OBJECTIVES: Joint modeling (JM) of longitudinal and time-to-event data is a powerful statistical technique that enables an understanding of how temporal changes in a biomarker relate to a time-to-event response. With increasing use of next generation sequencing (NGS) in cancer care, repeated measures of circulating tumor DNA (ctDNA) provide an opportunity to use JM to predict patient outcomes. This proof-of-concept study evaluates whether the evolution of two NGS biomarkers, allele frequency (AF) and tumor fraction (TF), is associated with survival in patients who have non-small cell lung cancer (NSCLC) with an EGFR L858R mutation.

METHODS: Patients were selected from the GuardantINFORM database which combines NGS information with clinical outcomes data. Patients with NSCLC and an EGFR L858R mutation who had a minimum of three serial ctDNA observations were included in the study. Analysis was performed using longitudinal and a time-to-event sub-models, where information from each was combined into a JM designed to evaluate the association between biomarker progression and patient survival.

RESULTS: JM results from the qualifying cohort of 252 patients indicate that the most recent change over time of each biomarker is associated with patient survival (AF: p-value = 0. 0139; TF: p-value = 0.0332). Through these associations, dynamic graphical renderings of patient-level survival curves can be displayed to assess clinical outcomes based on the patient’s unique biomarker evolution.

CONCLUSIONS: This research demonstrates the utility of JM in associating NGS biomarkers to patient survival. Although the study focuses on NSCLC and a specific EGFR mutation, the technique can be readily applied to other cancer types and NGS biomarkers. JM can offer a novel opportunity to assess patient-specific prognosis using serial testing NGS information and thus assist clinicians in contextualized decision making in cancer care.