Special Report:
Highlights of the NSABP Breast Cancer Prevention Trial

Bernard Fisher, MD, Operations Center, National Surgical Adjuvant Breast and Bowel Project (NSABP), Allegheny General Hospital, Pittsburgh, Pa, and Joseph Costantino, DrPH, Biostatistical Center, NSABP, Graduate School of Public Health, University of Pittsburgh, Pa

Introduction

Despite the benefits that have been achieved from current methods of detecting and treating breast cancer, there remains a need for more effective drug administration, for the examination of new therapeutic modalities, for the discovery of better methods of overcoming drug resistance, and for the evaluation of therapeutic modulation based on pharmacologic principles. Although some of these efforts may produce additional progress in the treatment of the disease, more speculative approaches based on new biological and clinical information also need to be explored.

One such approach is currently being investigated by the National Surgical Adjuvant Breast and Bowel Project (NSABP) in a trial to test the hypothesis that tamoxifen administered to women at "high" risk for breast cancer can prevent and/or can interfere with the progression of the disease so that it never becomes detectable in a woman's lifetime. The justification for formulating this hypothesis arose from findings obtained from extensive laboratory and clinical investigations of tamoxifen, particularly from those indicating that tamoxifen decreased the incidence of contralateral breast cancer in women with invasive breast cancer. The study to test the credibility of that hypothesis is referred to as the NSABP Breast Cancer Prevention Trial (BCPT). This invited commentary presents a brief overview of (1) the origins of the BCPT, (2) the justification, objectives, and major aspects of the trial, (3) the status of the study, and (4) comments regarding the study.

Origins of the NSABP BCPT

In 1958, when our laboratory began investigating the biology of tumor metastasis and the NSABP began to conduct clinical trials that would change the way breast cancer was treated, it would have been fanciful to have predicted that evaluating the worth of modalities that might prevent that disease could be justified. By the mid-1980s, however, several investigators became interested in conducting such studies. At that time, a few of them were promoting a study to evaluate whether a reduction in fat intake in postmenopausal women could lower their risk of breast cancer. Alternatively, in November 1984, we expressed interest via a grant application to the National Cancer Institute (NCI) in conducting a study to evaluate the worth of tamoxifen as a breast cancer prevention agent. In our proposal we hypothesized that there might be a link between the possible finding of a decrease in cancer of the contralateral breast in women who had invasive cancer and the potential of tamoxifen to prevent breast cancer in healthy women. That and a subsequent proposal were not approved. It was not until eight years later, on June 1, 1992, that we had the opportunity to initiate the BCPT, the first breast cancer prevention trial in the United States.

In June 1990, the NCI extended an invitation to the chairpersons of eight clinical cooperative groups and to seven cancer centers serving as NCI-funded Clinical Cancer Oncology Program research bases to submit, by October 1, 1990, a detailed proposal for a feasibility (pilot) study that, if approved, would permit the design of a protocol for a breast cancer prevention trial. The document accompanying the invitation indicated that the trial should be conducted using postmenopausal women and that there should be measurable endpoints for determining the effects of tamoxifen not only on breast cancer but also on cardiovascular disease and overall mortality. It was also recommended that the effect of tamoxifen on bone mineral density and osteoporosis be evaluated. Proposals were to be rigorously reviewed by the Cancer Control Protocol Review Committee in the NCI Division of Cancer Prevention and Control, by the Cancer Therapy Evaluation Program Protocol Review Committee, by representatives of the National Heart, Lung, and Blood Institute, and by other NCI/National Institutes of Health staff. In addition, peer review was to be conducted by an ad hoc Special Review Committee convened by the Division of Extramural Activities of the NCI. The final protocol was to be reviewed for approval by these same governmental agencies as well as by other peer reviewers external to the NCI.

