Journal of Clinical Oncology, Vol 17, Issue 5
(May), 1999: 1632
© 1999 American Society for Clinical Oncology
National Surgical Adjuvant Breast and Bowel Project Breast Cancer Prevention Trial: A Reflective Commentary
Bernard Fisher
From the National Surgical Adjuvant Breast and Bowel Project (NSABP) Operations and Statistical Centers, Pittsburgh, PA.
Address reprint requests to Bernard Fisher, MD, National Surgical Adjuvant Breast and Bowel Project, Allegheny University of the Health Sciences, 4 Allegheny Center, Suite 602, Pittsburgh, PA 15212-5234; email bfisher1{at}aherf.edu
 |
INTRODUCTION
|
|---|
ON SEPTEMBER 16, 1998, a comprehensive report of the results from the National Surgical Adjuvant Breast and Bowel Project (NSABP) randomized clinical trial P-1, which was designed to evaluate the worth of tamoxifen for the prevention of breast cancer, was published in the Journal of the National Cancer Institute (JNCI).1 That study demonstrated that the administration of a systemic agent decreased the incidence of both invasive and noninvasive breast cancer in women who were at increased risk for the disease. Unfortunately, the significance of the findings presented in that report were eclipsed by a barrage of criticisms that appeared in the print media and on television, radio, and the Internet. These negative reactions had their inception as early as July 3, 1991, when the Food and Drug Administration (FDA) granted the NSABP permission to proceed with the P-1 study and have continued to the present time. They have blurred the focus of physicians and the public with regard to the published results from P-1, as well as their clinical applications. At times there has been what seems to be an obsessive emphasis on the adverse effects of tamoxifen rather than on the benefits resulting from its use. There has also been a lack of clinical perspective about how to weigh the risks and benefits of the drug. Debate has arisen about whether or not the P-1 study actually demonstrated prevention of breast cancer. It has also been contended that the trial was discontinued prematurely, and differences in the P-1 findings and in the results of two European studies that were also designed to evaluate the role of tamoxifen in the prevention of breast cancer have resulted in confusion. Moreover, a preoccupation with a second prevention trial has tended to minimize the meaning, worth, and clinical significance of the P-1 findings by conveying the notion that the P-1 trial merely provided justification for the conduct of the new trial, and that although it may have presented proof of the concept of prevention in principle, the P-1 trial has failed to provide support for the clinical application of the findings.
In this commentary, I will discuss (1) how the fallout from events that occurred during the 6-month period just before publication of the P-1 findings affected the way they were received by physicians and the lay public; (2) the scientific credibility of the study; (3) putting into perspective the magnitude of the benefit achieved from tamoxifen; (4) how best to evaluate, from a physician's viewpoint, the significance of the adverse effects as they relate to the benefits of tamoxifen; and (5) whether the use of the term "prevention" is appropriate in the context of the P-1 trial. The reader should consult the published articles for information on the design and implementation of the trial, conditions for participant eligibility, breast cancer risk assessment, risk-benefit estimates, statistical methods, secondary aims, and the study findings.1,2
|
EVENTS PRECEDING PUBLICATION OF THE P-1 FINDINGS
|
|---|
On March 24, 1998, 6 months before the results of the P-1 trial were reported in the JNCI, members of an independent data monitoring committee charged with oversight of the study had concluded that the primary end point of the trial had been attained and that tamoxifen had been found to reduce the incidence of both invasive and noninvasive breast cancer in women at high risk for the disease. On that basis, the committee recommended that the study be unblinded and that participants be informed of whether they had received placebo or tamoxifen so that those women who had received placebo might consider either taking tamoxifen or participating in a second prevention trial that would be designed to compare tamoxifen with another selective estrogen receptor modifier, raloxifene. The action of the monitoring committee was based upon stopping rules that had been established before the onset of P-1—a process that is commonly used in the conduct of most clinical trials. Moreover, it was believed that there was an ethical imperative for prompt disclosure of the findings, particularly in view of the magnitude of the benefit observed. Consequently, the comment that has often been made by physicians in the United States, Canada, and Europe that the trial was discontinued prematurely has no basis in fact.
