PUK1 Synthesizing Evidence on Progression-Free Survival and Assessing the Feasibility of Network Meta-Analyses in Previously Untreated Advanced/Metastatic Renal Cell Carcinoma Therapies

Abstract

When direct comparisons are unavailable, contingent on satisfying foundational assumptions, network meta-analyses (NMA) can synthesize evidence for differences in relative treatment effects. The current study investigates the validity of conducting NMA for progression-free survival (PFS) for nivolumab+cabozantinib treatment versus relevant interventions in previously untreated advanced/metastatic renal cell carcinoma (aRCC), following the feasibility assessment framework by Cope et al. (2014). A systematic literature review (SLR) identified all published randomized controlled trials (RCTs) in aRCC. Available evidence was synthesized by evaluating whether the pre-defined relevant interventions (n=10) formed a network of evidence. Clinical heterogeneity was assessed for study and population characteristics that were potential treatment effect modifiers and for outcome and treatment characteristics. Finally, baseline risk and heterogeneity (or inconsistency) in observed treatment effects were evaluated to identify inconsistencies in the results. The SLR identified 18 individual RCTs that met our inclusion criteria. Prognostic risk status was considered a treatment effect modifier, and differed across trials in terms of baseline severity, reducing the NMA validity; networks of evidence were therefore stratified by risk status. For the all-risk network, trials were excluded when only subgroups were analyzed. The all-risk network included 15 studies, the intermediate/poor-risk included 11 studies (of which, 3 only included intermediate-risk patients), and the favorable-risk network included 10 studies. Most of the a-priori defined potential treatment effect modifiers were inconsistently reported. Heterogeneity was observed in ECOG score and nephrectomy in the all-risk and intermediate/poor-risk analyses. The few direct comparisons in the network supported by multiple studies did not show significant differences in their treatment effects. The current study showed that it is feasible to perform NMAs to compare PFS in aRCC. However, imbalances in prognostic risk score across trials are present and may bias results.