Progression-free survival (PFS) assessment by blinded independent central review (BICR) versus local investigator (LI) in metastatic melanoma (MM) randomized controlled trials (RCT): A systematic review and meta-analysis.

Abstract

9545 Background: Although BICR may reduce assessment variability, it introduces additional financial and logistical burdens to trial operations. This study analyzed the discrepancy indexes (DIs) to evaluate differences between PFS assessments evaluated by LIs and BICR in RCTs of patients (pts) with MM. Methods: A comprehensive literature search was conducted on PubMed, Embase, and Cochrane databases until June 30, 2023. Studies were eligible for inclusion in the meta-analysis if they were 1) Phase II or III RCTs with accessible, published data; 2) inclusion of pts diagnosed with MM; 3) availability of PFS data assessed through both LI and BICR; and 4) publication in English. The study complied with the PRISMA guidelines to ensure methodological rigor. Two independent researchers performed data extraction to minimize bias and ensure accuracy. A fixed-effects meta-analysis approach was applied to summarize treatment outcomes, producing pooled estimates and corresponding 95% confidence intervals (CIs). The primary outcome was DI, which was calculated for each trial as a ratio of the hazard ratios (HR) BICR by HR LI . The agreement between PFS HRs was also evaluated using the intraclass correlation coefficient (ICC) and Pearson’s correlation coefficient ( r ). The risk of bias was evaluated using the Cochrane Risk of Bias tool v.2 (RoB 2). Results: A total of 12 studies comprising 4,915 pts were included in the meta-analysis that spanned from 2012 to 2023. Of these, 10 studies (83%) were Phase III , 11 (92%) were cutaneous melanoma and all identified PFS as the primary endpoint. Most studies (n = 8, 75%) had a DI > 1 and the overall combined DI was 1.08 (95% CI: 1.01–1.15), indicating a statistically significant numerically small difference (8%) in PFS evaluations conducted by the two assessments, suggesting that BICR tended to be more conservative in PFS assessments. These results were primarily driven by the Phase II or double-blinded studies, which showed a higher median (inter-quartile range) (IQR) DI [1.14 (o.o8) and 1.16 (0.13), respectively] than phase III or open-label trials [DI 1.04 (0.12) or 1.0 (0), respectively] . However, there was an overall significant strong correlation [ICC: 0.87, p < 0.001); r = 0.89, 95% CI 0.67-0.96, p <0.0001)] between BICR and LI assessments and most (86%) of the PFS comparisons led to the same statistical inference. Finally, 10 studies had a low risk of systematic bias, and none had publication bias. Conclusions: This study demonstrated a slim statistically significant difference in PFS assessments between LI and BICR, but with strong agreements overall. These findings challenge the necessity of universally implementing BICR in all RCTs, supporting appropriate use in selected scenarios, primarily Phase II RCT. Also, supports the value of double-blinded studies.