Patent Application
20190032142
The present invention relates to methods and materials for classification of cancers and the identification of their tissue of origin. Specifically the invention relates to microRNA molecules associated with specific cancers, as well as various nucleic acid molecules relating thereto or derived therefrom.
microRNAs (miRs, miRNAs) are a novel class of non-coding, regulatory RNA genes1-3 which are involved in oncogenesis4 and show remarkable tissue-specificity5-7. They have emerged as highly tissue-specific biomarkers2,5,6 postulated to play important roles in encoding developmental decisions of differentiation. Various studies have tied microRNAs to the development of specific malignancies4. MicroRNAs are also stable in tissue, stored frozen or as formalin-fixed, paraffin-embedded (FFPE) samples, and in serum.
Hundreds of thousands of patients in the U.S. are diagnosed each year with a cancer that has already metastasized, without a clearly identified primary site. Oncologists and pathologists are constantly faced with a diagnostic dilemma when trying to identify the primary origin of a patient's metastasis. As metastases need to be treated according to their primary origin, accurate identification of the metastases' primary origin can be critical for determining appropriate treatment.
Once a metastatic tumor is found, the patient may undergo a wide range of costly, time consuming, and at times inefficient tests, including physical examination of the patient, histopathology analysis of the biopsy, imaging methods such as chest X-ray, CT and PET scans, in order to identify the primary origin of the metastasis.
Metastatic cancer of unknown primary (CUP) accounts for 3-5% of all new cancer cases, and as a group is usually a very aggressive disease with a poor prognosis10. The concept of CUP comes from the limitation of present methods to identify cancer origin, despite an often complicated and costly process which can significantly delay proper treatment of such patients. Recent studies revealed a high degree of variation in clinical management, in the absence of evidence based treatment for CUP11. Many protocols were evaluated12 but have shown relatively small benefit13. Determining tumor tissue of origin is thus an important clinical application of molecular diagnostics9.
Molecular classification studies for tumor tissue origin14-17 have generally used classification algorithms that did not utilize domain-specific knowledge: tissues were treated as a-priori equivalents, ignoring underlying similarities between tissue types with a common developmental origin in embryogenesis. An exception of note is the study by Shedden and co-workers18, that was based on a pathology classification tree. These studies used machine-learning methods that average effects of biological features (e.g., mRNA expression levels), an approach which is more amenable to automated processing but does not use or generate mechanistic insights.
Various markers have been proposed to indicate specific types of cancers and tumor tissue of origin. However, the diagnostic accuracy of tumor markers has not yet been defined. There is thus a need for a more efficient and effective method for diagnosing and classifying specific types of cancers.
The present invention provides specific nucleic acid sequences for use in the identification, classification and diagnosis of specific cancers and tumor tissue of origin. The nucleic acid sequences can also be used as prognostic markers for prognostic evaluation and determination of appropriate treatment of a subject based on the abundance of the nucleic acid sequences in a biological sample. The present invention provides a method for accurate identification of tumor tissue origin.
The invention is based in part on the development of a microRNA-based classifier for tumor classification. microRNA expression levels were measured in 1300 primary and metastatic tumor paraffin-embedded samples. microRNAs were profiled using a custom array platform. Using the custom array platform, a set of over 300 microRNAs was identified for the normalization of the array data and 65 microRNAs were used for the accurate classification of over 40 different tumor types. The accuracy of the assay exceeds 85%.
The findings demonstrate the utility of microRNA as novel biomarkers for the tissue of origin of a metastatic tumor. The classifier has wide biological as well as diagnostic applications.
According to a first aspect, the present invention provides a method of identifying a tissue of origin of a cancer, the method comprising obtaining a biological sample from a subject, measuring the relative abundance in said sample of nucleic acid sequences selected from the group consisting of SEQ ID NOS: 1-390, any combinations thereof, or a sequence having at least about 80% identity thereto; and comparing the measurement to a reference abundance of the nucleic acid by using a classifier algorithm, wherein the relative abundance of said nucleic acid sequences allows for the identification of the tissue of origin of said sample.
According to one aspect, the classifier algorithm is selected from the group consisting of decision tree classifier, K-nearest neighbor classifier (KNN), logistic regression classifier, nearest neighbor classifier, neural network classifier, Gaussian mixture model (GMM), Support Vector Machine (SVM) classifier, nearest centroid classifier, linear regression classifier and random forest classifier. According to one aspect, the sample is obtained from a subject with cancer of unknown primary (CUP), with a primary cancer or with a metastatic cancer. According to certain embodiments, the cancer is selected from the group consisting of adrenocortical carcinoma; anus or skin squamous cell carcinoma; biliary tract adenocarcinoma; Ewing sarcoma; gastrointestinal stromal tumor (GIST); gastrointestinal tract carcinoid; renal cell carcinoma: chromophobe, clear cell and papillary; pancreatic islet cell tumor; pheochromocytoma; urothelial cell carcinoma (TCC); lung, head & neck, or esophagus squamous cell carcinoma (SCC); brain: astrocytic tumor, oligodendroglioma; breast adenocarcinoma; uterine cervix squamous cell carcinoma; chondrosarcoma; germ cell cancer; sarcoma; colorectal adenocarcinoma; liposarcoma; hepatocellular carcinoma (HCC); lung large cell or adenocarcinoma; lung carcinoid; pleural mesothelioma; lung small cell carcinoma; B-cell lymphoma; T-cell lymphoma; melanoma; malignant fibrous histiocytoma (MFH) or fibrosarcoma; osteosarcoma; ovarian primitive germ cell tumor; ovarian carcinoma; pancreatic adenocarcinoma; prostate adenocarcinoma; rhabdomyosarcoma; gastric or esophageal adenocarcinoma; synovial sarcoma; non-seminomatous testicular germ cell tumor; seminomatous testicular germ cell tumor; thymoma/thymic carcinoma; follicular thyroid carcinoma; medullary thyroid carcinoma; and papillary thyroid carcinoma.
The invention further provides a method for identifying a cancer of germ cell origin, comprising measuring the relative abundance of SEQ ID NO: 55 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of germ cell origin. According to some embodiments the germ cell is selected from the group consisting of an ovarian primitive cell and a testis cell. According to some embodiments the group of nucleic acid furthers consists of SEQ ID NOS: 29, 62 or a sequence having at least about 80% identity thereto, and the abundance of said nucleic acid sequence is indicative of a testis cell cancer origin selected from the group consisting of seminomatous testicular germ cell and non-seminomatous testicular germ cell.
The invention further provides a method for identifying a cancer origin selected from the group consisting of biliary tract adenocarcinoma and hepatocellular carcinoma, comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 9, 29 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of biliary tract adenocarcinoma and hepatocellular carcinoma.
The invention further provides a method for identifying a cancer of brain origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 156, 66, 68 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of brain origin.
According to some embodiments the group of nucleic acid furthers consists of SEQ ID NOS: 40, 60 or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of a brain cancer origin selected from the group consisting of oligodendroglioma and astrocytoma.
The invention further provides a method for identifying a cancer of prostate adenocarcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of prostate adenocarcinoma origin.
The invention further provides a method for identifying a cancer of breast adenocarcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 27, 35, 14, 21, 32, 51, 7, 25, 50, 11, 148, 4, 49, 67 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of breast adenocarcinoma origin.
The invention further provides a method for identifying a cancer of ovarian carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 27, 35, 14, 21, 32, 51, 7, 25, 4, 39, 50, 11, 148, 49, 67, 57, 34 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of an ovarian carcinoma origin.
The invention further provides a method for identifying a cancer of thyroid carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21, 32, 51, 7, 11, 148, 4 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of thyroid carcinoma origin.
According to some embodiments the group of nucleic acid furthers consists of SEQ ID NOS: 17, 34 or a sequence having at least about 80% identity thereto, and wherein said thyroid carcinoma origin is selected from the group consisting of follicular and papillary.
The invention further provides a method for identifying a cancer origin selected from the group consisting of lung large cell and lung adenocarcinoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21, 32, 51, 7, 11, 148, 4, 49, 67, 57, 34 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of lung large cell and lung adenocarcinoma.
The invention further provides a method for identifying a cancer origin selected from the group consisting of lung large cell and lung adenocarcinoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21, 32, 51, 7, 11, 148, 4, 49, 67, 57, 34 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of lung large cell and lung adenocarcinoma.
The invention further provides a method for identifying a cancer of thymic carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21, 32, 51, 7, 50, 4, 39, 3, 34 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of a thymic carcinoma origin.
The invention further provides a method for identifying a cancer origin selected from the group consisting of a urothelial cell carcinoma and squamous cell carcinoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 14, 21, 32, 51, 7, 50, 4, 39, 3, 34, 69, 24, 44 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of is indicative of a cancer origin selected from the group consisting of urothelial cell carcinoma and squamous cell carcinoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 1, 5, 54 or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of squamous-cell-carcinoma origin selected from the group consisting of uterine cervix squamous-cell-carcinoma and non uterine cervix squamous cell carcinoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 11, 23 or a sequence having at least about 80% identity thereto in said sample, and wherein the abundance of said nucleic acid sequence is indicative of a non-uterine cervix squamous cell carcinoma origin selected from the group consisting of anus or skin squamous cell carcinoma; and lung, head & neck, and esophagus squamous cell carcinoma.
The invention further provides a method for identifying a cancer origin selected from melanoma and lymphoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 47, 50 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of melanoma and lymphoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 35, 48 or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of a lymphoma cancer origin selected from the group consisting of B-cell lymphoma and T-cell lymphoma.
The invention further provides a method for identifying a cancer of lung small cell carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 20, 45 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of lung small cell carcinoma origin.
The invention further provides a method for identifying a cancer of medullary thyroid carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 20, 45, 40, 67, 68 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of medullary thyroid carcinoma origin.
The invention further provides a method for identifying a cancer of lung carcinoid origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 20, 45, 40, 67, 68, 64, 53, 37 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of lung carcinoid origin.
The invention further provides a method for identifying a cancer origin selected from the group consisting of gastrointestinal tract carcinoid and pancreatic islet cell tumor, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 20, 45, 40, 67, 68, 64, 53, 37, 34, 18 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of gastrointestinal tract carcinoid and pancreatic islet cell tumor.
The invention further provides a method for identifying a cancer origin selected from the group consisting of gastric and esophageal adenocarcinoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 42, 36, 146 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin elected from the group consisting of gastric and esophageal adenocarcinoma.
The invention further provides a method for identifying a cancer of colorectal adenocarcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 42, 36, 146, 20, 43 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of colorectal adenocarcinoma origin.
The invention further provides a method for identifying a cancer origin selected from the group consisting of pancreatic adenocarcinoma and biliary tract adenocarcinoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 56, 65, 25, 27, 35, 42, 36, 146, 20, 4351, 49, 16, or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of pancreatic adenocarcinoma or biliary tract adenocarcinoma.
The invention further provides a method for identifying a cancer of renal cell carcinoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 66, 68, 19, 29 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of renal cell carcinoma origin.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 36, 147 or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of a chromophobe renal cell carcinoma origin.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 49, 9 or a sequence having at least about 80% identity thereto, and wherein the abundance of said nucleic acid sequence is indicative of a renal cell carcinoma origin selected from the group consisting of clear cell and papillary.
The invention further provides a method for identifying a cancer of pheochromocytoma origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 66, 68, 19, 29, 65, 56 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of pheochromocytoma origin.
The invention further provides a method for identifying a cancer of adrenocortical origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 66, 68, 19, 29, 65, 56, 31, 38, 61 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of adrenocortical origin.
The invention further provides a method for identifying a cancer of gastrointestinal stromal tumor origin, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 66, 68, 19, 29, 65, 56, 31, 38, 61, 14, 45 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer of gastrointestinal stromal tumor origin.
The invention further provides a method for identifying a cancer origin selected from the group consisting of pleural mesothelioma and sarcoma, the method comprising measuring the relative abundance of a nucleic acid sequence selected from the group consisting of SEQ ID NOS: 55, 6, 30, 46, 16, 2, 66, 68, 19, 29, 65, 56, 31, 38, 61, 14, 45, 35, 10, 5 or a sequence having at least about 80% identity thereto in said sample; wherein the abundance of said nucleic acid sequence is indicative of a cancer origin selected from the group consisting of pleural mesothelioma and sarcoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 3, 40, 15 or a sequence having at least about 80% identity thereto, and wherein said sarcoma is synovial sarcoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 3, 40, 15, 12, 58 or a sequence having at least about 80% identity thereto, and wherein said sarcoma is chondrosarcoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 3, 40, 15, 12, 58, 36, 26 or a sequence having at least about 80% identity thereto, and wherein said sarcoma is liposarcoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 3, 40, 15, 12, 58, 36, 26, 21, 25, 49 or a sequence having at least about 80% identity thereto and wherein said sarcoma is selected from the group consisting of Ewing sarcoma and osteosarcoma.
