Abstract
Late-stage cancer incidence has been proposed as a surrogate outcome for cancer-specific mortality in future screening trials. Two previous meta-analyses with 33 and 39 trials assessed trial-level surrogacy but provided inconsistent conclusions about the suitability of late-stage cancer endpoints replacing mortality. Our systematic review and meta-analysis (PROSPERO ID, CRD42023369320) investigated the association between the effect of cancer screening on the incidence of late-stage cancer and cancer-specific mortality. From 57 trials with 61 trial arm comparisons, correlation between late-stage incidence and mortality outcomes was 0.69 [95% confidence interval (CI), 0.47-0.84] for all cancers combined. Specifically, correlations were 0.58 (95% CI, 0.27-0.93) for bowel (N = 11 trials), 0.79 (95% CI, 0.49-0.94) for breast (N = 13), and 0.91 (95% CI, 0.84-0.96) for lung cancer (N = 14). Trial point estimates of the screening effect on mortality were within each trial's 95% CI late-stage incidence estimates in 56 of 61 (92%) trial arm comparisons and in 16 of 19 (84%) trial arm comparisons in which the entire 95% CI for screening effect on late-stage incidence was below one. Evidence suggests potential for late-stage cancer incidence as a key outcome in screening trials, but further research is needed to clarify when to measure late-stage outcomes, extrapolation for cancer types without trials, and the conditions when late-stage cancer does not accurately predict mortality.