Ovarian cancer diagnostics

Abstract

Methods for risk assessment, diagnosis, monitoring during therapy, or monitoring for recurrence for patients with known or suspected ovarian cancer are provided. These methods employ a statistical model and combine detection of microRNA biomarkers and protein biomarkers.

Description

OVARIAN CANCER DIAGNOSTICS FIELD

[0001] Methods for risk assessment, diagnosis, monitoring during therapy, or monitoring for recurrence for patients with known or suspected ovarian cancer are provided. CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of Polish Patent Application No. P.446295 filed October 3, 2023 and U.S. Provisional Application No. 63 / 588,146, filed October 5, 2023, which is incorporated herein by reference. SEQUENCE LISTING

[0003] This application contains a sequence listing that has been submitted in ASCII format via Patent Center and incorporated herein by reference in its entirety. Said ASCII copy, created on 1 October 2024, is named “3365W01WO_Sequence Listing.xml” and is 179,439 bytes in size. BACKGROUND

[0004] Multianalyte protein assays that combine several protein biomarkers with an individual’s menopausal status can aid in the preoperative prediction of ovarian cancer among women presenting with an adnexal mass. However, not all types of ovarian cancer are identified with high accuracy and / or sensitivity using protein biomarkers.

[0005] Serum microRNAs (miRNAs) are another class of biomarkers which have also shown potential to discriminate ovarian cancer cases from benign lesions or healthy controls.

[0006] In addition, existing diagnostic methods may not be optimal for suggesting appropriate treatment or predicting a response to treatment.

[0007] Therefore, a need remains for new diagnostic tests for women with known or suspected ovarian cancer for risk assessment, diagnosis, monitoring during therapy, or monitoring for recurrence.SUMMARY

[0008] A set of microRNAs (miRNAs) detectable in circulation has been identified and is described herein. The microRNAs can be combined with protein biomarkers associated with ovarian cancer to dramatically boost the sensitivity and accuracy of diagnoses of the same. In particular, complementarity of miRNA- and protein-based approaches in identifying a subject presenting with an ovarian mass for surgery is described.

[0009] The high sensitivity and high accuracy of the diagnostics assays described herein translate into the capability of the assays to identify serous cancers, which can have a lower survival rate than other types of ovarian cancers, if detected at a later stage. Early detection of serous cancers is expected to greatly improve survival in subject suffering from ovarian cancer. The methods described herein are especially effective in detecting early-stage serous ovarian cancer in a sample (e.g., a blood sample) collected from a subject who presents with an ovarian mass.

[0010] In one aspect, there is provided a method of diagnosing ovarian cancer in a subject, the method comprising: (a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein (i) the first panel comprises hsa-miR- 320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a- 3p, and hsa-miR-424-5p, and (i) the second panel comprises hsa-miR-320c, hsa-miR-200c- 3p, and hsa-miR-122-5p; (b) comparing the amounts obtained in step (a) to a mathematical model (e.g., a statistical model) based on corresponding amounts determined for samples collected from individuals with and / or without ovarian cancer; and obtaining a score (e.g., a probability score), wherein the score (e.g., the probability score) indicates the likelihood of the subject having ovarian cancer.

[0011] In another aspect, there is provided a method of diagnosing ovarian cancer in a subject, the method comprising: (a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein (i) the first panel comprises hsa-miR- 320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a- 3p, and hsa-miR-424-5p, and (ii) the second panel comprises hsa-miR-320c, hsa-miR-200c- 3p, and hsa-miR-122-5p; (b) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125)in the sample, wherein the amounts obtained in steps (a) and (b) are indicative of the subject having ovarian cancer.

[0012] In a further aspect, there is provided a method for determining the likelihood of a subject having ovarian cancer, the method comprising: (a) determining the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel in a sample collected from the subject, wherein (i) the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39- 3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and (ii) the second panel comprises hsa-miR- 320c, hsa-miR-200c-3p, and hsa-miR-122-5p; (b) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125) in the sample; (c) comparing the amounts obtained in steps (a) and (b) to a mathematical model (e.g., a statistical model) based on corresponding amounts determined for samples collected from individuals with or without ovarian cancer; and (d) obtaining a score (e.g., a probability score) from the mathematical model (e.g., the statistical model), wherein the score (e.g., the probability score) indicates the likelihood of the subject having ovarian cancer.

[0013] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject for surgery, e.g., for assigning the subject to an appropriate surgeon, such as a specialist oncology surgeon. The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer. The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either being at risk for recurrence of ovarian cancer or not being at risk for recurrence of ovarian cancer. The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either being responsive to therapy or as not being responsive to therapy.

[0014] Step (b) may comprise determining the amounts of two or more of proteins in the sample. Step (b) may further comprise determining the amounts of one, two, or three proteins selected from Apo-A1, PreAlb, and TRF. Step (b) may be performed using one or more antibodies capable of specifically binding to the one or more proteins. Step (b) may beperformed using ELISA. The amounts of the one or more proteins may be determined by proteomic profiling. The amounts of the one or more proteins may be determined by protein bead array.

[0015] The method may comprise determining the amount of two or more microRNAs of the first panel and / or two or more microRNAs of the second panel. The method may comprise determining the amount of two, three, four, five, six or seven microRNAs of the first panel and / or two or three microRNAs of the second panel.

[0016] The microRNAs selected from hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR- 39-3p, hsa-miR-122-5p, and / or hsa-miR-19a-3p may be correlated negatively with ovarian cancer in the mathematical model (e.g., the statistical model). The microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, hsa-miR-770-5p, and / or hsa-miR-424-5p may be correlated positively with ovarian cancer in the mathematical model (e.g., the statistical model).

[0017] In some instances, step (a) is performed by contacting of the sample with one or more probes, wherein the one or more probes are capable of specifically hybridizing to the one or more microRNAs or cDNAs thereof. Step (a) may be performed using a nucleic acid detection assay. The assay may be selected from the group consisting of microarray, next generation sequencing, RT-PCR, and RT-qPCR. Step (a) may be performed using RT-qPCR.

[0018] The one or more probes may comprise a detectable label. The one or more probes may be embedded in one or more hydrogel particles. The one or more microRNAs hybridized to the one or more probes may be amplified by PCR.

[0019] In further aspects, there is provided a method for identifying a subject presenting with an ovarian mass for surgery, the method comprising: (a) obtaining a sample collected from the subject; (b) determining the amounts of one or more microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample; (c) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125) in the sample; (d) comparing the amounts obtained in steps (b) and (c) to a statistical model; and (e) obtaining a score from the statistical model; wherein the score obtained in step (e) identifies the subject for surgery.

[0020] There is also provided a method for identifying a subject with a risk for developing ovarian cancer, the method comprising: (a) obtaining a sample collected from thesubject; (b) determining the amounts of one or more microRNAs selected from hsa-miR- 320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample; (c) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125) in the sample; (d) comparing the amounts obtained in steps (b) and (c) to a statistical model; and (e) obtaining a score from the statistical model; wherein the score obtained in step (e) identifies the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer.

[0021] There is also provided a method for monitoring a subject having been treated for ovarian cancer for recurrence of the ovarian cancer, the method comprising: (a) obtaining a sample collected from the subject; (b) determining the amounts of one or more microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample; (c) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125) in the sample; (d) comparing the amounts obtained in steps (b) and (c) to a mathematical model (e.g., a statistical model); and (e) obtaining a score (e.g., a probability score) from the mathematical model (e.g., the statistical model); wherein the score (e.g., the probability score) obtained in step (e) identifies the subject as either being at risk for recurrence of the ovarian cancer or not being at risk for recurrence of the ovarian cancer.

[0022] Further provided is a method for monitoring response to therapy in a subject who is being treated for ovarian cancer, the method comprising: (a) obtaining a sample collected from the subject; (b) determining the amounts of one or more microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample; (c) determining the amounts of one or more proteins associated with ovarian cancer (e.g., one or more proteins selected from B2M, HE4, FSH and CA125) in the sample; (d) comparing the amounts obtained in steps (b) and (c) to a mathematical model (e.g., a statistical model); and (e) obtaining a score (e.g., a probability score) from the mathematical model (e.g., the statistical model); wherein the score (e.g., the probability score) obtained in step (e) identifies the subject as either being responsive to therapy or as not being responsive to therapy.

[0023] Step (b) may comprise determining the amounts of two or more of microRNAs hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample. Step (b) may further comprises determining the amounts of one or more of, or eachof, microRNAs hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p hsa-miR-122-5p, hsa- miR-19a-3p, and hsa-miR-424-5p in the sample.

[0024] Step (c) may comprise determining the amounts of two or more of proteins B2M, HE4, FSH and CA125 in the sample. Step (c) may further comprises determining the amounts of one or more of, or each of, proteins ApoA1, PreAlb, and TRD in the sample.

[0025] Step (b) may comprise determining the amounts of each of the microRNAs hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p in the sample. Step (b) may comprise determining the amounts of each of the microRNAs hsa-miR-320e, hsa-miR- 320c, hsa-miR-200c-3p, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39- 3p hsa-miR-122-5p, hsa-miR-19a-3p, and hsa-miR-424-5p in the sample.

[0026] Step (c) may comprise determining the amounts of each of the proteins B2M, HE4, FSH and CA125 in the sample. Step (c) may comprise determining the amounts of each of the proteins B2M, HE4, FSH, CA125, ApoA1, PreAlb, and TRD in the sample.

[0027] The mathematical model (e.g., the statistical model) may take into account metadata about subject, such as the age of the subject or whether the subject has entered menopause.

[0028] Step (b) may be performed by contacting of the sample with one or more probes, wherein the one or more probes are capable of specifically hybridizing to the one or more microRNAs or cDNAs thereof.

[0029] Step (b) may be performed using a nucleic acid detection assay, optionally wherein the assay is selected from the group consisting of microarray, RT-PCR, and RT- qPCR. Step (b) may be performed using RT-qPCR.

[0030] The one or more probes may comprise a detectable label. The one or more probes may be embedded in one or more hydrogel particles.

[0031] The one or more microRNAs hybridized to the one or more probes may be amplified by PCR.

[0032] Step (c) may be performed using one or more antibodies capable of specifically binding to the one or more proteins. Step (c) may be performed using ELISA.

[0033] The mathematical model (e.g., the statistical model) may take into account the age of the subject. The mathematical model (e.g., the statistical model) may take into account whether the subject has entered menopause.

[0034] The method may be capable of detecting early-stage ovarian cancer. The method may be capable of detecting serous ovarian cancer. The method may be capable of detecting early-stage serous ovarian cancer.

[0035] In some instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 80% or more and a specificity of about 80% or more. In other instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 90% or more and a specificity of about 70% or more. In other instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 95% or more and a specificity of about 60% or more.

[0036] The mathematical model (e.g., the statistical model) may have an accuracy of about 85% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. The mathematical model (e.g., the statistical model) may have an accuracy of about 90% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer.

[0037] The sample may be selected from blood, lymph node tissue, or ovary tissue. The sample may be a blood sample, e.g., plasma, serum, and whole blood.

[0038] The sample may be tissue from a lymph node, e.g., a draining lymph node for an ovary.

[0039] The sample may be ovary tissue, e.g., wherein the ovary is suspected to contain a tumor.

[0040] The subject may be at the risk of developing ovarian cancer. The subject may have a higher level of CA-125 antigen in the blood than a diagnostic threshold level, optionally wherein the diagnostic threshold level is 35 U / mL. The subject may have at least one mutation in the BRCA1 or BRCA2 gene. The subject may be at least 55 years of age.