One of us (BF), then NSABP chairperson and principal investigator, and Dr. Carol Redmond, then director of the NSABP Biostatistical Center, decided that our combined clinical trials experience of more than two decades and, in particular, our experience with tamoxifen justified our submitting a proposal. We believed that there was scientific justification for conducting the trial and that such an endeavor represented a direct extension of the breast cancer research we had been conducting for 30 years. Moreover, because the NSABP comprised an established network of investigators who had demonstrated their ability to maintain and promote the compliance of patients on clinical trials, a mechanism existed for conducting such a study. Preparation of our proposal took from July to September 1990, and, on September 28, it was submitted to the NCI Division of Cancer Prevention and Control. A single-spaced document of more than 200 pages outlined the study aims, the justification for conducting the study and for using tamoxifen, and the design, administrative aspects, and conduct of the trial. In February 1991, our application was approved by the NCI and the National Cancer Advisory Board, and funding was recommended for five years. We could not foresee, however, that 16 more months would elapse before the first participant would be enrolled in the trial and that there would be turbulence from many fronts that, on occasion, would threaten to subvert our effort.

After approval of the grant, almost daily telephone discussions took place between the NSABP leadership and NCI officials regarding preparation of virtually every aspect of the protocol. A number of meetings between NSABP and NCI personnel were also convened. One such meeting occurred at the NCI in April 1991, when we expressed the view that premenopausal women should also be permitted to participate in the study. Our arguments for accruing younger women included the following: (1) we believed young women at increased risk had a right to choose whether to participate since their only alternative was bilateral mastectomy — not a very desirable or justifiable option, in our view, (2) the NSABP B-14 study had indicated that the reduction in incidence of contralateral breast cancer by tamoxifen was greater in premenopausal women, (3) premenopausal women with breast cancer who had first-degree relatives with breast cancer tended to experience their own breast cancers five to 10 years earlier than did the affected relatives, (4) protocol compliance was likely to be higher in younger women at increased risk, (5) breast cancer diagnosed at an earlier age was likely to be associated with a poorer prognosis than that diagnosed at a later age, (6) interventions to inhibit the development of breast cancer were likely to be more effective if applied at an earlier age, (7) one third of women who developed breast cancer had been diagnosed with the disease before menopause, (8) it was our view that the results of a study employing only postmenopausal women could not be generalized to a younger population unless younger women also were evaluated in such a study, (9) non–life-threatening toxicities were shown to be similar in premenopausal and postmenopausal patients enrolled in the B-14 study, and (10) life-threatening toxicities such as endometrial cancer and thromboembolism were rarely observed after tamoxifen administration in younger women. These reasons apparently satisfied NCI officials of the need to enroll younger women, and the age for entry into the trial was subsequently lowered to 35 years.

A major hurdle that we had to overcome was the Food and Drug Administration (FDA) hearing that took place on July 2, 1991, which was conducted for the purpose of deciding whether or not to approve our protocol. The NSABP and the NCI were in complete agreement with the content of the protocol, including the consent form, submitted to the FDA. Despite much criticism and behind-the-scenes lobbying by various advocacy groups against the trial before, during, and after the hearing, the proposal was approved the next day (on July 3, 1991), and we were given permission to proceed with the study provided that certain modifications would be made. We had finally achieved our goal one year and five days after the NCI's request for, and nine months after submission of, our proposal. However, there was little reason for celebration. As we were to subsequently learn, an additional 11 months of work would be necessary before the trial would be ready to begin.

During that time, under the direction of Dr. Redmond and one of us (BF), the NSABP Headquarters staff interacted with (1) members of the National Heart, Lung, and Blood Institute and its advisers regarding the cardiovascular component of the protocol, (2) officials of the National Institute of Arthritis and Musculoskeletal and Skin Diseases and their extramural colleagues regarding the development of an osteoporosis and bone density study, and (3) the vast cadre of individuals at all levels of authority in the NCI, FDA, Office of Protection of Research Risks, and other governmental agencies. At the same time, we were extensively involved with not only the pharmaceutical company supplying the tamoxifen but also the review boards of certain institutions who were to participate in the trial. We were required to display a full-time, passionate commitment to the task, as well as a willingness to negotiate and compromise with a diverse group of individuals in order to achieve the goal of implementing what we believed had the potential to be the most important breast cancer clinical trial ever undertaken. In addition, substudies were planned to evaluate quality-of-life issues, putative tamoxifen-induced toxicities affecting the eye, endometrium, and liver, and the creation of a tissue bank for future genetic investigations. Numerous committees were established, members were selected, and meetings were held. Key individuals from the various governmental agencies participated in the preparation of the BCPT consent form and approved the document. Plans were made for the recruitment and selection of investigators by a peer-review process similar to that used by the NCI. An extensive educational program was prepared for physicians, their associates, and potential trial participants. A great deal of time was devoted to the development and implementation of recruitment strategies, particularly those necessary for obtaining minority participation.

