Methods and systems for detecting colorectal cancer via nucleic acid methylation analysis

Abstract

The present disclosure provides methods and systems for screening or detecting a colorectal cancer or following colorectal disease progression that may be applied to cell-free nucleic acids such as cell-free DNA. The method may use detection of methylation signals within a single sequencing read in identified genomic regions as input features to train a machine learning model and generate a classifier useful for stratifying populations of individuals. The method may comprise extracting DNA from a cell-free sample obtained from a subject, converting the DNA for methylation sequencing, generating sequencing reads, and detecting colon proliferative cell disorder-associated signals in the sequencing information and training a machine learning model to provide a discriminator capable of distinguishing groups in a subject population such as healthy, cancer or distinguishing disease subtype or stage. The method may be used for, e.g., predicting, prognosticating, and/or monitoring response to treatment, tumor load, relapse, or colorectal cancer development.

1 - 49 . (canceled) 
     
     
         50 . A machine learning classifier capable of distinguishing a population of healthy individuals from individuals with colon cell proliferative disorder, comprising:
 a) sets of measured values representative of differentially-methylated genomic regions of claim  1  where the measured values are obtained from methylation sequencing data from healthy subjects and subjects having a colon cell proliferative disorder;   b) wherein the measured values are used to generate a set of features corresponding to properties of the differentially-methylated genomic regions and where the features are inputted to a machine learning or statistical model;   c) wherein the model provides a feature vector useful as a classifier capable of distinguishing a population of healthy individuals from individuals having a colon cell proliferative disorder.   
     
     
         51 . The classifier of  claim 50 , wherein the sets of measured values describe characteristics of the methylated regions selected from the group consisting of: base wise methylation percent for CpG, CHG, CHH, the count or rate of observing fragments with different counts or rates of methylated CpGs in a region, conversion efficiency (100-Mean methylation percent for CHH), hypomethylated blocks, methylation levels (global mean methylation for CPG, CHH, CHG, fragment length, fragment midpoint, number of methylated CpGs per fragment, fraction of CpG methylation to total CpG per fragment, fraction of CpG methylation to total CpG per region, fraction of CpG methylation to total CpG in panel, dinucleotide coverage (normalized coverage of dinucleotide), evenness of coverage (unique CpG sites at 1× and 10× mean genomic coverage (for S4 runs), mean CpG coverage (depth) globally and mean coverage at CpG islands, CGI shelves, and CGI shores. 
     
     
         52 . A system comprising a machine learning model classifier for detecting a colon cell proliferative disorder, comprising:
 a) a computer-readable medium comprising a classifier operable to classify subjects as having the colon cell proliferative disorder or not having the colon cell proliferative disorder based on a methylation signature panel; and   b) one or more processors for executing instructions stored on the computer-readable medium.   
     
     
         53 . The system of  claim 52 , comprising the classifier of  claim 50  loaded into a memory of a computer system, the machine learning model trained using training vectors obtained from training biological samples, a first subset of the training biological samples identified as having a colon cell proliferative disorder and a second subset of the training biological samples identified as not having a colon cell proliferative disorder. 
     
     
         54 . A method for determining a methylation profile of a cell-free deoxyribonucleic acid (cfDNA) sample from an individual, comprising:
 a) providing conditions capable of converting unmethylated cytosines to uracils in nucleic acid molecules of the cfDNA sample to produce a plurality of converted nucleic acids;   b) contacting the plurality of converted nucleic acids with nucleic acid probes complementary to a pre-identified methylation signature panel of at least two differentially methylated regions selected from the group consisting of Tables 1-11 to enrich for sequences corresponding to the signature panel;   c) determining nucleic acid sequences of the plurality of converted nucleic acid molecules; and   d) aligning the nucleic acid sequences of the plurality of converted nucleic acid molecules to a reference nucleic acid sequence, thereby determining the methylation profile of the individual.   
     
     
         55 . The method of  claim 54 , further comprising amplifying the plurality of converted nucleic acids. 
     
     
         56 . The method of  claim 55 , wherein the amplifying comprises polymerase chain reaction (PCR). 
     
