Although the article by Senie and Tenser reviewing some of the data relevant to whether operative timing within the menstrual cycle affects breast cancer outcome is reminiscent of a recent paper that appeared in the December 1996 issue of the Journal of Womens Health, the question it considers is potentially important enough that this issue should also be raised in Oncology. The article points out the experimental basis for believing that an important interaction may occur between the host-cancer-surgery and the mammalian reproductive cycle.[2,3] This is an important supposition because clinicians have routinely assumed that no experimental foundation underlaid the first and 31 subsequent analyses of relevant clinical data[4,5]an assumption that is false.
The contrary opinion, namely, that the early clinical data analyses were not based on biological findings, was first put forward by a leader who explicitly knew this not to be the case. In a cleverly titled but misleading editorial that accompanied Senies paper confirming the findings of others in the Annals of Internal Medicine, it was stated that the outcomes of these analyses performed on clinical data in vacuo may well have occurred by chance.[6,7] In fact, as pointed out by Senie and Tenser, the murine experiments predicted that resection during the high-progesterone, intermediate-estrogen concentration phase of the cycle would confer relative benefit. This finding has been, to a great extent, confirmed and reconfirmed in 11 subsequent series now, including at least 2,000 young women (Table 1).
Many other series have, however, failed to find differences, and a few have actually found other menstrual cycle phases to be superior. These data are, indeed, worrisome and inconsistent and demand additional fundamental research, as well as careful prospective clinical investigation, as urged by Senie and Tenser. The total lack of mechanistic understanding of the biology of the host-surgery-cancer-sex hormone interaction demands additional fundamental research. Gaining this understanding is a critical goal because the application of work often requires some tentative if not definite understanding of the mechanisms by which the observed phenomena occur. The fact that we are all but ignorant of the mechanisms by which the female reproductive cycle affects the metastatic potential of breast cancer continues to impede the development of prospective clinical trials, as well as application of this work to practice. The authors' plea for prospective trials that can help define the mechanisms of any forthcoming findings are well grounded and should be heeded by anyone planning such a study. Before these data can be accumulated, however, a conceptual framework must be established.
The strategy of optimally timed breast cancer resection and/or biopsy is conceptually different from adjuvant chemotherapy. Surgical timing, if it is relevant, will confer benefit without risk. Optimal resection timing will enhance survival probability with no extra cost and no extra suffering. Additionally, if individual retrospective studies are correct, the advantage of optimal resection timing may even exceed the average benefit conferred by adjuvant chemotherapy. At the same time, this benefit is potentially available to each premenopausal breast cancer patient rather than only some subset of women at high risk of death from breast cancer.
How Can the Timing of a Single Event Affect Outcome Years Later?
The retrospective clinical results outlined above indicate that the time at which surgery takes place within the fertility cycle affects whether or not lethal breast cancer metastases develop, usually several years after attempted surgical cure. Two nonexclusive scenarios can help explain how this can happen.
The First ScenarioIn the first case, consistent with the Fisher paradigm, widespread micrometastases are present very early in the course of the disease, even when early-stage breast cancer is discovered and resected. In this case, the humoral effects of wounding and/or primary tumor removal are either different or are perceived as different by the latent tumor cells and/or the host immune networks and support structures, depending on the extant hormonal milieu when this resection signal is generated. If the signal is sent and received during the early luteal phase, these micrometastases are more likely to remain dormant or be eliminated and are less likely to grow to a point at which they become clinically detectable and subsequently lethal.
The Second ScenarioA second possibility is that the frequency and efficiency of tumor cell shedding during the operation varies reproducibly depending on when in the cycle the operation is performed. Tumor cells are liberated either discontinuously or continuously and in greater numbers during surgical manipulation of the primary cancer. The operation, however, when performed at given points in the fertility cycle, results in either a different number and/or qualitatively different tumor cells and/or important variations in the endothelial capillary beds that receive those showers of tumor cells, depending on when in the cycle the tumor is resected. Fewer tumor cells, smaller clumps of tumor cells or fewer large clumps of less sticky, less motile, or biochemically less metastatically competent cells might be liberated into less hospitable microvascular beds when surgery is performed in the hormonal milieu associated with the early luteal phase of the cycle. There is also absolutely no reason why one of these two compatible scenarios should exclude the other.
Prospective Study Is Essential
In retrospective studies, it is difficult to accurately assess the extent of disease at diagnosis and whether or not irregularly cycling perimenopausal women or women using oral contraceptives have been appropriately excluded from the study. The accuracy of retrospective menstrual cycle phase localization is always suspect. Because of the necessary imprecision of these historical menstrual data, retrospective studies cannot tell us, with any degree of certainty, precisely the best time within the cycle for surgery. Nor can they accurately tell us for how long or how wide this window of opportunity is open.