The individuals comprising the leadership of the NCI Division of Cancer Prevention and Control are to be commended for inaugurating the process whereby the BCPT evolved. Moreover, their support during both the development of the protocol and the early years of the trial made it possible for the NSABP, through its established network of institutions and enthusiastic investigators, to successfully launch and conduct the study.

Justification, Objectives, and Major Aspects of the Trial

Justification for the Study

Before we submitted our proposal and prepared a protocol, we felt it necessary to obtain assurance that there was, indeed, adequate scientific and clinical information to justify the conduct of the trial. Our decision to proceed was based on a variety of justifications. The use of tamoxifen to treat patients with clinically detectable breast cancer was among the most successful therapeutic efforts of the 1980s. Studies demonstrated the benefit from tamoxifen in reducing tumor recurrence and in prolonging survival when the drug was used both as a single agent and in combination with chemotherapy for the treatment of advanced breast cancer and as postoperative adjuvant therapy in stages I and II disease. Of particular importance was the observation that patients who received tamoxifen had a significantly lower incidence of contralateral breast cancer than patients who received a placebo. Moreover, extensive information was available to indicate that most women could use tamoxifen safely with good compliance and minimal side effects. In addition, it was our conviction that this study provided an opportunity to obtain genetic information about breast cancer from a high-risk population. Even if the BCPT failed to demonstrate an advantage from the use of tamoxifen, the genetic studies planned for the trial could provide compelling findings, such as those indicating the incidence of BRCA-1 and BRCA-2 mutations in the study population and whether or not tamoxifen is preferentially effective in women with genetic alterations.

Most important for convincing us of the propriety for evaluating the worth of tamoxifen as a breast cancer prevention agent was the extensive literature related to its pharmacokinetics, metabolism, and antitumor effects in experimental animals as well as in humans. Moreover, tamoxifen had been shown to interfere with the initiation and promotion of tumors in experimental systems and to inhibit the growth of malignant cells by a variety of putative mechanisms. In the BCPT, however, a reduction in incidence of breast cancer might be due to none of those mechanisms but rather to tamoxifen's treatment effect on subclinical phenotypically expressed tumors.

Although it might be expected that a breast cancer prevention trial should concentrate solely on issues related to that disease, other information regarding the action of tamoxifen was sufficiently compelling to justify broadening the parameters of the study. Because tamoxifen perturbs lipid and lipoprotein metabolism and because those changes could influence a patient's risk of coronary artery disease, it seemed appropriate that the effects of tamoxifen on serum cholesterol, high-density lipoprotein, low-density lipoprotein, and incidence of mortality from coronary disease should be assessed simultaneously. Moreover, because of evidence indicating that tamoxifen might have a beneficial effect on osteoporosis, possibly due to its estrogen agonist activity, it was deemed appropriate to include a component in the BCPT that evaluated the effect of tamoxifen on osteoporosis.

Study Objectives

The BCPT was designed to determine whether tamoxifen therapy administered for at least five years (1) effectively prevents invasive breast cancer and reduces mortality attributable to that disease in women at increased risk, (2) lowers the incidence of fatal and nonfatal myocardial infarction, (3) reduces the incidence of bone fractures, and (4) evaluates tamoxifen toxicity and side effects in order to assess benefit versus risk from the drug. In keeping with recent advances, another objective was to obtain information regarding breast cancer genetics.

