curve
SOX2 Average 0.6 77 42 0.54
SOX2 Maximum 0.2 54 77 0.61
SOX2 Median 0.8 77 38 0.52
CD44Average 0.08 69 77 0.68
CD44 Maximum 0.09 69 65 0.66
CD44 Median 0.1 69 77 0.67
[0149] Table 19: Co-relation of marker expression with prognosis in ROC analysis
Markers p value 95% Confidence Interval Hazard ratio
SOX2 Median 0.2 0.5450 to 7.2036 1.9343
CD44 Median 0.001 1.2453 to 19.9270 4.9814
1.0238
[0150] 2. CD44 and SOX2 improved efficacy of clinic-pathological parameters in predicting recurrence
[0151] Multivariate analysis of clinical/pathological features (perineural invasion, extra-capsular spread, N stage, margin status and T stage) indicated that in combination, the model showed a sensitivity of 85%, while specificity was 56% (p=0.2) (Table 20). Importantly, addition of CD44 median and SOX2 maximum to this model made it statistically significant and increased the specificity from 56% to 78% (Sensitivity 85%) (p<0.05) (Table 21).
[0152] Table 20: Co-relation of clinical and pathological features with tumor recurrence
Variable Coefficient Std. Error P
Perineural_invasion -0.038126 0.85151 0.9643
Extra_capsular_spread 1.84255 1.30955 0.1594
N_Stage -1.8793 1.37013 0.1702
Margin_Status -0.76885 0.9103 0.3983
T="T2" 1.64964 1.44958 0.2551
T="T3" 17.6611 1900.38741 0.9926
T="T4" 0.10267 1.5281 0.9464
Constant -0.7926
[0153] Table 21: Co-relation of clinical and pathological features in combination with CSC markers with tumor recurrence
Variable Coefficient Std. Error P
CD44_Score_Median 0.027687 0.012731 0.0297
SOX2_score_Maximum -0.00151 0.007682 0.8437
Perineural_invasion -0.65308 1.03557 0.5283
Extra_capsular_spread 3.18043 1.68905 0.0597
N_Stage -3.35148 2.01834 0.0968
T="T2" 2.89443 1.81461 0.1107
T="T3" 12.971 1900.388 0.9946
T="T4" -0.18101 1.61057 0.9105
Constant -4.0287
[0154] 3. CD44/SOX 2 correlate with dysplasia and prognosis
[0155] Dysplasia grade was not an independent prognosticator (Odds ratio: 0.692; 95% CI: 0.1573 to 3.0436, AUC: 0.53; Sensitivity: 0%, Specificity: 100%) Survival analysis (K-M plot) indicated that high grade dysplasia was not an independent prognostic factor (p=0.1) (FIG.15). However, CD44 expression in field of cancer samples of recurrent patients (non-dysplasia: 80, mild dysplasia: 180, moderate dysplasia: 270; p<0.05) were higher than the non-recurrent cohort (no dysplasia: 60, mild dysplasia: 77.5, moderate dysplasia: 205; p<0.05). (FIG.16). SOX2 expression (no dysplasia: 70, mild dysplasia: 150, moderate dysplasia: 232.5; p<0.05) also correlated with dysplasia and recurrence at 1cm. However, SOX2 expression did not show any trend in non-recurrent patients. There was also no specific correlation of the marker profiles with samples collected at 2cm distance/additional sites (FIG.15).
[0156] Co-relation of markers, distance, dysplasia and prognosis of CD44. Median expression of CD44 (p<0.05) was highest in moderate dysplastic tissues in 1cm distance of both recurrent and non-recurrent patient. But it does not co-relate in 2cm and additional sites. The expression of CD44 in the moderate dysplastic tissues of recurrent patient was significantly higher than that of the moderate dysplastic tissue of non-recurrent patient (p<0.05) (*p<0.05, **p<0.001) are shown in FIG. 16(A), FIG. 16(B), FIG. 16(C), FIG. 16(D) FIG. 16(E), FIG. 16(F).
