PCN20 Can Real-World Data Replicate RCT Results? A Retrospective Comparative Effectiveness Study Of Docetaxel In Metastatic Hormone Sensitive Prostate Cancer Using German Claims Data

Abstract

Using real-world evidence to evaluate medical interventions is largely debated. Taking metastatic hormone-sensitive prostate cancer (mHSPC) as an example, we aimed to compare the real-world overall survival advantage of docetaxel over ADT monotherapy with the efficacy results from the CHAARTED trial. This was a retrospective observational study using German claims data. We identified mHSPC patients between 01/01/2014-31/12/2018, replicating the inclusion and exclusion criteria of the CHAARTED trial. Patients treated with docetaxel in combination with ADT at first line (docetaxel group) were compared with patients treated with ADT monotherapy (ADT group). Patients were matched using propensity score matching (PSM), and overall survival after start of therapy was compared between the two matched cohorts. Of 39,932 prostate cancer patients, 2,354 were mHSPC patients. Of these, 69 were included in the docetaxel group and 616 in the ADT group. After PSM, 56 patients were included in each group with no significant differences in baseline characteristics. Compared to patients in the CHAARTED trial, real-world patients included in the study were older (median age 68 years compared to 63 in the trial). Median survival was shorter for real-world patients (28.7 months in the ADT group, 36.7 months in the docetaxel group; median follow-up: 19.1 months) compared to patients in the trial (44 months in the ADT group, 57.6 months in the docetaxel group; median follow-up: 28.9 months). The hazard ratio was 0.58 (95% CI, 0.30-1.10) using real world data, compared to 0.61 (95% CI, 0.47-0.80) in the trials. Overall survival was shorter in the real-world compared to the trial, but the effect size was comparable despite differences in patient characteristics. While unobservable patient characteristics, low sample size, and regional bias are limitations of the study, results suggest that real-world data have the potential to replicate and complement results from CHAARTED.