Association between time to first subsequent therapy (TFST) and overall survival (OS) in previously untreated patients with ovarian cancer (OC): A SEER Medicare database analysis.

Abstract

e17095 Background: TFST may be a patient-relevant indicator of efficacy beyond progression in OC, however, the association between TFST and OS is unknown. We analyzed the SEER Medicare database to determine whether OS varied by TFST in previously untreated patients with OC. Methods: From the SEER-Medicare linked database, we identified newly diagnosed OC patients (≥66 years) who initiated first-line chemotherapy between 2009 and 2015. Patients were followed through the end of 2016. TFST was defined as the time from initiation of first-line chemotherapy until the start of subsequent therapy or death. Cox models summarized the overall association between TFST and OS through hazard ratios (HRs), and by TFST strata for patients who had a TFST event. In case hazards were non-proportional, we examined the fit of independent parametric survival curves. Results: In the cohort of 1407 patients, 1005 patients had a TFST event. Median age was 74 years and 34.9% were alive at the time of this analysis. For population with a TFST, the median TFST was 13.1 months and median OS was 30.88 months (HR = 0.29, 95% CI: 0.26, 0.32). Among patients < 75 years, the median TFST was 13.65 months and the median OS was 37.07 months (HR = 0.24, 95% CI: 0.21, 0.28). In the full cohort of 1407 patients, the best-fitting distributions accelerated failure time models under which the (unrestricted) mean TFST and OS were 44.9 and 73.8 months respectively. This indicated that on average TFST was an acceleration of OS by a factor of 39%. Conclusions: In this elderly population of previously untreated OC, longer TFST was associated with longer OS. Limitations of this analysis include censoring of OS and immortal time bias. [Table: see text]