The relative impact of the stage of disease at resection and the relative impact of the extent of the operation performed are likewise poorly assessable. These retrospective studies do not allow us to determine, when more than one surgical procedure is employed, the relative importance of the timing of an initial incisional or excisional biopsy or lumpectomy compared with the timing of the follow-up completion surgical procedure and/or concurrent or later lymph node dissection. Each of these important questions can be adequately addressed only by an adequately designed prospective clinical trial of breast cancer resection timing in regularly cycling premenopausal women.
Minimal Requirements for Prospective Study
The availability of prospective data per se will not magically settle this potentially important question. Each of four prospective studies initiated to date are unable, in principle, to accurately answer this potentially important question. The study of biological rhythms, like the menstrual cycle, is not without subtlety. In thinking about the strengths and weaknesses of the 31 studies published to date, three simple elements emerge as essential to any study of whether or not the menstrual cycle phase at operation affects breast cancer outcome. These are:
- The accurate triangulation of surgery within this cycle
- Temporal coordination of multiple surgical procedures when more than one operation is employed
- Protection of the distribution of operations within the cycle from potential cycle phase assignment bias.
All prospective studies initiated to date inadequately address each of these three necessities.
Accurate Biotemporal Triangulation of Surgical Intervention Within the CycleThe retrospective data collected to date rely entirely on retrieval of a single piece of historical information, namely, when was the first day of menstrual flow in the span of menses immediately prior to your breast cancer resection? Assuming, to begin with, that a single definitive surgical procedure serves as the sole operative intervention, it is clear that there are several potential sources of serious inaccuracy of menstrual cycle stage assignment based solely on this sort of historical information. If the patients recall is imperfect or the womans cycle is irregular or anovulatory, cycle stage assignment is potentially faulty. Other reasons intrinsic to the chronobiology of the menstrual cycle are more subtle and more difficult to avoid.
Woman-to-Woman and Cycle-to Cycle VariabilityAll such biological rhythm phase assignments have, to date, been made based upon the assumption of an ideal 28-day menstrual cycle, when it is well known that normal regular menstrual cycles range in duration between 21 and 36 days.[8,9] Information about the timing of an operation, expressed in days from the first day of the menstrual flow, will have a totally different biological meaning depending on the usual length of the cycle for that woman.
For example, an operation performed 8 days after the first day of menstrual flow, in a woman with a usual 21-day cycle duration, would place that surgical procedure within the putative early luteal phase of the womans cycle.
An operation performed 8 days after the first day of menstrual flow, in a woman with a usual 36-day cycle duration, would mean that her operation takes place during the putative follicular phase of her menstrual cycle.
These two simple specific examples demonstrate that historical information alone, whether collected retrospectively or prospectively, form an entirely inadequate basis for accurate assignment of putative physiologic menstrual cycle phase. Since a genuine understanding of the possible relationship between the woman, her cancer, and her operation depends completely upon accurate temporal biological information, any study not providing more certain physiologic triangulation of when resection should occur within the cycle is fundamentally flawed.
Within-Cycle Menstrual Cycle Phase Duration VariabilityThe physiologic events occurring within each menstrual cycle do not occur with perfect symmetry or precise within-cycle regularity. In fact, when there is cycle irregularity, the different physiologic segments of each cycle do not share this irregularity equally. The part of the cycle between luteinizing hormone/follicle-stimulating hormone (LH/FSH) surge-ovulation and the first day of subsequent menstrual bleeding span (when fertilization does not supervene)the putative luteal phaseis far more predictably fixed in timing and duration than the interval between the first day of menstrual bleeding and the subsequent LH/FSH surge-ovulationthe putative follicular phase. Almost all cycle-to-cycle duration variability (less than 80%) resides within this part (putative follicular phase) of the normal menstrual cycle.[8,9] This fact has profound implications for the value of historical menstrual cycle information.
It means that all retrospective data that depend on historical information for an assignment of cycle phase by counting backward from the date of operation to the first day of the immediately prior span of menstrual bleeding (virtually all published data) are assigning that phase in the most inaccurate historical way possible. An intrinsically more accurate method of triangulation would be to count forward to the first day of the next span of uterine bleeding. The most accurate historical method, however, would be to count both backward and forward.
It is quite disturbing to realize that all prospective studies instituted to date, assign an operative phase based solely on historical data obtained by counting backward. This methodology introduces unnecessary historical uncertainty. Using it as the primary method of triangulation of the operation within specific putative biological phases of the cycle cannot be justified. Assuming an ideal 28-day cycle length for every woman, when the opportunity exists to determine the real cycle length of each women being studied prospectively, is illogical and necessarily misleading.
Hormonal MeasurmentSingle blood samples obtained on or near the day of operation, measuring a variety of relevant hormone concentrations, each of which changes within the cycle (eg, FSH, LH, prolactin, melatonin(Drug information on melatonin), cortisol, estrogen and progesterone(Drug information on progesterone)), can enhance the likelihood of properly timing the physiologic phase of the breast cancer operation. Given the well-known predictable, daily (as well as menstrual) swings in the concentrations of each of these hormones, however, serial and time-of-day specified measurements throughout the entire cycle following the surgery would seem to be a reasonable minimum hormone sampling requirement. Even this strategy will not provide gold standard information about the cycle phase of operation. Obtaining a single serum sample at the time of operation, as is being done in at least one ongoing prospective study, even with suboptimal historical information, is unlikely to provide adequately precise information about the menstural cycle stage at operation.
