Impact of molecular profiling on overall survival of patients with advanced ovarian cancer

Abstract

Objective: Patients with recurrent epithelial ovarian cancer (EOC) have limited treatment options. Studies have reported that biomarker profiling may help predict patient response to available treatments. This study sought to determine the value of biomarker profiling in recurrent EOC.

Results: Patients in the Matched cohort had a median OS of 36 months compared to 27 months for patients in the Unmatched cohort (HR 0.62, 95% CI 0.41-0.96; p < 0.03). Individual biomarkers were analyzed, with TUBB3, and PGP prognostic for survival. Biomarker analysis also identified a molecular subtype (positive for at least two of the following markers: ERCC1, RRM1, TUBB3, PGP) with particularly poor overall survival.

Methods: 224 patients from a commercial registry (NCT02678754) with stage IIIC/IV EOC at diagnosis, or restaged to IIIC/IV EOC at the time of molecular profiling, were retrospectively divided into two cohorts based on whether or not the drugs they received matched their profile recommendations. The Matched cohort received no drugs predicted to be lack-of-benefit while the Unmatched cohort received at least one drug predicted to be lack-of-benefit. Profile biomarker/drug associations were based on multiple test platforms including immunohistochemistry, fluorescent in situ hybridization and DNA sequencing.

Conclusions: This report demonstrates the ability of multi-platform molecular profiling to identify EOC patients at risk of inferior survival. It also suggests a potential beneficial role of avoidance of lack-of-benefit therapies which, when administered, resulted in decreased survival relative to patients who received only therapies predicted to be of benefit.

Keywords: cancer; molecular; ovarian; profiling; survival.

Figure 1. Retrospective analytic schema

Figure 2. Plots showing duration of monitoring, duration of treatments received before and after profiling, and post-profiling survival for each patient in the study

Each column along the x-axis represents one patient. The y-axis is time (days). The zero point of the y-axis is the time of profiling. Patients are sorted left to right based upon survival time post-profiling. Grey bars represent the total time monitored, from diagnosis to either death or last follow-up. Black bars at the top of a column represent death. The colored bars represent drug treatments and are coded relative to their match status with the patient's molecular profile. Green bars represent time on a therapy associated with benefit. Red bars represent time on a therapy associated with lack-of-benefit. Yellow bars represent time on a combination regimen associated with both benefit and lack-of-benefit. Blue bars represent time on a therapy associated with neither benefit nor lack-of-benefit. Panel A shows patients in the Matched cohort, and Panel B shows patients in the Unmatched cohort.

Figure 3. The average number of drugs that were predicted to be of benefit (blue) or lack-of-benefit (red) for each cohort compared to the average number of drugs that the patients actually received (calculated from diagnosis)

Figure 4. Kaplan-Meier curves

A. Kaplan-Meier curve showing the increase in overall survival from time of profiling for those patients treated only with therapies predicted to be of benefit by their molecular profile compared to those patients who received at least one therapy predicted be lack-of-benefit (HR 0.62, p=0.0295). Kaplan-Meier curves for patients who were positive for TUBB3 (B) and PGP (C). D: Kaplan-Meier curves showing that patients with over-expression of multiple markers have decreased overall survival. The green line shows patients who were positive for two or more biomarkers from the set ERCC1, PGP, RRM1, and TUBB3. The red line shows patients who were positive for only one biomarker from this set. The black line shows patients who were not positive for any biomarkers in this set.