The potential of precision medicine to transform complex autoimmune disease treatment is often challenged by limited data availability and inadequate sample size when compared with the number of molecular features found in high-throughput multi-omics data sets. To address this issue, the novel framework PRoBeNet (Predictive Response Biomarkers using Network medicine) was developed. PRoBeNet operates under the hypothesis that the therapeutic effect of a drug propagates through a protein-protein interaction network to reverse disease states. PRoBeNet prioritizes biomarkers by considering i) therapy-targeted proteins, ii) disease-specific molecular signatures, and iii) an underlying network of interactions among cellular components (the human interactome). PRoBeNet helped discover biomarkers predicting patient responses to both an established autoimmune therapy (infliximab) and an investigational compound (a mitogen-activated protein kinase 3/1 inhibitor). The predictive power of PRoBeNet biomarkers was validated with retrospective gene-expression data from patients with ulcerative colitis and rheumatoid arthritis and prospective data from tissues from patients with ulcerative colitis and Crohn disease. Machine-learning models using PRoBeNet biomarkers significantly outperformed models using either all genes or randomly selected genes, especially when data were limited. These results illustrate the value of PRoBeNet in reducing features and for constructing robust machine-learning models when data are limited. PRoBeNet may be used to develop companion and complementary diagnostic assays, which may help stratify suitable patient subgroups in clinical trials and improve patient outcomes.
A key promise of precision medicine is the ability to match patient subgroups with the most appropriate treatments.1 This is achieved by discovering biomarkers that connect a patient's biological status with therapeutic outcomes for a specific therapy. In precision medicine, biomarkers can be discovered using machine-learning models that unveil complex, generalizable patterns from large molecular and clinical data sets, usually comprising data from hundreds to thousands of patients. For example, analyzing extensive molecular and clinical data sets from patients with cancer, machine-learning models found biomarkers that predict response to treatment in patients with diverse cancers.2–8 These models have substantially improved outcomes and survival rates for many cancer subtypes and greatly reduced the financial burden on health care payers.
Contact
Social