During the 6 months that followed the monitoring committee's actions, individuals from both the lay and medical communities publicized their views about the trial, despite the fact that the investigators who had designed, initiated, and conducted the study had not yet published a formal analytical appraisal of the findings in a peer-reviewed medical journal. Their views arose from a series of unusual events. Within days of the committee's decision, the NSABP and the National Cancer Institute (NCI) had notified officials of various government agencies, as well as thousands of study participants, physicians, and their staff members, of the committee's decision. Moreover, the data used for making the decision to stop the study were placed on the Internet by the NCI—the first time, to my knowledge, that such an action had ever been taken before publication of seminal medical findings. On April 5, 1998, a major newspaper disclosed information about the P-1 study, and the next day, the NCI convened a press conference to report the findings, an action that resulted in worldwide coverage. Two weeks later, on April 21, 1998, the trial was the subject of a United States Senate Appropriations Subcommittee hearing that was convened to provide yet another setting for discourse about and criticism of the P-1 findings. In addition, the results of two European trials that had failed to confirm the P-1 trial findings were published and attracted even more media attention, which prompted yet another round of speculation about the worth of these findings. On April 30, 1998, the manufacturers of tamoxifen filed an application with the FDA seeking approval to use the drug as a preventive agent in women who are at increased risk for breast cancer, and a meeting of the FDA Oncologic Drug Advisory Committee (ODAC) was convened on September 2, 1998, to debate the merits of that application. The proposal was subsequently approved with several caveats, which, as reported by the media, resulted in further uncertainty about the worth of the study. Two weeks after the ODAC meeting, the report of the P-1 study findings appeared in the JNCI. Readers unfamiliar with how data are obtained via the clinical trials process and the findings prepared for publication might wonder why the P-1 results were not published either before or when the study data were initially disclosed. The material that had been reviewed by the independent data monitoring committee was not ready for publication because of a need for the performance of a statistical reanalysis at a follow-up time longer than that which had transpired when the information was first released and for the preparation of a meticulous cognitive analysis of the information obtained—all of which required months of effort.
When they were finally published, the P-1 findings were viewed by many as anticlimactic, because, by that time, a diversity of conclusions about the study had already been formulated as a result of all the aforementioned publicity it had received. The sequence of events that occurred during the 6-month period between the initial disclosure and the publication of the findings was a clear indication that the usual process for disseminating scientific information had changed so dramatically that it resembled the trial of the Knave of Hearts in Alice's Adventures in Wonderland, during which, at one point, the king demanded, "Let the jury consider their verdict," and the queen replied, "No, no! Sentence first—verdict afterwards."
|
SCIENTIFIC CREDIBILITY OF THE STUDY
|
|---|
The credibility of the scientific process used to obtain the P-1 findings or the data themselves have not been challenged. Indeed, such objections would be difficult to justify. In designing the study, we first formulated a testable hypothesis using information derived from laboratory and clinical research that had been carried out during the previous 20 years. Certain well-known characteristics of tamoxifen provided a firm basis for our premise that it was likely to be effective as a breast cancer preventive agent. It had been proven to be of value when used to treat advanced breast cancer as well as stages I and II disease; it reduced the incidence of contralateral breast cancer and interfered with the initiation, promotion, and growth of tumors in experimental systems; and it could be used in most patients safely with good compliance and minimal adverse effects. Moreover, as a prospective randomized clinical trial involving 13,000 women, P-1 was designed and conducted in such a way as to provide information that could reject, modify, or verify the credibility of our hypothesis. In addition, all aspects of the trial were outlined in a protocol that had been developed before the trial began. Our study was not designed to obtain retrospectively generated information that would produce an anecdotal report, such as has too often dictated medical practice. The P-1 findings were governed by the boundaries of the study that had been outlined in the protocol—boundaries that might have differed either greatly or slightly from those that had dictated the conduct of other prevention trials and that may account for the differences in the findings generated from various studies, a circumstance that physicians and the lay public do not fully comprehend when presented with seemingly contradictory findings from diverse clinical trials. Thus, because appropriate methodology was used in the conduct of the P-1 study, its findings must stand on their own merit, and the only valid grounds for argument would lie in a difference in interpretation of the results.
