ABSTRACT: A number of recent studies have suggested that survival among premenopausal women after primary treatment of breast cancer may be affected by the estimated hormonal milieu at the time of surgery, especially in those with axillary lymph node metastases. The concept has created considerable controversy and has resulted in the publication of many negative reports. However, several biological mechanisms have been suggested for the observed survival advantage. These include cyclical patterns of immune function, as well as cell division and cell death, that correlate with hormonal fluctuations of the menstrual cycle. Comparisons among studies of timing have been complicated by differences in menstrual cycle divisions, variability in the sources of study populations, limited availability of menstrual history data, and changes over the past 2 decades in primary and adjuvant breast cancer therapy. Several recent publications have been enhanced by the availability of serum collected at the time of surgery that enables accurate measurement of the hormonal milieu. In these studies, the likelihood of misclassification by menstrual cycle phase is reduced, and dependence on recalled menstrual history is eliminated. High progesterone(Drug information on progesterone) levels have been associated with improved survival. These findings have encouraged some to suggest that perioperative administration of progesterone or tamoxifen(Drug information on tamoxifen) (Nolvadex) may provide a preventive avenue comparable to scheduling surgery during the luteal phase. Further multidisciplinary studies are needed, however, to clarify the influence of the naturally occurring or medically induced hormonal milieu at the time of breast cancer surgery on survival in premenopausal women. [ONCOLOGY 11(10):1509-1517, 1997]
The hormone dependency of breast cancer was first recognized more than a century ago by Beatson, who noted remission of metastatic disease following bilateral oophorectomy. This therapeutic modality has recently been confirmed as an adjuvant therapy by meta-analysis. Bilateral oophorectomy, which profoundly influences endogenous estrogen levels, significantly reduced the risk of metastatic spread of disease following primary breast cancer surgery. Although most of the recognized risk factors of breast cancer suggest the hormone dependency of the disease, the influence of estrogens(Drug information on estrogens) and other hormones on the biological mechanisms of tumor growth and dissemination have not been well characterized, especially among premenopausal women. The potential effect of cyclical hormone levels on detection, diagnosis, and treatment has only recently become a focus of research.
The phases of the menstrual cycle are characterized by sizable hormonal fluctuations. The follicular phase is defined by a preovulatory estrogen peak in the absence of progesterone. During the luteal phase, after ovulation, simultaneous increases of both hormones occur. The mean levels of estrogen and progesterone, plotted in Figure 1A, were estimated from two published reports.[3,4] Researchers studying normal breast tissue have documented the cyclical patterns of DNA synthesis and rates of cell division and cell death. However, the potential influence of the menstrual cycle on breast carcinoma survival was only recently suggested by Hrushesky and colleagues. Their initial publication was followed by numerous letters, commentaries, investigations, and reviews.[8-10] As additional data appear in the literature, the number of controversial reports increases, broadening the scope of the issues requiring further research. Therefore, this review includes research findings and biological considerations raised in several recent publications.
In a study of laboratory animals, Hrushesky and colleagues observed survival differences by phase of the estrous cycle at the time of tumor excision. These investigators suggested that the prognostic effect associated with the hormonal milieu at surgery may also influence breast cancer progression among premenopausal women. Their subsequent study of 41 patients revealed an improved prognosis among women with primary breast surgery between days 7 and 20 of their menstrual cycle (mid-cycle), when compared to the survival of those whose surgery occurred during the perimenstrual interval. The relationship of this division of the menstrual cycle to the mean levels of endogenous estrogen and progesterone is shown in Figure 1.
The report published in 1989 stimulated many investigators to survey their research databases or medical records to locate dates of last menstrual period and perform survival analyses of their case series. The rapid proliferation of letters and brief reports produced conflicting results[8-10] and considerable controversy. One additional team of investigators found marginally significant survival differences. However, others could not confirm a prognostic effect for tumor excision during perimenstrual vs midcycle intervals.[8-10,14-21]
Several research teams considered dividing the menstrual cycle to more adequately reflect established cyclical patterns of estrogen and progesterone. The traditional follicular and luteal phases of the cycle were applied by Senie and colleagues after their analyses of mid-cycle vs perimenstrual timing failed to confirm the results reported by Hrushesky et al. Date of last menstrual period was collected from 283 women interviewed at the time of initial breast cancer treatment (1976 to 1978) and followed annually for 10 years. Survival differences by menstrual phase and nodal status were observed. Cox proportional hazard models were also calculated to control for age at diagnosis, tumor size, nodal status, and adjuvant therapy. Surgery during the follicular phase was associated with a twofold increase in risk of recurrence only among patients with axillary lymph node metastases. Additional analyses assessed cycle day as a continuous variable; recurrence rates were highest among cases in which surgery was performed prior to the putative day of ovulation, the 15th day of the cycle.