[0157] Co-relation of markers, distance, dysplasia and prognosis. Median expression of SOX2 (p<0.01) was highest in moderate dysplastic tissues in 1cm distance of recurrent patient. Expression of SOX2 did not co-relate with dysplasia in non-recurrent patient at 1cm. There was no specific correlation observed with samples collected from 2cm distance/additional sites (*p<0.05). are shown in FIG. 17(A), FIG. 17(B), FIG. 17(C), FIG. 17(D) FIG. 17(E), FIG. 17(F).
[0158] 4. CD44/SOX2 could predict prognosis in patients without nodal metastasis
[0159] Assessment of the other clinical/pathological confounding factors such as treatment, tumor volume (><2.5 cm), margin status of tumor, risk habits, lymph node metastasis, angio-lymphatic invasion, extra-capsular spread and peri-neural invasion indicated that CD44 and SOX2 marker profile correlated with recurrence in patients without nodal metastasis (FIG 18 (A), (B), (C)), patients with tumor volume >2.5 cm, close margin and in patients negative for angio-lymphatic invasion and/or extra capsular spread.
[0160] Table 22: Co-relation of marker expression with prognosis in univariate Logistic regression analysis in (N0) and (N+) patients
Co-relation of marker expression with prognosis in (N0) patients

Markers p value AUC Sensitivity
Specificity
95% CI Odd's ratio
SOX2 Average 0.3 0.63 17 90 0.9897 to
1.0287 1.009
SOX2 Maximum 0.04 0.78 70 90 0.9960 to
1.0476 1.0215
SOX2 Median 0.5 0.58 17 90 0.9893 to
1.0207 1.0049
CD44Average 0.01 0.85 67 90 0.9987 to
1.0575 1.0277
CD44 Maximum 0.02 0.8 67 90 0.9972 to
1.0479 1.0223
CD44 Median 0.02 0.84 83 100 0.9975 to
1.0466 1.0218
Co-relation of marker expression with prognosis in ( N+) patients

Markers p value
AUC
Sensitivity
Specificity 95%
Confiden ce Interval Odd's ratio
SOX2 Average 0.6 0.58 0 100 0.9879 to
1.0186 1.0031
SOX2 Maximum 0.49 0.51 14 100 0.9803 to
1.0098 0.9949
SOX2 Median 0.9 0.51 0 100 0.9866 to
1.0120 0.9992
CD44Average 0.7 0.55 0 100 0.9837 to
1.0027 1.003
CD44 Maximum 0.9 0.54 0 100 0.9872 to
1.0136 1.0003
CD44 Median 0.4 0.54 0 100 0.9903 to
1.0211 1.0056
[0161] Independent analysis in the patients without nodal metastasis (N0; n=16) indicated that the sensitivity in predicting recurrence is 70%-83% with high hazard ratios (SOX2: 5.2(95% CI- 0.8724 to 31.1312; CD44 13.02(95% CI-2.1810 to 77.8280; p<0.05) (Table-22). Survival analysis indicated that patients with high SOX2 and CD44 levels in the field of cancer samples (decided by ROC cutoffs), had poor DFS (p<0.05) (Figure-18 (D), (E)). Multivariate logistic regression analysis indicated that a combination of CD44 median/SOX2 maximum was the best predictor of recurrence (AUC: 0.88, Sensitivity 83%, Specificity 100%, p=0.01) (Table 23). A categorization of N0 cohort based on treatment indicated that the CD44 and SOX2 positivity was associated with recurrence, irrespective of treatment (FIG.18 (F)). Analysis in the node positive cohort indicated that none of the markers were significantly associated with the poor prognosis (Table-22).