|
PUTTING INTO PERSPECTIVE THE MAGNITUDE OF THE BENEFIT ACHIEVED FROM TAMOXIFEN
|
|---|
The findings from the P-1 trial are straightforward and provide strong support for the hypothesis tested. There was a highly significant reduction (49%) in all age groups in the rate of occurrence of invasive breast cancer in women who received tamoxifen. Figure 1 puts into perspective the frequency with which breast cancer occurred among women in the placebo group who were either 49 years of age or younger or 50 years of age or older and portrays the extent to which their tumors were prevented by tamoxifen. Each dot in Fig 1 represents one woman among 1,000 recipients of either placebo or tamoxifen who demonstrated an invasive or noninvasive breast cancer during a 5-year period. There was a 44% decrease in the rate of breast cancer in women 49 years of age or younger and a 53% decrease in women 50 years of age or older. Invasive breast cancers also decreased in subgroups of participants at various levels of risk. Moreover, during each of the first six yearly intervals of follow-up, the occurrence of breast cancer was reduced in women who received tamoxifen—by 33% in the first year to somewhat more than 50% in the fifth and sixth years of follow-up.
View larger version (30K):
[in this window]
[in a new window]
|
Fig 1. Rate of breast cancer observed in the placebo and tamoxifen groups of participants in the P-1 study according to age. Each dot represents one event per 1,000 women over 5 years.
|
|
The P-1 findings also demonstrated a relationship between the estrogen receptor (ER) status of a pathologic lesion in the breast and the benefit achieved from tamoxifen. This was of particular importance because lesions that were ER-positive were most apt to be prevented from becoming clinically evident. Most intriguing was the extent of the increased risk for invasive breast cancer found in women with a history of lobular carcinoma-in-situ (LCIS) or atypical hyperplasia, lesions that are most often ER-positive. Of equal importance was the observation that tamoxifen dramatically reduced the incidence of invasive cancers in women who had previously harbored these lesions, a strong suggestion that the invasive cancers prevented were most likely ER-positive. Associated with those observations was one that demonstrated that tamoxifen administration decreased by 50% the clinical expression of LCIS and of ductal carcinoma-in-situ (DCIS), the latter being a precursor of invasive breast cancer that has also been demonstrated to contain ER-positive cells and to respond to tamoxifen therapy. How these pieces of the breast cancer puzzle fit together is not yet clear, but the P-1 findings have served to increase recognition of this biologic phenomenon and raise the question of when cells in the biologic cascade of events leading from tumor initiation to the phenotypic expression of invasive tumors express or lose their ER status and thus may be affected by tamoxifen.
At this juncture, it seems reasonable to attempt to determine, on the basis of the P-1 data, whether tamoxifen should be administered to a large population of eligible women, even though the vast majority of them would never develop breast cancer and even though not all tumors would be prevented in the women who would have had a tumor. Addressing that issue is perplexing, because it is not possible to identify a priori those women who either would or would not benefit from tamoxifen. Such a conundrum is not unique to the P-1 study. Similar uncertainties have occurred in previous studies conducted by the NSABP in the use of systemic adjuvant therapy and in evaluating the worth of postoperative breast irradiation. In those circumstances, many women who either did not require those therapies or who would not have received a benefit even if they did need them have undergone such treatment in order to ensure that the women who would benefit were not denied the opportunity to do so. Thus, despite continued uncertainties about who should or should not receive those treatments, the lives of countless women have been improved. Might not the same reasoning be applied when tamoxifen is considered for use as a breast cancer preventive agent? Perhaps the decision with regard to whether that course of action is appropriate might be facilitated by knowing how many women in a much larger population similar to that of P-1 would be likely to benefit from the drug.
According to NCI estimates, approximately 29 million women in the United States would have been potentially eligible for the P-1 trial and would thus have been expected to respond to tamoxifen in a manner similar to that of participants in that study. The number of women in that population who have the potential to benefit from receiving tamoxifen might be estimated by using the data reported in our recent publication of the P-1 findings. Because the average annual rate of the occurrence of invasive breast cancer in each 1,000 participants in the placebo group of P-1 was 6.76, it might be estimated that, in the population of 29 million women, almost 1 million would, during a 5-year period, have the potential for being diagnosed with such a tumor. In the tamoxifen group of P-1, where the rate of such tumors was 3.43 per 1,000 women per year, approximately 500,000 invasive breast cancers might be detected during that time. Thus almost one-half of a million invasive breast cancers would be prevented in the expanded population. A similar estimate indicates that almost 200,000 noninvasive tumors (ie, either DCIS or LCIS) would be prevented in that population. In view of these estimates, the benefits that might be achievable by more widespread use of tamoxifen cannot be viewed as trivial.