[0041] The mathematical model (e.g., the statistical model) may comprise one or more models selected from a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, supportvector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest.

[0042] The mathematical model (e.g., the statistical model) may be or may comprise a linear classification model.

[0043] In another aspect, there is provided a method of treating a subject suffering from ovarian cancer by surgery, wherein the subject has been identified for surgery using a method (e.g., a diagnostic method) described herein.

[0044] In a further aspect, there is provided a kit for use with a diagnostic method described herein.

[0045] In yet a further aspect, there is provided a kit comprising: (a) one or more probes capable of specifically hybridizing to one or more microRNA selected from hsa-miR- 320e, hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-770-5p or a cDNA thereof; and (b) one or more antibodies or antigen-binding fragments capable of specifically binding to one or more proteins selected from B2M, HE4, FSH and CA125.

[0046] In another aspect, there is provided a kit comprising: (a) one or more probes capable of specifically hybridizing to one or more microRNA selected from a first panel and one or more probes capable of specifically hybridizing to one or more microRNAs selected from a second panel, wherein (i) the first panel comprise hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and (ii) the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-122-5p; and (b) one or more antibodies or antigen-binding fragments capable of specifically binding to one or more proteins selected from B2M, HE4, FSH, CA125, Apo-A1, PreAlb, and TRF.

[0047] The one or more probes may comprise a detectable label. The one or more probes may be embedded in one or more hydrogel particles.

[0048] The kit may further comprise one or more reagents for detection and / or amplification of the one or more microRNAs.

[0049] The kit further may comprise one or more reagents for detecting the one or more proteins bound to the one or more antibodies or antigen-binding fragments.

[0050] In a further aspect, there is provided a computer program comprising computer program code configured to cause one or more physical computing devices toperform a method of analysis of sample results for a subject when the code is run on the one or more physical computing devices, the method of analysis comprising: (a) receiving first data indicative of amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel in a sample collected from the subject, wherein (i) the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa- miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and (ii) the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR-122-5p; (b) optionally receiving second data indicative of amounts of one or more proteins selected from B2M, HE4, FSH, CA125, Apo-A1, PreAlb, and TRF in the sample; (c) obtaining a mathematical model (e.g., a statistical model); (d) comparing the first data and optionally the second data to the mathematical model (e.g., the statistical model) to generate a score (e.g., a probability score); and (e) outputting the score (e.g., the probability score) from the mathematical model (e.g., the statistical model).

[0051] The first data may be indicative of the amounts of two or more microRNAs of the first panel and / or two or more microRNAs of the second panel.

[0052] The first data may be indicative of the amounts of two, three, four, five, six or seven microRNAs of the first panel and / or two or three microRNAs of the second panel.

[0053] The second data may be indicative of the amounts of two or more of the proteins.

[0054] The second data may be further indicative of the amounts of one, two, or three proteins selected from Apo-A1, PreAlb, and TRF.

[0055] The mathematical model (e.g., the statistical model) may characterise a joint distribution of the score with respect to the first data and the second data. The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject for surgery.

[0056] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer.

[0057] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either beingat risk for recurrence of the ovarian cancer or not being at risk for recurrence of the ovarian cancer.

[0058] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for identifying the subject as either being responsive to therapy or as not being responsive to therapy.

[0059] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for monitoring the response of the subject to the treatment of ovarian cancer.

[0060] The mathematical model (e.g., the statistical model) may be configured such that the score (e.g., the probability score) is suitable for determining the likelihood of the subject being responsive to ovarian cancer therapy.

[0061] The microRNAs selected from hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR- 39-3p, hsa-miR-122-5p, and / or hsa-miR-19a-3p may be correlated negatively with ovarian cancer in the statistical model. The microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, hsa-miR-770-5p, and / or hsa-miR-424-5p may be correlated positively with ovarian cancer in the statistical model.

[0062] The mathematical model (e.g., the statistical model) may take into account the age of the subject. The mathematical model (e.g., the statistical model) may take into account whether the subject has entered menopause.

[0063] The method may be capable of detecting early-stage ovarian cancer. The method may be capable of detecting serous ovarian cancer. The method may be capable of detecting early-stage serous ovarian cancer.

[0064] In some instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 80% or more and a specificity of about 80% or more. In other instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 90% or more and a specificity of about 70% or more. In other instances, the mathematical model (e.g., the statistical model) has a sensitivity of about 95% or more and a specificity of about 60% or more.

[0065] The mathematical model (e.g., the statistical model) may have an accuracy of about 85% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. The mathematical model (e.g., the statistical model) mayhave an accuracy of about 90% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer.

[0066] The mathematical model (e.g., the statistical model) may comprise one or more models selected from a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, support vector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest.

[0067] The mathematical model (e.g., the statistical model) may be or may comprise a linear classification model.

[0068] The computer program may be stored on a computer-readable storage medium, optionally a non-transitory computer-readable storage medium. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Further description, by way of example, is provided with reference to the following drawings.

[0070] FIG.1 shows a feature correlation matrix, indicating that the test proteins and test miRNAs described herein are largely linearly independent.

[0071] FIG. 2 shows that test proteins and test miRNAs described herein and metadata such as age and menopause status have a high correlation to ovarian cancer individually.

[0072] FIG. 3A shows a tSNE (t-distributed Stochastic Neighbor Embedding) plot that uses an miRNA feature set described herein to separate ovarian cancer cases and controls from the Aspira Women’s Health validation dataset.

[0073] FIG. 3B shows a tSNE plot that uses protein and metadata feature sets described herein to separate ovarian cancer cases and controls from the Aspira Women’s Health validation dataset.

[0074] FIG. 3C shows a tSNE plot that uses miRNA, protein and metadata feature sets described herein to separate ovarian cancer cases and controls from the Aspira Women’s Health validation dataset. Linear separation is improved compared to the separations achieved using miRNA feature set alone (FIG.3A) or protein and metadata feature sets alone (FIG.3B).

[0075] FIG. 3D shows a tSNE plot that uses an miRNA feature set described herein to separate ovarian cancer cases and controls from the Brigham and Women’s Hospital validation dataset.

[0076] FIG. 3E shows a tSNE plot that uses protein and metadata feature sets described herein to separate ovarian cancer cases and controls from the Brigham and Women’s Hospital validation dataset.

[0077] FIG. 3F shows a tSNE plot that uses miRNA, protein and metadata feature sets described herein to separate ovarian cancer cases and controls from the Brigham and Women’s Hospital validation dataset. Linear separation is improved compared to the separations achieved using miRNA feature set alone (FIG. 3D) or protein and metadata feature sets alone (FIG.3E).

[0078] FIG. 4A shows receiver operating characteristic (ROC) curves using the Aspira Women’s Health validation dataset, i.e., the internal dataset.

[0079] FIG. 4B shows ROC curves using the Brigham and Women’s Hospital validation dataset, i.e., the external dataset.

[0080] FIG.5A shows a bar graph demonstrating the accuracy of the combined model in identifying early-stage ovarian cancers, using the Aspira Women’s Health validation dataset, i.e., the internal dataset.

[0081] FIG.5B shows a bar graph demonstrating the accuracy of the combined model in identifying serous v. non serous ovarian cancers, using the Aspira Women’s Health validation dataset, i.e., the internal dataset.

[0082] FIG.5C shows a bar graph demonstrating the accuracy of the combined model in identifying early-stage ovarian cancers, using the Brigham and Women’s Hospital validation dataset, i.e., the external dataset.

[0083] FIG.5D shows a bar graph demonstrating the accuracy of the combined model in identifying serous v. non serous ovarian cancers, using the Brigham and Women’s Hospital validation dataset, i.e., the external dataset.

[0084] FIG.6 schematically illustrates an exemplary computer system configured to perform statistical analysis according to an embodiment.DETAILED DESCRIPTION

[0085] In order for the following description to be more readily understood, certain terms are first defined below. Additional definitions for the following terms and other terms are set forth throughout the specification.

[0086] Unless otherwise required by context, singular terms shall include pluralities and plural terms shall include the singular. Thus, the singular forms “a,” “an” and “the” include plural referents unless the context clearly dictates otherwise. As such, the terms “a” (or “an”), “one or more”, and “at least one” can be used interchangeably herein.

[0087] The words “have” and “comprise,” or variations such as “has,” “having,” “comprises,” or “comprising,” will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.

[0088] Unless specifically stated or obvious from context, as used herein, the term “or” is understood to be inclusive and covers both “or” and “and”. Furthermore, “and / or” is to be taken as specific disclosure of each of the two specified features or components with or without the other. Thus, the term “and / or” as used in a phrase such as “A and / or B” herein is intended to include “A and B”, “A or B”, “A” (alone), and “B” (alone). Likewise, the term "and / or" as used in a phrase such as “A, B, and / or C” is intended to include “A and / or B and / or C” and to thus encompass each of the following aspects: A, B, and C; A, B, or C; A or C; A or B; B or C; A and C; A and B; B and C; A (alone); B (alone); and C (alone).

[0089] It is understood that wherever aspects are described herein with the language “comprising”, otherwise analogous aspects described in terms of “consisting of” and / or “consisting essentially of” are also provided. In other words, if an aspect comprises A, B and C, aspects consisting essentially of A, B and C are also contemplated, as are aspects consisting of A, B and C.

[0090] The term “about” refers to an interval of accuracy that a person ordinarily skilled in the art will understand to still ensure the technical effect of the feature in question. The term indicates a deviation from the indicated numerical value of ±10%,±5%, or ±1%.

[0091] Unless otherwise defined herein, technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs and as commonly used in the art to which this application belongs. In case of conflict, the present specification, including definitions, will control.

[0092] Generally, nomenclature used in connection with, and techniques of, cell and tissue culture, molecular biology, virology, immunology, microbiology, genetics, analytical chemistry, synthetic organic chemistry, medicinal and pharmaceutical chemistry, and protein and nucleic acid chemistry and hybridization described herein are those well-known and commonly used in the art. Enzymatic reactions and purification techniques are performed according to manufacturer’s specifications, as commonly accomplished in the art or as described herein..

[0093] The term “probe” refers to a molecule or complex that is used to determine the presence or absence and / or amount of a microRNA in a sample (e.g. a blood sample). The probe may comprise a nucleic acid moiety (e.g., DNA, modified DNA, or modified RNA) that is capable of specifically hybridizing to the microRNA or a complementary DNA (cDNA) thereof. The probe may comprise a sequence of at least 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 contiguous nucleotides identical or complementary to the microRNA. The probe may further comprise a detectable label that is conjugated, covalently or non- covalently, to the nucleic acid moiety. Exemplary detectable labels include without limitation a fluorophore, a small molecule (e.g., a small molecule of the avidin family), an enzyme, an antibody or antibody fragment, or a nucleic acid sequence not present in the subject in a form that is linked to the microRNA (e.g., a barcode sequence). Accordingly, the probe may be a fluorophore-labeled nucleic acid having a nucleotide sequence that is complementary to a nucleotide sequence of a microRNA.

[0094] The term “specifically hybridizing” refers to non-covalent interaction between a first nucleic acid molecule (e.g., a nucleic acid probe having a certain nucleotide sequence) and a second nucleic acid molecule (e.g., a microRNA having a nucleotide sequence complementary to that of the nucleic acid probe). Hybridization conditions have been described in the art and are known to one of skill in the art. The condition for detecting the hybridization may be a suitable condition of a nucleic acid detection assay (e.g., microarray, RT-PCR, or RT-qPCR). The likelihood of hybridization between two nucleic acids correlates with the nucleotide sequence complementary between the two nucleic acids.