Primary Outcome Measures

Relative to breast cancer, the events of interest are time to development of invasive breast cancer and time to death due to invasive breast cancer. Relative to cardiovascular disease, the events of interest are time to nonfatal myocardial infarction and time to death from arteriosclerotic coronary heart disease. Insofar as osteoporosis is concerned, time to first fracture is the major event of interest. Total mortality is also an endpoint of interest. Non–life-threatening toxicities and life-threatening toxicities such as second primary cancers and thromboembolism are recorded.

Sample Size Determination

It was originally estimated that 16,000 women 35 years of age or older at increased risk for breast cancer would be randomized between placebo and tamoxifen. Several factors were considered in making the original sample size determination to ensure an adequate statistical power to test the hypothesis that tamoxifen prevents breast cancer. These included estimates of the average breast cancer risk of women who volunteered for the trial, their average rate of nonadherence to therapy, the duration of accrual, loss to follow-up, consent withdrawal, and duration of the effects of tamoxifen. The most influential factor in terms of sample size estimation is average breast cancer risk, as this provides the basis for estimating the number of breast cancer cases expected to occur in the study population.

Breast Cancer Risk

To be eligible for the BCPT, a woman must (1) be at least 60 years of age, (2) have a history of lobular carcinoma in situ, or (3) be between 35 and 59 years of age and have a five-year risk of developing breast cancer that is at least equivalent to that of the average 60-year-old woman (1.67%). A modified version of a multivariate logistic regression model developed by Gail et al[1] is used to determine the breast cancer risk of women screened for BCPT eligibility. This model incorporates information pertaining to the participant's age, race, family history of breast cancer, history of breast biopsy, and reproductive history. Each woman screened is provided with a risk profile that identifies her breast cancer risk eligibility status and displays a plot of her breast cancer risk over her lifetime compared with the lifetime risk of a woman of the same age and race who had no risk factors (Fig 1). In addition, each woman is provided with a risk/benefit analysis specific for her age, race, and breast cancer risk factor characteristics (Table 1). The risk/benefit analysis identifies the potential number of breast cancer cases and coronary artery disease cases that might be prevented by tamoxifen therapy and compares this with the cases of endometrial cancer and pulmonary embolism that might be caused by tamoxifen therapy. This information helps each woman make an informed decision about her participation in the trial.

Monitoring the Trial

To ensure propriety in the conduct of the BCPT and to comply with governmental regulations, the study has been monitored from its onset by an Endpoint Review, Safety Monitoring and Advisory Committee. This committee is composed of nine scientists with collective expertise in clinical trials methodology, oncology, gynecology, cardiology, research ethics, and biostatistics. These scientists are independent of both the operation of the study and the NSABP. The committee meets at least every six months to review endpoint and toxicity information as well as rates of adherence to therapy and delinquent data rates. The NSABP Headquarters Quality Assurance Program also monitors data submission, patient eligibility, compliance, and toxicity. In addition, a Steering Committee and numerous subject-focused committees play a role in the oversight of the study.

Status of the BCPT

Accrual rates and the characteristics of women who agreed to participate in the BCPT provide information for charting the progress of the study. Women were recruited for assessment of their breast cancer risk beginning in April 1992. By the time the study was opened for randomization on June 1, 1992, there were more than 11,000 participants whose risk had been assessed by investigators at the more than 300 participating institutions and their satellites in the United States, Canada, and Puerto Rico. Consequently, a group of women who met the criteria for entry into the trial was already identified as soon as the trial began, thus ensuring that the study would get off to a good start.

A more important indicator of the success of the study was the universal feeling of excitement among participating physicians, nurses, data managers, and institutions, as well as among NSABP and NCI personnel. An unprecedented esprit de corps existed, which was engendered by the harmony, passion, and commitment of the preceding two years of planning. Everyone believed in the importance of the trial. As predicted, accrual through the first 12 months was excellent (Fig 2). In only one year, nearly half (48%) of the 16,000 women — the number originally projected as being necessary to accomplish the study goals — were entered into the trial.