It must be emphasized that the magnitude of benefits observed in the P-1 study relates to the level of breast cancer risk in the women being evaluated. The greater that risk in women who comprise a population, the greater will be the number of breast cancers that occur and thus the greater the benefit from tamoxifen. A broad spectrum of risks existed among women who participated in the P-1 study. In some, the risk of developing an invasive breast cancer was just great enough to make them eligible for the trial, whereas in others, that risk was much greater. If, for example, the 5-year predicted risk in all of the women comprising the population of 29 million were  5.01%, it would be estimated that almost 2 million invasive cancers would have occurred in the placebo group and 650,000 in the tamoxifen group during a 5-year period. Thus approximately 1.2 million invasive breast cancers might have been prevented.
On the other hand, if the 5-year predicted risk in all 29 million women were  2.0%, then approximately 500,000 tumors might have been prevented. Thus expanding the findings from the P-1 trial to a larger population of putatively similar women vividly demonstrates the potential impact that the wider use of tamoxifen, or of a similar drug of proven efficacy, could have in diminishing the extent of the breast cancer problem. The findings support the axiom that small benefits attained in a frequently occurring disease can result in a major advance. It must be emphasized, however, that these statements do not imply that all 29 million women should necessarily receive tamoxifen.
|
EVALUATION OF ADVERSE EFFECTS AND COMPARISON WITH BENEFITS FROM TAMOXIFEN
|
|---|
Because the P-1 findings provide justification for the conclusion that, when used as a breast cancer preventive agent, tamoxifen results in a substantial advantage in women who are at increased risk for the disease, it is appropriate to consider the magnitude of the adverse effects of the drug and what their impact might be on any decision about how widespread its use should be. Numerous accounts (mostly anecdotal) of the dangers of liver damage, hepatoma, colon cancer, retinal toxicity, and of certain unfavorable quality-of-life issues resulting from tamoxifen administration have attracted a great deal of attention. The P-1 findings, however, have failed to support such concerns. The incidence of invasive cancers other than those in the breast and endometrium were equal among women in both the placebo and tamoxifen groups. In addition, no liver cancers have occurred to date, and there has been no increase in colon, rectal, ovarian, or other genitourinary tumors. A recent analysis of findings with regard to the effect of tamoxifen on the quality of life of P-1 participants has also failed to reveal any increases in adverse physical, emotional, or sexual conditions that were sufficiently troublesome to prevent women from dealing with them as part of their usual lifestyle.
The increased risk of endometrial cancer and vascular-related toxic effects, which occur most often in postmenopausal, tamoxifen-treated women, has been used by opponents of the P-1 study as a rationale for arguing first against the implementation of the trial, then for discontinuing it, and, recently, for discouraging administration of tamoxifen, despite its proven benefits. How justifiable are these criticisms? Recent reviews of the relationship between tamoxifen and endometrial cancer suggest that the concern about the level of excess risk of endometrial cancer may have been overestimated and that, when it does occur in women who receive tamoxifen, their prognosis is as favorable as that of women who have not taken the drug or of those who have received estrogen-replacement therapy. When our P-1 analyses were reported, all endometrial cancers in women who received tamoxifen were highly curable, localized stage I tumors. In addition, there have to date been no deaths from endometrial cancer in P-1 trial participants. It is difficult to convey to a woman and her physician what her chances are of developing endometrial cancer or a vascular event if she decides to take tamoxifen. Risk ratios with confidence intervals, cumulative incidence rates, or average annual incidence rates may fail to accurately indicate what might be expected in that regard. Being told that there is "a four- or five-fold increase" in the rate of such an event if one takes tamoxifen may not put a woman's risk into proper perspective. To simplify the task, we have prepared Fig 2, using the same format as that of Fig 1 in this commentary. Each dot in Fig 2 represents one woman among 1,000 recipients of tamoxifen who demonstrated an endometrial cancer (EC), a pulmonary embolus (PE), a stroke (St) or a deep-vein thrombosis (DVT) that could be attributed to having taken the drug over a 5-year period. The information in the figure shows that approximately seven of 1,000 women who received tamoxifen developed an EC over a 5-year period. For those who are more accustomed to information expressed in percentages, the findings demonstrate that the overall rate of EC related to tamoxifen was 0.7% over 5 years. This indicates that fewer than one woman per 100 will have an EC over a 5-year period. A similar depiction of the undesirable vascular events that resulted from tamoxifen administration in the P-1 study shows that, over 5 years, approximately 0.2% to 0.3% of participants developed a stroke, approximately 0.2% had a PE, and between 0.2% and 0.3% suffered from a DVT. In women 49 years of age, these events were less frequent, whereas in women 50 years, they occurred somewhat more frequently. In the older women, the rate of an EC that occurred in P-1 as a result of tamoxifen administration was approximately 1% over 5 years, whereas for each of the vascular-related events, the rate was approximately 0.5% for stroke, 0.3% for PE, and 0.3% for DVT. In women who had had a hysterectomy, the rate of vascular-related events was similar to that experienced in women who had not had a hysterectomy.
