Chen, James L.

Abstract Defining patient-to-patient similarity is essential for the development of precision medicine in clinical care and research. Conceptually, the identification of similar patient cohorts appears straightforward; however, universally accepted definitions remain elusive. Simultaneously, an explosion of vendors and published algorithms have emerged and all provide varied levels of functionality in identifying patient similarity categories. To provide clarity and a common framework for patient similarity, a workshop at the American Medical Informatics Association 2019 Annual Meeting was convened. This workshop included invited discussants from academics, the biotechnology industry, the FDA, and private practice oncology groups. Drawing from a broad range of backgrounds, workshop participants were able to coalesce around 4 major patient similarity classes: (1) feature, (2) outcome, (3) exposure, and (4) mixed-class. This perspective expands into these 4 subtypes more critically and offers the medical informatics community a means of communicating their work on this important topic.

BACKGROUNDBiliary tract cancers (BTCs) are uncommon and are associated with a dismal prognosis. Combinations of gemcitabine and platinum chemotherapy (gemcitabine and platinum–based therapy [GP]) form the standard approach for treating advanced BTC. To characterize the spectrum of mutations and to identify potential biomarkers for a GP response in BTC, this study evaluated the genomic landscape and assessed whether mutations affecting DNA repair were associated with GP resistance.METHODSPretreatment, formalin‐fixed, paraffin‐embedded samples from 183 BTC patients treated with GP were analyzed. Cox regression models were used to determine the association between mutations, progression‐free survival (PFS), and overall survival (OS).RESULTSWhen genes with an incidence > 10% were considered, no individual gene was independently predictive of a GP response. In patients with unresectable BTC who received GP as their first‐line therapy, the joint status of cyclin‐dependent kinase inhibitor 2A (CDKN2A), tumor protein 53 (TP53), and AT‐rich interaction domain 1A (ARID1A) was associated with PFS (P = .0004) and OS (P ≤ .0001). Patients with mutations in CDKN2A and TP53 were identified as a poor‐prognosis cohort with a median PFS of 2.63 months and a median OS of 5.22 months. Patients with mutant ARID1A, regardless of the single‐mutation status of TP53 or CDKN2A, had similar outcomes. A patient who exhibited mutations in all 3 genes had a median PFS of 20.37 months, and OS was not reached.CONCLUSIONSIn the largest exploratory analysis of this kind for BTC, 3 prevalent, mutually exclusive mutations represent distinct patient cohorts. These mutations are prognostic and may represent a predictive biomarker for a GP response. Prospective studies to validate these findings are needed, and they should include the incorporation of therapies that exploit the genomic instability observed with these mutations in BTC. Cancer 2016;122:3657‐66. © 2016 American Cancer Society.