[0095] The term “RT-PCR” refers to reverse transcription polymerase chain reaction, a process for amplifying RNA. RNA molecules are reverse transcribed to complementary DNA (cDNA) using reverse transcriptase and then using PCR to amplify the resulting cDNA.

[0096] The term “RT-qPCR” refers to reverse transcription quantitative polymerase chain reaction, a variant of RT-PCR in which amplification of cDNA during the RT-PCR process is quantitatively detected in real time using a probe that detects amplified target DNA. For example, self-quenching nucleic acid probes may be added to the reaction mixture. The self-quenching nucleic acid probes only fluoresce when they bind a target sequence. As each cycle of PCR is completed, the self-quenching probes bind to the amplified DNA, unquench and fluoresce with exposure to a light excitation source. As DNA is amplified, increased probe and target binding results in increased fluorescence of the self- quenching nucleic acid probe. Detection of the fluorescing probes after each amplification cycle allows real-time measurement of the amplification process, as increasing amounts of the nucleic acid probe bind with amplified target DNA and fluoresce. An intercalating dye probe may be added to the reaction mixture that fluoresces upon interaction with double- stranded nucleic acids. The increase in dye fluorescence during the amplification process allows the measurement of DNA amplification in real-time, as increasing amounts of the dye probe intercalate with the increasing amounts of target DNA being amplified.

[0097] The term "next-generation sequencing" refers to high-throughput parallel sequencing of short fragments of single-stranded nucleic acids attached to slides or beads, such as techniques by ILLUMINA®, ROCHE (454sequencing), ION TORRENT®, or THERMO FISHER®. The incorporation of individual nucleotides onto single-stranded nucleic acids may be detected optically (via fluorescence of incorporated nucleotides) or by detection of hydrogen ions released during nucleotide incorporation (e.g. ion semiconductor sequencing).

[0098] The term “ovarian cancer”, as used herein, refers to a group of malignancies affecting the ovary, that have developed from epithelial cells, sex cord-stromal cells (e.g. granulosa, theca, and hilus cells), or germ cells (e.g. oocytes). About 60% of ovarian tumors are of epithelial origin and account for 90% of ovarian cancers (see Karst et al., “Ovarian Cancer Pathogenesis: A Model in Evolution,” Journal of Oncology, vol. 2010, Article ID 932371, 13 pages, 2010. doi:10.1155 / 2010 / 932371, incorporated herein by reference). Such epithelium-derived ovarian carcinomas are heterogeneous in character, with differences in tumor morphology, clinical symptoms, and genetic alterations. The World Health Organization (WHO) lists eight different tumor histologies, including serous, endometrioid, mucinous, clear cell, transitional cell, squamous cell, mixed epithelial, and undifferentiatied.Tumors of each of these subtypes may be classified as benign (having low malignant potential and / or indolence), malignant, or borderline, as well as low-grade (Type I) or high-grade (Type II). Ovarian cancers

[0099] Existing methods of detecting ovarian cancer perform poorly at correctly identifying early-stage ovarian cancers as ovarian cancers. These methods are also ineffective at correctly identifying ovarian cancers of serous subtype. Early diagnosis of ovarian cancer is important for improving patient outcomes, e.g., survival. Development of methods which have high sensitivity and accuracy are particularly important in the identification of serous ovarian cancer, the most deadly form of ovarian cancer.

[0100] Type I cancers (such as endometrioid and mucinous carcinomas) are of low histological grade and exhibit low-grade nuclei with infrequent mitotic figures (see Vang et al., Adv Anat Pathol. 2009 Sep; 16(5): 267–282, incorporated herein by reference). Type I ovarian serous cancers often have mutations of the KRAS, BRAF or ERBB2 genes, and lack TP53 mutations. Type I tumors are usually indolent and have better patient outcomes, although they can develop into Type II malignancies. In contrast, Type II ovarian serous cancers (such as serous carcinoma) are considered a higher histological grade, having high- grade nuclei and numerous mitotic figures, and are characterized by rapid tumor development. Genetically, Type II ovarian serous tumors usually include TP53 mutation and lack mutations of KRAS, BRAF, or ERBB2.

[0101] Ovarian tumors may also be surgically staged to indicate malignancy. Stage I indicates confinement to the ovary. Stage IA indicates that the cancer is completely inside one ovary, and stage IB indicates that the cancer is completely inside both ovaries. Stage IC indicates that the cancer is in both ovaries and one of the following: some cancer is on the surface of an ovary; cancer cells are present in fluid inside the abdomen (fluid may be taken during surgery); or that an ovary ruptures before or during surgery. Stage IC1 indicates that the tumor capsule broke during surgery to cause a surgical spill of cancer cells into the abdomen and / or pelvis. Stage IC2 indicates that at least one ovary or fallopian tube, or the tumor capsule ruptured before surgery. Stage IC3 indicates detection of cancer cells in fluid or washings from the abdomen.

[0102] Stage II indicates tumor extension beyond the ovary to other pelvic structures (e.g., fallopian tube or uterus). Stage IIA indicates that the cancer has grown into the fallopian tubes or into the uterus. Stage IIB indicates that the cancer has grown into other pelvic tissues, such as the bladder or rectum. Stage IIC indicates that the cancer has grown into pelvic tissues and that cancer cells are present in fluid taken from inside the abdomen.

[0103] Stage III indicates metastasis to the peritoneum and / or regional lymph nodes. Stage IIIA indicates that cancer growths are visible microscopically in tissue from the abdominal lining. Stage IIIA1 indicates that the cancer is in one or both ovaries or fallopian tubes and may have spread or grown into the pelvis or nearby organs. Stage IIIA1(i) indicates that the areas of cancer spread in the lymph nodes are 10mm across or smaller. Stage IIIA1(ii) indicates that the areas of cancer spread in the lymph nodes is great than 10mm across. Stage IIIA2 indicates that the cancer is in one or both ovaries or fallopian tubes and may have spread or grown into the pelvis or nearby organs, and that microscopic examination of biopsies from the upper lining of the abdomen show small cancer deposits. Stage IIIB indicates that tumor growths 2cm or smaller are visible on the abdominal lining or the capsule of the liver or spleen. Stage IIIC indicates that tumor growths 2cm or larger are visible on the abdominal lining, and / or that cancer is detected in the lymph nodes of the upper abdomen, groin, or behind the uterus.

[0104] Stage IV indicates metastasis of the ovarian cancer past the peritoneum. Stage IVA indicates that the cancer has caused pleural effusion (build up fluid in the lining of the lungs. Stage IVB indicates that the cancer has spread inside the liver or spleen, to lymph nodes in the groin, or outside the abdomen and / or to other organs (e.g., the lungs).

[0105] As used herein, the term “treating” or “treatment” refers to relieving, reducing, or alleviating at least one symptom in a subject or effecting a delay of progression of a disease. For example, treatment can be the diminishment of one or several symptoms of a disorder or complete eradication of a disorder, such as cancer. The term “treat” also denotes to arrest and / or reduce the risk of worsening a disease, or prevention of at least one symptom associated with or caused by the state, disease or disorder being prevented. For example, treatments may relieve, reduce or alleviate at least one symptom of ovarian cancer.

[0106] The course of treatment of ovarian cancer is often similar for many patients in order to account for the heterogeneity of response by the different cancer types. Such treatments include cytoreductive surgery (e.g., debulking to remove the tumor, salpingo-oophorectomy, hysterectomy, lymphadenectomy, omentectomy), followed by platinum- based chemotherapy (e.g., cisplatin or carboplatin, alone or in combination with one or more other chemotherapeutic agents, such as taxanes). In some instances, other chemotherapeutic agents may be used if the ovarian cancer is resistant to platinum-based drugs either alone or in combination, such as liposomal doxorubicin, paclitaxel, docetaxel, nab-paclitaxel, gemcitabine, etoposide, pemetrexed, cyclophosphamide, topotecan, vinorelbine, or irinotecan.

[0107] Other treatments of ovarian cancer used in place of or in addition to surgery and chemotherapy may include administration of a targeted antibody, such as bevacizumab or olaparib. In some instances, radiation therapy may be used to treat ovarian cancer, such as external beam radiation therapy or brachytherapy (implantation of a radioactive material near the affected site) in order to target the affected area. In some instances, hormone therapy may be used to treat ovarian cancer, such as administration of luteinizing-hormone-releasing hormone (LHRH) agonists (e.g., goserelin and / or leuprolide), tamoxifen, or aromatase inhibitors (e.g., letrozole, anastrozole, and exemestane). miRNA detection

[0108] MicroRNAs (miRNAs) are detectable in bodily fluids including but not limited to blood, lymph, saliva, and urine, as well as biopsied tissue. MiRNAs circulate in the blood and lymph. The profile of miRNAs circulating in the blood and lymph differs between healthy subjects and subjects with cancer, e.g., ovarian cancer. Therefore, a sample of blood or lymphatic fluid can be obtained from a subject, the miRNAs can be profiled, and this information can be compared to values established in a statistical model to give a score (e.g., a probability score) that can be used to determine whether or not a subject has ovarian cancer.

[0109] Detection of the miRNAs is performed by contacting the sample with one or more probes, wherein the one or more probes are capable of specifically hybridizing to the one or more microRNAs or cDNAs thereof. Each microRNA-specific probe may be coupled (e.g., covalently linked) to a substrate such as a particle. Each microRNA-specific probe may be embedded in a hydrogel particle (e.g., a polyethylene glycol particle).

[0110] Each microRNA-specific probe may comprise universal adapter sequences that flank the microRNA-specific sequence. The adapter sequences may specifically hybridize to universal adapters.

[0111] Detection of the miRNAs may comprise (i) contacting the sample with a plurality of nucleic acid probes, each nucleic acid probe comprising a capturing sequence for binding a target microRNA and an adjacent adapter sequence for binding a universal adapter; (ii) incubating the plurality of probes and the sample, in the presence of one or more universal adapters, under conditions that permit binding of both an individual target nucleic acid and an individual universal adapter to a same individual nucleic acid probe; and (iii) carrying out a reaction (e.g., a ligation reaction) that allows coupling of the individual universal adapter to the individual target microRNA when hybridized to the same individual nucleic acid probe.

[0112] The one or more target microRNAs may be amplified in a reaction mixture comprising a detectable entity such that the amplified one or more target nucleic acids are labeled with the detectable entity. The amplification product may be contacted with the plurality of nucleic acid probes such that the amplified target microRNAs are re-hybridized to the plurality of the nucleic acid probes.

[0113] The universal adapters are labeled with detectable moieties or other labeling groups to facilitate detection. Primers may be added to amplify the adapter-ligated microRNAs by PCR. The primers may be labeled with detectable moieties or other labeling groups to facilitate detection. The detectable moiety may be a fluorescent moiety. Detecting a signal generated by the detectable moieties associated with the re-hybridized amplified target microRNA(s) may indicate the amount of the one or more target microRNAs in the sample. The amounts of the one or more microRNAs may be determined by quantifying the amplified adapter-ligated microRNAs, e.g., using fluorescent detection.

[0114] Capturing microRNAs of interest by specifically hybridizing them to microRNA-specific probes is advantageous because it removes PCR inhibitors such as heparin that can reduce detection sensitivity. Employing microRNA-specific probes linked to particles (e.g., embedded in hydrogel particles) facilitates detection of microRNAs (e.g., circulating microRNAs) directly from biological fluids such as whole blood, plasma, or serum.

[0115] Various methods are available in the art that allow for the detection and profiling of miRNAs. Suitable methods for determining the amounts of one or more microRNAs are described in, e.g., WO 2011 / 156434 and WO 2014 / 17657, both of which being hereby incorporated herein by reference in their entirety.