View larger version (30K):
[in this window]
[in a new window]
|
Fig 2. Adverse events from tamoxifen in P-1 participants overall, according to age, and in women who had a hysterectomy. Each dot represents one event per 1,000 women over 5 years.
|
|
Thus, in P-1, tamoxifen administration resulted in an increase in endometrial and vascular events that was greater than that observed in women who received placebo. Despite this increase, the incidence of those adverse events is likely to be judged as being low. However, the incidence can be viewed in its proper perspective only when it is related to some other parameter, in this case, to the decreased incidence of breast cancer in women at increased risk for the disease, a decrease that is likely to be viewed as "substantial." Although there is ample justification for relating these two independent sets of observations to one another, how the connection between them should be established is unclear. To estimate the net benefit from tamoxifen by merely subtracting the number of adverse events from the number of cancers prevented and then using the resulting value to formulate conclusions about whether the drug should be given is questionable. Such an approach assumes that all events are equivalent relative to their significance. As has been previously pointed out, the morbidity and mortality resulting from EC in women who receive tamoxifen is apt to be lower than that associated with an unprevented breast cancer, particularly when there is appropriate patient follow-up in women who receive the drug to enable ECs to be detected at an early stage. Similarly, if women are carefully selected to receive tamoxifen so that they are free of comorbid conditions that predispose them to vascular complications, then the incidence of those events will be reduced so that the adverse effects from surgery, radiation therapy, chemotherapy, and tamoxifen used to treat the breast cancers that would have occurred if tamoxifen had not been given would be apt to exceed those associated with the vascular events. Consequently, using the number of adverse events noted to cancel out an equal number of tumors prevented is questionable.
Other factors to be taken into consideration when the incidence of adverse events is related to the incidence of breast cancers prevented are those indicating that rates of breast cancer occurrence vary according to a woman's risk of the disease, that the rates of EC and vascular events relate to a woman's age, and that the occurrence of EC also relates to whether or not a woman has had a hysterectomy. Consequently, if one is inclined to compare the rate of tumors prevented with the rate of adverse events in a risk-benefit evaluation, then any attempt at a quantitative estimate requires that the appropriate risk profile predicting adverse events be associated with the one indicating a woman's risk for breast cancer. When such a correlation is established, based on the findings from the P-1 study, women younger than 50 years of age who meet the eligibility requirements of the trial are likely to be considered highly eligible for tamoxifen, because their risk of having an adverse effect is practically nil, and the reduction in the incidence of breast cancer for the group overall is reduced by almost one half. Moreover, the greater their risk, the greater the benefit. Postmenopausal women who have had a hysterectomy are also favorable candidates for tamoxifen because most of their risk of an adverse event is eliminated because they cannot develop EC. Because women with a history of LCIS or atypical hyperplasia are at particularly high risk for breast cancer, and because tamoxifen markedly reduces that risk, the level of benefit achieved markedly outweighs the adverse effects that might result from the drug. Also, because the risk of invasive breast cancer in women with localized DCIS is at least as high if not higher than that for women with a history of LCIS or atypical hyperplasia, the benefit they would receive from tamoxifen, insofar as reducing their rate of invasive breast cancer is concerned, is likely to eclipse the consequences of the adverse drug effect. Consequently, they too may be viewed as candidates for the drug. Although no information is as yet available to indicate whether women who are at increased risk for breast cancer because they carry BRCA1 or BRCA2 mutations should be considered candidates for tamoxifen, they should nonetheless be afforded that option, particularly if they are contemplating having bilateral mastectomy to prevent breast cancer.