According to some embodiments the group of nucleic acid further consists of SEQ ID NOS: 3, 40, 15, 12, 58, 36, 26, 21, 59, 39, 33 or a sequence having at least about 80% identity thereto and wherein said sarcoma is selected from the group consisting of rhabdomyosarcoma; and malignant fibrous histiocytoma and fibrosarcoma.
According to another aspect, the present invention provides a method of distinguishing between cancers of different origins, said method comprising:
(a) obtaining a biological sample from a subject;
(b) measuring the relative abundance in said sample of nucleic acid sequences selected from the group consisting of SEQ ID NOS: 1-390 or a sequence having at least about 80% identity thereto; and
(c) comparing said measurement to a reference abundance of said nucleic acid by using a classifier algorithm;
wherein the relative abundance of said nucleic acid sequence in said sample allows for distinguishing between cancers of different origins.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 372, 233, 55, 200, 201 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from a germ-cell tumor and a cancer originating from the group consisting of non-germ-cell tumors.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 6, 30, 13 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from hepatobiliary tumors and a cancer originating from the group consisting of non-germ-cell non-hepatobiliary tumors.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 28, 29, 231, 9 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from liver tumors and a cancer originating from biliary-tract carcinomas.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 46, 5, 12, 30, 29, 28, 32, 13, 152, 49 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from the group consisting of tumors from an epithelial origin and a cancer originating from the group consisting of tumors from a non-epithelial origin.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 164, 168, 170, 16, 198, 50, 176, 186, 11, 158, 20, 155, 231, 4, 8, 46, 3, 2, 7 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from the group consisting of melanoma and lymphoma and a cancer originating from the group consisting of all other non-epithelial tumors.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 159, 66, 225, 187, 162, 161, 68, 232, 173, 11, 8, 174, 155, 231, 4, 182, 181, 37 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from brain tumors and a cancer originating from the group consisting of all non-brain, non-epithelial tumors.
According to some embodiments the measurement of the relative abundance of SEQ ID NOS: 40, 208, 60, 153, 230, 228, 147, 34, 206, 35, 52, 25, 229, 161, 187, 179 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from astrocytoma and a cancer originating from oligodendroglioma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 56, 65, 25, 175, 152, 155, 32, 49, 35, 181, or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from the group consisting of neuroendocrine tumors and a cancer originating from the group consisting of all non-neuroendocrine, epithelial tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 27, 177, 4, 32, 35 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from the group consisting of gastrointestinal epithelial tumors and a cancer originating from the group consisting of non-gastrointestinal epithelial tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 56, 199, 14, 15, 165, 231, 36, 154, 21, 49 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from prostate tumors and a cancer originating from the group consisting of all other non-gastrointestinal epithelial tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 222, 62, 29, 28, 211, 214, 227, 215, 218, 152, 216, 212, 224, 13, 194, 192, 221, 217, 205, 219, 32, 193, 223, 220, 210, 209, 213, 163, 30 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from seminoma and a cancer originating from the group consisting of non-seminoma testis-tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 42, 32, 36, 178, 243, 242, 49, 240, 57, 11, 46, 17, 47, 51, 7, 8, 154, 190, 157, 196, 197, or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from the group consisting of squamous cell carcinoma, transitional cell carcinoma and thymoma, and a cancer originating from the group consisting of non gastrointestinal adenocarcinoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 56, 46, 25, 152, 50, 45, 191, 181, 179, 49, 32, 42, 184, 40, 147, 236, 57, 203, 36, or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from breast adenocarcinoma, and a cancer originating from the group consisting of squamous cell carcinoma, transitional cell carcinoma, thymomas and ovarian carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 253, 32, 4, 39, 10, 46, 5, 226, 2, 195, 32, 185, 11, 168, 184, 16, 242, 12, 237, 243, 250, 49, 246, 167 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from ovarian carcinoma, and a cancer originating from the group consisting of squamous cell carcinoma, transitional cell carcinoma and thymomas.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 11, 147, 17, 157, 40, 8, 49, 9, 191, 205, 207, 195, 51, 46, 45, 52, 234, 231, 21, 169, 43, 3, 196, 154, 390, 171, 255, 197, 190, 189, 39, 7, 48, 47, 32, 36, 4, 178, 37, 181, 25, 183, 182, 35, 240, 57, 242, 204, 236, 176, 158, 148, 206, 50, 20, 34, 186, 239, 251, 244, 24, 188, 172, 238 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from thyroid carcinoma, and a cancer originating from the group consisting of breast adenocarcinoma, lung large cell carcinoma, lung adenocarcinoma and ovarian carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 249, 180, 65, 235, 241, 248, 254, 247, 160, 243, 245, 252, 17, 49, 166, 225, 168, 34 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from follicular thyroid carcinoma and a cancer originating from papillary thyroid carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 32, 56, 50, 45, 25, 253, 152, 9, 46, 191, 178, 49, 40, 10, 147, 4, 36, 228, 236, 230, 189, 240, 67, 202, 17 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from breast adenocarcinoma and a cancer originating from the group consisting of lung adenocarcinoma and ovarian carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 56, 11, 168, 16, 237, 21, 52, 12, 154, 279, 9, 39, 47, 23, 50, 167, 383, 34, 35, 388, 5, 359, 245, 254, 10, 240, 236, 202, 4, 25, 203, 231, 20, 158, 186, 258, 244, 172, 2, 235, 256, 28, 277, 296, 374, 153, 181 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from lung adenocarcinoma and a cancer originating from ovarian carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 161, 164, 22, 53, 285, 3, 152, 191, 154, 21, 206, 174, 19, 45, 171, 179, 8, 296, 284, 18, 51, 258, 49, 184, 35, 34, 37, 42, 228, 15, 14, 242, 230, 253, 36, 182, 293, 292, 4, 294, 297, 354, 377, 189, 30, 386, 249, 5, 274 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from thymic carcinoma and a cancer originating from the group consisting of transitional cell carcinoma and squamous cell carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 69, 28, 280, 13, 191, 152, 29, 175, 30, 204, 4, 24, 5, 329, 273, 170, 184, 26, 231, 368, 37, 16, 169, 155, 35, 40, 17 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from transitional cell carcinoma and a cancer originating from the group consisting of squamous cell carcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 164, 5, 231, 54, 1, 242, 372, 249, 167, 254, 354, 381, 380, 245, 358, 364, 240, 11, 378 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between squamous cell carcinoma cancers originating from the uterine cervix, and squamous cell carcinoma cancers originating from the group consisting of anus and skin, lung, head & neck and esophagus.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 305, 184, 41, 183, 49, 382, 235, 291, 181, 5, 296, 289, 206, 338, 334, 25, 11, 19, 198, 23 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between squamous cell carcinoma cancers originating from the group consisting of anus and skin, and between squamous cell carcinoma cancers originating from the group consisting of lung, head & neck and esophagus.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 4, 11, 46, 8, 274, 169, 36, 47, 363, 231, 303, 349, 10, 7, 3, 16, 164, 170, 168, 198, 50, 245, 365, 45, 382, 259, 296, 364, 314, 12 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from melanoma and a cancer originating from lymphoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 11, 191, 48, 35, 228 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from B-cell lymphoma and a cancer originating from T-cell lymphoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 158, 20, 176, 186, 148, 36, 51, 172, 260, 265, 67, 188, 277, 284, 302, 68, 168, 242, 204, 162, 177, 27, 65, 263, 155, 191, 190, 45, 59, 43, 56, 266, 14, 15, 8, 7, 39, 189, 249, 231, 293, 2 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from lung small cell carcinoma and a cancer originating from the group consisting of lung carcinoid, medullary thyroid carcinoma, gastrointestinal tract carcinoid and pancreatic islet cell tumor.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 159, 40, 147, 11, 311, 4, 8, 231, 301, 297, 68, 67, 265, 36 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from medullary thyroid carcinoma and a cancer originating from other neuroendocrine tumors selected from the group consisting of lung carcinoid, gastrointestinal tract carcinoid and pancreatic islet cell tumor.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 331, 162, 59, 326, 306, 350, 317, 155, 325, 318, 339, 264, 332, 262, 336, 324, 322, 330, 321, 263, 309, 53, 320, 275, 352, 312, 355, 367, 269, 64, 308, 175, 190, 54, 302, 152, 301, 266, 47, 313, 359, 65, 307, 191, 242, 4, 147, 40, 372, 168, 16, 182, 167, 356, 148, 382, 37, 364, 35 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from lung carcinoid tumors, and a cancer originating from gastrointestinal neuroendocrine tumors selected from the group consisting of gastrointestinal tract carcinoid and pancreatic islet cell tumor.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 263, 288, 18, 286, 162, 225, 287, 206, 205, 296, 258, 313, 377, 373, 256, 153, 259, 265, 303, 268, 267, 165, 15, 272, 14, 202, 236, 203, 4, 168, 310, 298, 27, 29, 34, 228, 3, 349, 35, 26 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from pancreatic islet cell tumors and a Gastrointestinal neuroendocrine carcinoid cancer originating from the group consisting of small intestine and duodenum; appendicitis, stomach and pancreas.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 36, 267, 268, 165, 15, 14, 356, 167, 372, 272, 370, 42, 41, 146 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between adenocarcinoma tumors of the gastrointestinal system originating from:
the group consisting of gastric and esophageal adenocarcinoma, and
the group consisting of cholangiocarcinoma or adenocarcinoma of the extrahepatic biliary tract, pancreatic adenocarcinoma and colorectal adenocarcinoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 42, 184, 67, 158, 20, 186, 284, 389, 203, 240, 236, 146, 204, 43, 176, 202, 49, 46, 38, 363 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from colorectal adenocarcinoma and a cancer originating from the group consisting of adenocarcinoma of biliary tract or pancreas.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 49, 11, 13, 373, 154, 5, 30, 45, 178, 147, 274, 16, 40, 21, 43, 253, 245, 256, 12, 374, 379, 180, 153, 51, 52, 1, 295, 257, 385, 293, 294 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from pancreatic adenocarcinoma, and a cancer originating from the group consisting of cholangiocarcinoma or adenocarcinoma of the extrahepatic biliary tract.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 29, 28, 30, 46, 49, 195, 152, 175, 47, 4, 387, 196, 177, 375, 27, 304, 40, 191, 147, 35, 16, 34, 5, 155, 181, 312, 183, 182, 320, 59, 38, 324, 323, 37, 322, 325, 19, 42, 334, 265, 22 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from:
renal cell tumors selected from the group consisting of chromophobe renal cell carcinoma, clear cell renal cell carcinoma and papillary renal cell carcinoma, and
the group consisting of sarcomas, adrenal tumors and pleural mesothelioma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 65, 56, 11, 162, 59, 331, 350, 155, 335, 159, 336, 332, 263, 306, 339, 337, 275, 301, 276, 330, 317, 309, 45, 318, 324, 352, 191, 262, 269, 313, 19, 367, 326, 325, 322, 327, 190, 261, 321, 360, 353, 312, 371, 5, 328, 205, 183, 38, 181, 37, 40, 182, 147, 17, 42, 382, 34, 18, 3 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from pheochromocytoma, and a cancer originating from the group consisting of all sarcoma, adrenal carcinoma and mesothelioma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 61, 333, 31, 347, 346, 344, 345, 387, 334, 351, 324, 326, 269, 155, 320, 322, 59, 318, 325, 245, 254, 331, 275, 180, 355, 370, 323, 312, 178, 249, 183, 181, 38, 182, 37, 3, 25 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from adrenal carcinoma and a cancer originating from the group consisting of mesothelioma and sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 165, 14, 15, 333, 272, 270, 45, 301, 191, 46, 195, 266, 190, 19, 334, 155, 25, 147, 40, 34 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from a gastrointestinal stromal tumor and a cancer originating from the group consisting of mesothelioma and sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 13, 30, 361, 280, 362, 147, 40, 291, 387, 290, 299, 152, 178, 303, 242, 49, 11, 35, 34, 36, 206, 16, 170, 177, 17 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from a chromophobe renal cell carcinoma tumor and a cancer originating from the group consisting of clear cell and papillary renal cell carcinoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 344, 382, 9, 338, 29, 49, 28, 195, 46, 4, 11, 254 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a renal carcinoma cancer originating from a clear cell tumor and a cancer originating from a papillary tumor.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 49, 35, 17, 34, 25, 36, 168, 170, 26, 4, 190, 46, 10, 240, 43, 39, 385, 63, 202, 181, 37, 5, 183, 182, 38, 206, 296, 1 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from pleural mesothelioma and a cancer originating from the group consisting of sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 152, 29, 159, 28, 339, 275, 352, 19, 320, 155, 262, 38, 37, 182, 331, 317, 323, 355, 3, 282, 312, 181, 269, 318, 59, 266, 322, 8, 324, 10, 40, 147, 169, 205, 34, 168, 14, 15, 12, 46, 255, 39, 23, 190, 236, 386, 379, 202 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from a synovial sarcoma and a cancer originating from the group consisting of other sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 12, 271, 206, 333, 11, 58, 36, 18, 178, 293, 189, 382, 381, 240, 249, 5, 377, 235, 17, 20, 385, 384, 46, 283 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from chondrosarcoma and a cancer originating from the group consisting of other non-synovial sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 295, 205, 25, 26, 231, 183, 42, 254, 168, 64, 14, 178, 15, 39, 36, 154, 265, 174, 384, 67 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from liposarcoma and a cancer originating from the group consisting of other non chondrosarcoma and non synovial sarcoma tumors.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 22, 154, 21, 174, 205, 158, 186, 148, 20, 59, 8, 183, 231 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from:
the group consisting of Ewing sarcoma and osteosarcoma, and
the group consisting of rhabdomyosarcoma, malignant fibrous histiocytoma and fibrosarcoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 155, 179, 43, 208, 278, 17, 385, 174, 5, 52, 257, 366, 48, 49, 12, 25, 169, 34, 35, 23, 384, 189, 377, 265, 294, 293, 292 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from Ewing sarcoma and a cancer originating from osteosarcoma.
