Estrogen Plus Progestin and Colorectal Cancer: Long-Term Findings From the Women's Health Initiative Randomized Clinical Trial

Abstract

Clinical trials frequently include multiple end points that mature at different times. The initial report, typically based on the primary end point, may be published when key planned co-primary or secondary analyses are not yet available. Clinical Trial Updates provide an opportunity to disseminate additional results from studies, published in JCO or elsewhere, for which the primary end point has already been reported.We report long-term colorectal cancer findings from the Women's Health Initiative trial where 16,608 postmenopausal women with a uterus were randomly assigned to daily conjugated equine estrogen (CEE) 0.625 mg, plus medroxyprogesterone acetate (MPA) 2.5 mg, or placebo. When intervention ended after 5.6 years, although there were 44% fewer colorectal cancers in the intervention group (43 v 72, P = .003), the cancers were more commonly lymph node-positive (59.0% v 29.4%, P = .003). Now after cumulative 24-year follow-up, with 431 colorectal cancers, CEE plus MPA no longer influenced colorectal cancer incidence (215 [0.15, annualized rate %] v 216 [0.15], hazard ratio [HR], 0.95 [95% CI, 0.79 to 1.15]). Although not statistically significant, there were more colorectal cancer deaths with CEE plus MPA (87 [0.049] v 69 [0.041] deaths, HR, 1.20 [95% CI, 0.87 to 1.65], P = .26). Vaginal bleeding (54.1% v 5.2% at 6 months) and breast changes were more frequent in the intervention group. After adjusting for postrandomization vaginal bleeding and breast changes, bowel examinations were significantly delayed in intervention group participants (P = .005), potentially contributing to diagnostic delay. Taken together, the findings suggest no clinically meaningful benefit for about 5 years of CEE plus MPA use on colorectal cancer outcome.

Trial registration: ClinicalTrials.gov NCT00000611.

Figure 1.

Kaplan-Meier estimates, complemented by cumulative hazard ratios, for colorectal cancer incidence in the CEE + MPA trial. Cumulative HRs (95%CI; lower panel) were computed under the proportional hazards assumption, using increasingly longer cumulative follow-up elapsed from randomization. For example, if the trial had ended after only 4 years of follow-up, the resulting HR (95%CI) = 0.63 (0.41, 0.96) for 36 vs 54 events. Corresponding Cox regression models were stratified by age group, randomization status in the dietary trial, prior history of colorectal cancer and study phase (time-dependent). The HR = 0.95 for total cumulative follow-up (blue reference line) is essentially a weighted average of time-varying hazard ratios, which were less than 1 earlier in follow-up, HR (95%CI) = 0.61 (0.43, 0.87) for the intervention, and greater than 1 for subsequent follow-up, HR (95%CI) = 1.14 (0.91, 1.43) for postintervention. There was compelling evidence against proportionality (P < 0.001).

Figure 2.

Kaplan-Meier estimates, complemented by cumulative hazard ratios, for colorectal cancer mortality in the CEE + MPA trial. Cumulative HRs (95%CI; lower panel) were computed under the proportional hazards assumption, using increasingly longer cumulative follow-up elapsed from randomization. For example, if the trial had ended after only 4 years of follow-up, the resulting HR (95%CI) = 1.75 (0.59, 5.23) for 9 vs 5 events. Corresponding Cox regression models were stratified by age group, randomization status in the dietary trial and study phase (time-dependent). The HR = 1.20 for total cumulative follow-up (blue reference line) is essentially a weighted average of time-varying hazard ratios, which were less than 1 earlier in follow-up, HR (95%CI) = 0.87 (0.38, 1.98) for the intervention, and greater than 1 for subsequent follow-up, HR (95%CI) = 1.27 (0.90, 1.79) for postintervention. There was no evidence against proportionality (P = 0.48).