During the remaining seven months of 1993 (June through December) and during the first three months of 1994, nearly 3,300 more women were accrued to the study. Almost 11,000 of the 16,000 women required to achieve the study goals had been enrolled. Risk assessments continued to be received at a steady rate. During the latter part of 1993 and during the beginning of 1994, however, accrual plateaued at a lower level than it had in the preceding time period but still remained satisfactory (more than 350 women per month in 1993 compared to 250 to 300 per month in the first part of 1994). A major reason for the decline in accrual early in 1994 related to unwarranted, widely publicized attacks on the trial. As a result, the task of recruitment at individual institutions became more difficult. Nonetheless, the participants' resolve to complete the study remained undiminished.

Unfortunately, at the end of March 1994, after approximately 11,100 women either had been randomized or had accepted participation and were waiting to be randomized, another set of unjustified accusations led to a series of catastrophic events that profoundly affected the participants in NSABP studies, as well as the leadership, staff, and investigators of the NSABP, and even the clinical trials process itself. Accrual to all NSABP clinical trials, including the BCPT, was suspended on March 25, 1994 (Fig 2). Although accrual to the BCPT was reopened in November of 1994, active enrollment was not resumed until March 1995. In the 21-month interval between March 1995 and November 1996, accrual to the BCPT has varied from 50 to 91 women per month. As of November 30, 1996, over 86,000 women had been screened for breast cancer risk eligibility, and 12,542 had been randomized to the trial.

The distribution of participants by selected demographic characteristics (Table 2) indicates that 40% of women are under 50 years of age, 30% are between 50 and 59 years of age, and 30% are 60 years of age or older. Almost 58% have one first-degree relative (either a mother or a sister) with a history of breast cancer, and over 19% have two or more relatives with the disease. When classified according to their risk of breast cancer as compared with the expected risk for women of the same age and risk with no other risk factors, over 76% of the participants have a relative risk of 3 or more. When examined according to age, the relative risk ranges from approximately 3.9 for women who are 70 years of age or older to 9.6 for women 35 through 39 years of age (Table 3). Forty-four percent of the participants have an estimated probability of being diagnosed with breast cancer in the next five years that is greater than 3% (Table 4). At the time of entry, 37% of the participants had undergone a hysterectomy: 27% of women less than 50 years of age compared to 43% of women 50 years of age or older.

Several scenarios unfold from the accrual information and from the characteristics and breast cancer risks of the participants. Had the quarterly accrual continued at the same rate as that existing at the end of 1993, it could reasonably have been anticipated that all 16,000 women would have been enrolled by the end of either the summer or fall of 1995 (Fig 2). Due to negative publicity and suspension of randomization, the rate of accrual during the two-year period since enrollment was resumed has been substantially less than that previously observed. Fortunately, however, the estimated risk of women participating in the BCPT is at more than twice the level of risk assumed when the original sample size for the trial was developed. Based on these findings, it has been determined that the proposed target sample size can be reduced to about 13,000 participants. According to current projections, that number will be attained by the end of June 1997. On the other hand, since most of the participants had already been entered by March 1994, the increased risk of the trial population had already been established. Consequently, if the events that occurred subsequent to March 1994 had not happened, accrual could have been completed by the fall of that year — more than two and one-half years before the current projected time for completion of the trial.

Comments Regarding the Study

The events of March 1994 that had such a destructive effect on the NSABP were related to the discovery of data discrepancies recorded prior to randomization in 0.3% of patients in NSABP treatment trials. These discrepancies, which were created by a single investigator from Canada who did not participate in the BCPT, did not affect the outcome of any NSABP study. The response to the disclosure of this event affected the difficulties encountered in the conduct of the BCPT. Prior to March 1994, the issues that plagued the BCPT most often represented an effort to force the discontinuance of the trial. The publicity surrounding the Canadian problem afforded opponents of the study an unexpected opportunity to expand the platform for their opposition. They linked the mistrust against clinical trials created by the Canadian problem with issues that were specific to the BCPT, so that the attack on the study was intensified.