[0116] Detection of the miRNAs may be performed by reverse-transcribing the microRNA molecules in the sample, thereby obtaining a cDNA sample; and sequencing the cDNA sample. Step (a) may further comprise amplifying the DNA molecules in the cDNA sample before sequencing the cDNA sample. The sequencing may be performed by next- generation sequencing. miRNAs associated with ovarian cancer

[0117] miRNAs associated with ovarian cancer include, but are not limited to, those shown in Table 1. Table 1. Ovarian cancer-associated miRNAs miRNAhsa-mir-15b-5p hsa-mir-34b-3p e amounts of eight or more, nine or more, ten or more, eleven or more, ore e o o e miRNAs from Table 1 may be determined in a sample collected from a subject.

[0119] The amounts of ten or more miRNAs (e.g., the first ten miRNAs listed in Table 1) may be determined in a sample collected from a subject. The amounts of thirteen or more miRNAs (e.g., the first thirteen miRNAs listed in Table 1) may be determined in a sample collected from a subject.

[0120] The amounts of ten or more miRNAs listed in Table 3 (provided below in connection with Example 1 may be determined in a sample collected from a subject. The amounts of thirteen or more miRNAs listed in Table 3 may be determined in a sample collected from a subject.

[0121] The amounts of one or more miRNAs selected from a first panel of miRNAs, listed in Table 1A, may be determined in a sample collected from a subject.

[0122] Table 1A. First Panel, Ovarian cancer-associated miRNAs miRNA

[0123] The amounts of one or more miRNAs selected from a second panel of miRNAs, listed in Table 1B, may be determined in a sample collected from a subject.

[0124] Table 1B. Second Panel, Ovarian cancer-associated miRNAmiRNA hsa-mir-320c e amounts of two or more miRNAs selected from the first panel of miRNAss e n a e A and / or the amounts of two or more miRNAs selected from the second panel of miRNAs listed in Table 1B may be determined in a sample collected from a subject.

[0126] The amounts of two, three, four, five, six or seven miRNAs selected from the first panel of miRNAs listed in Table 1A and / or the amounts of two or three miRNAs selected from the second panel of miRNAs listed in Table 1B may be determined in a sample collected from a subject. Proteins associated with ovarian cancer

[0127] Proteins are detectable in bodily fluids including but not limited to blood, lymph, saliva, and urine, as well as biopsied tissue. Proteins circulate in the blood and lymph. The profile of proteins circulating in the blood and lymph differs between healthy subjects and subjects with cancer, e.g., ovarian cancer. Therefore, a sample of blood or lymphatic fluid can be obtained from a subject, the proteins can be profiled and this information, optionally along with the information obtained from profiling miRNAs, can be compared to values established in a mathematical model (e.g., a statistical model) to give a score (e.g., a probability score) that can be used to determine whether or not a subject has ovarian cancer. Proteins may also be detectable in tissues, such as in lymph node tissue or ovarian tissue.

[0128] Proteins associated with ovarian cancer include, but are not limited to, those shown in Table 2 (Nolen and Lokshin (2012) Future Oncol. Jan;8(1):55-71). Table 2. Ovarian cancer-associated proteins Biological function Biomarkers Role in cancer h, ne(CCL13), MIG (CXCL9), MIP-1δ (CCL15), MIP3a (CCL20), MIP-4 h, h,GLP-1, amylin, glucagon, peptide YY, follistatin, hepcidin or s h,

[0129] Particularly suitable proteins for use with the methods disclosed herein include those that are easily detectable in a blood, plasma, or serum sample collected from a subject. These include, but are not limited to, CA125, soluble IR-2α, Prostatin, ApoA-1, transthyretin (TTR), inter-alpha-trypsin inhibitor heavy chain H4 (ITIH4), apolipoproteinA1, leptin, prolactin, IGF-2, Osteopontin, amyloid A1, catabolic fragments of complement factors, elastin microfibril interfacer 2 (EMILIN2), VWF, PEBP1, β-hemoglobin, transferrin, Hp, Hp-1 precursor, Hp-α subunit, fibrinopeptide-A, HE4, and claudine-4. Exemplary proteins Exemplary proteins associated with ovarian cancer for use with the diagnostic methods described herein include one or more of the following: Follicle-stimulating hormone (FSH)

[0130] FSH is a 128 amino acid protein (NCBI Accession number NP 000501). Antibodies to FSH can be purchased from, for example, Santa Cruz Biotechnology, Inc. (e.g., Catalog Number SC-57149) (www.scbt.com, Santa Cruz, CA). FSH is typically upregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer. Human Epididymis Protein 4 (HE4)

[0131] HE4 is a 124 amino acid protein (NCBI Accession number NP 006094). Antibodies to HE4 can be purchased from, for example, Santa Cruz Biotechnology, Inc. (Catalog Number sc-27570) (www.scbt.com, Santa Cruz, CA). HE4 is typically upregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer. Cancer Antigen 125 (CA125)

[0132] CA 125 is a 22152 amino acid protein (Swiss-Prot Accession number Q8WXI7). Antibodies to CA 125 can be purchased from, for example, Santa Cruz Biotechnology, Inc. (Catalog Number sc-52095) (www.scbt.com, Santa Cruz, CA). CA 125 is typically upregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer. Prealbumin (PreAlb)

[0133] Prealbumin is a 147 amino acid protein (Swiss Prot Accession number P02766). Antibodies to prealbumin can be purchased from, for example, Santa Cruz Biotechnology, Inc. (Catalog Number sc- 13098) (www.scbt.com, Santa Cruz, CA). Prealbumin is typically downregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer.Transferrin (TRF)

[0134] Transferrin is a 698 amino acid protein (UniProtKB / TrEMBL Accession number Q06AH7). Antibodies to transferrin can be purchased from, for example, Santa Cruz Biotechnology, Inc. (Catalog Number sc-52256) (www.scbt.com, Santa Cruz, CA). Transferrin is typically downregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer. Apolipoprotein A1 (ApoA1)

[0135] Apo Al is a 267 amino acid protein (Swiss Prot Accession number P02647). Antibodies to Apolipoprotein Al can be made using any method well known in the art, or can be purchased from, for example, Santa Cruz Biotechnology, Inc. (Catalog Number SC- 130503) (www.scbt.com, Santa Cruz, CA). Apo A1 is typically downregulated in subjects with ovarian cancer as compared to subjects that do not have ovarian cancer. b-2 microglobulin (B2M)

[0136] The mature form of p2-microglobulin is a 99 amino acid protein derived from a 119 amino acid precursor (GI: 179318; SwissProt Accession No. P61769). Antibodies to b-2 microglobulin can be purchased from, e.g., Abcam (catalog AB759) Protein detection

[0137] The one or more proteins can be measured by immunoassay. Immunoassays typically utilize one or more antibodies, or one or more antigen-binding fragment, that is capable of specifically binding to the one or more proteins of interest to detect the presence or amount of the one or more proteins in a sample. Antibodies can be produced by methods well known in the art, e.g., by immunizing animals with the one or more proteins.

[0138] Suitable immunoassays include, for example, Western blot, sandwich immunoassays including ELISA and other enzyme immunoassays, fluorescence-based or chemiluminescence-based immunoassays. Other forms of immunoassay include magnetic immunoassay, and radioimmunoassay.

[0139] Immunoassays can be carried out on solid substrates (e.g., chips, beads, microfluidic platforms, membranes) or on any other forms that supports binding of the antibody to the protein of interest and subsequent detection. A single protein may be detected at a time or a multiplex format may be used. Multiplex immunoanalysis may involve planarmicroarrays (protein chips or arrays) and bead-based microarrays (protein bead or suspension arrays).

[0140] Proteomic profiling can also be used to identify and quantify one or more proteins of interest in a sample. Surface-enhanced laser desorption ionization–time of flight mass spectrometry (SELDI-TOF MS) has been used in screening and early detection of many diseases. SELDI-TOF-MS detects one or more proteins selectively adsorbed onto the surface of a protein chip array after non-specifically bound proteins are washed off by stringent buffers, and has been shown to be sensitive, rapid and reliable. A capture reagent for the one or more proteins of interest (e.g., one or more antibodies or antigen-binding fragments capable of specifically binding to the one or more proteins of interest) are attached to the surface of the array. The captured proteins are then detected by mass spectrometry.

[0141] A sample may be analyzed by means of a protein array (e.g., a protein bead array). Such arrays are useful in high-throughput low-cost screens to identify alterations in the amounts of the one or more proteins of interest. Protein arrays generally comprise solid substrates to which a capture reagent (also called an adsorbent or affinity reagent, e.g., an antibody or antigen-binding fragment that is capable of specifically binding to the protein of interest) is attached. Suitable solid supports include membranes (e.g., membranes composed of nitrocellulose, paper, or other material), polymer-based films (e.g., polystyrene), beads, or glass slides. Mathematical model

[0142] In a mathematical model (e.g., a statistical model as described herein), a data matrix comprising miRNA, protein and / or metadata variables is mapped onto cancer labels (e.g., “cancerous”, “malignant” and “benign”). Suitable models may include, e.g., neural networks or linear classifiers.

[0143] The cancer labels may comprise information on cancer grade. The cancer grade may comprise stage I, stage II, stage III, and / or stage IV ovarian cancers. Stage I may comprise stage IA, stage IB, and / or stage IC ovarian cancers. Stage II may comprise stage IIA, stage IIB, and / or stage IIC ovarian cancers. Stage III may comprise stage IIIA, stage IIIB, and / or stage IIIC ovarian cancers. Stage IIIA may comprise stage IIIA1 and / or stage IIIA2 ovarian cancers. Stage IIIA1 may comprise IIIA(i) and / or stage IIIA(ii) ovarian cancers. Stage IV may comprise stage IVA and / or stage IVB ovarian cancers.

[0144] The cancer labels may comprise information on cancer subtype. The cancer subtype may be categorized according to serous or non-serous phenotype. The information may be binary. Building a model for miRNA in ovarian cancer

[0145] MiRNA profiles differ between healthy subjects and subjects with ovarian cancer. One or more miRNAs can be selected from these profiles to be used in a statistical model, e.g. a neural network or linear classifier, to predict ovarian cancer. To identify the optimal miRNAs to be used in the statistical model, forward regression is used to rank one or more miRNAs according to their power to predict ovarian cancer. This can be a positive prediction, i.e., the miRNA correlates with ovarian cancer, or a negative prediction, i.e. the miRNA correlates with the absence of ovarian cancer. The one or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir-424-5p, hsa-mir-19a-3p, hsa-mir- 106b-5p, hsa-mir-148b-3p, hsa-mir-320d, hsa-mir-199a-5p, hsa-mir-29c-3p, hsa-mir-423- 5p, hsa-mir-1237-3p, hsa-mir-15b-5p, hsa-mir-34b-3p, and / or hsa-mir-497-5p.

[0146] The forward regression technique, which is used in certain examples provided herein for selecting the miRNAs, is described in Abhimanyu Das and David Kempe, arXiv preprint arXiv:1102.3975, 2011 and Alan Miller CRC Press 2002. The first miRNA is selected according to its power to predict ovarian cancer. According to one example, the first miRNA selected in the forward regression is hsa-mir-103a-3p. The statistical model may comprise a first miRNA, wherein the first miRNA is hsa-mir-103a-3p.