According to some embodiments, measurement of the relative abundance of SEQ ID NOS: 33, 268, 267, 333, 276, 319, 306, 320, 334, 323, 300, 281, 59, 339, 316, 176, 348, 352, 349, 67, 357, 315, 343, 342, 355, 340, 344, 10, 341, 331, 20, 277, 318, 158, 265, 284, 36, 183, 40, 63, 147, 43, 289, 52, 190, 4, 5, 39, 169, 208 or a sequence having at least about 80% identity thereto in said sample allows for distinguishing between a cancer originating from rhabdomyosarcoma and a cancer originating from the group consisting of malignant fibrous histiocytoma and fibrosarcoma.
According to some aspects of the invention the biological sample is selected from the group consisting of bodily fluid, a cell line, a tissue sample, a biopsy sample, a needle biopsy sample, a fine needle biopsy (FNA) sample, a surgically removed sample, and a sample obtained by tissue-sampling procedures such as endoscopy, bronchoscopy, or laparoscopic methods. According to some embodiments, the tissue is a fresh, frozen, fixed, wax-embedded or formalin-fixed paraffin-embedded (FFPE) tissue.
According to additional aspects of the invention the nucleic acid sequence relative abundance is determined by a method selected from the group consisting of nucleic acid hybridization and nucleic acid amplification. According to some embodiments, the nucleic acid hybridization is performed using a solid-phase nucleic acid biochip array or in situ hybridization. According to some embodiments, the nucleic acid amplification method is real-time PCR. According to some embodiments, the real-time PCR comprises forward and reverse primers. According to additional embodiments, the real-time PCR method further comprises a probe. According to additional embodiments, the probe comprises a sequence selected from the group consisting of a sequence that is complementary to a sequence selected from SEQ ID NOS: 1-390; a fragment thereof and a sequence having at least about 80% identity thereto.
According to another aspect, the present invention provides a kit for cancer origin identification, the kit comprising a probe comprising a sequence selected from the group consisting of a sequence that is complementary to a sequence selected from SEQ ID NOS: 1-390; a fragment thereof and a sequence having at least about 80% identity thereto.
These and other embodiments of the present invention will become apparent in conjunction with the figures, description and claims that follow.
Identification of the tissue-of-origin of a tumor is vital to its management. The present invention is based in part on the discovery that specific nucleic acid sequences can be used for the identification of the tissue-of-origin of a tumor. The present invention provides a sensitive, specific and accurate method which can be used to distinguish between different tumor origins. A new microRNA-based classifier was developed for determining tissue origin of tumors based on 65 microRNAs markers. The classifier uses a specific algorithm and allows a clear interpretation of the specific biomarkers.
According to the present invention each node in the classification tree may be used as an independent differential diagnosis tool, for example in the identification of different types of lung cancers. The possibility to distinguish between different tumor origins facilitates providing the patient with the best and most suitable treatment.
The present invention provides diagnostic assays and methods, both quantitative and qualitative for detecting, diagnosing, monitoring, staging and prognosticating cancers by comparing the levels of the specific microRNA molecules of the invention. Such levels are preferably measured in at least one of biopsies, tumor samples, fine-needle aspiration (FNA), cells, tissues and/or bodily fluids. The methods provided in the present invention are particularly useful for discriminating between different cancers.
All the methods of the present invention may optionally further include measuring levels of additional cancer markers. The cancer markers measured in addition to said microRNA molecules depend on the cancer being tested and are known to those of skill in the art.
Assay techniques can be used to determine levels of gene expression, such as genes encoding the nucleic acids of the present invention in a sample derived from a patient. Such assay methods, which are well known to those of skill in the art, include, but are not limited to, nucleic acid microarrays and biochip analysis, reverse transcriptase PCR (RT-PCR) assays, immunohistochemistry assays, in situ hybridization assays, competitive-binding assays, northern blot analyses and ELISA assays.
According to one embodiment, the assay is based on expression level of 65 microRNAs in RNA extracted from FFPE metastatic tumor tissue.
The expression levels are used to infer the sample origin using analysis techniques such as, but not limited to, decision-tree classifier, K nearest neighbors classifier, logistic regression classifier, linear regression classifier, nearest neighbor classifier, neural network classifier and nearest centroid classifier.
In use of the decision tree classifier the expression levels are used to make binary decisions (at each relevant node) following the pre-defined structure of the binary decision-tree (defined using a training set).
At each node, the expressions of one or several microRNAs are combined together using a function of the form P=exp (β0+β1*miR1+β2*miR2+β3*miR3 . . . )/(1−exp (β0+β1*miR1+β2*miR2+β3*miR3 . . . )), where the values of (30, (31, (32 . . . and the identities of the microRNAs have been pre-determined (using a training set). The resulting P is compared to a probability threshold level (PTH, which was also determined using the training set), and the classification continues to the left or right branch according to whether P is larger or smaller than the PTH for that node. This continues until an end-point (“leaf”) of the tree is reached. According to some embodiments, PTH=0.5 for all nodes, and the value of (30 is adjusted accordingly. According to further embodiments, β0, β1, β2, . . . are adjusted so that the slope of the log of the odds ratio function is limited.
Training the tree algorithm means determining the tree structure—which nodes there are and what is on each side, and, for each node: which miRs are used, the values of β0, β1, β2 . . . and the PTH. These are determined by a combination of machine learning, optimization algorithm, and trial and error by experts in machine learning and diagnostic algorithms
An arbitrary threshold of the expression level of one or more nucleic acid sequences can be set for assigning a sample to one of two groups. Alternatively, in a preferred embodiment, expression levels of one or more nucleic acid sequences of the invention are combined by a method such as logistic regression to define a metric which is then compared to previously measured samples or to a threshold. The threshold is treated as a parameter that can be used to quantify the confidence with which samples are assigned to each class. The threshold can be scaled to favor sensitivity or specificity, depending on the clinical scenario. The correlation value to the reference data generates a continuous score that can be scaled and provides diagnostic information on the likelihood that a sample belongs to a certain class of cancer origin or type. In multivariate analysis the microRNA signature provides a high level of prognostic information.
In another preferred embodiment, expression level of nucleic acids is used to classify a test sample by comparison to a training set of samples. In this embodiment, the test sample is compared in turn to each one of the training set samples. The comparison is performed by comparing the expression levels of one or multiple nucleic acids between the test sample and the specific training sample. Each such pairwise comparison generates a combined metric for the multiple nucleic acids, which can be calculated by various numeric methods such as correlation, cosine, Euclidian distance, mean square distance, or other methods known to those skilled in the art. The training samples are then ranked according to this metric, and the samples with the highest values of the metric (or lowest values, according to the type of metric) are identified, indicating those samples that are most similar to the test sample. By choosing a parameter K, this generates a list that includes the K training samples that are most similar to the test sample. Various methods can then be applied to identify from this list the predicted class of the test sample. In a favored embodiment, the test sample is predicted to belong to the class that has the highest number of representative in the list of K most-similar training samples (this method is known as the K Nearest Neighbors method). Other embodiments may provide a list of predictions including all or part of the classes represented in the list, those classes that are represented more than a given minimum number of times, or other voting schemes whereby classes are grouped together.
It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. It must be noted that, as used in the specification and the appended claims, the singular forms “a,” “an” and “the” include plural referents unless the context clearly dictates otherwise.
For the recitation of numeric ranges herein, each intervening number there between with the same degree of precision is explicitly contemplated. For example, for the range of 6-9, the numbers 7 and 8 are contemplated in addition to 6 and 9, and for the range 6.0-7.0, the number 6.0, 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9 and 7.0 are explicitly contemplated.
About
As used herein, the term “about” refers to +/−10%.
Attached
“Attached” or “immobilized”, as used herein, to refer to a probe and a solid support means that the binding between the probe and the solid support is sufficient to be stable under conditions of binding, washing, analysis, and removal. The binding may be covalent or non-covalent. Covalent bonds may be formed directly between the probe and the solid support or may be formed by a cross linker or by inclusion of a specific reactive group on either the solid support or the probe or both molecules. Non-covalent binding may be one or more of electrostatic, hydrophilic, and hydrophobic interactions. Included in non-covalent binding is the covalent attachment of a molecule, such as streptavidin, to the support and the non-covalent binding of a biotinylated probe to the streptavidin. Immobilization may also involve a combination of covalent and non-covalent interactions.
Baseline
“Baseline”, as used herein, means the initial cycles of PCR, in which there is little change in fluorescence signal.
Biological Sample
“Biological sample”, as used herein, means a sample of biological tissue or fluid that comprises nucleic acids. Such samples include, but are not limited to, tissue or fluid isolated from subjects. Biological samples may also include sections of tissues such as biopsy and autopsy samples, FFPE samples, frozen sections taken for histological purposes, blood, blood fraction, plasma, serum, sputum, stool, tears, mucus, hair, skin, urine, effusions, ascitic fluid, amniotic fluid, saliva, cerebrospinal fluid, cervical secretions, vaginal secretions, endometrial secretions, gastrointestinal secretions, bronchial secretions, cell line, tissue sample, or secretions from the breast. A biological sample may be provided by fine-needle aspiration (FNA), pleural effusion or bronchial brushing. A biological sample may be provided by removing a sample of cells from a subject but can also be accomplished by using previously isolated cells (e.g., isolated by another person, at another time, and/or for another purpose), or by performing the methods described herein in vivo. Archival tissues, such as those having treatment or outcome history, may also be used. Biological samples also include explants and primary and/or transformed cell cultures derived from animal or human tissues.
