Recent trends in reporting and methodological rigor of non-inferiority trials in oncology.

Abstract

e23026 Background: Non-inferiority (NI) trials, which investigate if a new treatment is not significantly worse than the standard treatment, are increasingly popular. Selecting an appropriate NI margin is crucial as a larger margin might conceal the inferiority of a treatment option; also, intention-to-treat analyses (ITT)-only analysis can bias results towards demonstrating non-inferiority. Reporting of NI trial design has been inconsistent in the literature, and recent patterns have not clearly delineated. Here we aim to describe trial factors that might influence NI margin reporting. Methods: We searched PubMed, ClinicalTrials.gov, and HemOnc.org for published manuscripts of NI trials from 2015-2025 focusing on systemic therapy comparisons. We then conducted multivariable logistic regression (MLR) to assess the association between NI margin reporting and trial sponsorship, journal tier, analysis type, and trial period. Results: The 93 included studies collectively enrolled 82,638 patients. Of these, 70 (75%) met the primary endpoint of NI of the experimental arm. 56 (60%) did not report a rationale for selecting the NI margin. Among the studies that provided a justification, the rationale was: based on previous literature (57%), author discussion (16%), clinical judgment (8%), or a combination of these factors (13.5%). 47 studies (50.5%) conducted ITT analysis only, 4 (4%) performed per protocol (PP) analysis only, and 42 (45%) conducted both ITT and PP analysis. 46 studies (49%) were published in high-tier journals. Industry sponsorship was noted in 51 studies (55%), while 39 (42%) were supported by government agencies, cooperative groups, or academic institutions. For the 36 studies that proved NI based on the ITT analysis alone, 22 (61%) did not provide any justification for choosing the margin. Of these, 23 (64%) were industry-supported; 15 studies (42%) were multi-national. None of the predictors in MLR analysis reached statistical significance. Conclusions: NI trial design is important for success and interpretation of NI trials; however, our analysis shows most NI trials do not report the rationale for their selected margin, and more than half use ITT-only analysis. No analyzed variable was statistically significant in MLR for the prediction of NI margin reporting, although there was a trend towards an association between ITT-only analysis and non-reporting of margin selection rationale. NI is an appealing means to introduce new treatments, such as biosimilars, for cost-effectiveness, less toxicity, or shorter alternative options. Stricter adherence to guidelines and transparent reporting are needed to ensure the reliability and impact of the NI design. MLR results. Variable OR (95% CI) Industry-sponsored vs not 0.61 (0.25-1.48) Mid/low vs high journal tier 0.82 (0.42-1.57) ITT-only vs PP/both analyses 0.51 (0.23-1.10) Year of publication (2021-2025 vs 2015-2019) 1.42 (0.59-3.47)