Reviewers: John Han-Chih Chang, MD and Kenneth Blank, MD
Source: Journal of the American Medical Association Nov 5, 1997; Volume 278 (Number 17): pp 1407 - 11.
In most previous reports of the association of breast cancer and weight, an inverse proportionality has been found in premenopausal women. Premenopausal breast cancer risk is alsoinversely proportional to weight at 18 years of age. The relationship in postmenopausal breast cancer is less well defined. The issue becomes even more perplexing when one considers that heavier women have, on average, 50% - 100% higher endogenous estrogen (associated with increase risk of breast cancer) levels than leaner women. Few have looked at weight change in relation to breast cancer incidence. This report contains the results of a cohort study performed by Havard University along with the Brigham's and Women's Hospital on the subject of weight and its relationship with breast cancer risk.
From 1976 to 1992, a cohort study was performed. A health questionnaire was sent to female registered nurses aged 30 to 55 years. The data on 95,256 respondents to the Nurses' Health Study were analyzed. Follow-up questionnaires were sent every two years thereafter. The response rates were 95%. Body mass index (BMI) was the measure of obesity that was used. This was calculated by using the weight in kilograms and dividing by the height squared yielding a number with kg/m2 as its unit. They looked at a couple of subsets of respondents to determine the accuracy of self-reported weights. In both, they found that 87% - 96% correlation with actual documented weight with the mean difference being 1.5 kg.
Two thousand five hundred and seventeen invasive breast cancers were recorded of which approximately 60% were postmenopausal.
For current BMI, premenopausal women with 26 kg/m2 or less had a higher relative risk (RR) of incident breast cancer than those with less BMI. For postmenopausal women, there were no significant trends, except when there was no history of hormonal replacement therapy (HRT). In these respondents, current BMI and breast cancer was positively linked, meaning that the greater the BMI the higher the RR.
At the age 18, the risk of breast cancer was inversely proportional to BMI. For premenopausal breast cancer, a BMI of 18.2 kg/m2 or less had a RR 1.64 times more than those with a greater than 25kg/m2 BMI. Respectively, for postmenopausal breast cancer the RR was increased by 1.39 times.
For weight gain after age 18, the respondents who gained 20 kg or more had an increase RR of 1.4 for postmenopausal breast cancer over those that only fluctuated 2 kg or less from their 18 year-old baseline. The argument is even stronger in those that had not ever used HRT. For those that had weight gains of 10 to 20 kg and over 20 kg, the RR increased 1.61 and 1.99 times those that had "minimal" weight gain.
For breast cancer mortality, 201 deaths due to primary malignancy were recorded in premenopausal women. They were positively correlated with weight gain since age 18 years. In those that gained more than 20 kg, increased RR of death due to breast cancer was 1.27 times those that only fluctuated 2 kg from their weight at age 18 years. In the postmenopausal population, 279 deaths due to breast cancer were reported. Current BMI greater than 28kg/m2 had a RR of 1.9 times that of 21 kg/m2 or less. The RR for weight gain greater than 20 kg versus change of 2 kg or less was 2.17. This was even stronger for those not on HRT: 2.17 for current BMI greater than 28 kg/m2 and 3.8 for gain greater than 20 kg.
The results of this cohort study suggest that adiposity is detrimental to ones health in multiple ways, one of which is related to a increased risk for breast cancer. It has long been hypothesized that increased estradiol levels increase the risk for breast cancer. Obesity in a premenopausal women, have been shown to reduce estradiol and progesterone levels. In postmenopausal women, adipose tissue supplies the estrogen for the body. Obese women have decreased levels of sex-hormone binding globulin, thus making less bound to the protein and more biologically active. This cohort study seems to readily support these proposals. Before readily accepting these results as the truths, one must step back and look at how the data was obtained. They achieved statistical power based on the enormous numbers in the study. Their main contentions rely on the fact that what the respondent reports was accurate. In two small subset analysis of less than 200 respondents each, the authors found a 86% to 96% correlation with the true weight. This was such a small sampling of some 95,000 respondents. How could it be representative of the entire analysis population? How accurate can one be about her weight at age 18 years when the questionnaires went out the nurses, who were already at least 30 years-old? One must bear these issues in mind when deciding how to advise patients about the risk factors for breast cancer in light of this study's findings.