The decision regarding whether women 50 years of age or older who have stopped menstruating, have not had a hysterectomy, and have no history of LCIS, DCIS, or atypical hyperplasia should receive tamoxifen is less clear. Because the incidence of adverse events remains constant regardless of the cancer risk in such women, it is evident that the greater the risk, the less controversial the issue, because of the greater likelihood that the mortality and morbidity associated with the breast cancers that might have been prevented by tamoxifen would exceed that resulting from the adverse events associated with the drug. A precise level of risk above and below which such a woman should or should not be considered a candidate for tamoxifen has not yet been determined and is likely to be one on which it is difficult to agree. It is important that a woman for whom tamoxifen is being considered be viewed as being more than just someone at risk for breast cancer. Her general health and well-being must also be assessed. The probability that an adverse event from tamoxifen will occur in women who are 50 years of age or older should not prevent the use of tamoxifen in this age group. This contention is supported by two major considerations: (1) the P-1 study was unblinded by an independent group of investigators from a variety of disciplines in order to afford study participants taking placebo (including women aged 50 years or older) the opportunity to receive tamoxifen if they wished or to participate in the new NSABP P-2 study, in which women 50 years of age and older who receive tamoxifen will serve as the standard group against which the benefits and adverse effects from raloxifene will be measured, and (2) after the report of the P-1 findings, British investigators did not alter their prevention trial because of the adverse EC and vascular events reported in P-1. These circumstances clearly indicate that, in the general population, there must be a substantial number of women 50 years of age and older for whom tamoxifen administration would be appropriate.
|
THE MEANING OF THE WORD "PREVENTION"
|
|---|
In my view, both the certainty and the magnitude of the effect of tamoxifen on the prevention of breast cancer are remarkable aspects of the P-1 findings. In more than three decades of research, I have rarely, if ever, observed such a major response to a breast cancer therapeutic agent. Although there has been little or no dispute over the authenticity of the P-1 findings, there has been discord over how best to interpret them. In that regard, one of the caveats associated with the FDA approval of tamoxifen was that it not be designated as a breast cancer preventive agent but, rather, that it be classified as a drug to be used to decrease the risk of the disease. For a decade, neither the NCI, the FDA, other governmental agencies, nor any of the institutional review boards that had approved the P-1 study had challenged the use of the term "prevention" as it related to tamoxifen and the occurrence of breast cancer, ischemic heart disease, or osteoporosis. Before the September 1998 FDA meeting, none of these organizations had ever suggested that the term be eliminated or modified. In our 1991 grant application, and again in our recently published article, I explained our use of the term as follows: "Whether the benefit achieved from tamoxifen in the P-1 study was due to the drug's interference with the initiation [genetic changes in normal cells resulting from carcinogenic agents], promotion [clonal expression and genetic changes resulting in preneoplastic lesions progressing to noninvasive and malignant tumors], or to hindrance of the growth of [phenotypically expressed] occult tumors is unknown. Because it is likely that a broad spectrum of molecular-biologic and pathologic changes in breast tissue existed among participants at the time of their entry into the trial, it might be assumed that both mechanisms are responsible for the finding. [Our study, as well as other prevention trials that are currently being conducted or that are in the planning stages, are unlikely to be capable of providing information to resolve the conundrum about precisely where tamoxifen exerts its effect.] Nonetheless, the absence of specific information to resolve the issue does not detract from the evidence indicating that tamoxifen did, in fact, prevent the clinical expression of tumors, ie, the goal of primary disease prevention." Moreover, we have never indicated that our use of the term prevention in the P-1 study implied that the initiation of breast cancers had been prevented or that all tumors that had been prevented from becoming detectable during the course of the trial were permanently eliminated. Nonetheless, I continue to consider the term prevention, as we have defined it, as being appropriate. Furthermore, I disagree with the member of the ODAC committee who asserted that the beneficial effects observed in the P-1 study are not prevention in the way that scientists understand the meaning of the word. That opinion played a part in influencing the FDA's decision to ban use of the word and to replace it with the term "risk reduction." It is not my impression that scientists involved in prevention research have negated the propriety of using the terms prevention or chemoprevention to describe what occurred in our study. Moreover, the idea that