Methods of treating a subject suffering from rheumatoid arthritis with anti-TNF therapy based in part on a trained machine learning classifier
Inventors
Ghiassian, Susan • Mellors, Theodore R. • Santolini, Marc • Ameli, Asher • Schoenbrunner, Nancy • Akmaev, Viatcheslav R. • Johnson, Keith J.
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12062415-B2
Publication Date
2024-08-13
Expiration Date
Abstract
Presented herein are systems and methods for developing classifiers useful for predicting response to particular treatments. For example, in some embodiments, the present disclosure provides a method of treating subjects suffering from an autoimmune disorder, the method comprising administering an alternative to anti-TNF therapy to subjects who have been determined to be non-responsive via a classifier established to distinguish between responsive and non-responsive prior subjects in a cohort who have received the anti-TNF therapy.
Core Innovation
The invention relates to a method of treating a subject suffering from rheumatoid arthritis with anti-TNF therapy, where the subject is predicted to be responsive to the anti-TNF therapy. The prediction is based at least in part on a trained machine learning classifier that distinguishes between responsive subjects and non-responsive subjects who have received the anti-TNF therapy, based at least in part on analyzing an expression level of a set of genes.
The prediction framework supports cross-cohort and cross-platform model development and validation for predicting rheumatoid arthritis non-response to anti-TNF therapy. Baseline whole-blood gene expression is used, with gene-selection and machine learning workflows, and models trained on microarray-derived genes are retrained on RNASeq data.
The document further describes classifier performance and validation concepts, including independent cohort validation, area under the curve, negative predictive value, true negative rate, accuracy, and confusion-matrix-based evaluation for a clinical endpoint defined using ACR50 at 6 months. A finalized gene set is described for validation in withheld samples from CORRONA, and robustness is characterized using feature selection frequency.
Claims Coverage
The claims cover 2 inventive features directed to classifier-guided anti-TNF treatment of rheumatoid arthritis and gene-expression-based distinction between responsive and non-responsive subjects.
Treating rheumatoid arthritis with classifier-predicted anti-TNF responsiveness
Administering anti-TNF therapy to a subject suffering from rheumatoid arthritis where the subject has been predicted to be responsive to the anti-TNF therapy based at least in part on a trained machine learning classifier that distinguishes between responsive subjects and non-responsive subjects who have received the anti-TNF therapy.
Gene-expression-based responsive versus non-responsive distinction
Distinguishing between responsive subjects and non-responsive subjects based at least in part on analyzing an expression level of a set of genes of the responsive subjects and the non-responsive subjects.
The claims focus on treating rheumatoid arthritis with anti-TNF therapy guided by a trained machine learning classifier and on using gene expression levels from a set of genes to distinguish responsive from non-responsive subjects.
Stated Advantages
Enables predicting which rheumatoid arthritis subjects are responsive to anti-TNF therapy before treatment based on gene expression levels of a set of genes.
Supports improving prediction by incorporating SNPs and/or clinical characteristics into the trained machine learning classifier.
Directs non-responders to alternative therapies rather than administering anti-TNF therapy to those predicted to be non-responsive.
Provides predictive model validation concepts using independent cohort validation and quantitative performance thresholds such as AUC, NPV, TNR, and accuracy.
Supports cross-platform development concepts, including microarray discovery and RNA-seq retraining, with modular RNA-seq bioinformatics pipeline traceability and validation.
Improved prediction performance for anti-TNF non-response using gene-expression-based machine learning, reported via AUC, negative predictive value, and true negative rate, including RNA-only models outperforming baseline covariates.
Documented Applications
Treating a subject suffering from rheumatoid arthritis using anti-TNF therapy after prediction of responsiveness versus non-responsiveness by a trained machine learning classifier based at least in part on gene expression levels.
Using the prediction to direct non-responders to alternative therapies such as rituximab, sarilumab, tofacitinib, leflunomide, vedolizumab, tocilizumab, anakinra, abatacept.
Developing and validating classifier models using independent cohort validation and performance thresholds (e.g., AUC/NPV/TNR/accuracy), including cross-platform development concepts from microarray discovery to RNA-seq retraining.
Predicting rheumatoid arthritis non-response to anti-TNF therapy using baseline whole-blood gene expression and machine learning classifier workflows, including cross-cohort and cross-platform retraining and validation in CORRONA withheld samples.
Interested in licensing this patent?