Plasma retinol and prognosis of postmenopausal breast cancer patients.

BACKGROUND The role of retinol (vitamin A) in breast cancer prognosis has never been investigated in postmenopausal women. We prospectively assessed the long-term prognostic role of retinol plasma levels in a cohort of postmenopausal breast cancer patients. PATIENTS AND METHODS We investigated 208 women self-reported as postmenopausal operated on for T(1-2)N(0)M(0) breast cancer who participated in a chemoprevention trial as controls and never received chemotherapy or hormone therapy. Plasma samples were collected 3 months (median) after surgery and assayed within 3 weeks for retinol. Minimum and median potential follow-up were 12 and 15 years, respectively. The main analyses were on all women and on a subgroup ages >or=55 years, assumed too old to be in perimenopause. The main end point was breast cancer death. Breast cancer survival was estimated by the Kaplan-Meier method. The hazard ratios of breast cancer death by retinol level were estimated by Cox models stratified for age, where relevant, and recruitment period, and adjusted for tumor size and histology. RESULTS At 12 years, patients with low retinol (<2.08 micromol/L, median of distribution) had lower breast cancer survival than those with high retinol (log-rank P = 0.052); the difference was significant for women >or=55 years (log-rank P = 0.006). The adjusted hazard ratios for low versus high retinol were 2.11 (95% confidence interval, 1.08-4.14) for all women and 3.58 (95% confidence interval, 1.50-8.57) for those >or=55 years. CONCLUSIONS Low plasma retinol strongly predicts poorer prognosis in postmenopausal breast cancer patients. Retinol levels should be determined as part of the prognostic workup.