OBJECTIVES: Surrogate outcomes (SO), which are intended to predict a final clinical endpoint, are commonly used to ensure early access to medicines. However, their role in reimbursement evaluations is debated owing to concerns about how reliably they predict treatment benefits on the final outcome and whether such predictive ability holds across populations and interventions. This work explored whether progression-free survival [PFS] and overall response rate [ORR] of immuno-oncology (IO) monotherapies are reliable SOs of overall survival (OS) across four indications.

METHODS: We conducted a systematic literature review (SLR) of randomised controlled trials (RCTs) published since 2014 that assessed an IO-monotherapy arm (atezolizumab, avelumab, durvalumab, nivolumab, pembrolizumab) in head-and-neck cancer (HNC), melanoma, non-small-cell lung cancer (NSCLC), or urothelial cancer (UC). Bivariate random-effect meta-analyses were conducted for PFS and ORR as surrogates of OS. Hazard ratios (PFS and OS) and odds ratios (ORR) from the latest data cuts were used for comparisons of IO-monotherapies (reference) with other treatments. Dose comparisons of the same IO-monotherapy were excluded. Sensitivity analyses by comparator treatment (chemotherapy/BSC/placebo and IO-combinations) were also conducted.

RESULTS: The SLR identified a total of 38 RCTs that could be included in the analyses (NSCLC [n=20], melanoma [n=7], UC [n=6], HNC [n=5]). Across indications, the estimated correlations of OS-PFS and OS-ORR were numerically positive (0.02 to 0.84) and negative (-0.56 to -0.07), respectively. However, a significant correlation between PFS and OS was only found for melanoma when comparing chemotherapy/placebo/IO-combinations vs IO-monotherapies (rho=0.84; 95%CI: 0.34–0.99). No significant correlations were found between ORR and OS in any indication. The impact of effect-modifiers (e.g., line of treatment) on these correlations could not be investigated owing to insufficient data.

CONCLUSIONS: The findings suggest that for IO-monotherapies the strength of surrogacy between PFS and OS may be indication-specific, which may have implications in regulatory and reimbursement assessments.