Cancer
The term “cancer” is meant to include all types of cancerous growths or oncogenic processes, metastatic tissues or malignantly transformed cells, tissues, or organs, irrespective of histopathologic type or stage of invasiveness. Examples of cancers include, but are not limited, to solid tumors and leukemias, including: apudoma, choristoma, branchioma, malignant carcinoid syndrome, carcinoid heart disease, carcinoma (e.g., Walker, basal cell, basosquamous, Brown-Pearce, ductal, Ehrlich tumor, non-small cell lung (e.g., lung squamous cell carcinoma, lung adenocarcinoma and lung undifferentiated large cell carcinoma), oat cell, papillary, bronchiolar, bronchogenic, squamous cell, and transitional cell), histiocytic disorders, leukemia (e.g., B cell, mixed cell, null cell, T cell, T-cell chronic, HTLV-II-associated, lymphocytic acute, lymphocytic chronic, mast cell, and myeloid), histiocytosis malignant, Hodgkin disease, immunoproliferative small, non-Hodgkin lymphoma, plasmacytoma, reticuloendotheliosis, melanoma, chondroblastoma, chondroma, chondrosarcoma, fibroma, fibrosarcoma, giant cell tumors, histiocytoma, lipoma, liposarcoma, mesothelioma, myxoma, myxosarcoma, osteoma, osteosarcoma, Ewing sarcoma, synovioma, adenofibroma, adenolymphoma, carcinosarcoma, chordoma, craniopharyngioma, dysgerminoma, hamartoma, mesenchymoma, mesonephroma, myosarcoma, ameloblastoma, cementoma, odontoma, teratoma, thymoma, trophoblastic tumor, adeno-carcinoma, adenoma, cholangioma, cholesteatoma, cylindroma, cystadenocarcinoma, cystadenoma, granulosa cell tumor, gynandroblastoma, hepatoma, hidradenoma, islet cell tumor, Leydig cell tumor, papilloma, Sertoli cell tumor, theca cell tumor, leiomyoma, leiomyosarcoma, myoblastoma, myosarcoma, rhabdomyoma, rhabdomyosarcoma, ependymoma, ganglioneuroma, glioma, medulloblastoma, meningioma, neurilemmoma, neuroblastoma, neuroepithelioma, neurofibroma, neuroma, paraganglioma, paraganglioma nonchromaffin, angiokeratoma, angiolymphoid hyperplasia with eosinophilia, angioma sclerosing, angiomatosis, glomangioma, hemangioendothelioma, hemangioma, hemangiopericytoma, hemangiosarcoma, lymphangioma, lymphangiomyoma, lymphangiosarcoma, pinealoma, carcinosarcoma, chondrosarcoma, cystosarcoma, phyllodes, fibrosarcoma, hemangiosarcoma, leimyosarcoma, leukosarcoma, liposarcoma, lymphangiosarcoma, myosarcoma, myxosarcoma, ovarian carcinoma, rhabdomyosarcoma, sarcoma (e.g., Ewing, experimental, Kaposi, and mast cell), neurofibromatosis, and cervical dysplasia, and other conditions in which cells have become immortalized or transformed.
Classification
The term classification refers to a procedure and/or algorithm in which individual items are placed into groups or classes based on quantitative information on one or more characteristics inherent in the items (referred to as traits, variables, characters, features, etc.) and based on a statistical model and/or a training set of previously labeled items. A “classification tree” places categorical variables into classes.
Complement
“Complement” or “complementary” is used herein to refer to a nucleic acid may mean Watson-Crick (e.g., A-T/U and C-G) or Hoogsteen base pairing between nucleotides or nucleotide analogs of nucleic acid molecules. A full complement or fully complementary means 100% complementary base pairing between nucleotides or nucleotide analogs of nucleic acid molecules. In some embodiments, the complementary sequence has a reverse orientation (5′-3′).
Ct
Ct signals represent the first cycle of PCR where amplification crosses a threshold (cycle threshold) of fluorescence. Accordingly, low values of Ct represent high abundance or expression levels of the microRNA.
In some embodiments the PCR Ct signal is normalized such that the normalized Ct remains inversed from the expression level. In other embodiments the PCR Ct signal may be normalized and then inverted such that low normalized-inverted Ct represents low abundance or low expression levels of the microRNA.
Data Processing Routine
As used herein, a “data processing routine” refers to a process that can be embodied in software that determines the biological significance of acquired data (i.e., the ultimate results of an assay or analysis). For example, the data processing routine can determine a tissue of origin based upon the data collected. In the systems and methods herein, the data processing routine can also control the data collection routine based upon the results determined. The data processing routine and the data collection routines can be integrated and provide feedback to operate the data acquisition, and hence provide assay-based judging methods.
Data Set
As use herein, the term “data set” refers to numerical values obtained from the analysis. These numerical values associated with analysis may be values such as peak height and area under the curve.
Data Structure
As used herein, the term “data structure” refers to a combination of two or more data sets, an application of one or more mathematical manipulation to one or more data sets to obtain one or more new data sets, or a manipulation of two or more data sets into a form that provides a visual illustration of the data in a new way. An example of a data structure prepared from manipulation of two or more data sets would be a hierarchical cluster.
Detection
“Detection” means detecting the presence of a component in a sample. Detection also means detecting the absence of a component. Detection also means determining the level of a component, either quantitatively or qualitatively.
Differential Expression
“Differential expression” means qualitative or quantitative differences in the temporal and/or spatial gene expression patterns within and among cells and tissue. Thus, a differentially expressed gene may qualitatively have its expression altered, including an activation or inactivation, in, e.g., normal versus diseased tissue. Genes may be turned on or turned off in a particular state, relative to another state, thus permitting comparison of two or more states. A qualitatively regulated gene may exhibit an expression pattern within a state or cell type which may be detectable by standard techniques. Some genes may be expressed in one state or cell type, but not in both. Alternatively, the difference in expression may be quantitative, e.g., in that expression is modulated: up-regulated, resulting in an increased amount of transcript, or down-regulated, resulting in a decreased amount of transcript. The degree to which expression differs needs only to be large enough to quantify via standard characterization techniques such as expression arrays, quantitative reverse transcriptase PCR, northern blot analysis, real-time PCR, in situ hybridization and RNase protection.
Epithelial Tumors
“Epithelial tumors” is meant to include all types of tumors from epithelial origin. Examples of epithelial tumors include, but are not limited to cholangioca or adenoca of extrahepatic biliary tract, urothelial carcinoma, adenocarcinoma of the breast, lung large cell or adenocarcinoma, lung small cell carcinoma, carcinoid, lung, ovarian carcinoma, pancreatic adenocarcinoma, prostatic adenocarcinoma, gastric or esophageal adenocarcinoma, thymoma/thymic carcinoma, follicular thyroid carcinoma, papillary thyroid carcinoma, medullary thyroid carcinoma, anus or skin squamous cell carcinoma, lung, head&neck, or esophagus squamous cell carcinoma, uterine cervix squamous cell carcinoma, gastrointestinal tract carcinoid, pancreatic islet cell tumor and colorectal adenocarcinoma.
Non Epithelial Tumors
“Non epithelial tumors” is meant to include all types of tumors from non epithelial origin. Examples of non epithelial tumors include, but are not limited to adrenocortical carcinoma, chromophobe renal cell carcinoma, clear cell renal cell carcinoma, papillary renal cell carcinoma, pleural mesothelioma, astrocytic tumor, oligodendroglioma, pheochromocytoma, B-cell lymphoma, T-cell lymphoma, melanoma, gastrointestinal stromal tumor (GIST), Ewing Sarcoma, chondrosarcoma, malignant fibrous histiocytoma (MFH) or fibrosarcoma, osteosarcoma, rhabdomyosarcoma, synovial sarcoma and liposarcoma.
Expression Profile
The term “expression profile” is used broadly to include a genomic expression profile, e.g., an expression profile of microRNAs. Profiles may be generated by any convenient means for determining a level of a nucleic acid sequence, e.g., quantitative hybridization of microRNA, labeled microRNA, amplified microRNA, cDNA, etc., quantitative PCR, ELISA for quantitation, and the like, and allow the analysis of differential gene expression between two samples. A subject or patient tumor sample, e.g., cells or collections thereof, e.g., tissues, is assayed. Samples are collected by any convenient method, as known in the art. Nucleic acid sequences of interest are nucleic acid sequences that are found to be predictive, including the nucleic acid sequences provided above, where the expression profile may include expression data for 5, 10, 20, 25, 50, 100 or more of the nucleic acid sequences, including all of the listed nucleic acid sequences. According to some embodiments, the term “expression profile” means measuring the relative abundance of the nucleic acid sequences in the measured samples.
Expression Ratio
“Expression ratio”, as used herein, refers to relative expression levels of two or more nucleic acids as determined by detecting the relative expression levels of the corresponding nucleic acids in a biological sample.
FDR (False Discovery Rate)
When performing multiple statistical tests, for example in comparing between the signal of two groups in multiple data features, there is an increasingly high probability of obtaining false positive results, by random differences between the groups that can reach levels that would otherwise be considered statistically significant. In order to limit the proportion of such false discoveries, statistical significance is defined only for data features in which the differences reached a p-value (by two-sided t-test) below a threshold, which is dependent on the number of tests performed and the distribution of p-values obtained in these tests.
Fragment
“Fragment” is used herein to indicate a non-full-length part of a nucleic acid. Thus, a fragment is itself also a nucleic acid.
Gastrointestinal Tumors
“gastrointestinal tumors” is meant to include all types of tumors from gastrointestinal origin. Examples of gastrointestinal tumors include, but are not limited to cholangioca. or adenoca of extrahepatic biliary tract, pancreatic adenocarcinoma, gastric or esophageal adenocarcinoma, and colorectal adenocarcinoma.
Gene
“Gene”, as used herein, may be a natural (e.g., genomic) or synthetic gene comprising transcriptional and/or translational regulatory sequences and/or a coding region and/or non-translated sequences (e.g., introns, 5′- and 3′-untranslated sequences). The coding region of a gene may be a nucleotide sequence coding for an amino acid sequence or a functional RNA, such as tRNA, rRNA, catalytic RNA, siRNA, miRNA or antisense RNA. A gene may also be an mRNA or cDNA corresponding to the coding regions (e.g., exons and miRNA) optionally comprising 5′- or 3′-untranslated sequences linked thereto. A gene may also be an amplified nucleic acid molecule produced in vitro, comprising all or a part of the coding region and/or 5′- or 3′-untranslated sequences linked thereto.
Germ Cell Tumors
“Germ cell tumors” as used herein, include, but are not limited, to non-seminomatous testicular germ cell tumors, seminomatous testicular germ cell tumors and ovarian primitive germ cell tumors.
Groove Binder/Minor Groove Binder (MGB)
“Groove binder” and/or “minor groove binder” may be used interchangeably and refer to small molecules that fit into the minor groove of double-stranded DNA, typically in a sequence-specific manner Minor groove binders may be long, flat molecules that can adopt a crescent-like shape and thus fit snugly into the minor groove of a double helix, often displacing water. Minor groove binding molecules may typically comprise several aromatic rings connected by bonds with torsional freedom such as furan, benzene, or pyrrole rings. Minor groove binders may be antibiotics such as netropsin, distamycin, berenil, pentamidine and other aromatic diamidines, Hoechst 33258, SN 6999, aureolic anti-tumor drugs such as chromomycin and mithramycin, CC-1065, dihydrocyclopyrroloindole tripeptide (DPI3), 1,2-dihydro-(3H)-pyrrolo[3,2-e]indole-7-carboxylate (CDPI3), and related compounds and analogues, including those described in Nucleic Acids in Chemistry and Biology, 2nd ed., Blackburn and Gait, eds., Oxford University Press, 1996, and PCT Published Application No. WO 03/078450, the contents of which are incorporated herein by reference. A minor groove binder may be a component of a primer, a probe, a hybridization tag complement, or combinations thereof. Minor groove binders may increase the Tm of the primer or a probe to which they are attached, allowing such primers or probes to effectively hybridize at higher temperatures.
High Expression miR-205 Tumors
“High expression miR-205 tumors” as used herein include, but are not limited, to urothelial carcinoma (TCC), thymoma/thymic carcinoma, anus or skin squamous cell carcinoma, lung, head&neck, or esophagus squamous cell carcinoma and uterine cervix squamous cell carcinoma.
Low Expression 205 Tumors
“Low expression miR-205 tumors” as used herein include, but are not limited, to lung, large cell or adenocarcinoma, follicular thyroid carcinoma and papillary thyroid carcinoma.
Host Cell
“Host cell”, as used herein, may be a naturally occurring cell or a transformed cell that may contain a vector and may support replication of the vector. Host cells may be cultured cells, explants, cells in vivo, and the like. Host cells may be prokaryotic cells, such as E. coli, or eukaryotic cells, such as yeast, insect, amphibian, or mammalian cells, such as CHO and HeLa cells.