[0147] Furthermore, miRNAs can be then subsequently selected by the forward regression method, for inclusion in the statistical model, according to their power to predict ovarian cancer and also according to their lack of correlation with each miRNA that was selected before. Two or more microRNAs may be selected using forward regression. The two or more miRNAs may comprise hsa-mir-103a-3p and hsa-mir-320c. The statistical model may comprise a first miRNA and a second miRNA, wherein the first miRNA is hsa-mir- 103a-3p and the second miRNA is hsa-mir-320c. The statistical model may comprise a first miRNA selected from the first panel of miRNAs, shown in Table 1A above and a second miRNA selected from the second panel of miRNAs, shown in Table 1B, above. The first miRNA selected from the first panel of miRNAs of Table 1A may be hsa-mir-103a-3p andthe second miRNA selected from the second panel of miRNAs of Table 1B may be hsa-mir- 320c.

[0148] Three or more microRNAs may be selected using forward regression. The three or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, and hsa-mir-200c-3p. The statistical model may comprise a first miRNA, a second miRNA, and a third miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, and the third miRNA is hsa-mir-200c-3p. The statistical model may comprise a first miRNA selected from the first panel of miRNAs, shown in Table 1A above and a second and third miRNA selected from the second panel of miRNAs, shown in Table 1B, above. The first miRNA selected from the first panel of miRNAs of Table 1A may be hsa-mir-103a-3p, the second miRNA selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, and the third miRNA selected from the second panel of miRNAs of Table 1B may be hsa-mir- 200c-3p.

[0149] Four or more microRNAs may be selected using forward regression. The four or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, and hsa- mir-195-5p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, and a fourth miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, and the fourth miRNA is hsa- mir-195-5p. The statistical model may comprise two miRNAs selected from the first panel of miRNAs, shown in Table 1A above and two miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The two miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a-3p and hsa-mir-195-5p, and the two miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c and hsa-mir-200c-3p.

[0150] Five or more microRNAs may be selected using forward regression. The five or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir- 195-5p, and hsa-mir-320e. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, and a fifth miRNA wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, and the fifth miRNA is hsa-mir-320e. The statistical model may comprise three miRNAs selected from the first panel of miRNAs, shown in Table 1A above and two miRNAs selected from the second panel of miRNAs, shown in Table 1B,above. The three miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a-3p, hsa-mir-195-5p, and hsa-mir-320e and the two miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c and hsa-mir-200c-3p.

[0151] Six or more microRNAs may be selected using forward regression. The six or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir- 195-5p, hsa-mir-320e, and hsa-mir-122-5p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, and a sixth miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, and the sixth miRNA is hsa-mir-122-5p. The statistical model may comprise three miRNAs selected from the first panel of miRNAs, shown in Table 1A above and three miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The three miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a- 3p, hsa-mir-195-5p, and hsa-mir-320e and the three miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, hsa-mir-200c-3p and hsa-mir-122-5p.

[0152] Seven or more microRNAs may be selected using forward regression. The seven or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, and hsa-mir-770-5p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, and a seventh miRNA, wherein the first miRNA is hsa-mir-103a- 3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir- 122-5p, and the seventh miRNA is hsa-mir-770-5p. The statistical model may comprise four miRNAs selected from the first panel of miRNAs, shown in Table 1A above and three miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The four miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a- 3p, hsa-mir-195-5p, hsa-mir-320e and hsa-mir-770-5p and the three miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, hsa-mir-200c-3p and hsa- mir-122-5p.

[0153] Eight or more microRNAs may be selected using forward regression. The eight or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa- mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, and cel-mir-39-3p. Thestatistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, and an eighth miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, and the eighth miRNA is cel-mir-39-3p. The statistical model may comprise five miRNAs selected from the first panel of miRNAs, shown in Table 1A above and three miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The five miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-770-5p and cel-mir-39-3p and the three miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, hsa-mir-200c-3p and hsa-mir-122-5p.

[0154] Nine or more microRNAs may be selected using forward regression. The nine or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir- 195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, and hsa-mir-424-5p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, an eighth miRNA, and a ninth miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, the eighth miRNA is cel-mir-39-3p, and the ninth miRNA is hsa-mir-424-5p. The statistical model may comprise six miRNAs selected from the first panel of miRNAs, shown in Table 1A above and three miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The six miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-770-5p, cel-mir-39-3p and hsa- mir-424-5p and the three miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, hsa-mir-200c-3p and hsa-mir-122-5p.

[0155] Ten or more microRNAs may be selected using forward regression. The ten or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir- 195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir-424-5p, and hsa-mir-19a-3p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, an eighth miRNA, a ninth miRNA, and a tenth miRNA, wherein the first miRNA is hsa-mir-103a-3p,the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, the eighth miRNA is cel-mir-39-3p, the ninth miRNA is hsa-mir-424-5p, and the tenth miRNA is hsa-mir-19a-3p. The statistical model may comprise seven miRNAs selected from the first panel of miRNAs, shown in Table 1A above and three miRNAs selected from the second panel of miRNAs, shown in Table 1B, above. The seven miRNAs selected from the first panel of miRNAs, shown in Table 1A may be hsa-mir-103a- 3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir-424-5p and hsa- mir-19a-3p and the three miRNAs selected from the second panel of miRNAs of Table 1B may be hsa-mir-320c, hsa-mir-200c-3p and hsa-mir-122-5p.

[0156] The amounts / concentrations of the miRNAs identified using forward regression are concatenated with the amounts / concentrations of protein and the metadata variables to form a new data matrix. (Each row of this data matrix corresponds to one patient sample.) The data matrix is used to train a statistical model, e.g., neural network classification model or linear classification model, to map the data matrix to the cancer labels. According to some examples, the classification is binary. That is, the model is trained to classify samples as either (i) cancerous or (ii) benign (non-cancerous). However, this is non-limiting. In other examples the statistical model may be trained to classify into a greater number of classes – for instance, classifying into predictions of specific types of tumor. In an example, one or more miRNAs identified by forward regression for use in the statistical model comprise(s) one or more miRNAs selected from a first panel (Table 1A, above) and one or more miRNAs selected from a second panel (Table 1B, above), wherein the first panel comprises hsa-mir- 103a-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir-424-5p, and hsa-mir-19a-3p and the second panel comprises hsa-mir-320c, hsa-mir-200c-3p, and hsa- mir-122-5p. The amounts of miRNAs hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa- mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir-424-5p, and hsa-mir-19a-3p in a sample obtained from a subject at risk of developing, or presenting with, an ovarian mass may be compared to a statistical model. The statistical model is trained in an earlier training phase based on corresponding amounts determined for samples collected from individuals with or without ovarian cancer. Furthermore, a probability score may be obtained using the statistical model, wherein the probability score indicates the likelihood of the subject having ovarian cancer.

[0157] Furthermore, miRNAs can be identified using forward regression for use in the statistical model, e.g,. neural network or linear classifier.

[0158] Eleven or more microRNAs may be selected using forward regression. The eleven or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir- 424-5p, hsa-mir-19a-3p, and hsa-mir-106b-5p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, an eighth miRNA, a ninth miRNA, a tenth miRNA, and an eleventh miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, the eighth miRNA is cel-mir-39-3p, the ninth miRNA is hsa-mir-424-5p, the tenth miRNA is hsa-mir-19a-3p, and the eleventh miRNA is hsa-mir-106b-5p.

[0159] Twelve or more microRNAs may be selected using forward regression. The twelve or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir- 424-5p, hsa-mir-19a-3p, hsa-mir-106b-5p, and hsa-mir-148b-3p. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, an eighth miRNA, a ninth miRNA, a tenth miRNA, an eleventh miRNA, and a twelfth miRNA, wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa-mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, the eighth miRNA is cel-mir-39-3p, the ninth miRNA is hsa-mir-424-5p, the tenth miRNA is hsa-mir-19a-3p, the eleventh miRNA is hsa-mir-106b- 5p, and the twelfth miRNA is hsa-mir-148b-3p.

[0160] Thirteen or more microRNAs may be selected using forward regression. The thirteen or more miRNAs may comprise hsa-mir-103a-3p, hsa-mir-320c, hsa-mir-200c-3p, hsa-mir-195-5p, hsa-mir-320e, hsa-mir-122-5p, hsa-mir-770-5p, cel-mir-39-3p, hsa-mir- 424-5p, hsa-mir-19a-3p, hsa-mir-106b-5p, hsa-mir-148b-3p, and hsa-mir-320d. The statistical model may comprise a first miRNA, a second miRNA, a third miRNA, a fourth miRNA, a fifth miRNA, a sixth miRNA, a seventh miRNA, an eighth miRNA, a ninth miRNA, a tenth miRNA, an eleventh miRNA, a twelfth miRNA, and a thirteenth miRNA,wherein the first miRNA is hsa-mir-103a-3p, the second miRNA is hsa-mir-320c, the third miRNA is hsa-mir-200c-3p, the fourth miRNA is hsa-mir-195-5p, the fifth miRNA is hsa- mir-320e, the sixth miRNA is hsa-mir-122-5p, the seventh miRNA is hsa-mir-770-5p, the eighth miRNA is cel-mir-39-3p, the ninth miRNA is hsa-mir-424-5p, the tenth miRNA is hsa-mir-19a-3p, the eleventh miRNA is hsa-mir-106b-5p, the twelfth miRNA is hsa-mir- 148b-3p, and the thirteenth miRNA is hsa-mir-320d. Computational implementation

[0161] Examples may, in part, be implemented on a computer. The computer may receive first data indicative of the amounts of one or more microRNAs in a sample collected from the subject, and second data indicative of the amounts of one or more proteins in a sample collected from the subject. The computer may also receive meta-data, such as the age of the subject and / or whether the subject has entered menopause.

[0162] The computer may obtain a statistical model, compare the first data and second data with the statistical model and provide, as an output of the statistical model, a score. The statistical model may characterize a joint distribution of the score with respect to the first data and the second data. The statistical model may take into account the meta-data. The statistical model may comprise one or models selected from: a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, support vector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest.

[0163] The score may be suitable for at least one of: identifying the subject for surgery; identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer; identifying the subject as either being at risk for recurrence of the ovarian cancer or not being at risk for recurrence of the ovarian cancer; and identifying the subject as either being responsive to therapy or as not being responsive to therapy.

[0164] FIG. 6 of the accompanying drawings schematically illustrates an exemplary computer system 100 upon which a computer program provided herein may run. The exemplary computer system 100 comprises a computer-readable storage medium 102, a memory 104, a processor 106 and one or more interfaces 108, which are all linked together over one or more communication busses 110. The exemplary computer system 100 may takethe form of a conventional computer system, such as, for example, a desktop computer, a personal computer, a laptop, a tablet, a smart phone, a server, a mainframe computer, and so on.

[0165] The computer-readable storage medium 102 and / or the memory 104 may store one or more computer programs (or software or code) and / or data (including but not limited to: the first data, the second data, and the statistical model). The computer programs stored in the computer-readable storage medium 102 and / or the memory 104 may include computer programs that, when executed by the processor 106, cause the processor 106 to carry out a method provided herein. The computer-readable storage medium 102 and / or the memory 104 may be a non-transitory computer readable storage medium. The computer- readable storage medium 102 and / or the memory 104 may be one or more memory chips or memory blocks implemented within the processor, magnetic media such as hard disk or floppy disks, or optical media such as for example DVD and the data variants thereof, CD.

