Indirect comparison of survival data based on the Shiny method: the role of control groups in the assessment of heterogeneity

Objective: New techniques have recently been developed to reconstruct patient-level data from Kaplan-Meier curves. Based on these “reconstructed patients”, indirect comparisons can be made to rank the effectiveness of the main treatments available. In these studies, the assessment of heterogeneity is a crucial phase. Our objective was to set up an approach to evaluate heterogeneity when reconstructed patients represent the clinical material. Results: Two data sets are analyzed: the first regards treatments for triple-negative breast cancer (3 trials, 3 treatments), while the second focuses on nonmetastatic prostate cancer (3 trials, 3 treatments and placebo). Heterogeneity has been quantified according to the likelihood ratio test. In the first case, a significant degree of heterogeneity is found (likelihood ratio test, 12.94; df, 1; p=0.0003); consequently, the results of indirect comparisons risk being misleading and therefore require a quite complex interpretation. By contrast, in the second case, the lack of heterogeneity (likelihood ratio test, 1.17; df 2, p=0.60) suggests that the results of indirect comparisons are reliable. Further research is needed to define the most appropriate cut-off in the likelihood ratio test to identify cases where heterogeneity has a relevant impact.