Indirect treatment comparisons including network meta-analysis: Lenvatinib plus everolimus for the second-line treatment of advanced/metastatic renal cell carcinoma.

• Main Authors: Gabriel Tremblay, Heather J McElroy, Tracy Westley, Genevieve Meier, Derek Misurski, Matthew Guo
• Format: Article
• Language: English
• Published: Public Library of Science (PLoS) 2019-01-01
• Series: PLoS ONE
• Online Access: https://doi.org/10.1371/journal.pone.0212899

BACKGROUND:In the absence of clinical trials providing direct efficacy results, this study compares different methods of indirect treatment comparison (ITC), and their respective impacts on efficacy estimates for lenvatinib (LEN) plus everolimus (EVE) combination therapy compared to other second-line treatments for advanced/metastatic renal cell carcinoma (a/mRCC). METHODS:Using EVE alone as the common comparator, the Bucher method for ITC compared LEN + EVE with cabozantinib (CAB), nivolumab (NIV), placebo (PBO) and axitinib (AXI). Hazard ratios (HR) for overall survival (OS) and progression-free survival (PFS) estimated the impact of applying three versions of the LEN+EVE trial data in separate ITCs. Last, to overcome exchangeability bias and potential violations to the proportional hazards assumption, a network meta-analysis using fractional polynomials was performed. RESULTS:Bucher ITCs demonstrated LEN + EVE superiority over EVE for PFS, indirect superiority to NIV, AXI, and PBO, and no difference to CAB. For OS, LEN + EVE was superior to EVE and indirectly superior to PBO, applying original HOPE 205 data. Using European Medicines Agency data, LEN + EVE was directly superior to EVE for OS. Fractional polynomial HRs for PFS and OS substantially overlapped with Bucher estimates, demonstrating LEN+EVE superiority over EVE, alone, NIV, and CAB. However, there were no statistically significant results as the credible intervals for HR crossed 1.0. CONCLUSIONS:Comparing three Bucher ITCs, LEN + EVE demonstrated superior PFS when indirectly compared to NIV, AXI, and PBO, and mixed results for OS. While fractional polynomial modelling for PFS and OS failed to find statistically significant differences in LEN + EVE efficacy, the overall HR trends were comparable.