HRT and Risk of Breast Cancer With a Favorable Histology
The iowa women’s health study is a population-based random sample of postmenopausal women who were aged 55-69 in 1986. A total of 1520 incident breast cancers have occurred in the at-risk cohort of 37,105 women. Gapstur and colleagues utilized this population sample to determine if past or current hormone replacement theory (HRT) use was a risk factor for the development of breast cancer. The survey instrument was a questionnaire, but the type of HRT used was not elicited. To make this analysis more informative, Gapstur et al asked whether the relationship between postmenopausal hormone use and breast cancer varied by tumor type. Based on standard histologic criteria, the breast cancers were put into one of four categories. Ductal carcinoma in situ (DCIS) accounted for 11.5%, 5.4% were deemed invasive breast cancer with a favorable histology, 76.6% were called infiltrating ductal or lobular carcinoma, and 7% were other. The last group was not
included in the analysis. Age at menarche, age at menopause, and type of menopause were not related to the age-adjusted incidence of any tumor type. A positive association between age at first birth and breast cancer risk was seen for all histological tumor types. Family history of breast cancer increased the risk of DCIS and invasive
ductal or lobular carcinoma with an unfavorable prognosis. Overall,Gapstur et al found a positive association between HRT use andincidence of breast cancer only for those with breast cancers deemed "favorable." This subgroup accounted for only 5% of the tumors, and the risk was largely confined to current use. Interestingly, in this subset, the risk of breast cancer was less in those who used hormones more than five years than in those who used hormones postmenopausally for less than five years. Conversely, the incidence of DCIS and invasive ductal or lobular carcinoma were not related to past or current use, regardless of duration. (Gapstur SM, et al. JAMA 1999;281:2091-2097.)
COMMENT by Sarah L. Berga, MD
In seeking to determine if postmenopausal hormone use increases the risk of breast cancer, Gapstur et al sought to improve upon the usual and customary design by stratifying breast cancer into histologic types and then considering the effect of HRT use upon these subtypes. Otherwise, the study is a standard epidemiological trial in which a large cohort of women were observed prospectively. I have long felt that it makes little sense to lump all breast cancers together. Certainly, not all breast cancers are the same, one must consider stage, host, histology, or causal molecular and cellular derangements. The logic of subtyping is compelling, but it is not clear what criteria should be used to subtype. For instance, it has been shown that mutations in tumor suppressor genes [p53, BRCA, BRCA 2, ATM (ataxia-telangiectasia, mutated)] or overexpression of oncogenes (such as Her-2-Neu) are causally related to the development of breast cancer. Ideally, one would subtype according to relevant molecular features. Since molecular fingerprinting of breast cancers was not possible, Gapstur et al resorted to the best criteria that could bedevised in this setting. It is debatable if the subtypes rep- resent relevant biological groups. If they do not, one can argue that the data must remain aggregated rather than segregated. If the data remain aggregated, then this is a large study showing that current and past HRT use does not increase the risk of breast cancer. If one allows that
the subtyping is biologically valid, then the study suggests, as have others like it, that HRT use may promote the development of breast cancers with a favorable prognosis in a small minority of women. Put another way, cancers that develop in women "due" to HRT have a good prognosis. However, given the tentative nature of the subgroups, the most conservative conclusion to draw from this data set is that HRT use does not promote the development of breast cancer. Most breast cancers occur
independently of past or current HRT use. I tell patients that we do know a little about what causes breast cancer, and it does not appear to be estrogen exposure.
Which of the following is not a likely risk factor for the development of breast cancer?
a. Mutations in the tumor suppressor gene p53
b. Testing positive for the mutations BRCA1 and BRCA2
c. High expression of the oncogene Her-2-Neu
d. Estrogen use for more than five years
e. Mother who died of breast cancer