INTRODUCTION

Ductal carcinoma in situ (DCIS) accounts for approximately 20% of all screening-detected breast cancers.¹ Nearly 53,000 cases of DCIS are expected to be diagnosed in the United States in 2007.² Treatment options for DCIS include mastectomy and breast-conserving surgery (BCS) with or without radiation therapy. Compared with mastectomy, BCS with radiation therapy is associated with a higher rate of local recurrences, approximately half of which are invasive.³

The National Surgical Adjuvant Breast and Bowel Project (NSABP) B-24 study was designed to investigate the value of adjuvant tamoxifen therapy after BCS and radiation therapy in patients with DCIS. In this double-blind, placebo-controlled study, 1,804 women with DCIS treated with lumpectomy and 50 Gy of whole-breast radiation therapy were randomly assigned to receive tamoxifen 20 mg daily or placebo daily for 5 years.⁴ The initial results of the B-24 trial were first presented publicly in December 1998, and the full manuscript was published in June 1999. At a median follow-up time of 74 months, tamoxifen reduced the absolute 5-year rates of ipsilateral and contralateral breast cancer events by 3.3% (P = .04) and 1.4% (P = .01), respectively.⁴ An update reported in August 2001, when median follow-up had reached 83 months, confirmed these benefits, showing that tamoxifen significantly decreased the 7-year cumulative incidence of all breast cancer events (relative risk, 0.63; P = .0003), and of ipsilateral invasive breast cancer in particular (relative risk, 0.53; P = .01).⁵ However, there was no effect on survival. On the strength of these results, in July 2000, the US Food and Drug Administration approved the use of tamoxifen for the adjuvant treatment of DCIS after surgery and radiotherapy.

After the publication of the NSABP B-24 trial results, the breast cancer guidelines panel of the National Comprehensive Cancer Network (NCCN) amended the DCIS guideline in 2000 to suggest consideration of adjuvant tamoxifen in patients treated with BCS.⁶ Little is known about the impact of these emerging clinical trial data and the resulting clinical practice guideline modifications on patterns of care for DCIS. Therefore, we examined the temporal trends and determinants of tamoxifen use in the NCCN breast cancer outcomes database after surgery for DCIS.

PATIENTS AND METHODS

RESULTS

Median age in the study cohort was 53.4 years (range, 23.2 to 89.1 years). Eighty-five percent of the patients (n = 1,377) were white, 23% (n = 378) had undergone a hysterectomy, 48% (n = 773) were premenopausal, and 64% (n = 1,030) had been treated with BCS (Table 1).

Overall, 41% of the patients (665 of 1,622) received tamoxifen after surgery for DCIS. The proportion of patients receiving tamoxifen increased over time, from 24% (87 of 367) before July 1, 1999, to 46% (578 of 1,255) on or after July 1, 1999 (Fig 1). The increase in tamoxifen use appears to have begun around the time of the initial presentation of the NSABP B-24 results in December 1998. After the June 1999 publication of the B-24 results, the proportion of patients receiving tamoxifen reached a plateau at approximately 50%. Of patients diagnosed in 2003, 55% received tamoxifen.

The proportion of DCIS patients receiving tamoxifen according to various demographic, clinical, surgical, and pathologic factors is summarized in Table 1. In univariate analysis, patients who were younger (< 70 years), had a comorbidity score of 1 or less, had no history of cerebrovascular or peripheral vascular disease or other comorbid conditions, or had a history of a hysterectomy were significantly more likely to receive tamoxifen. There was a significant association between smaller pathologic DCIS size and tamoxifen therapy. This association remained significant when patients with missing values were excluded from the analysis. Twenty-two percent of patients (357 of 1,622) did not have information on tumor size and margin status, given that this information was not collected for patients presenting before July 1, 1999. Other factors associated with receipt of tamoxifen in univariate analysis included BCS, receipt of radiation therapy after surgery, NCCN center, and diagnosis on or after July 1, 1999. Menopausal status was not found to be associated with receipt of tamoxifen.

In multivariable logistic regression analysis (Table 2), several factors were found to be associated independently with the receipt of tamoxifen. Patients who were diagnosed on or after July 1, 1999, had an almost four-fold increased odds of receiving tamoxifen. Patients who did not have a history of cerebrovascular or peripheral vascular disease had an approximately three-fold increased odds of receiving tamoxifen. Similarly, patients who underwent radiation therapy or who had undergone hysterectomy had a 1.82- and 1.34-fold increased odds of receiving tamoxifen, respectively. In addition, significant interactions were noted between type of breast surgery and age at diagnosis and between type of breast surgery and NCCN center. As shown in Table 2 and Figure 2, BCS patients younger than age 70 years were significantly more likely to receive tamoxifen than BCS patients age 70 years and older (P = .0073). In contrast, there was no significant difference in tamoxifen use across age groups for mastectomy patients. Furthermore, as shown in Table 2 and Figure 3, regardless of breast surgery type, there was a wide variation in the use of tamoxifen across NCCN centers (P = .002). Overall, the rate of tamoxifen use was higher after BCS than after total mastectomy, with absolute rates ranging from 34% to 74% after BCS and from 17% to 53% after mastectomy.

