Immortal time bias in the association between toxicity and response for immune checkpoint inhibitors: a meta-analysis

Only a minority of studies addressing the association between immune-related adverse events (IrAE) and outcome considered the immortal time bias, with conflicting results. PubMed, Embase, Web of Science and Scopus were searched through 2 January 2020. Studies reporting the impact of IrAE on outcome in patients treated with antiprogrammed death receptor 1 or PD-L1 were included. Twenty nine articles were included. IrAEs were associated with improved outcomes with high heterogeneity. With a landmark approach, the association between IrAE and outcome remains significant but the effect size was smaller (hazard ratio 0.61 vs 0.41 for overall survival; p = 0.015; hazard ratio 0.66 vs 0.47 for progression-free survival, p = 0.029; odds ratio 2.59 vs 6.77 for overall response rate; p < 0.001), no significant heterogeneity. Our analysis suggests a confounding effect of immortal time bias and a real effect of IrAE on outcome.

Automated summary

A minority of studies did not account for immortal time bias and showed conflicting results. It was found that IrAEs in patients were significantly associated with outcomes, although the effect size was smaller in comparison.