[0162] Table 23 Multivariate analysis of the markers
Combinations of markers AUC Sensitivity Specificity P Values
CD44 average+ SOX2 maximum 0.88 83 100 0.01
CD44 median + SOX2 maximum 0.88 83 100 0.03
CD44 maximum+ SOX2 maximum 0.86 66 80 0.02
CD44 average+ SOX2 average 0.84 66 90 0.02
CD44 median + SOX2 median 0.84 66 90 0.03
[0163] The field of cancer map incorporating the SOX2 and CD44 profile along with histological status and prognosis emphasized that 70% of the samples (23/33) showed high expression of either marker or both in the recurrent patients indicating an extensive spread of the field. In the disease-free cohort, 3/10 showed high expression of both markers. However, in these patients only 40% (26/64) samples expression of the markers, indicating that the spread of the field is also an important prognostic factor (FIG. 19(A)). Further, the map indicated the lack of a clear pattern in the marker profile of the recurrent/non-recurrent patients in the N+ cohort, both cohorts showing high expression of markers (recurrent: 6/7; non-recurrent: 11/16). The spread of marker profile was also similar (non-recurrent patients: 61/95 (64%); recurrent: 21/31 (74%)), indicating a widespread field of the markers in the N+ patients irrespective of outcome (FIG. 19(B)).
[0164] Each block (brown dotted lines P1 to P 23) represents one patient. The central red area represents the tumor with the yellow border represents the tumor margin. The black line represents the surgical margin. The blue line represents 1cm away from the surgical line and yellow line represents the 2cm from the surgical line. The outer black line represents the contra lateral sites. Each small circle represents each sample. Red represents both the marker positive, yellow represents CD44 positive, green represents SOX2 and white represents both the marker negative.
[0165] Higher marker expression identified 6 out of 7 recurrent cases correctly. However, the marker profile also identified 9 false positive cases.
[0166] EXAMPLE. 4: SYSTEM FOR IDENTIFYING A RESECTION MARGIN IN AN ORAL CAVITY
A system for identifying a resection margin in an oral cavity of a subject, wherein the system comprises: a processor comprising a memory, wherein the memory stores a set of instructions to perform, obtaining (i) histology images of a mapped area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44, (ii) images of frozen section staining and immunohistochemistry profiling, wherein the frozen section staining and immunohistochemistry profiling validates the combined expression of the cancer stem cell biomarkers comprising SOX2 and CD44, wherein the combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject, wherein the three zones comprise a first zone, a second zone, and a third zone, wherein the first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity; and analyzing the histology images of the mapped area, the frozen section staining images, the immunohistochemistry profiling images using a trained machine learning model to predict cells with combined expression of the biomarkers in the adjacent tissues around the tumor region for the identification of the resection margin in the oral cavity of the subject.
[0167] Experimental Data on the field of cancer patent
[0168] One application of the findings, the primary approaches being adopted is to combined clinical and IHC score of patient’s data developed machine learning model towards predicting outcome. The model was developed patient-wise and site wise data. The logistic regression and machine learning model gave thel given best sensitivity and specificity of 80% and 89% respectively.
[0169] The correlation between features (patient-wise) which is used for the developing the machine learning model in which CD44 features highly correlated to recurrence is illustrated in FIG. 20.
[0170] The Visualization of patient-wise features after PCA dimension reduction, 3 PC showed the two-cluster distribution of recurrence subjects according to lymph node metastasis is illustrated in FIG. 21.
[0171] The correlation between features (site-wise) which is used for the developing the machine learning model in which CD44 features highly correlated to recurrence is illustrated in FIG. 22.
[0172] The Visualization of site-wise features after PCA dimension reduction, explain non-linear separation of recurrence and non-recurrence patients is illustrated in FIG. 23.
[0173] Multiple machine learning models were developed using histological and marker profile after outlier correction and normalization.
[0174] In the patient-wise mode (Table 25), 3-cross validation is performed for training data, the best models were Random Forest and Logistic regression.
[0175] In the site-wise model (Table 25), each site of patients with IHC scores were considered as each subset of patients.
[0176] Eighty percentage of data was used to develop model and validated on 20% of data the Random Forest was given the best results with best features as CD44 marker expression, SOX2 expression, tumor size, Lymph Node status (Table 26).
