Systems and methods of diagnosing idiopathic pulmonary fibrosis on transbronchial biopsies using machine learning and high dimensional transcriptional data

Abstract

The present invention provides systems, methods, and classifiers for differentiating between samples as usual interstitial pneumonia (UIP) or non-UIP.

What is claimed is: 
     
         1 . A method of detecting whether a lung tissue sample is positive for usual interstitial pneumonia (UIP) or non-usual interstitial pneumonia (non-UIP), comprising:
 Assaying the expression level of each of a first group of transcripts and a second group of transcripts in a test sample of a subject, wherein the first group of transcripts includes one or more sequence corresponding to any one of the genes overexpressed in UIP and listed in any of Tables 5, 7, 9, 10, 11, and 12 and the second group of transcripts includes one or more sequence corresponding to any one of the genes under-expressed in UIP and listed in any of Tables 5, 8, 9, 10, 11, or 12; and   comparing the expression level of each of the first group of transcripts and the second group of transcripts with reference expression levels of the corresponding transcripts to (1) classify said lung tissue as usual interstitial pneumonia (UIP) if there is (a) an increase in an expression level corresponding to the first group and/or (b) a decrease in an expression level corresponding to the second group as compared to the reference expression levels, or (2) classify the lung tissue as non-usual interstitial pneumonia (non-UIP) if there is (c) an increase in the expression level corresponding to the second group and/or (d) a decrease in the expression level corresponding to the first group as compared to the reference expression levels.   
     
     
         2 . A method of detecting whether a lung tissue sample is positive for usual interstitial pneumonia (UIP) or non-usual interstitial pneumonia (non-UIP), comprising:
 assaying by sequencing, array hybridization, or nucleic acid amplification the expression level of each of a first group of transcripts and a second group of transcripts in a test sample from a lung tissue of a subject, wherein the first group of transcripts includes one or more sequence corresponding to any one of the genes overexpressed in UIP and listed in any of Tables 5, 7, 9, 10, 1, and 12 and the second group of transcripts includes one or more sequence corresponding to any one of the genes under-expressed in UIP and listed in any of Tables 5, 8, 9, 10, 11, or 12; and   comparing the expression level of each of the first group of transcripts and the second group of transcripts with reference expression levels of the corresponding transcripts to (1) classify said lung tissue as usual interstitial pneumonia (UIP) if there is (a) an increase in an expression level corresponding to the first group and/or (b) a decrease in an expression level corresponding to the second group as compared to the reference expression levels, or (2) classify the lung tissue as non-usual interstitial pneumonia (non-UIP) if there is (c) an increase in the expression level corresponding to the second group and/or (d) a decrease in the expression level corresponding to the first group as compared to the reference expression levels.   
     
     
         3 . A method of detecting whether a lung tissue sample is positive for UIP or non-UIP, comprising:
 measuring the expression level of two or more transcripts expressed in the sample; and   using a computer generated classifier to classify the sample as UIP and non-UIP;   wherein the classifier was trained using a heterogeneous spectrum of non-UIP pathology subtypes comprising HP, NSIP, sarcoidosis, RB, bronchiolitis, and organizing pneumonia (OP).   
     
     
         4 . The method of any one of  claims 1 - 3 , wherein the test sample is a biopsy sample or a bronchoalveolar lavage sample. 
     
     
         5 . The method of any one of  claims 1 - 3 , wherein the test sample is fresh-frozen or fixed. 
     
     
         6 . The method of any one of  claims 1 - 3 , wherein the expression levels are determined by RT-PCR, DNA microarray hybridization, RNASeq, or a combination thereof. 
     
     
         7 . The method of any one of  claims 1 - 3 , wherein the method comprises detecting cDNA produced from RNA expressed in the test sample. 
     
     
         8 . The method of  claim 7 , wherein prior to the detecting step, the cDNA is amplified from a plurality of cDNA transcripts. 
     
     
         9 . The method of any one of  claims 1 - 3 , wherein one or more of the transcripts is labeled. 
     