[0166] The processor 106 may be any data processing unit suitable for executing one or more computer readable program instructions, such as those belonging to computer programs stored in the computer-readable storage medium 102 and / or the memory 104. The processor 106 may include one or more of: general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs), application specific integrated circuits (ASIC), gate level circuits and processors based on multi-core processor architecture, as non-limiting examples. As part of the execution of one or more computer-readable program instructions, the processor 106 may store data to and / or read data from the computer- readable storage medium 102 and / or the memory 104. The processor 106 may comprise a single data processing unit or multiple data processing units operating in parallel or in cooperation with each other. The processor 106 may, as part of the execution of one or more computer readable program instructions, store data to and / or read data from the computer- readable storage medium 102 and / or the memory 104.

[0167] The one or more interfaces 108 may comprise a network interface enabling the computer system 100 to communicate with other computer systems across a network. In some examples, the computer system 100 may obtain the first data, the second data, and / or the statistical model via the network. The network may be any kind of network suitable for transmitting or communicating data from one computer system to another. For example, the network could comprise one or more of a local area network, a wide area network, ametropolitan area network, the internet, a wireless communications network, and so on. The computer system 100 may communicate with other computer systems over the network via any suitable communication mechanism / protocol. The processor 106 may communicate with the network interface via the one or more communication busses 110 to cause the network interface to send data and / or commands to another computer system over the network. Similarly, the one or more communication busses 110 enable the processor 106 to operate on data and / or commands received by the computer system 100 via the network interface from other computer systems over the network.

[0168] The interface 108 may alternatively or additionally comprise a user input interface and / or a user output interface. The user input interface may be arranged to receive input from a user, or operator, of the system 100. The user may provide this input via one or more user input devices (not shown), such as a mouse (or other pointing device, track-ball or keyboard. The user output interface may be arranged to provide a graphical / visual output to a user or operator of the system 100 on a display (or monitor or screen) (not shown). The processor 106 may instruct the user output interface to form an image / video signal which causes the display to show a desired graphical output. The display may be touch-sensitive enabling the user to provide an input by touching or pressing the display.

[0169] The interface 108 may alternatively or additionally comprise an interface to a measurement system, for measuring amounts of one or more microRNAs in a sample and / or amounts of one or more proteins in a sample.

[0170] In general, the various examples may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. For example, some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device, although these are not limiting examples. While various aspects described herein may be illustrated and described as block diagrams, it is well understood that these blocks may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.

[0171] A single processor or other unit may fulfil the functions of several items recited in the claims. The functions may be performed in a single integrated electronic device, or the functions may be distributed across different discrete devices. For example, somefunctions may be performed by a remote service accessed via a wired or wireless network connection. A computer program may be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0172] It will be readily apparent to those skilled in the art that other suitable modifications and adaptations of the methods, kits and computer programs described herein may be made using suitable equivalents without departing from the scope disclosed herein. Having now described certain aspects in detail, the same will be more clearly understood by reference to the following examples, which are included for purposes of illustration only and are not intended to be limiting. NUMBERED EMBODIMENTS

[0173] The following numbered embodiments are provided: 1. A method of diagnosing ovarian cancer in a subject and / or identifying a subject having ovarian cancer for surgery, the method comprising: (a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR- 122-5p; (b) comparing the amounts obtained in step (a) to a statistical model based on corresponding amounts determined for samples collected from individuals with and / or without ovarian cancer; and (c) obtaining a score; wherein the score indicates the likelihood of the subject having ovarian cancer. 2. A method of diagnosing ovarian cancer in a subject and / or identifying a subject having ovarian cancer for surgery, the method comprising:(a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR- 122-5p; (b) determining the amounts of one or more proteins selected from B2M, HE4, FSH and CA125 in the sample; wherein the amounts obtained in steps (a) and (b) are indicative of the subject having ovarian cancer. 3. A method for determining the likelihood of a subject having ovarian cancer, the method comprising: (a) determining the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel in a sample collected from the subject, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR- 122-5p; (b) determining the amounts of one or more proteins selected from B2M, HE4, FSH and CA125 in the sample; (c) comparing the amounts obtained in steps (a) and (b) to a statistical model based on corresponding amounts determined for samples collected from individuals with or without ovarian cancer; and (d) obtaining a score from the statistical model; wherein the score indicates the likelihood of the subject having ovarian cancer. 4. The method any one of embodiments 1-3, wherein the statistical model is configured such that the score is suitable for at least one of: 1. identifying the subject for surgery;2. identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer; 3. identifying the subject as either being at risk for recurrence of ovarian cancer or not being at risk for recurrence of ovarian cancer; and 4. identifying the subject as either being responsive to therapy or as not being responsive to therapy. 5. The method of any one of embodiments 2-4, wherein step (b) comprises determining the amounts of two or more of proteins in the sample. 6. The method of any one of embodiments 2-5, wherein step (b) further comprises determining the amounts of one, two, or three proteins selected from Apo-A1, PreAlb, and TRF. 7. The method of any one of embodiments 2-6, wherein step (b) is performed using one or more antibodies capable of specifically binding to the one or more proteins. 8. The method of embodiment 6, wherein step (b) is performed using ELISA. 9. The method of any one of embodiments 2-6, wherein the amounts of the one or more proteins are determined by proteomic profiling. 10. The method of any one of embodiments 2-6, wherein the amounts of the one or more proteins are determined by protein bead array. 11. The method of any one of the preceding embodiments, wherein the method comprises determining the amount of two or more microRNAs of the first panel and / or two or more microRNAs of the second panel. 12. The method of any one of the preceding embodiments, wherein the method comprises determining the amount of two, three, four, five, six or seven microRNAs of the first panel and / or two or three microRNAs of the second panel.13. The method of any one of the preceding embodiments, wherein the microRNAs selected from hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-122-5p, and / or hsa-miR-19a-3p are correlated negatively with ovarian cancer in the statistical model. 14. The method of any one of the preceding embodiments, wherein the microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, hsa-miR-770-5p, and / or hsa- miR-424-5p are correlated positively with ovarian cancer in the statistical model. 15. The method of any one of the preceding embodiments, wherein the statistical model takes into account the age of the subject. 16. The method of any one of the preceding embodiments, wherein the statistical model takes into account whether the subject has entered menopause. 17. The method of any one of the preceding embodiments, wherein the method is capable of detecting early-stage ovarian cancer. 18. The method of any one of the preceding embodiments, wherein the method is capable of detecting serous ovarian cancer. 19. The method of any one of the preceding embodiments, wherein the method is capable of detecting early-stage serous ovarian cancer. 20. The method of any one of embodiments 3-19, wherein the statistical model has a sensitivity of about 80% or more and a specificity of about 80% or more. 21. The method of any one of embodiments 3-20, wherein the statistical model has a sensitivity of about 90% or more and a specificity of about 70% or more. 22. The method of any one of embodiments 3-21, wherein the statistical model has a sensitivity of about 95% or more and a specificity of about 60% or more.23. The method of any one of embodiments 3-22, wherein the statistical model has an accuracy of about 85% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. 24. The method of any one of embodiments 3-23, wherein the statistical model has an accuracy of about 90% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. 25. The method of any one of the preceding embodiments, wherein step (a) is performed by contacting of the sample with one or more probes, wherein the one or more probes are capable of specifically hybridizing to the one or more microRNAs or cDNAs thereof. 26. The method of any one of the preceding embodiments, wherein step (a) is performed using a nucleic acid detection assay, 27. The method of embodiment 26, wherein the assay is selected from the group consisting of microarray, next generation sequencing, RT-PCR, and RT-qPCR. 28. The method of embodiment 27, wherein step (a) is performed using RT-qPCR. 29. The method of any one of embodiments 25-28, wherein the one or more probes comprises a detectable label. 30. The method of any one of embodiments 25-29, wherein the one or more probes are embedded in one or more hydrogel particles. 31. The method of any one of embodiments 25-30, wherein the one or more microRNAs hybridized to the one or more probes is amplified by PCR. 32. The method of any one of the preceding embodiments, wherein the sample is selected from a sample of blood, lymph node tissue, or ovary tissue.33. The method of embodiment 32, wherein the sample is a blood sample, optionally wherein the blood sample is selected from the group consisting of plasma, serum, and whole blood. 34. The method of embodiment 32, wherein the sample is lymph node tissue, optionally wherein the lymph node is a draining lymph node for an ovary. 35. The method of embodiment 32, wherein the sample is ovary tissue, optionally wherein the ovary tissue is suspected to contain a tumor. 36. The method of any one of the preceding embodiments, wherein the subject is at risk of developing ovarian cancer. 37. The method of embodiment 36, wherein the subject has a higher level of CA-125 antigen in the blood than a diagnostic threshold level, optionally wherein the diagnostic threshold level is 35 U / mL. 38. The method of embodiment 36 or 37, wherein the subject has at least one mutation in the BRCA1 or BRCA2 gene. 39. The method of any one of embodiments 36-38, wherein the subject is at least 55 years of age. 40. The method of any one of the preceding embodiments, wherein the statistical model comprises one or more models selected from a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, support vector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest. 41. The method of embodiment 40, wherein the statistical model is or comprises a linear classification model.42. A method of treating a subject suffering from ovarian cancer by surgery, wherein the subject has been identified for surgery using a method in accordance with any one of the preceding embodiments. 43. A kit for use with the method of any one of embodiments 1-41. 44. A kit comprising: (a) one or more probes capable of specifically hybridizing to one or more microRNA selected from a first panel and one or more probes capable of specifically hybridizing to one or more microRNAs selected from a second panel, wherein: i. the first panel comprise hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR- 122-5p; and (b) one or more antibodies or antigen-binding fragments capable of specifically binding to one or more proteins selected from B2M, HE4, FSH, CA125, Apo-A1, PreAlb, and TRF. 45. The kit of embodiment 44, wherein the one or more probes comprise a detectable label. 46. The kit of embodiment 44 or 45, wherein the one or more probes are embedded in one or more hydrogel particles. 47. The kit of any one of embodiments 44-46, further comprising one or more reagents for detection and / or amplification of the one or more microRNAs. 48. The kit of any one of embodiments 44-47, further comprising one or more reagents for detecting the one or more proteins bound to the one or more antibodies or antigen-binding fragments. 49. A computer program comprising computer program code configured to cause one or more physical computing devices to perform a method of analysis of sample results for asubject when the code is run on the one or more physical computing devices, the method of analysis comprising: (a) receiving first data indicative of amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, in a sample collected from the subject, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424-5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa-miR- 122-5p; (b) optionally receiving second data indicative of amounts of one or more proteins selected from B2M, HE4, FSH, CA125, Apo-A1, PreAlb, and TRF in the sample; (c) obtaining a statistical model; (d) comparing the first data and optionally the second data to the statistical model to generate a score; and (e) outputting the score from the statistical model. 50. The computer program of embodiment 49, wherein the first data is indicative of the amounts of two or more microRNAs of the first panel and / or two or more microRNAs of the second panel. 51. The computer program of embodiment 49 or 50, wherein the first data is indicative of the amounts of two, three, four, five, six or seven microRNAs of the first panel and / or two or three microRNAs of the second panel. 52. The computer program of any one of embodiments 49-51, wherein the second data is indicative of the amounts of two or more of the proteins. 53. The computer program of any one of embodiments 49-52, wherein second data is further indicative of the amounts of one, two, or three proteins selected from Apo-A1, PreAlb, and TRF.54. The computer program of any one of embodiments 49-53, wherein the statistical model characterizes a joint distribution of the score with respect to the first data and the second data. 55. The computer program of any one of embodiments 49-54, wherein the statistical model is configured such that the score is suitable for at least one of: 1. identifying the subject for surgery; 2. identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer; 3. identifying the subject as either being at risk for recurrence of the ovarian cancer or not being at risk for recurrence of the ovarian cancer; and 4. identifying the subject as either being responsive to therapy or as not being responsive to therapy. 56. The computer program of any one of embodiments 49-55, wherein the microRNAs selected from hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-122-5p, and / or hsa-miR-19a-3p are correlated negatively with ovarian cancer in the statistical model. 57. The computer program of any one of embodiments 49-56, wherein the microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR-200c-3p, hsa-miR-770-5p, and / or hsa- miR-424-5p are correlated positively with ovarian cancer in the statistical model. 58. The computer program of any one of embodiments 49-57, wherein the statistical model takes into account the age of the subject. 59. The computer program of any one of embodiments 49-58, wherein the statistical model takes into account whether the subject has entered menopause. 60. The computer program of any one of embodiments 49-59, wherein the method is capable of detecting early-stage ovarian cancer. 61. The computer program of any one of embodiments 49-60, wherein the method is capable of detecting serous ovarian cancer.62. The computer program of any one of embodiments 49-61, wherein the method is capable of detecting early-stage serous ovarian cancer. 63. The computer program of any one of embodiments 49-62, wherein the statistical model has a sensitivity of about 80% or more and a specificity of about 80% or more. 64. The computer program of any one of embodiments 49-63, wherein the statistical model has a sensitivity of about 90% or more and a specificity of about 70% or more. 65. The computer program of any one of embodiments 49-64, wherein the statistical model has a sensitivity of about 95% or more and a specificity of about 60% or more. 66. The computer program of any one of embodiments 49-65, wherein the statistical model has an accuracy of about 85% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. 67. The computer program of any one of embodiments 49-66, wherein the statistical model has an accuracy of about 90% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer. 68. The computer program of any one of embodiments 49-67, wherein the statistical model comprises one or more models selected from a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, support vector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest. 69. The computer program of embodiment 68, wherein the statistical model is or comprises a linear classification model.EXAMPLES