     
         57 . The method of  claim 54 , further comprising determining the nucleic acid sequences of the converted nucleic acid molecules at a depth of greater than 1000×, greater than 2000×, greater than 3000×, greater than 4000×, or greater than 5000×. 
     
     
         58 . The method of  claim 54 , wherein the reference nucleic acid sequence is at least a portion of a human reference genome. 
     
     
         59 . The method of  claim 58 , wherein the human reference genome is hg18. 
     
     
         60 . The method of  claim 54 , wherein the pre-identified methylation signature panel includes three or more methylated genomic regions in Tables 1-11, four or more methylated genomic regions in Tables 1-11, five or more methylated genomic regions in Tables 1-11, six or more methylated genomic regions in Tables 1-11, seven or more methylated genomic regions in Tables 1-11, eight or more methylated genomic regions in Tables 1-11, nine or more methylated genomic regions in Tables 1-11, ten or more methylated genomic regions in Tables 1-11, eleven or more methylated genomic regions in Tables 1-11, twelve or more methylated genomic regions in Tables 1-11, or thirteen or more methylated genomic regions in Tables 1-11. 
     
     
         61 . The method of  claim 54 , wherein the pre-identified methylation signature panel comprises methylated genomic regions selected from the group consisting of Table 1, Table 2, Table 3, Table 4, Table 5, Table 6, Table 7, Table 8, Table 9, Table 10, and Table 11. 
     
     
         62 . The method of  claim 54 , wherein the pre-identified methylation signature panel comprise methylated regions selected from the group consisting of IKZF1, KCNQ5, ELMO1, CHST2, PRKCB, and FLIT. 
     
     
         63 . The method of  claim 54 , wherein the pre-identified methylation signature panel comprise methylated regions selected from the group consisting of IKZF1, KCNQ5, and ELMO1. 
     
     
         64 . The method of  claim 54 , wherein the pre-identified methylation signature panel comprise methylated regions selected from the group consisting of IKZF1, KCNQ5, ELMO1, CHST2, PRKCB, FLU, CLIP4, ELOVL5, FAM72B, and ST3GAL1. 
     
     
         65 . The method of  claim 54 , wherein the methylation profile is indicative of a presence or an absence of a colon cell proliferative disorder in the individual. 
     
     
         66 . The method of  claim 65 , wherein the colon cell proliferative disorder is selected from the group consisting of adenoma (adenomatous polyps), sessile serrated adenoma (SSA), advanced adenoma, colorectal dysplasia, colorectal adenoma, colorectal cancer, colon cancer, rectal cancer, colorectal carcinoma, colorectal adenocarcinoma, carcinoid tumors, gastrointestinal carcinoid tumors, gastrointestinal stromal tumors (GISTs), lymphomas, and sarcomas. 
     
     
         67 . The method of  claim 65 , wherein the colon cell proliferative disorder is selected from the group consisting of stage 1 colorectal cancer, stage 2 colorectal cancer, stage 3 colorectal cancer, or stage 4 colorectal cancer. 
     
     
         68 . The method of  claim 65 , further comprising applying a trained machine learning classifier to the methylation profile, wherein the trained machine learning classifier is trained to be capable of distinguishing between healthy individuals and individuals with the colon cell proliferative disorder to provide an output value associated with presence of the colon cell proliferative disorder, thereby detecting a presence or an absence of the colon cell proliferative disorder in the subject. 
     
     
         69 . The method of  claim 68 , further comprising administering a treatment to the individual for the colon cell proliferative disorder based on detecting the presence of the colon cell proliferative disorder in the individual. 
     
     
         70 . The method of  claim 68 , wherein the trained machine learning classifier is selected from the group consisting of a deep learning classifier, a neural network classifier, a linear discriminant analysis (LDA) classifier, a quadratic discriminant analysis (QDA) classifier, a support vector machine (SVM) classifier, a random forest (RF) classifier, a linear kernel support vector machine classifier, a first or second order polynomial kernel support vector machine classifier, a ridge regression classifier, an elastic net algorithm classifier, a sequential minimal optimization algorithm classifier, a naive Bayes algorithm classifier, and a principal component analysis classifier.