Presentation (CTI)

OBJECTIVES: Linking temporal biomarker information with patient outcomes is fundamental to clinical decision-making in patient monitoring. To accomplish this, we leveraged an advanced statistical approach, joint modeling of longitudinal and time-to-event data (JM), to associate circulating tumor DNA (ctDNA) dynamics with real-world progression-free survival (rwPFS) and real-world overall survival (rwOS).
METHODS: RADIOHEAD is a prospective study of 1070 patients receiving immune checkpoint inhibitor therapy for cancer treatment. Patients were tested at pre-treatment and on-treatment timepoints with Guardant Reveal, a tissue-free epigenomic assay that detects and quantifies ctDNA, reported as a methylation-based tumor fraction (TF). The TF was analyzed for a cohort of 251 patients with advanced non-small cell lung cancer (NSCLC) from the overall RADIOHEAD study. JM is comprised of two sub-models. A hierarchical cubic spline random effects sub-model evaluated serial TFs and a Cox-regression sub-model examined rwOS and rwPFS respectively. Consequently, a JM was created for each time-to-event outcome. Study covariates included age, gender, smoking status, cancer stage, and comorbidities. For each JM, association structures that consolidate sub-model information were investigated and included the biomarker’s estimated value, rate of change, and cumulative effect.
RESULTS: Results indicate that, after including covariates, the most recent estimated TF was associated with rwOS (p-value<0.0001) and the biomarker’s cumulative effect was associated with rwPFS (p-value<0.0001). Various dynamic predictions are presented graphically, each accentuating how evolving TF modifies outcome prediction. 
CONCLUSIONS: Using JM to associate TF with progression and outcome in NSCLC patients is a novel approach. We have shown that patient-specific dynamic predictions can be generated to provide a comprehensive understanding of how changes in TF associate with outcome. JM results can equip healthcare providers with viable options to monitor patient progression and update prognosis as new TF information becomes available, establishing a foundation for integrating TF measurements to enhance patient care.