In the beginning, much of the opposition to the BCPT related to tamoxifen toxicity. Reports in the media and in the medical literature about the possible danger of liver damage, hepatoma, and colon cancers that could result from tamoxifen administration created anxiety in women and their physicians. Thus far, those reports have not been substantiated. In addition, the risk of endometrial cancer in tamoxifen-treated women was used by almost every opponent of the BCPT as a rationale for discontinuing the trial. The fact that tamoxifen-treated patients are at increased risk for endometrial cancer was known by all those involved with the BCPT prior to or early in its planning stage. In fact, this knowledge was used to estimate risk vs benefit in designing the BCPT and was acknowledged in the original consent form. The best information relative to the magnitude of the risk was that obtained from the NSABP B-14 study, in which node-negative breast cancer patients received either placebo or tamoxifen.[2,3] The relative risk of endometrial cancer in such patients is estimated to be about 3, a level similar to that of patients receiving estrogen replacement therapy. As new information about endometrial cancer risk became available from that trial and from other sources, it was conveyed to everyone involved with the BCPT. In addition, BCPT consent forms were continually changed to reflect the new findings, and the protocol was modified accordingly. The question of how appropriate it is to use information obtained from patients with breast cancer for predicting what might occur in tamoxifen-treated women without the disease remains unanswered.

Substantial damage to the BCPT resulted from the contention that endometrial cancer in tamoxifen-treated breast cancer patients is apt to be more aggressive and to result in death more often than is such cancer in non–tamoxifen-treated patients. Moreover, the accusation was incorrectly made that deaths from endometrial cancer in the B-14 study were not disclosed in a timely fashion. Information has been presented indicating that the outcome from endometrial cancer in tamoxifen-treated patients is no different from that in non–tamoxifen-treated patients[2]; the assertion that there was a delay in reporting deaths was incorrect and resulted from a misunderstanding of the clinical trials process.[4]

Conclusions

In keeping with scientific principles, the BCPT was initiated to test the hypothesis that tamoxifen is a breast cancer preventive agent in women at increased risk for that disease. That premise was formulated from the results of substantive laboratory and clinical investigations. Accrual to the BCPT is nearing completion, and if compliance with therapy and follow-up continue, information will be obtained that could provide support for the hypothesis. Should a reduction in the incidence rate of breast cancer among tamoxifen-treated participants be observed, there will be corroboration of the thesis. If an advantage from tamoxifen is demonstrated, it will then be appropriate to consider whether the benefit from tamoxifen is sufficient to allow for any undesirable effects of the drug. Too often, arguments regarding risk vs benefit are generated before firm information regarding one or the other, or even both, parameters is available.

The scientific process allows for and encourages valid criticism and refutation of the justification of a study and the methodology employed in its design and conduct. If they are based on valid evidence, those expressions of opposition give rise to scientific issues that are most often resolved by the results of the study being conducted. It must be recognized that, in spite of criticisms directed toward the BCPT and the temporary suspension of accrual to the study, the trial continues because of its scientific validity. Appreciation is expressed to Dr Beth Fisher for her contributions, to Tanya Spewock for editorial assistance, and to Mary Hof for preparation of the manuscript.

References

1. Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst. 1989;81:1879-1886.
2. Fisher B, Costantino JP, Redmond CK, et al. Endometrial cancer in tamoxifen-treated breast cancer patients: findings from the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-14. J Natl Cancer Inst. 1994;86:527-537.
3. Fisher B, Dignam J, Bryant J, et al. Five versus more than five years of tamoxifen therapy for breast cancer patients with negative lymph nodes and estrogen receptor-positive tumors. J Natl Cancer Inst. 1996;88:1529-1542.
4. Fisher B. Commentary on endometrial cancer deaths in tamoxifen-treated breast cancer patients. J Clin Oncol. 1996;14:1027-1039.

Appreciation is expressed to Dr Beth Fisher for her contributions, to Tanya Spewock for editorial assistance, and to Mary Hof for preparation of the manuscript.

Address reprint requests to Dr Fisher at the NSABP Scientific Director's Office, Room 914, 3550 Terrace St, Pittsburgh, PA 15261

Back to Cancer Control Journal Volume 4 Number 1