[0177] Table 25: Machine learning model combing clinical, histological and marker features –Patient-wise and site-wise test validation
Test Result (Cross Validation) Sensitivity Specificity
(Patient- Wise)
Random Forest 80 89
Logistic Regression 80 88
(Site –wise)
Random Forest 81 93
Logistic Regression 75 70
PCA Logistic regression 81 62
[0178] Table 26: Important features selected by Random Forest model
VARIABLE IMPORTANCE
CD44 Score Median 0.379908
Age 0.179198
CD44 score Average 0.161765
Max_dimension 0.089638
SOX2 score Maximum 0.048263
SOX2 score Average 0.044249
T3T4 (Tumor Size) 0.035091
CD44 score Maximum 0.019748
LN 0.016968
ECS 0.012172
WDCC 0.010847
[0179] EXMAPLE. 5: KIT FOR IDENTIFYING A RESECTION MARGIN IN AN ORAL CAVITY
[0180] A kit for identifying a resection margin in an oral cavity of a subject, wherein the kit comprising reagent/s for identifying the cancer stem cell biomarkers comprising SOX2 and CD44 and a combination thereof, wherein the reagent/s comprising anti- SOX2 polyclonal antibody and anti- CyclinD1 or anti- CD44 antibodies or a fragment or derivative thereof labelled with a detectable label, wherein an area is mapped around a tumor region in the oral cavity of the subject with a combined expression of the SOX2 and CD44, wherein the biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the subject.
[0181] DIGITAL PROCESSING SYSTEM
[0182] The flowchart of digital processing system is illustrated in FIG. 24.
[0183] Digital processing system may correspond to each of user system: local system or remote and server noted above. Digital processing system may contain one or more processors (such as a central processing unit (CPU) (2402), random access memory (RAM) (2404), secondary memory (2406), graphics controller (GPU) (2412), primary display unit (2414), network interfaces like (WLAN) (2416), and input interfaces (2418). All the components except display unit (2414) may communicate with each other over communication path (2410) which may contain several buses as is well known in the relevant arts.
[0184] CPU executes instructions stored in RAM to provide several features of the disclosed embodiment CPU may contain multiple processing units, with each processing unit potentially being designed for a specific task. Alternatively, CPU may contain only a single general purpose processing unit. RAM may receive instructions from secondary/system memory.
[0185] IMPORTANT ATTRIBUTES OF THE DISCLOSED EMBODIMENT
[0186] At present the methods used to study field of cancer are loss of heterozygosity, microsatellite alteration, chromosomal instability, mutation in p53 genes. Several tumor biomarkers have been reported in various types of cancer, including head and neck cancer and these markers include genomic alterations, growth factor receptors, vascular markers, proliferation markers, apoptotic markers. Recently cancer stem cells were also investigated for their involvement in the progression of field to carcinoma. Yang et al had observed that the expression of Podoplanin and ABCG2 within a single pre-neoplastic oral erythroplakia lesion significantly correlates with subsequent development of multiple and multifocal carcinomas. Fenf et al have observed that the expression of ALDH1 and BMI1 in oral erythroplakia lesion significantly correlate with subsequent development of multiple and multifocal carcinoma, which shows the effect of field of cancer. However, these studies did not map the markers to the field of cancer. Braakhuis et al have studied field of cancer by LOH in the samples collected in a similar way from tumor adjacent normal area, but they did not use the CSC-based molecular markers.
[0187] Studies have also looked into ways of margin delineation and mapping of field of cancer by other methods. Bugter et al have shown field of cancer of esophageal cancer can be detected by multi-diameter single-fibre reflectance (MDSFR) spectroscopy in the buccal mucosa of patients. However, this system does not look into molecular markers as adjuncts.
