Purpose: We assessed changes in advanced cancer incidence and cancer mortality in eight randomized trials of breast cancer screening.
Patients and methods: Depending on published data, advanced cancer was defined as cancer > or = 20 mm in size (four trials), stage II+ (four trials), and > or = one positive lymph node (one trial). For each trial, we obtained the estimated relative risk (RR) and 95% CI between the intervention and control groups, for both breast cancer mortality and diagnosis of advanced breast cancer. Using a meta-regression approach, log(RR-mortality) was regressed on log(RR-advanced cancer), weighting each trial by the reciprocal of the square of the standard error of log(RR) for mortality.
Results: RR for advanced breast cancer ranged from 0.69 (95% CI, 0.61 to 0.78) in the Swedish Two-County Trial to 0.97 (95% CI, 0.97 to 1.25) in the Canadian National Breast Screening Study-1 (NBSS-1) trial. Log(RR)s for advanced cancer were highly predictive of log(RR)s for mortality (R(2) = 0.95; P < .0001), and the linear regression curve had a slope of 1.00 (95% CI, 0.76 to 1.25) after fixing the intercept to zero. The slope changed only slightly after excluding the Two-County Trial and the Canadian NBSS-1 and NBSS-2 trials.
Conclusion: In trials on breast cancer screening, for each unit decrease in incidence of advanced breast cancer, there was an equal decrease in breast cancer mortality. Monitoring of incidence of advanced breast cancer may provide information on the current impact of screening on breast cancer mortality in the general population.