Identity
“Identical” or “identity”, as used herein, in the context of two or more nucleic acids or polypeptide sequences mean that the sequences have a specified percentage of residues that are the same over a specified region. The percentage may be calculated by optimally aligning the two sequences, comparing the two sequences over the specified region, determining the number of positions at which the identical residue occurs in both sequences to yield the number of matched positions, dividing the number of matched positions by the total number of positions in the specified region, and multiplying the result by 100 to yield the percentage of sequence identity. In cases where the two sequences are of different lengths or the alignment produces one or more staggered ends and the specified region of comparison includes only a single sequence, the residues of single sequence are included in the denominator but not the numerator of the calculation. When comparing DNA and RNA sequences, thymine (T) and uracil (U) may be considered equivalent. Identity may be performed manually or by using a computer sequence algorithm such as BLAST or BLAST 2.0.
In Situ Detection
“In situ detection”, as used herein, means the detection of expression or expression levels in the original site, hereby meaning in a tissue sample such as biopsy.
K-Nearest Neighbor
The phrase “K-nearest neighbor” refers to a classification method that classifies a point by calculating the distances between it and points in the training data set. It then assigns the point to the class that is most common among its K-nearest neighbors (where K is an integer).
Leaf
A leaf, as used herein, is the terminal group in a classification or decision tree.
Label
“Label”, as used herein, means a composition detectable by spectroscopic, photochemical, biochemical, immunochemical, chemical, or other physical means. For example, useful labels include 32P, fluorescent dyes, electron-dense reagents, enzymes (e.g., as commonly used in an ELISA), biotin, digoxigenin, or haptens and other entities which can be made detectable. A label may be incorporated into nucleic acids and proteins at any position.
Logistic Regression
Logistic regression is part of a category of statistical models called generalized linear models. Logistic regression can allow one to predict a discrete outcome, such as group membership, from a set of variables that may be continuous, discrete, dichotomous, or a mix of any of these. The dependent or response variable can be dichotomous, for example, one of two possible types of cancer. Logistic regression models the natural log of the odds ratio, i.e., the ratio of the probability of belonging to the first group (P) over the probability of belonging to the second group (1−P), as a linear combination of the different expression levels (in log-space). The logistic regression output can be used as a classifier by prescribing that a case or sample will be classified into the first type if P is greater than 0.5 or 50%. Alternatively, the calculated probability P can be used as a variable in other contexts, such as a 1D or 2D threshold classifier.
Metastasis
“Metastasis” means the process by which cancer spreads from the place at which it first arose as a primary tumor to other locations in the body. The metastatic progression of a primary tumor reflects multiple stages, including dissociation from neighboring primary tumor cells, survival in the circulation, and growth in a secondary location.
Neuroendocrine Tumors
“Neuroendocrine tumors” is meant to include all types of tumors from neuroendocrine origin. Examples of neuroendocrine tumors include, but are not limited to lung small cell carcinoma, lung carcinoid, gastrointestinal tract carcinoid, pancreatic islet cell tumor and medullary thyroid carcinoma.
Node
A “node” is a decision point in a classification (i.e., decision) tree. Also, a point in a neural net that combines input from other nodes and produces an output through application of an activation function.
Nucleic Acid
“Nucleic acid” or “oligonucleotide” or “polynucleotide”, as used herein, mean at least two nucleotides covalently linked together. The depiction of a single strand also defines the sequence of the complementary strand. Thus, a nucleic acid also encompasses the complementary strand of a depicted single strand. Many variants of a nucleic acid may be used for the same purpose as a given nucleic acid. Thus, a nucleic acid also encompasses substantially identical nucleic acids and complements thereof. A single strand provides a probe that may hybridize to a target sequence under stringent hybridization conditions. Thus, a nucleic acid also encompasses a probe that hybridizes under stringent hybridization conditions.
Nucleic acids may be single-stranded or double-stranded, or may contain portions of both double-stranded and single-stranded sequences. The nucleic acid may be DNA, both genomic and cDNA, RNA, or a hybrid, where the nucleic acid may contain combinations of deoxyribo- and ribo-nucleotides, and combinations of bases including uracil, adenine, thymine, cytosine, guanine, inosine, xanthine hypoxanthine, isocytosine and isoguanine. Nucleic acids may be obtained by chemical synthesis methods or by recombinant methods.
A nucleic acid will generally contain phosphodiester bonds, although nucleic acid analogs may be included that may have at least one different linkage, e.g., phosphoramidate, phosphorothioate, phosphorodithioate, or O-methylphosphoroamidite linkages and peptide nucleic acid backbones and linkages. Other analog nucleic acids include those with positive backbones, non-ionic backbones and non-ribose backbones, including those described in U.S. Pat. Nos. 5,235,033 and 5,034,506, which are incorporated herein by reference. Nucleic acids containing one or more non-naturally occurring or modified nucleotides are also included within one definition of nucleic acids. The modified nucleotide analog may be located for example at the 5′-end and/or the 3′-end of the nucleic acid molecule. Representative examples of nucleotide analogs may be selected from sugar- or backbone-modified ribonucleotides. It should be noted, however, that also nucleobase-modified ribonucleotides, i.e., ribonucleotides, containing a non-naturally occurring nucleobase instead of a naturally occurring nucleobase such as uridine or cytidine modified at the 5-position, e.g., 5-(2-amino) propyl uridine, 5-bromo uridine; adenosine and guanosine modified at the 8-position, e.g., 8-bromo guanosine; deaza nucleotides, e.g., 7-deaza-adenosine; O- and N-alkylated nucleotides, e.g., N6-methyl adenosine are suitable. The 2′-OH-group may be replaced by a group selected from H, OR, R, halo, SH, SR, NH2, NHR, NR2 or CN, wherein R is C1-C6 alkyl, alkenyl or alkynyl and halo is F, Cl, Br or I. Modified nucleotides also include nucleotides conjugated with cholesterol through, e.g., a hydroxyprolinol linkage as described in Krutzfeldt et al., Nature 2005; 438:685-689, Soutschek et al., Nature 2004; 432:173-178, and U.S. Patent Publication No. 20050107325, which are incorporated herein by reference. Additional modified nucleotides and nucleic acids are described in U.S. Patent Publication No. 20050182005, which is incorporated herein by reference. Modifications of the ribose-phosphate backbone may be done for a variety of reasons, e.g., to increase the stability and half-life of such molecules in physiological environments, to enhance diffusion across cell membranes, or as probes on a biochip. The backbone modification may also enhance resistance to degradation, such as in the harsh endocytic environment of cells. The backbone modification may also reduce nucleic acid clearance by hepatocytes, such as in the liver and kidney. Mixtures of naturally occurring nucleic acids and analogs may be made; alternatively, mixtures of different nucleic acid analogs, and mixtures of naturally occurring nucleic acids and analogs may be made.
Probe
“Probe”, as used herein, means an oligonucleotide capable of binding to a target nucleic acid of complementary sequence through one or more types of chemical bonds, usually through complementary base pairing, usually through hydrogen bond formation. Probes may bind target sequences lacking complete complementarity with the probe sequence depending upon the stringency of the hybridization conditions. There may be any number of base pair mismatches which will interfere with hybridization between the target sequence and the single-stranded nucleic acids described herein. However, if the number of mutations is so great that no hybridization can occur under even the least stringent of hybridization conditions, the sequence is not a complementary target sequence. A probe may be single-stranded or partially single- and partially double-stranded. The strandedness of the probe is dictated by the structure, composition, and properties of the target sequence. Probes may be directly labeled or indirectly labeled such as with biotin to which a streptavidin complex may later bind.
Reference Value
As used herein, the term “reference value” or “reference expression profile” refers to a criterion expression value to which measured values are compared in order to identify a specific cancer. The reference value may be based on the abundance of the nucleic acids, or may be based on a combined metric score thereof.
In preferred embodiments the reference value is determined from statistical analysis of studies that compare microRNA expression with known clinical outcomes.
Sarcoma
Sarcoma is meant to include all types of tumors from sarcoma origin. Examples of sarcoma tumors include, but are not limited to gastrointestinal stromal tumor (GIST), Ewing sarcoma, chondrosarcoma, malignant fibrous histiocytoma (MFH) or fibrosarcoma, osteosarcoma, rhabdomyosarcoma, synovial sarcoma and liposarcoma.
Sensitivity
“Sensitivity”, as used herein, may mean a statistical measure of how well a binary classification test correctly identifies a condition, for example, how frequently it correctly classifies a cancer into the correct class out of two possible classes. The sensitivity for class A is the proportion of cases that are determined to belong to class “A” by the test out of the cases that are in class “A”, as determined by some absolute or gold standard.
Specificity
“Specificity”, as used herein, may mean a statistical measure of how well a binary classification test correctly identifies a condition, for example, how frequently it correctly classifies a cancer into the correct class out of two possible classes. The specificity for class A is the proportion of cases that are determined to belong to class “not A” by the test out of the cases that are in class “not A”, as determined by some absolute or gold standard.
Stringent Hybridization Conditions
“Stringent hybridization conditions”, as used herein, mean conditions under which a first nucleic acid sequence (e.g., probe) will hybridize to a second nucleic acid sequence (e.g., target), such as in a complex mixture of nucleic acids. Stringent conditions are sequence-dependent and will be different in different circumstances. Stringent conditions may be selected to be about 5-10° C. lower than the thermal melting point (Tm) for the specific sequence at a defined ionic strength pH. The Tm may be the temperature (under defined ionic strength, pH, and nucleic concentration) at which 50% of the probes complementary to the target hybridize to the target sequence at equilibrium (as the target sequences are present in excess, at Tm, 50% of the probes are occupied at equilibrium). Stringent conditions may be those in which the salt concentration is less than about 1.0 M sodium ion, such as about 0.01-1.0 M sodium ion concentration (or other salts) at pH 7.0 to 8.3 and the temperature is at least about 30° C. for short probes (e.g., about 10-50 nucleotides) and at least about 60° C. for long probes (e.g., greater than about 50 nucleotides). Stringent conditions may also be achieved with the addition of destabilizing agents such as formamide. For selective or specific hybridization, a positive signal may be at least 2 to 10 times background hybridization. Exemplary stringent hybridization conditions include the following: 50% formamide, 5×SSC, and 1% SDS, incubating at 42° C., or, 5×SSC, 1% SDS, incubating at 65° C., with wash in 0.2×SSC, and 0.1% SDS at 65° C.
Substantially Complementary
“Substantially complementary”, as used herein, means that a first sequence is at least 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 97%, 98% or 99% identical to the complement of a second sequence over a region of 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100 or more nucleotides, or that the two sequences hybridize under stringent hybridization conditions.
Substantially Identical
“Substantially identical”, as used herein, means that a first and a second sequence are at least 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 97%, 98% or 99% identical over a region of 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100 or more nucleotides or amino acids, or with respect to nucleic acids, if the first sequence is substantially complementary to the complement of the second sequence.
Subject
As used herein, the term “subject” refers to a mammal, including both human and other mammals. The methods of the present invention are preferably applied to human subjects.
Target Nucleic Acid
“Target nucleic acid”, as used herein, means a nucleic acid or variant thereof that may be bound by another nucleic acid. A target nucleic acid may be a DNA sequence. The target nucleic acid may be RNA. The target nucleic acid may comprise a mRNA, tRNA, shRNA, siRNA or Piwi-interacting RNA, or a pri-miRNA, pre-miRNA, miRNA, or anti-miRNA.
The target nucleic acid may comprise a target miRNA binding site or a variant thereof. One or more probes may bind the target nucleic acid. The target binding site may comprise 5-100 or 10-60 nucleotides. The target binding site may comprise a total of 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30-40, 40-50, 50-60, 61, 62 or 63 nucleotides. The target site sequence may comprise at least 5 nucleotides of the sequence of a target miRNA binding site disclosed in U.S. patent application Ser. Nos. 11/384,049, 11/418,870 or 11/429,720, the contents of which are incorporated herein.
1D/2D Threshold Classifier
“1D/2D threshold classifier”, as used herein, may mean an algorithm for classifying a case or sample such as a cancer sample into one of two possible types such as two types of cancer. For a 1D threshold classifier, the decision is based on one variable and one predetermined threshold value; the sample is assigned to one class if the variable exceeds the threshold and to the other class if the variable is less than the threshold. A 2D threshold classifier is an algorithm for classifying into one of two types based on the values of two variables. A threshold may be calculated as a function (usually a continuous or even a monotonic function) of the first variable; the decision is then reached by comparing the second variable to the calculated threshold, similar to the 1D threshold classifier.
Tissue Sample
As used herein, a tissue sample is tissue obtained from a tissue biopsy using methods well known to those of ordinary skill in the related medical arts. The phrase “suspected of being cancerous”, as used herein, means a cancer tissue sample believed by one of ordinary skill in the medical arts to contain cancerous cells. Methods for obtaining the sample from the biopsy include gross apportioning of a mass, microdissection, laser-based microdissection, or other art-known cell-separation methods.