Histologic Examination of Breast and Endometrial TissueHistopathological examination of normal breast tissue surrounding a breast cancer can be a helpful adjunct, because expert breast pathologists can reliably identify the histologic footprints of up to 12 specific menstrual cycle phases within normal breast tissue. This strategy, while useful, is also not definitive. It is possible, for example, that the malignancy within the resected breast tissue might diminish the reliability or otherwise affect the capacity of breast tissue to accurately reflect ovarian cycle stage.
The gold standard for determining menstrual cycle stage is clearly endometrial sampling. This procedure is usually performed in the offices of gynecologists, without anesthesia, using a simple manual aspiration technique. One can make a strong argument that any study asking about menstrual cycle phase must minimally determine that phase with a combination of optimal historical information, multiple hormone concentration measurements, and endo- metrial sampling. Despite the fact that endometrial sampling is the gold standard, it is not perfect. The combination of any two of the methods described above and most especially all three together is necessary to assure accurate timing. If these or comparably precise methods are not employed, the results of that study will be suspect. No ongoing or planned study employs such methodology.
Multiple Surgical Interventions
A palpably bothersome complication of the need for accurate triangulation of operation within the menstrual cycle is the nagging fact that many patients with breast cancer undergo more than one surgical resection for treatment of a single breast cancer. Furthermore, the results of at least two retrospective series indicate that the timing of preresection biopsy may also impact outcome. [11,12] Given this very real possibility, it must be concluded that the timing, within the menstrual cycle, of each and every surgical intervention must be known. This means that either the above minimum triangulation procedures must be employed serially for each and every potentially relevant operative procedure, or that all operative procedures planned for each of the patients studied must be temporally coordinated with one another; ie, they must be done in rapid succession with only a few days between them.
In fact, if you consider, just for a moment, the analytic and statistical complexity of determining the effect of the timing of two (or more) separate surgical interventions on a single outcome, only the latter strategy may actually be a practical option. Therefore, any prospective study that does not closely coordinate all surgical procedures in time will unlikely be able, in principle, to come to a definitive answer about whether or not the timing of surgery within the menstrual cycle influences breast cancer outcome. No ongoing or planned study employs such methodology.
Guarding Against Bias
A great deal of both medical-scientific and public information has been made available over the last 5 years about the possibility that the timing of surgery may impact breast cancer spread. Any study designed to answer this question prospectively must, therefore, protect the assignment of the timing of biopsy and/or operation from the possibility of conscious or subconscious manipulation by patient, staff, or surgeon. Study designs that do not provide foolproof assurance of protection from such bias will fail to yield definitive information. The design of no prospective study undertaken to date provides any such assurance.
It is unlikely that any amount of additional retrospective information will definitively answer this potentially important question. Not only will inadequately designed prospective studies fail to shed light on whether or not there is an exploitable relationship between the womans menstrual cycle and the surgical curability of breast cancer, but also such studies are actually likely to provide misleading information. The recent American Society of Clinical Oncology (ASCO) report of a Canadian study, with less than 18 months median follow-up, is a perfect example of the potential of inadequate chronobiological design to mislead.
This study was constructed by adding a menstrual history questionnaire to National Cancer Institute of Canada adjuvant trials. Early results appear to indicate that there may be an optimal time for cancer resection; however, this time may be in the first half of the cycle (actually day 8 after the first day of bleeding of the last menses). This timing, day 8, would correspond to putative early luteal phase in a woman with a 21-day cycle and to the follicular phase in a woman with a 28-day cycle. No data on individual cycle length are available. Furthermore, careful reading of the abstract reveals that these authors used the initial biopsy as their reference point. Therefore, if surgical resection timing is also important, the effect of its timing is not retrievable.
In summary, we do not now know, with adequate certainty, whether the menstrual timing of biopsy and/or resection of breast cancer meaningfully impacts outcome. If it does, we do not know how much it matters, nor do we know, with any degree of certainty, when within the cycle, the operation(s) should be performed. We must not further delay finding these things out. We have virtually no chance to answer these questions unless the necessary prospective studies are adequately designed to be able, in principle, to deliver a reliable answer. So far, they are not.
Why Bother Learning Whether the Timing of Breast Cancer Resection Impacts Breast Cancer Outcome?
At least 60,000 premenopausal women are diagnosed with breast cancer each year in the United States alone. The average absolute 10-year survival advantage for women resected during the early luteal phase of the cycle (from FSH/LH surge and ovulation and for approximately the next week) is 25%. Three-quarters of randomly resected premenopausal women are, therefore, resected at other phases of their menstrual cycles. This means that if the timing of resection does affect outcome, as many as 11,000 young women may be destined to die unnecessarily each year, by virtue of the random timing of breast cancer resection within their menstrual cycles.