Veronesi and colleagues recently published similar results from one of the largest retrospective investigations in the literature. In their series of 1,175 patients, day in the menstrual cycle at the time of surgery was treated as both a continuous and dichotomous variable (follicular and luteal phases). The smoothed rates of recurrence by day revealed a preovulatory peak, with the lowest risk late in the cycle—a pattern similar to the plotted recurrence rates of Senie and colleagues. Analyses following categorization by menstrual phase indicated that survival differences were limited to patients with axillary lymph node metastases. The hazard rate (1.4) of recurrence in this subgroup of patients who had surgery during the follicular phase indicated that the risk of recurrence was 40% greater than for those with the same stage of disease whose surgery occurred during the luteal phase (P < .03).
Investigators from Finland studied 267 patients treated by mastectomy between 1980 and 1987 whose last menstrual period date before surgery had been recorded. Analyses were performed after four cycle phases (early and late follicular and luteal) and the most significant findings were found when last menstrual period was treated as a continuous variable. These investigators also observed a decrease in breast cancer mortality as the interval between last menses and surgery increased.
Badwe et al created a third division of the menstrual cycle based on mean estrogen and progesterone fluctuations, as shown in Figure 1A. Metastatic spread of disease occurred significantly more frequently among the 249 patients whose surgery was performed during the interval of unopposed estrogens (days 3 to 12) than among those who had surgery at other times in the cycle when estrogen and progesterone were at similar levels (P < .001). As in other studies, survival differences were restricted to those with positive lymph node metastases.
These researchers also assessed death rates at 10 years in relation to day of the cycle at time of surgery. Mortality was highest when surgery occurred between days 3 and 14. The graph plotted resembled that published by other researchers.[21,22] Saad and colleagues, who applied a fourth cycle division, also observed greater disease-free survival in patients treated later in the menstrual cycle, especially those with positive axillary nodes. As previously noted, the positive findings of these studies were supported by some but not confirmed in other reports and letters.[8-10,25]
Summary of Findings
Table 1 presents a summary of the retrospective studies performed to date. Several of these research teams tested the prognostic effect of two or three menstrual cycle divisions in their reports so as to confirm or refute published findings. Rageth et al assessed the survival of 224 patients after stratification by the cycle divisions of Hrushesky and Senie. Although their results were not significant, the trend was opposite of that in the original publications. Analyses performed by Jager and Sauerbrei applied three different cycle divisions to data from 562 patients treated over 11 years (1980 to 1990). Survival did not differ regardless of the cycle division applied or stratification by nodal status. Among 192 patients with positive lymph nodes, Gnant et al found no significant difference in survival based on the “unopposed” estrogen hypothesis of Badwe and colleagues.
Kroman and colleagues tested two cycle divisions in a population of 1,635 patients who had been stratified by nodal status. None of their survival analyses showed any effect of menstrual timing. Additional reports in the literature testing the cycle divisions of Hrushesky, Badwe, Senie and their colleagues have presented conflicting findings. Also, numerous reports with less detailed analyses (published as letters or brief reports) found no significant difference in survival associated with menstrual timing of surgery.[8-10]
Two meta-analyses of selected publications with adequate data have been performed by researchers at Guy’s Hospital.[29, 30] The first report, published in 1992, tested the hypothesis of unopposed estrogen by estimating risk of recurrence from survival curves in 10 studie; a significant overall effect of timing was observed. The second meta-analysis included 21 reports and again found a protective effect for surgery performed late in the cycle.
Among studies reporting survival differences by menstrual timing of breast cancer surgery, statistically significant results have been limited to patients with axillary lymph node metastases. The data presented in Table 2 and Table 3, therefore, were selected to provide comparisons of node-positive cases. Reports divided by the follicular and luteal phases are demonstrated in Table 2. Three of the five studies indicated superior survival when surgery occurred after the putative day of ovulation. Studies included in Table 3 contrasted survival in patients treated between days 3 and 12 of the menstrual cycle with survival in those treated during days 0 to 2 or days 13 to 32. Of the six reports listed, four indicated the significant effect of menstrual timing on survival. Additional published data lacking multivariate analyses or stratification by nodal status were excluded from this table.