Tumor
“Tumor”, as used herein, refers to all neoplastic cell growth and proliferation, whether malignant or benign, and all pre-cancerous and cancerous cells and tissues.
Variant
“Variant”, as used herein, referring to a nucleic acid means (i) a portion of a referenced nucleotide sequence; (ii) the complement of a referenced nucleotide sequence or portion thereof; (iii) a nucleic acid that is substantially identical to a referenced nucleic acid or the complement thereof; or (iv) a nucleic acid that hybridizes under stringent conditions to the referenced nucleic acid, complement thereof, or a sequence substantially identical thereto.
Wild Type
As used herein, the term “wild-type” sequence refers to a coding, a non-coding or an interface sequence which is an allelic form of sequence that performs the natural or normal function for that sequence. Wild-type sequences include multiple allelic forms of a cognate sequence, for example, multiple alleles of a wild type sequence may encode silent or conservative changes to the protein sequence that a coding sequence encodes.
The present invention employs miRNAs for the identification, classification and diagnosis of specific cancers and the identification of their tissues of origin.
1. microRNA Processing
A gene coding for microRNA (miRNA) may be transcribed leading to production of a miRNA primary transcript known as the pri-miRNA. The pri-miRNA may comprise a hairpin with a stem and loop structure. The stem of the hairpin may comprise mismatched bases. The pri-miRNA may comprise several hairpins in a polycistronic structure.
The hairpin structure of the pri-miRNA may be recognized by Drosha, which is an RNase III endonuclease. Drosha may recognize terminal loops in the pri-miRNA and cleave approximately two helical turns into the stem to produce a 60-70 nt precursor known as the pre-miRNA. Drosha may cleave the pri-miRNA with a staggered cut typical of RNase III endonucleases yielding a pre-miRNA stem loop with a 5′ phosphate and ˜2 nucleotide 3′ overhang. Approximately one helical turn of stem (˜10 nucleotides) extending beyond the Drosha cleavage site may be essential for efficient processing. The pre-miRNA may then be actively transported from the nucleus to the cytoplasm by Ran-GTP and the export receptor Ex-portin-5.
The pre-miRNA may be recognized by Dicer, which is also an RNase III endonuclease. Dicer may recognize the double-stranded stem of the pre-miRNA. Dicer may also cut off the terminal loop two helical turns away from the base of the stem loop, leaving an additional 5′ phosphate and a ˜2 nucleotide 3′ overhang. The resulting siRNA-like duplex, which may comprise mismatches, comprises the mature miRNA and a similar-sized fragment known as the miRNA*. The miRNA and miRNA* may be derived from opposing arms of the pri-miRNA and pre-miRNA. MiRNA* sequences may be found in libraries of cloned miRNAs, but typically at lower frequency than the miRNAs.
Although initially present as a double-stranded species with miRNA*, the miRNA may eventually become incorporated as a single-stranded RNA into a ribonucleoprotein complex known as the RNA-induced silencing complex (RISC). Various proteins can form the RISC, which can lead to variability in specificity for miRNA/miRNA* duplexes, binding site of the target gene, activity of miRNA (repress or activate), and which strand of the miRNA/miRNA* duplex is loaded in to the RISC.
When the miRNA strand of the miRNA:miRNA* duplex is loaded into the RISC, the miRNA* may be removed and degraded. The strand of the miRNA:miRNA* duplex that is loaded into the RISC may be the strand whose 5′ end is less tightly paired. In cases where both ends of the miRNA:miRNA* have roughly equivalent 5′ pairing, both miRNA and miRNA* may have gene silencing activity.
The RISC may identify target nucleic acids based on high levels of complementarity between the miRNA and the mRNA, especially by nucleotides 2-7 of the miRNA. Only one case has been reported in animals where the interaction between the miRNA and its target was along the entire length of the miRNA. This was shown for miR-196 and Hox B8 and it was further shown that miR-196 mediates the cleavage of the Hox B8 mRNA (Yekta et al. Science 2004; 304:594-596). Otherwise, such interactions are known only in plants (Bartel & Bartel 2003; 132:709-717).
A number of studies have looked at the base-pairing requirement between miRNA and its mRNA target for achieving efficient inhibition of translation (reviewed by Bartel 2004; 116:281-297). In mammalian cells, the first 8 nucleotides of the miRNA may be important (Doench & Sharp GenesDev 2004; 18:504-511). However, other parts of the microRNA may also participate in mRNA binding. Moreover, sufficient base pairing at the 3′ can compensate for insufficient pairing at the 5′ (Brennecke et al., PloS Biol 2005; 3:e85). Computation studies, analyzing miRNA binding on whole genomes have suggested a specific role for bases 2-7 at the 5′ of the miRNA in target binding but the role of the first nucleotide, found usually to be “A” was also recognized (Lewis et al. Cell 2005; 120:15-20). Similarly, nucleotides 1-7 or 2-8 were used to identify and validate targets by Krek et al. (Nat Genet 2005; 37:495-500).
The target sites in the mRNA may be in the 5′ UTR, the 3′ UTR or in the coding region. Interestingly, multiple miRNAs may regulate the same mRNA target by recognizing the same or multiple sites. The presence of multiple miRNA binding sites in most genetically identified targets may indicate that the cooperative action of multiple RISCs provides the most efficient translational inhibition.
miRNAs may direct the RISC to down-regulate gene expression by either of two mechanisms: mRNA cleavage or translational repression. The miRNA may specify cleavage of the mRNA if the mRNA has a certain degree of complementarity to the miRNA. When a miRNA guides cleavage, the cut may be between the nucleotides pairing to residues 10 and 11 of the miRNA. Alternatively, the miRNA may repress translation if the miRNA does not have the requisite degree of complementarity to the miRNA. Translational repression may be more prevalent in animals since animals may have a lower degree of complementarity between the miRNA and binding site.
It should be noted that there may be variability in the 5′ and 3′ ends of any pair of miRNA and miRNA*. This variability may be due to variability in the enzymatic processing of Drosha and Dicer with respect to the site of cleavage. Variability at the 5′ and 3′ ends of miRNA and miRNA* may also be due to mismatches in the stem structures of the pri-miRNA and pre-miRNA. The mismatches of the stem strands may lead to a population of different hairpin structures. Variability in the stem structures may also lead to variability in the products of cleavage by Drosha and Dicer.
2. Nucleic Acids
Nucleic acids are provided herein. The nucleic acids comprise the sequences of SEQ ID NOS: 1-390 or variants thereof. The variant may be a complement of the referenced nucleotide sequence. The variant may also be a nucleotide sequence that is substantially identical to the referenced nucleotide sequence or the complement thereof. The variant may also be a nucleotide sequence which hybridizes under stringent conditions to the referenced nucleotide sequence, complements thereof, or nucleotide sequences substantially identical thereto.
The nucleic acid may have a length of from about 10 to about 250 nucleotides. The nucleic acid may have a length of at least 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 125, 150, 175, 200 or 250 nucleotides. The nucleic acid may be synthesized or expressed in a cell (in vitro or in vivo) using a synthetic gene described herein. The nucleic acid may be synthesized as a single-strand molecule and hybridized to a substantially complementary nucleic acid to form a duplex. The nucleic acid may be introduced to a cell, tissue or organ in a single- or double-stranded form or capable of being expressed by a synthetic gene using methods well known to those skilled in the art, including as described in U.S. Pat. No. 6,506,559, which is incorporated herein by reference.
SEQ ID NOs 1-34 are in accordance with Sanger database version 10; SEQ ID NOs 35-390 are in accordance with Sanger database version 11;
3. Nucleic Acid Complexes
The nucleic acid may further comprise one or more of the following: a peptide, a protein, a RNA-DNA hybrid, an antibody, an antibody fragment, a Fab fragment, and an aptamer.
4. Pri-miRNA
The nucleic acid may comprise a sequence of a pri-miRNA or a variant thereof. The pri-miRNA sequence may comprise from 45-30,000, 50-25,000, 100-20,000, 1,000-1,500 or 80-100 nucleotides. The sequence of the pri-miRNA may comprise a pre-miRNA, miRNA and miRNA*, as set forth herein, and variants thereof. The sequence of the pri-miRNA may comprise any of the sequences of SEQ ID NOS: 1-390 or variants thereof.
The pri-miRNA may comprise a hairpin structure. The hairpin may comprise a first and a second nucleic acid sequence that are substantially complimentary. The first and second nucleic acid sequence may be from 37-50 nucleotides. The first and second nucleic acid sequence may be separated by a third sequence of from 8-12 nucleotides. The hairpin structure may have a free energy of less than −25 Kcal/mole, as calculated by the Vienna algorithm with default parameters, as described in Hofacker et al. (Monatshefte f. Chemie 1994; 125:167-188), the contents of which are incorporated herein by reference. The hairpin may comprise a terminal loop of 4-20, 8-12 or 10 nucleotides. The pri-miRNA may comprise at least 19% adenosine nucleotides, at least 16% cytosine nucleotides, at least 23% thymine nucleotides and at least 19% guanine nucleotides.
5. Pre-miRNA
The nucleic acid may also comprise a sequence of a pre-miRNA or a variant thereof. The pre-miRNA sequence may comprise from 45-90, 60-80 or 60-70 nucleotides. The sequence of the pre-miRNA may comprise a miRNA and a miRNA* as set forth herein. The sequence of the pre-miRNA may also be that of a pri-miRNA excluding from 0-160 nucleotides from the 5′ and 3′ ends of the pri-miRNA. The sequence of the pre-miRNA may comprise the sequence of SEQ ID NOS: 1-390 or variants thereof.
6. miRNA
The nucleic acid may also comprise a sequence of a miRNA (including miRNA*) or a variant thereof. The miRNA sequence may comprise from 13-33, 18-24 or 21-23 nucleotides. The miRNA may also comprise a total of at least 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 or 40 nucleotides. The sequence of the miRNA may be the first 13-33 nucleotides of the pre-miRNA. The sequence of the miRNA may also be the last 13-33 nucleotides of the pre-miRNA. The sequence of the miRNA may comprise the sequence of SEQ ID NOS: 1-69, 146-148, 152-390 or variants thereof.
7. Probes
A probe comprising a nucleic acid described herein is also provided. Probes may be used for screening and diagnostic methods, as outlined below. The probe may be attached or immobilized to a solid substrate, such as a biochip.
The probe may have a length of from 8 to 500, 10 to 100 or 20 to 60 nucleotides. The probe may also have a length of at least 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 120, 140, 160, 180, 200, 220, 240, 260, 280 or 300 nucleotides. The probe may further comprise a linker sequence of from 10-60 nucleotides. The probe may comprise a nucleic acid that is complementary to a sequence selected from the group consisting of SEQ ID NOS: 1-390 or variants thereof.
8. Biochip
A biochip is also provided. The biochip may comprise a solid substrate comprising an attached probe or plurality of probes described herein. The probes may be capable of hybridizing to a target sequence under stringent hybridization conditions. The probes may be attached at spatially defined addresses on the substrate. More than one probe per target sequence may be used, with either overlapping probes or probes to different sections of a particular target sequence. The probes may be capable of hybridizing to target sequences associated with a single disorder appreciated by those in the art. The probes may either be synthesized first, with subsequent attachment to the biochip, or may be directly synthesized on the biochip.
The solid substrate may be a material that may be modified to contain discrete individual sites appropriate for the attachment or association of the probes and is amenable to at least one detection method. Representative examples of substrates include glass and modified or functionalized glass, plastics (including acrylics, polystyrene and copolymers of styrene and other materials, polypropylene, polyethylene, polybutylene, polyurethanes, TeflonJ, etc.), polysaccharides, nylon or nitrocellulose, resins, silica or silica-based materials including silicon and modified silicon, carbon, metals, inorganic glasses and plastics. The substrates may allow optical detection without appreciably fluorescing.
The substrate may be planar, although other configurations of substrates may be used as well. For example, probes may be placed on the inside surface of a tube, for flow-through sample analysis to minimize sample volume. Similarly, the substrate may be flexible, such as flexible foam, including closed cell foams made of particular plastics.
The biochip and the probe may be derivatized with chemical functional groups for subsequent attachment of the two. For example, the biochip may be derivatized with a chemical functional group including, but not limited to, amino groups, carboxyl groups, oxo groups or thiol groups. Using these functional groups, the probes may be attached using functional groups on the probes either directly or indirectly using a linker. The probes may be attached to the solid support by either the 5′ terminus, 3′ terminus, or via an internal nucleotide.