tamoxifen reduces the "risk" of breast cancer is both confusing and imprecise, as it is difficult to devise a mechanism of action that could account for how a drug interacts with a risk so as to reduce it! In the end, it matters little whether the term prevention, risk reduction, or reduction in tumor incidence is used; the P-1 data continue to stand on their own merit. Considered at face value, they provide the first evidence that the administration of a systemic agent can perturb the natural history of seemingly healthy women who are at increased risk for developing invasive or noninvasive breast cancer. And, in that regard, they have a historic relationship to the findings of another study that we reported in 1975, which demonstrated for the first time that postoperative adjuvant chemotherapy could alter the outcome of women with invasive breast cancer.3
|
CONCLUDING REMARKS
|
|---|
It has been almost a decade since my colleagues and I decided to respond to an NCI request for a proposal to conduct a study aimed at determining whether tamoxifen could serve as an effective breast cancer preventive agent. At that time, we decided that there was sufficient scientific justification for formulating a firm hypothesis that was testable by using tamoxifen as a probe and a clinical trial as the mechanism for estimating its effect. Consequently, since its inception, I have viewed the P-1 trial as a scientific investigation, ie, an experiment aimed at establishing the credibility of the concept being evaluated. The study was not about drug testing or about estimating the extent of the undesirable side effects of tamoxifen. Nor was it about the process used to conduct the study or about the political and social issues that somehow became entwined with it. It is, therefore, disconcerting that the role played by science in the conduct of P-1 and the scientific and clinical significance of the findings have been largely ignored or obscured by diverse opinions, criticisms, and interpretations that have been made by individuals other than those responsible for the investigation.
The P-1 experiment obtained data that supported the concept being tested. Moreover, it expands knowledge about pathologic entities in the breast that have escaped detection by any means currently available. From a clinical standpoint, of even greater importance is the evidence that those aberrant entities can be manipulated so that their clinical appearance is delayed or prevented. In addition, the biologic and clinical findings from P-1 have provided the justification for the conduct of multiple prevention studies that will expand on the P-1 findings and, it is hoped, provide answers to the innumerable questions that have arisen as a result. Other trials may demonstrate a greater or lesser prevention of breast cancer, depending on the strength of the hypotheses established, the scientific validity of the probes, and the credibility of the clinical trials used for assessing their effect. Obviously, in these studies, the risk of breast cancer and the characteristics in the population of women being evaluated will influence the extent of the benefit achieved. Despite the need for additional research to eliminate the breast cancer problem, the fact remains that, because an appropriate pathway of science was traversed in the conduct of the P-1 trial, the impressive reduction in the incidence of both invasive and noninvasive cancer that occurred cannot be negated by any criticism directed against the study.
Although female breast cancer remains a public health problem, notable improvements have been made in its diagnosis and treatment. Lumpectomy followed by breast irradiation is now recommended for most patients, and radical mastectomy has become of only historic importance. Adjuvant systemic chemotherapy and hormonal therapy are now major components of treatment strategies, and early-stage breast cancer is being detected with increasing frequency by mammography and other procedures. Of particular significance are the recently reported findings demonstrating that the concept of breast cancer prevention has become a reality. It must be emphasized that each of these advances, and others that have occurred during the past 30 years, has become part of the current treatment paradigm despite the fact that the initial study findings gave rise to a multitude of questions, many of which are still unresolved. Uncertainty regarding adverse events, duration of therapy, impact on survival, which patients should or should not receive therapy, and what represents an acceptable risk-benefit relation has not prevented the use of those agents, which have improved the lives of countless women. Questions also remain to be answered relative to the use of tamoxifen for the prevention of breast cancer. One particular issue relates to the methodology to be used for more specifically quantifying the risk of women such as those who participated in the P-1 study. Disagreement among epidemiologists, biostatisticians, and clinicians as to what might constitute the best way to define such risks should not, at this time, impede the use of tamoxifen as a breast cancer preventive agent.
|
ACKNOWLEDGMENTS
|
|---|
Supported by Public Health Service grants no. U10-CA37377 and U10-CA69974 from the National Cancer Institute, National Institutes of Health, U.S. Department of Health and Human Services.