[0188] The novelty of disclosed embodiment involves identification of specific CSC markers that specify neoplastic regions in the tumor-adjacent normal and the area of field of cancer. Subsequently, the mapping of the CSC-specific markers was carried out in the area surrounding the tumor tissue in the patients. Use of CSCs markers to study the field spread can be an easy technique, thereby it is easy to find out the resection margin of tumor during surgery. Additionally, the markers could accurately prognosticate in a clinically distinct cohort of patients with oral cancer.
[0189] USES, APPLICATIONS, AND BENEFITS OF THE DISCLOSED EMBODIMENT
[0190] Following are the uses, applications and benefits of these markers
[0191] In prognosis of tumor
[0192] In the lymph node-negative cohort of OSCC patients, CD44 and SOX2 can independently predict tumor recurrence with high sensitivity and specificity. CD44, SOX2 markers expression improve the accuracy of prognosis, when added to the existing clinical and pathological factors. In N0 cases though the patients, if the CD44 and SOX2 expressions are high in the surgical margins the patients would be susceptible for recurrence/poor prognosis.
[0193] AI-based algorithm based on the imaging in combination with markers for prognosis
[0194] In estimating field spread
[0195] This study explored the mapping the field around the tumor with CSC markers. SOX2 and CD44 were used to map the field in the adjacent normal tissues from 1cm, 2cm distance around the tumor in 4 different sites and in additional samples such as two sides of tongue and upper and lower buccal mucosa of contralateral sites in the patients without lymph-node metastasis; this indicated that the molecular field of cancer is far more extensive in oral cancer and that it can be specified by CSCs. This field illustrates primarily dysplastic changes correlated with CSCs marker expression. The combined expression of SOX2 and CD44 could map the field accurately and correlate with disease outcome of the patients with high efficacy (Sensitivity and specificity>80%).
[0196] In accessing surgical margin of tumor
[0197] These two markers in combination can be used in finding the field spread in the lymph node negative patients, thereby enabling accurate delineation of the surgical margin of the tumor in intra-operative assays.
[0198] BEST MODE TO PRACTICE
[0199] Subject to large cohort validations, the immediate clinical applications include
[0200] Possible development of point of care assays (qPCR, IHC based) to predict recurrence or for accurate prognosis in N0 patients.
[0201] Intra-operative modes of detection of the markers to delineate the margins during surgery. This will require the use of fluorescent/nanoparticle tags to enable detection.
[0202] Merely for illustration, only representative number/type of graph, chart, block, and sub-block diagrams were shown. Many environments often contain many more block and sub-block diagrams or systems and sub-systems, both in number and type, depending on the purpose for which the environment is designed.
[0203] While specific embodiments of the invention have been shown and described in detail to illustrate the inventive principles, it will be understood that the invention may be embodied otherwise without departing from such principles.
[0204] Reference throughout this specification to “one embodiment”, “an embodiment”, or similar language means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, appearances of the phrases “in one embodiment”, “in an embodiment” and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment.
[0205] It should be understood that the figures and/or screen shots illustrated in the attachments highlighting the functionality and advantages of the present invention are presented for example purposes only. The present invention is sufficiently flexible and configurable, such that it may be utilized in ways other than that shown in the accompanying figures.
[0206] It should be understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview of this application and scope of the appended claims. All publications, patents, and patent applications cited herein are hereby incorporated by reference in their entirety for all purposes.
,CLAIMS:I/WE CLAIM:
1. A method for identifying a resection margin in an oral cavity of a subject and predicting recurrence in the subject, wherein the method comprises:
mapping an area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44, wherein the biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the excised specimen of the subject;
wherein the combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject, wherein the three zones comprise a first zone, a second zone, and a third zone,
wherein the first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity,
validating the combined expression of the markers using paraffin-embedded sections and Immuno-histochemical profiling;
obtaining histology images of the mapped area, images of the stained paraffin-embedded sections, images of the immunohistochemistry profiling; and
analyzing the histology images of the mapped area, immunohistochemistry profiling images using a machine learning model trained to predict the resection margin and the recurrence using the combined expression of the biomarkers in the adjacent tissues around the tumor region.