The probe may also be attached to the solid support non-covalently. For example, biotinylated oligonucleotides can be made, which may bind to surfaces covalently coated with streptavidin, resulting in attachment. Alternatively, probes may be synthesized on the surface using techniques such as photopolymerization and photolithography.
9. Diagnostics
As used herein, the term “diagnosing” refers to classifying pathology, or a symptom, determining a severity of the pathology (e.g., grade or stage), monitoring pathology progression, forecasting an outcome of pathology and/or prospects of recovery.
As used herein, the phrase “subject in need thereof” refers to an animal or human subject who is known to have cancer, at risk of having cancer (e.g., a genetically predisposed subject, a subject with medical and/or family history of cancer, a subject who has been exposed to carcinogens, occupational hazard, environmental hazard) and/or a subject who exhibits suspicious clinical signs of cancer (e.g., blood in the stool or melena, unexplained pain, sweating, unexplained fever, unexplained loss of weight up to anorexia, changes in bowel habits (constipation and/or diarrhea), tenesmus (sense of incomplete defecation, for rectal cancer specifically), anemia and/or general weakness). Additionally or alternatively, the subject in need thereof can be a healthy human subject undergoing a routine well-being check up.
Analyzing presence of malignant or pre-malignant cells can be effected in vivo or ex vivo, whereby a biological sample (e.g., biopsy, blood) is retrieved. Such biopsy samples comprise cells and may be an incisional or excisional biopsy. Alternatively, the cells may be retrieved from a complete resection.
While employing the present teachings, additional information may be gleaned pertaining to the determination of treatment regimen, treatment course and/or to the measurement of the severity of the disease.
As used herein the phrase “treatment regimen” refers to a treatment plan that specifies the type of treatment, dosage, follow-up plans, schedule and/or duration of a treatment provided to a subject in need thereof (e.g., a subject diagnosed with a pathology). The selected treatment regimen can be an aggressive one which is expected to result in the best clinical outcome (e.g., complete cure of the pathology) or a more moderate one which may relieve symptoms of the pathology yet results in incomplete cure of the pathology. It will be appreciated that in certain cases the treatment regimen may be associated with some discomfort to the subject or adverse side effects (e.g., damage to healthy cells or tissue). The type of treatment can include a surgical intervention (e.g., removal of lesion, diseased cells, tissue, or organ), a cell replacement therapy, an administration of a therapeutic drug (e.g., receptor agonists, antagonists, hormones, chemotherapy agents) in a local or a systemic mode, an exposure to radiation therapy using an external source (e.g., external beam) and/or an internal source (e.g., brachytherapy) and/or any combination thereof. The dosage, schedule and duration of treatment can vary, depending on the severity of pathology and the selected type of treatment, and those of skill in the art are capable of adjusting the type of treatment with the dosage, schedule and duration of treatment.
A method of diagnosis is also provided. The method comprises detecting an expression level of a specific cancer-associated nucleic acid in a biological sample. The sample may be derived from a patient. Diagnosis of a specific cancer state in a patient may allow for prognosis and selection of therapeutic strategy. Further, the developmental stage of cells may be classified by determining temporarily expressed specific cancer-associated nucleic acids.
In situ hybridization of labeled probes to tissue arrays may be performed. When comparing the fingerprints between individual samples the skilled artisan can make a diagnosis, a prognosis, or a prediction based on the findings. It is further understood that the nucleic acid sequences which indicate the diagnosis may differ from those which indicate the prognosis and molecular profiling of the condition of the cells or exosomes may lead to distinctions between responsive or refractory conditions or may be predictive of outcomes.
10. Kits
A kit is also provided and may comprise a nucleic acid described herein together with any or all of the following: assay reagents, buffers, probes and/or primers, and sterile saline or another pharmaceutically acceptable emulsion and suspension base. In addition, the kits may include instructional materials containing directions (e.g., protocols) for the practice of the methods described herein. The kit may further comprise a software package for data analysis of expression profiles.
For example, the kit may be a kit for the amplification, detection, identification or quantification of a target nucleic acid sequence. The kit may comprise a poly (T) primer, a forward primer, a reverse primer, and a probe.
Any of the compositions described herein may be comprised in a kit. In a non-limiting example, reagents for isolating miRNA, labeling miRNA, and/or evaluating a miRNA population using an array are included in a kit. The kit may further include reagents for creating or synthesizing miRNA probes. The kits will thus comprise, in suitable container means, an enzyme for labeling the miRNA by incorporating labeled nucleotide or unlabeled nucleotides that are subsequently labeled. It may also include one or more buffers, such as reaction buffer, labeling buffer, washing buffer, or a hybridization buffer, compounds for preparing the miRNA probes, components for in situ hybridization and components for isolating miRNA. Other kits of the invention may include components for making a nucleic acid array comprising miRNA, and thus may include, for example, a solid support.
The following examples are presented in order to more fully illustrate some embodiments of the invention. They should, in no way be construed, however, as limiting the broad scope of the invention.
1300 primary and metastatic tumor FFPE were used in the study. Tumor samples were obtained from several sources. Institutional review approvals were obtained for all samples in accordance with each institute's institutional review board or IRB equivalent guidelines. Samples included primary tumors and metastases of defined origins, according to clinical records. Tumor content was at least 50% for >95% of samples, as determined by a pathologist based on hematoxylin-eosin (H&E) stained slides.
For FFPE samples, total RNA was isolated from seven to ten 10-μm-thick tissue sections using the miR extraction protocol developed at Rosetta Genomics. Briefly, the sample was incubated a few times in xylene at 57° C. to remove paraffin excess, followed by ethanol washes. Proteins were degraded by proteinase K solution at 45° C. for a few hours. The RNA was extracted with acid phenol:chloroform followed by ethanol precipitation and DNAse digestion. Total RNA quantity and quality was checked by spectrophotometer (Nanodrop ND-1000).
Custom microarrays (Agilent Technologies, Santa Clara, Calif.) were produced by printing DNA oligonucleotide probes to: 982 miRs sequences, 17 negative controls, 23 spikes, and 10 positive controls (total of 1032 probes). Each probe, printed in triplicate, carried up to 28-nucleotide (nt) linker at the 3′ end of the microRNA's complement sequence. 17 negative control probes were designed using as sequences which do not match the genome. Two groups of positive control probes were designed to hybridize to miR array: (i) synthetic small RNAs were spiked to the RNA before labeling to verify the labeling efficiency; and (ii) probes for abundant small RNA (e.g., small nuclear RNAs (U43, U24, Z30, U6, U48, U44), 5.8 s and 5 s ribosomal RNA are spotted on the array to verify RNA quality.
4. Cy-Dye Labeling of miRNA for miR Array
One μg of total RNA were labeled by ligation (Thomson et al. Nature Methods 2004; 1:47-53) of an RNA-linker, p-rCrU-Cy/dye (Eurogentec or equivalent), to the 3′ end with Cy3 or Cy5. The labeling reaction contained total RNA, spikes (0.1-100 fmoles), 400 ng RNA-linker-dye, 15% DMSO, lx ligase buffer and 20 units of T4 RNA ligase (NEB), and proceeded at 4° C. for 1 h, followed by 1 h at 37° C., followed by 4° C. up to 40 min.
The labeled RNA was mixed with 30 μl hybridization mixture (mixture of 45 μL of the 10× GE Agilent Blocking Agent and 246 μL of 2×Hi-RPM Hybridization). The labeling mixture was incubated at 100° C. for 5 minutes followed by ice incubation in water bath for 5 minutes. Slides were Hybridize at 54° C. for 16-20 hours, followed by two washes. The first wash was conducted at room temperature with Agilent GE Wash Buffer 1 for 5 min followed by a second wash with Agilent GE Wash Buffer 2 at 37° C. for 5 min.
Arrays were scanned using an Agilent Microarray Scanner Bundle G2565BA (resolution of 5 μm at XDR Hi 100%, XDR Lo 5%). Array images were analyzed using Feature Extraction 10.7 software (Agilent).
Triplicate spots were combined to produce one signal for each probe by taking the logarithmic mean of reliable spots. All data were log 2-transformed and the analysis was performed in log 2-space. A reference data vector for normalization R was calculated by taking the median expression level for each probe across all samples. For each sample data vector S, a 2nd degree polynomial F was found so as to provide the best fit between the sample data and the reference data, such that R≈F(S). Remote data points (“outliers”) were not used for fitting the polynomial F. For each probe in the sample (element Si in the vector S), the normalized value (in log-space) Mi was calculated from the initial value Si by transforming it with the polynomial function F, so that Mi=F(Si).
The aim of a logistic regression model is to use several features, such as expression levels of several microRNAs, to assign a probability of belonging to one of two possible groups, such as two branches of a node in a binary decision-tree. Logistic regression models the natural log of the odds ratio, i.e., the ratio of the probability of belonging to the first group, for example, the left branch in a node of a binary decision-tree (P) over the probability of belonging to the second group, for example, the right branch in such a node (1−P), as a linear combination of the different expression levels (in log-space). The logistic regression assumes that:
where β0 is the bias, Mi is the expression level (normalized, in log 2-space) of the i-th microRNA used in the decision node, and βi its corresponding coefficient. βi>0 indicates that the probability to take the left branch (P) increases when the expression level of this microRNA (Mi) increases, and the opposite for βi<0. If a node uses only a single microRNA (M), then solving for P results in:
The regression error on each sample is the difference between the assigned probability P and the true “probability” of this sample, i.e., 1 if this sample is in the left branch group and 0 otherwise. The training and optimization of the logistic regression model calculates the parameters β and the p-values [for each microRNA by the Wald statistic and for the overall model by the χ2 (chi-square) difference], maximizing the likelihood of the data given the model and minimizing the total regression error
The probability output of the logistic model is here converted to a binary decision by comparing P to a threshold, denoted by PTH i.e., P>PTH then the sample belongs to the left branch (“first group”) and vice versa. Choosing at each node the branch which has a probability >0.5, i.e., using a probability threshold of 0.5, leads to a minimization of the sum of the regression errors. However, as the goal was the minimization of the overall number of misclassifications (and not of their probability), a modification which adjusts the probability threshold (PTH) was used in order to minimize the overall number of mistakes at each node (Table 2). For each node the threshold to a new probability threshold PTH was optimized such that the number of classification errors is minimized. This change of probability threshold is equivalent (in terms of classifications) to a modification of the bias β0, which may reflect a change in the prior frequencies of the classes. Once the threshold was chosen β0 was modified such that the threshold will be shifted back to 0.5. In addition, β0, β1, β2, . . . were adjusted so that the slope of the log of the odds ratio function is limited.
The original data contain the expression levels of multiple microRNAs for each sample, i.e., multiple of data features. In training the classifier for each node, only a small subset of these features was selected and used for optimizing a logistic regression model. In the initial training this was done using a forward stepwise scheme. The features were sorted in order of decreasing log-likelihoods, and the logistic model was started off and optimized with the first feature. The second feature was then added, and the model re-optimized. The regression error of the two models was compared: if the addition of the feature did not provide a significant advantage (a χ2 difference less than 7.88, p-value of 0.005), the new feature was discarded. Otherwise, the added feature was kept. Adding a new feature may make a previous feature redundant (e.g., if they are very highly correlated). To check for this, the process iteratively checks if the feature with lowest likelihood can be discarded (without losing χ2 difference as above). After ensuring that the current set of features is compact in this sense, the process continues to test the next feature in the sorted list, until features are exhausted. No limitation on the number of features was inserted into the algorithm.
The stepwise logistic regression method was used on subsets of the training set samples by re-sampling the training set with repetition (“bootstrap”), so that each of the 20 runs contained somewhat different training set. All the features that took part in one of the 20 models were collected. A robust set of 1-3 features per each node was selected by comparing features that were repeatedly chosen in the bootstrap sets to previous evidence, and considering their signal strengths and reliability. When using these selected features to construct the classifier, the stepwise process was not used and the training optimized the logistic regression model parameters only.
The KNN algorithm (see e.g., Ma et al., Arch Pathol Lab Med 2006; 130:465-73) calculates the distance (Pearson correlation) of any sample to all samples in the training set, and classifies the sample by the majority vote of the k samples which are most similar (k being a parameter of the classifier). The correlation is calculated on the pre-defined set of microRNAs (the microRNAs that were used by the decision-tree). KNN algorithms with k=1;10 were compared, and the optimal performer was selected, using k=5. The KNN was based on comparing the expression of all 65 microRNAs in each sample to all other samples in the training database.