I recognize Joseph P. Costantino, DrPH (NSABP statistician), and Carol K. Redmond, ScD (former director of the NSABP Biostatistical Center), for their efforts which made the P-1 study a reality, and thank Tanya Spewock for editorial assistance and Mary Hof for preparation of the manuscript.
|
REFERENCES
|
|---|
1.
Fisher B, Costantino J: Highlights of the NSABP Breast Cancer Prevention Trial. Cancer Control 4:78-86, 1997[Medline]
2.
Fisher B, Costantino JP, Wickerham DL, et al: Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 study. J Natl Cancer Inst 90:1371-1388, 1998[Abstract/Free Full Text]
3.
Fisher B, Carbone P, Economou SG, et al: L-phenylalanine mustard (L-PAM) in the management of primary breast cancer: A report of early findings. N Engl J Med 292:117-122, 1975[Abstract]
Submitted February 16, 1999;
accepted February 26, 1999.
This article has been cited by other articles:
|
|
|
|
 
K. Anderson, J. S. Jacobson, D. F. Heitjan, J. G. Zivin, D. Hershman, A. I. Neugut, and V. R. Grann
Cost-effectiveness of preventive strategies for women with a BRCA1 or a BRCA2 mutation.
Ann Intern Med,
March 21, 2006;
144(6):
397 - 406.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
B. Fisher, J. P. Costantino, D. L. Wickerham, R. S. Cecchini, W. M. Cronin, A. Robidoux, T. B. Bevers, M. T. Kavanah, J. N. Atkins, R. G. Margolese, et al.
Tamoxifen for the Prevention of Breast Cancer: Current Status of the National Surgical Adjuvant Breast and Bowel Project P-1 Study
J Natl Cancer Inst,
November 16, 2005;
97(22):
1652 - 1662.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. N. Freedman, B. I. Graubard, S. R. Rao, W. McCaskill-Stevens, R. Ballard-Barbash, and M. H. Gail
Estimates of the Number of U.S. Women Who Could Benefit From Tamoxifen for Breast Cancer Chemoprevention
J Natl Cancer Inst,
April 2, 2003;
95(7):
526 - 532.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. J Cersosimo
Tamoxifen for Prevention of Breast Cancer
Ann. Pharmacother.,
February 1, 2003;
37(2):
268 - 273.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. Sato, T. Fukutomi, S. Akashi-Tanaka, K. Miyakawa, N. Yamamoto, and T. Hasegawa
Accuracy in Estimating Tumor Extension According to Mammographic Subtypes in Patients with Ductal Carcinoma In Situ
Jpn. J. Clin. Oncol.,
May 1, 2002;
32(5):
157 - 161.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
D. Hershman, V. Sundararajan, J. S. Jacobson, D. F. Heitjan, A. I. Neugut, and V. R. Grann
Outcomes of Tamoxifen Chemoprevention for Breast Cancer in Very High-Risk Women: A Cost-Effectiveness Analysis
J. Clin. Oncol.,
January 1, 2002;
20(1):
9 - 16.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
S. M. Lippman and P. H. Brown
RESPONSE: Re: Tamoxifen Prevention of Breast Cancer: an Instance of the Fingerpost
J Natl Cancer Inst,
April 19, 2000;
92(8):
658 - 658.
[Full Text]
[PDF]
|
|
|
|
|
|
|
B. Fisher
RESPONSE: Re: Tamoxifen Prevention of Breast Cancer: an Instance of the Fingerpost
J Natl Cancer Inst,
April 19, 2000;
92(8):
659 - 659.
[Full Text]
[PDF]
|
|
|
|
TOP
INTRODUCTION
EVENTS PRECEDING PUBLICATION OF...