[0174] Although methods and materials similar or equivalent to those described herein can be used, suitable methods and materials are described below. The following examples are included for illustrative purposes only and are not intended to be limiting. Example 1. Circulating miRNA

[0175] 204 microRNAs (miRNAs) were identified that can be detected in circulation. They are listed in Table 3 below: Table 3: Circulating miRNAs SEQ ID #Name Nucleotide sequence1 hsa-mir-185-5p UGGAGAGAAAGGCAGUUCCUGA GSEQ ID # Name Nucleotide sequence 30 hsa-mir-151a-3p CUAGACUGAAGCUCCUUGAGGSEQ ID # Name Nucleotide sequence 71 hsa-mir-589-5p UGAGAACCACGUCUGCUCUGAGSEQ ID # Name Nucleotide sequence 112 hsa-mir-361-5p UUAUCAGAAUCUCCAGGGGUACSEQ ID # Name Nucleotide sequence 153 hsa-mir-382-5p GAAGUUGUUCGUGGUGGAUUCGSEQ ID # Name Nucleotide sequence 194 hsa-mir-22-5p AGUUCUUCAGUGGCAAGCUUUAExample 2. Protein, miRNA, and metadata are largely linearly independent.

[0176] This example demonstrates that protein, miRNA, and metadata are largely linearly independent, and also have a high correlation to ovarian cancer individually.

[0177] The feature correlation matrix shown in FIG. 1 has an approximate block diagonal structure. The miRNAs are self-correlated, and so are the proteins and metadata, but the miRNA and proteins do not exhibit a high degree of inter-correlation.

[0178] FIG.2 shows that many of the miRNAs (e.g., miR-200c-3p) and proteins (e.g., HE4) have high correlation to the cancer labels. Thus, the miRNAs and proteins are well- suited to train a linear classification model.

[0179] This example demonstrates miRNA and protein are linearly independent, each showing high correlation to ovarian cancer individually. They can be used in combination to train a linear classification model. Example 3. Modified triage test statistical model

[0180] This example explains how the original triage test statistical model was modified to include the assessment of miRNAs.

[0181] The original triage test is based on seven proteins (CA125, ApoA1, Pre- albumin, beta 2 microglobulin, TRF, HE4, and FSH) and metadata such as age and menopausal status. A panel of 180 miRNA were introduced into the test. These 180 miRNAs were a subset of the 204 miRNAs identified in Example 1 and are listed in Table 4.Table 4. miRNAs introduced into triage test hsa-mir-185-5p hsa-mir-320c hsa-mir-34b-3p hsa-mir-19a-3p hsa-mir-320e hsa-mir-494-3p hsa-mir-33a-5p hsa-mir-134-5phsa-mir-18a-5p hsa-mir-423-3p hsa-mir-99a-5p hsa-mir-4732-5p hsa-mir-106b-3p hsa-mir-181d-5p hsa-mir-532-3p hsa-mir-140-5p

[0182] The 180 miRNA listed in Table 4 were selected based on signal-to-noise ratio. To reduce the miRNA dimension, a forward regression algorithm was used (Miller, Alan. Subset selection in regression. chapman and hall / CRC, 2002; Das, Abhimanyu, and David Kempe. Submodular meets spectral: Greedy algorithms for subset selection, sparse approximation and dictionary selection. arXiv preprint arXiv:1102.3975 (2011).

[0183] ^^ ൌ ^ ^^^, … , ^^^^ ∈ ℝ^ൈ^denotes the miRNA data, and ^^ ∈ ^0,1^^denotes the cancer labels, where 0 indicates a control, and 1, a case. The forward regression algorithm can be summarized as follows: 1) Initialize ^^ ൌ ∅ and the number of features, ^^. 2) for every integer ^^ ∈ ^0, ^^ െ 1^ do 3) Let ^^^ೕbe a variable maximizing ^^^ଶ,ௌೕ∪ ^ೕ, and set ^^^ା^ൌ ^^^∪ ^^^. 4) Output ^^^In the algorithm above, ^^^ଶ,ௌdenotes the squared multiple correlation

[0184] The output, ^^^, is a set of ^^ indices identifying the optimal miRNA features within the index range 1 ^ ^^ ^ ^^. The forward regression algorithm starts by selecting the feature, ^^^, most highly correlated to ^^, and then adds more features in an iterative sense to maximize linear separation of the control and case group.

[0185] Forward regression has advantages when compared to oblivious, correlation- based feature selection (i.e., selecting the top ^^ most highly correlated features), and methods such as fold-change, commonly applied in the literature on cancer prediction, as forward regression also factors in correlation between the predictors, and is thus less likely to add redundant features.

[0186] Once ^^ ^ 180 miRNA features were chosen by forward regression, the miRNA, protein, and metadata variables were combined to form a new data matrix ^^ ∈ ℝ^ൈ^^ାଽ^. ^^ was formed simply by concatenating the selected ^^ miRNA features, 7 proteins, and 2 metadata variables. A softmax function (Gao, Bolin, and Lacra Pavel. On the properties of the softmax function with application in game theory and reinforcement learning. arXiv preprint arXiv:1704.00805 (2017)) was trained to map ^^ to the cancer labels ^^: where ^^ ∈ ^0,1^ is the class(i.e., one row of ^^), and the ^ ^^^, ^^^^ are weights and biases to be trained. Here ^^ denotes the class label assigned to ^^. The class with the highest probability ^^ was then chosen for membership.

[0187] The data was split into three groups, one for training the classification model (sample size ^^ ൌ 468, with 191 cancer patients and 277 controls), one internal validation set (44 cancers and 56 controls) drawn from the same distribution / population as the training data, and one external validation set (51 cancers and 59 controls), which was collected independently of the training data and from a different population. All model fitting and hyperparameter selection was done using only the training data. Then, the internal and external validation sets were inputted into the trained model for evaluation. To choose ^^, i.e., the number of miRNA features, 10-fold cross-validation on the training data was used. Specifically, the ^^ ∈^1, … , 20^which maximized the AUC score after 10-fold validation was chosen. After 10-fold validation on the training set, ^^ ൌ 10 miRNA offered optimal AUC score.

[0188] Table 5 lists the 10 miRNAs selected by forward regression and the corresponding p-values. The p-values were calculated using a two-tailed t-test under the assumption of non-equal variance and sample size. Only the combined internal and external validation sets were used for this calculation. The training data were not included in the calculation to avoid spurious correlation and p-value, as the top ^^ miRNA were selected using the training data. Table 5: Panel of selected miRNAs miRNA p-value hsa-miR-320e 0.02

[0189] 5 of the 10 listed miRNA have insignificant p-values ^ ^^ ^ .05^ on the validation data. The miRNA panel has high dimension (180 miRNA), and thus spurious correlation is likely to occur, especially given the relatively small training sample size ^ ^^ ൌ468^. For example, cel-miR-39-3p was chosen as part of the ^^ ൌ 10 miRNA set usingforward regression. The corresponding p-value on the validation set is ^^ ൌ .59 ^ .05, which likely indicates a spurious relation was detected in the training data. This makes sense as cel- miR-39-3p has low expression in humans and was included in the panel for normalization and baseline. The remaining 5 miRNA listed in Table 5 yield significant p-values ( ^^ ^ .05) on the validation data.

[0190] Multifactorial intervention (MFI) analysis confirmed the selection of the 10 miRNAs. The result of this analysis is shown in Table 6. The 10 selected miRNAs were ranked first.Table 6. Ranking of miRNAs by MFI ranking miRNA 1 hsa-mir-103a-3p

[0191] The proteins and metadata variables are shown in Table 7. The p-values were calculated in the same way as for Table 3. There is no significant difference detected in age ^ ^^ ൌ 0.35^ and menopausal status ^ ^^ ൌ 0.06^ when comparing cancer and control patients. 6 of 7 proteins yielded significant p-values. Table 7: Panel of metadata and proteins Variable p-valueCA-125 1e-4 Apo-A1 1e-13 d to evaluate model.

[0192] The datasets shown in Table 8 were used in the modified triage test. The internal validation dataset refers to that provided by Aspira Women’s Health (AWH) and the external validation dataset refers to that provided by Brigham and Women’s Hospital (BWH). Table 8. Datasets used in the modified triage test. Aspira Women’s Health (AWH); Brigham and Women’s Hospital (BWH) Cohort n (total sample number) Controls Cases cts.The training set was provided by Aspira Women’s Health, comprising n=568 women undergoing preoperative evaluation for an adnexal mass (n=333 benign and n=235 malignant). An independent, external validation set (n=110) comprised study subjects enrolled in prospective collection protocols at Brigham and Women’s Hospital, including n=59 subjects with benign masses or healthy controls and n=51 ovarian cancer cases. Among the 51 cases in the validation set, n=20 were FIGO Stage I / II and n=31 were FIGO Stage III / IV. Histologies included n=34 serous and n=17 non-serous. All samples underwent miRNA profiling using a custom 179-miRNA panel optimized for serum analysis using the Fireplex® circulating miRNA assay (Abcam, Cambridge, MA). Proteins were measured using ELISA kits by Aspira. Seven proteins (CA125, ApoA1, Pre-albumin, beta 2 microglobulin, TRF, HE4, and FSH) were provided for each patient, as well as age and menopausal status.Example 5. The modified triage test statistical model is improved

[0194] This example demonstrates that the modified triage test statistical model is an improvement on the original triage test statistical model.

