Immortal time bias in the association between toxicity and response for immune checkpoint inhibitors: A systematic review and meta-analysis.

Abstract

Listen

Translate article

e15080 Background: A number of studies have addressed the association between Immune related Adverse Events (IrAE) and outcome in patients treated with Immune Checkpoint Inhibitors (ICI). Only a minority of them considered the effect of immortal time bias (ITB), and results of these papers are conflicting. The aim of our meta-analysis was to assess the relationship between IrAE and outcomes for patients treated with ICIs, with a focus on ITB in weighing the role of this association. Methods: PubMed, Embase (Ovid), and Scopus were searched through January 2, 2020. Studies reporting the prognostic impact of IrAE development on outcome in advanced cancer patients treated with anti PD-1 or PD-L1.Two investigators independently extracted data from each study following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guideline. Author name, cancer type, PD-1 and PD-L1 inhibitor used, number of patients, number and type of all adverse events, presence of a landmark analysis, median OS and PFS and difference between patient with or without IrAE in terms of OS and PFS (Hazard Ratio) and ORR (Odds Ratio) were extracted. Data were pooled using a random-effects model. Main outcome were Overall Survival (OS), Progression Free Survival (PFS) and Overall Response Rate (ORR). Analysis was performed with dedicated software (R studio version 1.2.1335, metafor package). Results: A total of 29 articles comprising 7430 patients were included. 7 papers published the results of survival analysis according to landmark analysis, 2 papers published only the naïve analysis while stating that a landmark analysis resulted non significant, and the others did not perform a landmark analysis. IrAEs were associated with improved outcomes with high heterogeneity ( I2 70.8%, p < 0.001 for OS; 65.7%, p 0.002 for PFS; 62.1%, p 0.001 for ORR). When subgroup analysis was performed according to the adoption of a landmark approach, the association between IrAE and outcome remains significant for OS, PFS and ORR but the effect size was smaller ( HR 0.61 vs 0.41 for OS, ANOVA Q-test for difference between subgroup p = 0.015; HR 0.66 vs 0.47 for PFS, ANOVA Q-test p 0.029; OR 2.59 vs 6.77 for ORR, ANOVA Q-test p < 0.001) while heterogeneity was substantially reduced (I2 for papers using LM 43% for OS, p 0.115; 9.7% for PFS, p 0.355; 0.0% for ORR, p 0.556). Conclusions: Our analysis suggests a significant confounding effect of ITB as well as a real effect of IrAE development on outcome.