The relationship between blood prolactin levels and risk of breast cancer in premenopausal women.

Single specimens of blood have been taken from over 5000 normal volunteer women in each of two sequential (1967-1976, 1977-1984) population-based studied on the Island of Guernsey. Multivariate analysis was used to determine the relationship between prolactin levels and risk factors in breast cancer in 2591 and 1959 premenopausal women in whom blood prolactin had been measured. In both populations the prolactin concentrations appeared to be log-normally distributed and therefore all analyses have been done on log-transformed data. Initially the variables in the statistical model were age at menarche, ages at first and last baby, parity, ponderosity (Quetelet Index), mammographic pattern (as graded by Wolfe), family history of breast cancer, age, menstrual cycle status, time of day of blood sampling, oral contraceptive use, history of breast feeding and methodological changes in the laboratory measurement of prolactin. Of these variables age at menarche, ages at first and last child and family history of breast cancer were found not to be significant and were excluded from the final model. The main finding to emerge was that after standardizing for all the other variables, prolactin levels in the follicular phase were significantly lower than those found at midcycle or during the luteal phase of the menstrual cycle. A peak level of prolactin was found at day 12 of the cycle. Increasing parity was related to a steady decrease in prolactin concentration. Increasing ponderosity was associated with an increased prolactin level as was a DY compared to an N1 mammographic pattern. Women with a history of oral contraceptive use had lowered prolactin concentration. All these effects occurred evenly over the menstrual cycle and were generally found for both data sets. Thus body weight, parity and, indirectly, age at first baby might influence breast cancer risk by being associated with changes in blood prolactin concentration.