2. The method as claimed in claim 1, wherein the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.
3. The method as claimed in claim 1, wherein the cells comprise oral squamous carcinoma cells.
4. The method as claimed in claim 1, wherein the combined expression of biomarkers in the adjacent tissues is used to predict the transforming clonal events in the resection margin in the subject without nodal metastasis.
5. The method as claimed in claim 1, wherein the machine learning model comprises one or more of a random forest (RF) model, a Logistic Regression model, or PCA Logistic regression model.
6. The method as claimed in claim 1, wherein the machine learning model predicts the neoplastic outcome of the cells with combined expression of the biomarkers comprising the CD44 and SOX2 with a sensitivity of 83% and a specificity of 100%.
7. The method as claimed in claim 1, wherein the machine learning model is trained using labeled histology images annotated with cancer stem cell biomarkers comprising SOX2 and CD44, clinical and histological variables comprising tumor size, node size, margin status, perineural invasion (PNI), Extracapsular spread (ECS).
8. A system for identifying a resection margin in an oral cavity of a subject, wherein the system comprises:
a processor comprising a memory, wherein the memory stores a set of instructions to perform,
obtaining (i) histology images of a mapped area around a tumor region in the oral cavity of the subject with a combined expression of cancer stem cell biomarkers comprising SOX2 and CD44, (ii) images of stained paraffin-embedded sections and immunohistochemistry profiling, wherein the histology stained sections and immunohistochemistry profiling validates the combined expression of the cancer stem cell biomarkers comprising SOX2 and CD44, wherein the combined expression of the biomarkers are mapped at three zones in the adjacent tissues around the tumor region in the subject, wherein the three zones comprise a first zone, a second zone, and a third zone, wherein the first zone comprises a distance of one centimeter from the tumor region, the second zone comprises a distance of two centimeters from the tumor region, the third zone comprises at least four contralateral sites in the oral cavity; and iii) clinical parameters
analyzing the histology images of the mapped area, the immunohistochemistry profiling images using a trained machine learning model to predict cells with combined expression of the biomarkers in the adjacent tissues around the tumor region for the identification of the resection margin in the oral cavity of the subject.
9. The system as claimed in claim 8, wherein the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to on left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.
10. The system as claimed in claim 8, wherein the cells comprise oral squamous carcinoma cells.
11. The system as claimed in claim 8, wherein the combined expression of biomarkers in the adjacent tissues facilitates the prediction of transforming clonal events in the resection margin in the subject without nodal metastasis.
12. The system as claimed in claim 8, wherein the machine learning model comprises one or more of a random forest (RF) model, a Logistic Regression model, or PCA Logistic regression model.
13. The system as claimed in claim 8, wherein the machine learning model predicts the neoplastic outcome of the cells with combined expression of the biomarkers comprising the CD44 and SOX2 with a sensitivity of 83% and a specificity of 100%.
14. The system as claimed in claim 10, wherein the machine learning model is trained using labeled histology images annotated with cancer stem cell biomarkers comprising SOX2 and CD44, clinical and histological variables comprising tumor size, node size, margin status, perineural invasion (PNI), extracapsular spread (ECS).
15. A kit for identifying a resection margin in an oral cavity of a subject, wherein the kit comprising reagent/s for identifying the cancer stem cell biomarkers comprising SOX2 and CD44 and a combination thereof, wherein the reagent/s comprising anti-SOX2 polyclonal antibody and anti- CyclinD1 or anti- CD44 antibodies or a fragment or derivative thereof labeled with a detectable label, wherein an area is mapped around a tumor region in the oral cavity of the subject with a combined expression of the SOX2 and CD44, wherein the biomarkers are mapped in the area comprising adjacent tissues around the tumor region in the subject.
16. The kit as claimed in claim 15, wherein the combined expression of biomarkers are mapped in different zones of the oral cavity comprising, but not limited to left upper buccal sulcus, right upper buccal sulcus, right lower buccal sulcus, and left upper buccal sulcus.