     
         10 . The method of any one of  claims 1 - 3 , further comprising measuring the expression level of at least one control nucleic acid in the test sample. 
     
     
         11 . The method of any one of  claims 1 - 3 , wherein the lung tissue is classified as any one of interstitial lung diseases (ILD), a particular type of ILD, a non-ILD, or non-diagnostic. 
     
     
         12 . The method of any one of  claims 1 - 3 , wherein the lung tissue is classified as either idiopathic pulmonary fibrosis (IPF) or Nonspecific interstitial pneumonia (NSIP). 
     
     
         13 . The method of  claim 1  or  2 , wherein the method comprises assaying the test sample for the expression level of one or more transcripts of any one of SEQ ID NOS: 1-22. 
     
     
         14 . The method of  claim 13 , further comprising assaying the test sample for the expression level of from 1 to 20 other genes. 
     
     
         15 . The method of  claim 3 , wherein the method comprises assaying the test sample for the expression level of one or more transcripts of any one of SEQ ID NOS: 1-22. 
     
     
         16 . The method of any one of  claims 1 - 2 , further comprising using smoking status as a covariate to the classification step of (1) or (2). 
     
     
         17 . The method of  claim 16 , wherein smoking status is determined by detecting an expression profile indicative of the subject's smoker status. 
     
     
         18 . The method of any one of the preceding claims, wherein classification of the sample comprises detection of the expression levels of one or more transcripts that are susceptible to smoker status bias, and wherein the transcripts that are susceptible to smoker status bias are weighted differently than transcripts that are not susceptible to smoker bias. 
     
     
         19 . The method of any one of the preceding claims, wherein classification of the sample comprises detection of the expression levels of one or more transcripts that are susceptible to smoker status bias, and wherein the transcripts that are susceptible to smoker status bias are excluded from the classification step. 
     
     
         20 . The method of any one of  claims 1 - 2 , wherein the first group comprises 2 or more different transcripts, or 3 or more, 4 or more, 5 or more, 10 or more, 15 or more, 20 or more, or more than 20 different transcripts. 
     
     
         21 . The method of any one of  claims 1 - 2 , wherein the second group comprises 2 or more different transcripts, or 3 or more, 4 or more, 5 or more, 10 or more, 15 or more, 20 or more, or more than 20 different transcripts. 
     
     
         22 . The method of  claim 13  or  15 , comprising detecting 2 or more different transcripts of any one of SEQ ID NOS: 1-22, or 3 or more, 4 or more, 5 or more, 10 or more, 15 or more, 20 or more, or more than 20 different transcripts of any one of SEQ ID NOS: 1-22. 
     
     
         23 . The method of  claim 13  or  15 , comprising assaying the test sample for the expression level of all of the transcripts of SEQ ID NOS: 1-22. 
     
     
         24 . The method of  claim 15 ,  22 , or  23 , further comprising assaying the test sample for the expression level of from 1 to 20 other genes. 
     
     
         25 . The method of  claim 24 , wherein the other genes comprise or consist of HMCN2, ADAMTSL1, CD79B, KEL, KLHL14, MPP2, NMNAT2, PLXDC1, CAPN9, TALDO1, PLK4, IGHV3-72, IGKV1-9, and CNTN4. 
     
     
         26 . The method of  claim 3 , further comprising using smoking status as a covariate to the classification step. 
     
     
         27 . The method of  claim 16  or  27 , wherein the method uses smoking status as a covariate prior to the classification step. 
     
     
         28 . The method of any one of the preceding claims, comprising implementing a classifier trained using one or more feature selected from gene expression, variants, mutations, fusions, loss of heterozygoxity (LOH), and biological pathway effect. 
     
     
         29 . The method of  claim 29 , wherein the classifier is trained using features including gene expression, sequence variants, mutations, fusions, loss of heterozygoxity (LOH), and biological pathway effect. 
     
     
         30 . The method of any one of the preceding claims, wherein the classification step further comprises detecting sequence variants in the test sample and comparing the sequence variants to the respective sequences in a reference sample to classify the sample as UIP or non-UIP.