Cancers, mainly male, as population biomarkers for breast cancer mortality

Background. BCM correlates significantly with mortality from other site-spe cific cancers and cardiovascular diseases for both sexes. These other morta lities could be used as independent biomarkers (predictors) of BCM allowing an evaluation of the importance of common etiological factors. Methods. BCM (age-adjusted, 45-74 years), obtained around 1992 from 37 coun tries worldwide, was estimated in multivariate regression to identify the b est predictors. Results. Male and female biomarkers predicted BCM with a R-2 of 0.80 and 0. 69, respectively. Strongest correlation was obtained with male colon, prost ate, lung and rectum cancer and female esophagus cancer (R-2 = 0.84, P < 0. 0001). The estimated independent mean percentage contribution +/- SD to BCM was 40 +/- 7 from prostate cancer, 38 +/- 9 from male colon, 13 +/- 6 from male l ung and rectum cancer combined, and 9 +/- 3 from female esophagus cancer. T he regression equation (1992 data) predicted mean BCM in 28 available count ries from 1967 to 1991 with a mean error of 5%. BCM in individual countries was also reliably predicted from 1967 to 1991, r = 0.86 to 0.90 (P < 0.000 1). In 1953, r was 0.74 (P < 0.0001). Conclusions. The evidence suggests a major influence of modifiable environm ental factors common to BCM, its biomarkers, and both sexes: most likely nu trition, smoking, and alcohol intake. The results obtained with male data s uggest a minor impact of sex-linked risk factors and, until recently, of tr eatment and early detection on BCM at the population level. (C) 2000 Americ an Health Foundation and Academic Press.