The decision-tree and KNN each return a predicted tissue of origin and histological type where applicable. The tissue of origin and histological type may be one of the exact origins and types in the training or a variant thereof. For example, whereas the training includes brain oligodendrioglioma and brain astrocytoma, the answer may simply be brain carcinoma. In addition to the tissue of origin and histological type, the KNN and decision-tree each return a confidence measure. The KNN returns the number of samples within the K nearest neighbors that agreed with the answer reported by the KNN (denoted by V), and the decision-tree returns the probability of the result (P), which is the multiplication of the probabilities at each branch point made on the way to that answer. The classifier returns the two different predictions or a single prediction in case the predictions concur, can be unified into a single answer (for example into the prediction brain if the KNN returned brain oligodendrioglioma and the decision-tree brain astrocytoma), or if based on V and P, one answer is chosen to override the other.
A tumor classifier was built using the microRNA expression levels by applying a binary tree classification scheme (
For each of the individual nodes logistic regression models were used, a robust family of classifiers which are frequently used in epidemiological and clinical studies to combine continuous data features into a binary decision (
Expression of miRs Provides for Distinguishing Between Tumors
hsa-miR-372 (SEQ ID NO: 55) is used at node 1 of the binary-tree-classifier detailed in the invention to distinguish between germ-cell tumors and all other tumors.
hsa-miR-122 (SEQ ID NO: 6) is used at node 2 of the binary-tree-classifier detailed in the invention to distinguish between hepatobiliary tumors and non germ-cell non-hepatobiliary tumors.
hsa-miR-126 (SEQ ID NO: 9) and hsa-miR-200b (SEQ ID NO: 29) are used at node 3 of the binary-tree-classifier detailed in the invention to distinguish between liver tumors and biliary-tract carcinoma.
A combination of the expression level of any of the miRs detailed in table 6 with the expression level of any of hsa-miR-30a (SEQ ID NO: 46), hsa-miR-10b (SEQ ID NO: 5) and hsa-miR-140-3p (SEQ ID NO: 12) also provides for distinguishing between tumors from epithelial origins and tumors from non-epithelial origins. This is demonstrated at node 4 of the binary-tree-classifier detailed in the invention with hsa-miR-200c (SEQ ID NO: 30) and hsa-miR-30a (SEQ ID NO: 46) (
hsa-miR-146a (SEQ ID NO: 16), hsa-let-7e (SEQ ID NO: 2) and hsa-miR-30a (SEQ ID NO: 46) are used at node 5 of the binary-tree-classifier detailed in the invention to distinguish between the group consisting of melanoma and lymphoma, and the group consisting of all other non-epithelial tumors.
hsa-miR-9* (SEQ ID NO: 66) and hsa-miR-92b (SEQ ID NO: 68) are used at node 6 of the binary-tree-classifier detailed in the invention to distinguish between brain tumors and the group consisting of all non-brain, non-epithelial tumors.
A combination of the expression level of any of the miRs detailed in table 9 with the expression level of hsa-miR-497 (SEQ ID NO: 208) or hsa-let-7d (SEQ ID NO: 153) also provides for classification of brain tumors as astrocytic tumors or oligodendrogliomas. This is demonstrated at node 7 of the binary-tree-classifier detailed in the invention with hsa-miR-222 (SEQ ID NO: 40) and hsa-miR-497 (SEQ ID NO: 208). In another embodiment of the invention, the expression levels of hsa-miR-222 (SEQ ID NO: 40) and hsa-let-7d (SEQ ID NO: 153) are combined to distinguish between astrocytic tumors and oligodendrogliomas.
hsa-miR-375 (SEQ ID NO: 56), hsa-miR-7 (SEQ ID NO: 65) and hsa-miR-193a-3p (SEQ ID NO: 25) are used at node 8 of the binary-tree-classifier detailed in the invention to distinguish between the group consisting of neuroendocrine tumors and the group consisting of all non-neuroendocrine, epithelial tumors.
hsa-miR-143 (SEQ ID NO: 14) and hsa-miR-181a (SEQ ID NO: 21) are used at node 10 of the binary-tree-classifier detailed in the invention to distinguish between prostate tumors and all other non-GI epithelial tumors.
A combination of the expression level of any of the miRs detailed in table 13 with the expression level of hsa-miR-200b (SEQ ID NO: 29), hsa-miR-200a (SEQ ID NO: 28), hsa-miR-516a-5p (SEQ ID NO: 211), hsa-miR-767-5p (SEQ ID NO: 227), hsa-miR-518a-3p (SEQ ID NO: 215), hsa-miR-520d-5p (SEQ ID NO: 222), hsa-miR-519a (SEQ ID NO: 218) and hsa-miR-517c (SEQ ID NO: 214) also provides for classification of seminoma and non-seminoma testis-tumors.
hsa-miR-516a-5p (SEQ ID NO: 211) and hsa-miR-200b (SEQ ID NO: 29) are used at node 12 of the binary-tree-classifier detailed in the invention to distinguish between seminoma and non-seminoma testis-tumors.
Node 13 of the binary-tree-classifier separates tissues with high expression of miR-205 (SCC marker) such as SCC, TCC and thymomas from adenocarcinomas.
Breast adenocarcinoma and ovarian carcinoma are excluded from this separation due to a wide range of expression of miR-205.
A combination of the expression level of any of the miRs detailed in table 14 with the expression level of hsa-miR-331-3p (SEQ ID NO: 197) also provides for this classification.
hsa-miR-205 (SEQ ID NO: 32), hsa-miR-345 (SEQ ID NO: 51) and hsa-miR-125a-5p (SEQ ID NO: 7) are used at node 13 of the binary-tree-classifier detailed in the invention.
hsa-miR-193a-3p (SEQ ID NO: 25), hsa-miR-375 (SEQ ID NO: 56) and hsa-miR-342-3p (SEQ ID NO: 50) are used at node 14 of the binary-tree-classifier detailed in the invention. According to another embodiment, hsa-miR-193a-3p (SEQ ID NO: 25), hsa-miR-375 (SEQ ID NO: 56) and hsa-miR-224 (SEQ ID NO: 42) may be used at node 14 of the binary-tree-classifier detailed in the invention.
hsa-miR-205 (SEQ ID NO: 32), hsa-miR-10a (SEQ ID NO: 4) and hsa-miR-22 (SEQ ID NO: 39) are used at node 15 of the binary-tree-classifier detailed in the invention.
hsa-miR-934 (SEQ ID NO: 69), hsa-miR-191 (SEQ ID NO: 24) and hsa-miR-29c (SEQ ID NO: 191) are used at node #21 of the binary-tree-classifier detailed in the invention to distinguish between TCC and SCC.
hsa-miR-10b (SEQ ID NO: 5), hsa-miR-138 (SEQ ID NO: 11) and hsa-miR-185 (SEQ ID NO: 23) are used at node 23 of the binary-tree-classifier detailed in the invention to distinguish between anus or skin SCC and upper SCC tumors.
hsa-miR-30e (SEQ ID NO: 48) and hsa-miR-21* (SEQ ID NO: 35) are used at node 25 of the binary-tree-classifier detailed in the invention to distinguish between B-cell lymphoma and T-cell lymphoma.
hsa-miR-17 (SEQ ID NO: 20) and hsa-miR-29c* (SEQ ID NO: 45) are used at node #26 of the binary-tree-classifier detailed in the invention to distinguish between lung small cell carcinoma and other neuroendocrine tumors.
hsa-miR-652 (SEQ ID NO: 64), hsa-miR-34c-5p (SEQ ID NO: 53) and hsa-miR-214 (SEQ ID NO: 37) are used at node 28 of the binary-tree-classifier detailed in the invention to distinguish between lung carcinoid tumors and GI neuroendocrine tumors.
hsa-miR-21 (SEQ ID NO: 34), and hsa-miR-148a (SEQ ID NO: 18) are used at node 29 of the binary-tree-classifier detailed in the invention to distinguish between pancreatic islet cell tumors and GI neuroendocrine carcinoid tumors.
hsa-miR-345 (SEQ ID NO: 51), hsa-miR-31 (SEQ ID NO: 49) and hsa-miR-146a (SEQ ID NO: 16) are used at node #32 of the binary-tree-classifier detailed in the invention to distinguish between cholangio cancer or adenocarcinoma of extrahepatic biliary tract and pancreatic adenocarcinoma.
hsa-miR-202 (SEQ ID NO: 31), hsa-miR-509-3p (SEQ ID NO: 61) and hsa-miR-214* (SEQ ID NO: 38) are used at node 35 of the binary-tree-classifier detailed in the invention to distinguish between adrenal carcinoma and sarcoma or mesothelioma tumors.
hsa-miR-29C* (SEQ ID NO: 45) and hsa-miR-143 (SEQ ID NO: 14) are used at node 36 of the binary-tree-classifier detailed in the invention to distinguish between GIST and sarcoma or mesothelioma tumors.
hsa-miR-210 (SEQ ID NO: 36) and hsa-miR-221 (SEQ ID NO: 147) are used at node #37 of the binary-tree-classifier detailed in the invention to distinguish between chromophobe renal cell carcinoma tumors and clear cell or papillary renal cell carcinoma tumors.
hsa-miR-31 (SEQ ID NO: 49) and hsa-miR-126 (SEQ ID NO: 9) are used at node 38 of the binary-tree-classifier detailed in the invention to distinguish between renal clear cell and papillary cell carcinoma tumors.
hsa-miR-21* (SEQ ID NO: 35) hsa-miR-130a (SEQ ID NO: 10) and hsa-miR-10b (SEQ ID NO: 5) are used at node 39 of the binary-tree-classifier detailed in the invention to distinguish between pleural mesothelioma tumors and sarcoma tumors.
hsa-miR-100 (SEQ ID NO: 3) hsa-miR-145 (SEQ ID NO: 15) and hsa-miR-222 (SEQ ID NO: 40) are used at node 40 of the binary-tree-classifier detailed in the invention to distinguish between synovial sarcoma tumors and other sarcoma tumors.
hsa-miR-140-3p (SEQ ID NO: 12) and hsa-miR-455-5p (SEQ ID NO: 58) are used at node 41 of the binary-tree-classifier detailed in the invention to distinguish between chondrosarcoma tumors and other non-synovial sarcoma tumors.
hsa-miR-210 (SEQ ID NO: 36) and hsa-miR-193a-5p (SEQ ID NO: 26) are used at node 42 of the binary-tree-classifier detailed in the invention to distinguish between liposarcoma tumors and other non-chondrosarcoma and non-synovial sarcoma tumors.
hsa-miR-181a (SEQ ID NO: 21) is used at node 43 of the binary-tree-classifier detailed in the invention to distinguish between Ewing sarcoma or osteosarcoma tumors and rhabdomyosarcoma, malignant fibrous histiocytoma (MFH) or fibrosarcoma tumors.
The foregoing description of the specific embodiments so fully reveals the general nature of the invention that others can, by applying current knowledge, readily modify and/or adapt for 5 various applications such specific embodiments without undue experimentation and without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be 10 apparent to those skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and broad scope of the appended claims.
It should be understood that the detailed description and specific examples, while indicating preferred embodiments of the invention, are given by way of illustration only, since various changes and modifications within the spirit and scope of the invention will become 15 apparent to those skilled in the art from this detailed description.
| Number | Date | Country | |
|---|---|---|---|
| 60907266 | Mar 2007 | US | |
| 60929244 | Jun 2007 | US | |
| 61024565 | Jan 2008 | US | |
| 61140642 | Dec 2008 | US | |
| 61415875 | Nov 2010 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 15654168 | Jul 2017 | US |
| Child | 15909145 | US | |
| Parent | 14746487 | Jun 2015 | US |
| Child | 15654168 | US | |
| Parent | 13856190 | Apr 2013 | US |
| Child | 14746487 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 13167489 | Jun 2011 | US |
| Child | 13856190 | US | |
| Parent | 12532940 | Sep 2009 | US |
| Child | 13167489 | US | |
| Parent | PCT/IL2009/001212 | Dec 2009 | US |
| Child | 13167489 | US | |
| Parent | PCT/IL2011/000849 | Nov 2011 | US |
| Child | 13856190 | US |