[0195] As described in Example 2, to reduce the dimensionality of the miRNA data, a forward regression algorithm was employed, which selects a subset of miRNA to optimize linear separation between cases and controls. The miRNA panel (hsa-miR-320e, hsa-miR- 320c, hsa-miR-200c-3p, hsa-miR-770-5p, hsa-miR-195-5p, hsa-miR-103a-3p, cel-miR-39- 3p hsa-miR-122-5p, hsa-miR-19a-3p, and hsa-miR-424-5p) was combined with the protein (CA125, ApoA1, Pre-albumin, beta 2 microglobulin, TRF, HE4, and FSH) and metadata (age, menopause status) to train a neural network to diagnose ovarian cancer. The datasets described in Example 4 were used.

[0196] TSNE plots of the validation datasets were produced, shown in FIG.3A to 3F. FIG.3A to 3C were produced using the internal (Aspira Women’s Health (AWH)) validation dataset. FIG. 3D to 3F were produced using the external (Brigham and Women’s Hospital (BWH)) validation dataset. These plots show the combined features offer the best linear separation between cases and controls, regardless of validation dataset.

[0197] Receiver operating characteristic (ROC) curves were produced to compare the effectiveness of models trained on miRNA + protein + metadata (joint) vs miRNA alone vs protein + metadata, assessed by the area under the curve (AUC). The ROC curves generated using the internal (Aspira Women’s Health (AWH)) validation dataset are shown in FIG.4A. The ROC curves generated using the external (Brigham and Women’s Hospital (BWH)) validation dataset are shown in FIG.4B.

[0198] The combined model offers the best AUC score on the internal (Aspira Women’s Health (AWH)) and external (Brigham and Women’s Hospital (BWH)) validation sets. The sensitivity and specificity scores are tabulated in Table 9. The combined features offer a highly sensitive model (while maintaining good specificity). In particular, on the external validation set (i.e., data collected independently from the training set), the combined model offers an AUC ൌ 0.95, which is 5% greater than the AUC offered by the next best performing model (i.e., the protein + metadata model). The AUC confidence interval of the combined model has width 0.09, with lower bound 0.89. The protein + metadata model AUCconfidence interval has width 0.14, with lower bound 0.81. Thus, there is greater confidence in the combined model when compared to the protein + metadata model.

[0199] Table 9 gives sensitivity and specificity scores for each model. The best scores are highlighted in bold. The cancer probability threshold used to calculate the sensitivity and specificity corresponds to the point on the training ROC curve closest to ^0,1^, as highlighted in FIG.4A for the combined model. It should be understood that this is non-limiting – a bias for either greater sensitivity (at the expense of some loss of specificity) or greater specificity (at the expense of some loss of sensitivity) could be selected. As the training ROC curve differs for each model (FIG.4A), the cancer probability threshold in this example is different for each model. The combined model offers the best specificity and sensitivity scores on the internal validation set, when compared to the miRNA and protein + metadata models.

[0200] On the external validation set, the combined model offers high sensitivity (.96), with .66 specificity. The protein + metadata model offers higher specificity (.78) at the cost of lower sensitivity (.82), when compared to the combined model.

[0201] The preferred result in this case is situational. For example, if one were to use the model to decide upon surgical interventions, a low false negative rate is desired. If the model were to be used for first-pass screening of ovarian cancer, then higher specificity would be optimal. The sensitivity and specificity scores presented are based on the cancer threshold which corresponds to the point on the training ROC closest to ^0,1^. Depending on the application, the threshold applied to the score / probability output by the statistical model can be adjusted based on the ROC to favor sensitivity or specificity. (In the present example, a sample with a score / probability greater than or equal to the threshold is classified as cancerous, and a sample with a score / probability less than or equal to the threshold is classified as benign.) Table 9. Sensitivity and specificity scores with cancer probability threshold corresponding to the point on the training ROC curve closest to (0, 1). Aspira Women’s Health (AWH); Brigham and Women’s Hospital (BWH) Feature set Specificity Sensitivity Specificity SensitivitymiRNA 0.75 0.80 0.49 0.90 Protein + 0.84 0.86 0.66 0.96cer was serous or non-serous. The data are shown in FIG. 5A to 5D. FIG. 5A and 5C show the high predictive value of the model when testing for stage, for the internal and external validation datasets, respectively. FIG.5B and 5D show the high predictive value of the model when identifying serous v. non serous ovarian cancer, for the internal and external validation datasets, respectively.

[0203] The classification accuracy scores for each model among serous and non- serous cancers are shown in Table 10. The serous / non-serous status was known for 16 out of 44 cancers ^36%^ in the internal validation group, and for 50 out of 51^98%^cancers in the external group. The combined model offered greater than or equal accuracy on serous tumors, when compared to the protein + metadata model and the microRNA model, on both the external and internal validation sets. The microRNA model offered the best accuracy among non-serous tumors on the external validation set, although the accuracy among serous tumors was 12% lower than the combined model.

[0204] The joint model offered the highest sensitivity among cancers which were serous and early-stage (89%) and among early-stage cancers overall (90%) when compared to miRNA alone (56% and 80%) or protein + metadata (67% and 65%). These results are shown in Table 10. Table 10. Brigham and Women’s Hospital dataset - accuracy scores broken down by cancer stage and serous / non-serous ovarian cancer (OC). Feature set Early OC Late OC Serous OC Non-serous Serous andProtein + 0.90 1 0.97 0.94 0.89 metadata +hly sensitive model is desired. The combined data model is most optimal in this regard.

[0206] This example demonstrates that the combined model, which uses a neural network trained with miRNA data, is a highly sensitive model. In particular, it has higher sensitivity than the model that was trained using protein and metadata only.

[0207] All publications and other reference materials referenced herein are hereby incorporated by reference in their entirety. Reference to these materials does not constitute an admission that any of these documents forms part of the common general knowledge in the art.

[0208] It should be understood that the particular embodiments described herein are given by way of illustration only, not limitation. Other features, objects, and advantages are apparent from the above detailed description, drawings and examples. Various changes and modifications will be apparent to those skilled in the art.

Claims

CLAIMS What is claimed is:

1. A method of diagnosing ovarian cancer in a subject and / or identifying a subject having ovarian cancer for surgery, the method comprising: (a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195- 5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424- 5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa- miR-122-5p; (b) comparing the amounts obtained in step (a) to a statistical model based on corresponding amounts determined for samples collected from individuals with and / or without ovarian cancer; and (c) obtaining a score; wherein the score indicates the likelihood of the subject having ovarian cancer.

2. A method of diagnosing ovarian cancer in a subject and / or identifying a subject having ovarian cancer for surgery, the method comprising: (a) determining in a sample collected from the subject the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195- 5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424- 5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa- miR-122-5p; (b) determining the amounts of one or more proteins selected from B2M, HE4, FSH and CA125 in the sample; wherein the amounts obtained in steps (a) and (b) are indicative of the subject having ovarian cancer.

3. A method for determining the likelihood of a subject having ovarian cancer, the method comprising: (a) determining the amounts of one or more microRNAs selected from a first panel and of one or more microRNAs selected from a second panel in a sample collected from the subject, wherein: i. the first panel comprises hsa-miR-320e, hsa-miR-770-5p, hsa-miR-195- 5p, hsa-miR-103a-3p, cel-miR-39-3p, hsa-miR-19a-3p, and hsa-miR-424- 5p, and ii. the second panel comprises hsa-miR-320c, hsa-miR-200c-3p, and hsa- miR-122-5p; (b) determining the amounts of one or more proteins selected from B2M, HE4, FSH and CA125 in the sample; (c) comparing the amounts obtained in steps (a) and (b) to a statistical model based on corresponding amounts determined for samples collected from individuals with or without ovarian cancer; and (d) obtaining a score from the statistical model; wherein the score indicates the likelihood of the subject having ovarian cancer.

4. The method of claim 1 or 3, wherein the statistical model is configured such that the score is suitable for at least one of:

1. identifying the subject for surgery; 2. identifying the subject as either being at risk for developing ovarian cancer or not being at risk for developing ovarian cancer; 3. identifying the subject as either being at risk for recurrence of ovarian cancer or not being at risk for recurrence of ovarian cancer; and 4. identifying the subject as either being responsive to therapy or as not being responsive to therapy.

5. The method of any one of claims 2-4, wherein step (b) comprises determining the amounts of two or more of proteins in the sample.

6. The method of any one of claims 2-5, wherein step (b) further comprises determining the amounts of one, two, or three proteins selected from Apo-A1, PreAlb, and TRF.

7. The method of any one of claims 2-6, wherein step (b) is performed using one or more antibodies capable of specifically binding to the one or more proteins, e.g., using ELISA.

8. The method of any one of claims 2-6, wherein the amounts of the one or more proteins are determined by proteomic profiling or protein bead array.

9. The method of any one of the preceding claims, wherein the method comprises determining the amount of two or more microRNAs of the first panel and / or two or more microRNAs of the second panel.

10. The method of any one of the preceding claims, wherein the method comprises determining the amount of two, three, four, five, six or seven microRNAs of the first panel and / or two or three microRNAs of the second panel.

11. The method of any one of the preceding claims, wherein (a) the microRNAs selected from hsa-miR-195-5p, hsa-miR-103a-3p, cel- miR-39-3p, hsa-miR-122-5p, and / or hsa-miR-19a-3p are correlated negatively with ovarian cancer in the statistical model; and / or (b) the microRNAs selected from hsa-miR-320e, hsa-miR-320c, hsa-miR- 200c-3p, hsa-miR-770-5p, and / or hsa-miR-424-5p are correlated positively with ovarian cancer in the statistical model.

12. The method of any one of the preceding claims, wherein the statistical model takes into account the age of the subject and / or whether the subject has entered menopause.

13. The method of any one of the preceding claims, wherein the method is capable of detecting early-stage ovarian cancer and / or serous ovarian cancer.

14. The method of any one of claims 3-13, wherein the statistical model has (a) a sensitivity of about 80% or more and a specificity of about 80% or more;(b) a sensitivity of about 90% or more and a specificity of about 70% or more; and / or (c) a sensitivity of about 95% or more and a specificity of about 60% or more.

15. The method of any one of claims 3-14, wherein the statistical model has (a) an accuracy of about 85% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer; and / or (b) an accuracy of about 90% or more of detecting early-stage ovarian cancer, serous ovarian cancer and / or early-stage serous ovarian cancer.

16. The method of any one of the preceding claims, wherein step (a) is performed by contacting of the sample with one or more probes, wherein the one or more probes are capable of specifically hybridizing to the one or more microRNAs or cDNAs thereof.

17. The method of any one of the preceding claims, wherein step (a) is performed using a nucleic acid detection assay, optionally wherein the assay is selected from the group consisting of microarray, next generation sequencing, RT-PCR, and RT-qPCR.

18. The method of claim 16 or 17, wherein the one or more probes comprises a detectable label and / or are embedded in one or more hydrogel particles.

19. The method of any one of claims 16-18, wherein the one or more microRNAs hybridized to the one or more probes is amplified by PCR.

20. The method of any one of the preceding claims, wherein the sample is selected from a sample of blood (e.g., plasma, serum, and whole blood), lymph node tissue (e.g., a draining lymph node for an ovary), or ovary tissue (e.g., ovary tissue is suspected to contain a tumor).

21. The method of any one of the preceding claims, wherein the subject is at risk of developing ovarian cancer, e.g., wherein the subject has a higher level of CA-125 antigen in the blood than a diagnostic threshold level (e.g., higher than 35 U / mL) and / or has at least one mutation in the BRCA1 or BRCA2 gene and / or is at least 55 years of age.

22. The method of any one of the preceding claims, wherein the statistical model comprises one or more models selected from a linear classification model, linear discriminant analysis, logistic regression, multivariate adaptive regression splines, naive Bayes, neural network, support vector machine, functional tree, LAD tree, Bayesian network, elastic net regression, and random forest.

Images