This publication from the WHI data evaluates the relationship of vasomotor symptoms (VMS) to breast cancer incidence and mortality . It was observed that there was a statistically significant increase of breast cancer incidence in those women with persistent vasomotor symptoms compared to those without vasomotor symptoms HR 1.13 CI (1.02-1.27) yet no statistically significant impact on survival was noted (HR 1.33 CI 0.88-2.02). Vasomotor symptoms were assessed via a self-administered baseline questionnaire in 40 centres across the US to 161,808 women, ages 50-79, between 1993 and 1998. After applying exclusion criteria, such as use of hormone therapy, 25,499 postmenopausal women remained in the analytic sample. Duration of symptoms was determined by VMS history before baseline enrolment. “No vasomotor symptoms” was defined as “never experienced and none experienced over the last 4 weeks. “ Persistent VMS” included moderate or severe symptoms ever or within the 4 weeks before WHI entry. Breast cancers that developed during follow-up for 8.5 years were verified by review of medical records. Cause of death was ascertained via the National Death Index. Breast cancer incidence, mortality and the association with VMS were calculated using time dependent Cox regression analyses adjusted by breast cancer risk factors. 1,399 breast cancers occurred over 17.9 years of follow-up. 9,715 women with persistent VMS for 10+ years had a hazard ratio of 1.13 ( CI 1.02-1.27) for developing breast cancer compared to 15,784 women without VMS. ER – cancers, both regional and metastatic, occurred more frequently in patients with persistent symptoms. Although deaths from breast cancer were higher in the women with persistent VMS, this result was not statistically significant (HR 1.33 (CI 0.88-2.02)) .
This paper attempts to determine if vasomotor symptoms, that have a complex neuroendocrine etiology  and are associated with a reduction of cerebral estrogen, could be a bio-marker for the risk of developing and dying from breast cancer. Breast cancer risk factors may reflect an increased lifetime exposure to estrogen, such as early menarche, lower parity, obesity, late menopause, use of hormone therapy, etc. . This paper from the WHI surprisingly found an increased risk of developing the disease in women with persistent VMS. The outcome here contrasts with the SWAN cohort, where 3,098 women age 42-52 were followed longitudinally over an average of 11.7 years with annual exacting questionnaires about vasomotor symptoms, and fewer breast cancers were found . Earlier literature including 2 case control studies had shown a 50% lower risk of breast cancer with VMS [5, 6]. A meta-analysis of 6 studies of endocrine treatments for breast cancer had shown an association with lower breast cancer recurrence when women had experienced treatment-related estrogen deficiency symptoms . One non-confirming prospective cohort of 11,297 women evaluated every 3 years for 13.7 years showed a similar incidence of breast cancer in women with and without VMS . The study strengths include the large number of patients and verified breast cancer cases and mortality information. The limitations are significant and relate to the reliance on a non-validated retrospective and single questionnaire for differentiating patient groups as well as the large number of exclusion criteria in the protocol. Women with mild symptoms were allocated to the no flash group arbitrary based on an older study that showed a 33% false positive rate for this symptom . More recent research using prospective digital diaries shows that the majority of hot flashes are in fact mild (41%) or moderate (44%) with only 13% severe or very severe (2%) . There must also have been significant recall bias affecting the accuracy of the baseline data (11% were in their 70’s) obtained from participants. Recall of hot flashes has been shown to be limited by variables such as anxiety, mood and negative affect . There were also differences in the baseline characteristics of the 2 groups of women all of which could have influenced the outcomes. The fact that the breast cancers associated with VMS were poorer prognosis, both regional and metastatic ER- and not ER+ types, might implicate another mechanism other than hormonal in this study. More research with validated detailed, prospective, frequently administered, questionnaires is required to establish accurate comparison groups. To associate a recalled history of vasomotor symptoms as a possible marker for an increased breast cancer risk suggested by this research is therefore not reliable. I would emphasize that more research is needed to determine the relationship between endogenous estrogen levels, VMS, breast cancer and this puzzling result.
Professor Department of Obstetrics and Gynaecology, University of Toronto
Director Menopause and POI Units, Mt. Sinai Hospital, Toronto, Canada