DISCUSSION

The results of this study indicate that between July 1997 and December 2003, 41% of patients who underwent surgery for DCIS at NCCN centers received tamoxifen after surgery. There was a rapid increase in the proportion of DCIS patients treated with tamoxifen, beginning around the time of the first public presentation of the NSABP B-24 trial results in December 1998. Since the June 1999 publication of those results, the proportion of patients receiving tamoxifen has reached a plateau at approximately 50%.

A number of factors were associated with adjuvant tamoxifen therapy for DCIS in the period after the dissemination of the NSABP B-24 results. Patients with a history of cerebrovascular or peripheral vascular disease were less likely to receive tamoxifen, as were women with an intact uterus. This suggests that physicians are weighing the likelihood of tamoxifen-associated vascular and endometrial toxicity in formulating their adjuvant therapy recommendations for DCIS. Interestingly, patients who received adjuvant radiation therapy were also more likely to receive tamoxifen. The NSABP trial was restricted to women who had undergone lumpectomy and radiotherapy; there is no high-quality evidence that tamoxifen reduces recurrence risk in DCIS patients who have not received postlumpectomy radiotherapy. Therefore, physicians may be reluctant to extrapolate the results of the NSABP trial to patients who have been treated with BCS alone. It is also possible that the likelihood of receiving tamoxifen simply increases with the number of breast cancer physicians involved in a patient's care. Finally, it may be that patients who were believed to be at higher risk of recurrence for reasons not captured in our model were treated more aggressively, such that they received both radiation therapy and tamoxifen.

We also found that younger age was associated with tamoxifen use, but only among women treated with BCS. This finding is consistent with three prior single-institution studies of patterns of care for DCIS, which all demonstrated that BCS was an independent predictor of tamoxifen use.^9-11 Thus it appears that physicians are using tamoxifen in this setting with true adjuvant intent, and not simply as chemoprevention for the contralateral breast. The fact that physicians do so selectively in women younger than age 70 years may well be explained by the fact that the risks of thromboembolic events and endometrial cancer from tamoxifen use are greater in postmenopausal women,¹² making the risk-benefit ratio for tamoxifen use in younger women more appealing.

Finally, regardless of breast surgery type, there was substantial variation in the use of tamoxifen across NCCN centers, with absolute rates ranging from 34% to 74% after BCS and from 17% to 53% after mastectomy. This suggests that in the absence of any proven survival benefit, clinical opinion leaders in these institutions differ in how they weigh the modest reduction in breast cancer events conferred by tamoxifen against its potential adverse effects.^5,12,13 To investigate further the factors that might be associated with this striking institutional variation, we examined the association between active participation in NSABP B-24, defined by enrollment of at least 10 patients, and use of tamoxifen after the trial results were released. In the four sites that had participated actively in the trial, tamoxifen was administered to 43% of study patients versus 38% in nonparticipating centers (P = .0214). This suggests the possibility that physicians in centers that participated in the trial became comfortable prescribing tamoxifen for DCIS, and as a result, were likely to use it off protocol after the B-24 results were released. Alternatively, active B-24 trial participation may simply have been a marker for a pre-existing proclivity to use tamoxifen in this setting.

A strength of our study is that all of the data were entered prospectively into the database and were subject to rigorous quality assurance to ensure accuracy. However, our study also has several limitations. The NCCN database does not include information on whether patients were offered tamoxifen but declined or patients' reasons for declining. As a result, we cannot comment on the extent to which observed treatment rates reflect physician recommendations as opposed to patient preferences. In addition, we did not have information on hormone receptor status for most of the study patients. Therefore, we could not examine the impact on patterns of care of the 2002 reanalysis of NSABP B-24, suggesting that the benefits of tamoxifen are greatest in patients who have estrogen receptor (ER) –positive DCIS.¹⁴ Finally, because complete data are not available for treatment initiated after 2004, we cannot fully assess the impact on patterns of care of the United Kingdom, Australia, and New Zealand trial, published in July 2003, which concluded that the addition of tamoxifen to patients with DCIS treated with BCS with or without radiotherapy did not reduce total and ipsilateral invasive breast cancer events.¹⁵

The majority of breast cancer patients are not treated at NCCN centers. The patients who choose to receive their care at NCCN institutions, as well as the physicians who practice in these centers, may differ in systematic ways. Ideally, population-based studies should be done to determine whether our findings are generalizable. Unfortunately, currently available national data sources such as Surveillance, Epidemiology, and End Results data linked to Medicare claims have serious limitations for this kind of analysis, including incomplete capture of hormonal therapy use, insufficient clinical detail to permit a thorough analysis of factors driving physicians' treatment decisions, and inability to examine patterns of care across the full range of patient age.

The findings of our study clearly indicate the need for future prospective studies that address alternative adjuvant therapies and breast cancer chemoprevention in patients with DCIS. Among postmenopausal women with early invasive breast cancer, aromatase inhibitors are associated with better disease-free survival and greater reduction in contralateral breast cancer risk than tamoxifen.^16,17 Whether they offer a similar advantage among postmenopausal women with DCIS is currently being investigated in two large randomized trials. The NSABP has closed protocol B-35, in which 3,104 postmenopausal women with hormone receptor–positive DCIS treated with BCS and radiation therapy were randomly assigned to receive either tamoxifen 20 mg daily or anastrozole 1 mg daily for 5 years; results are not yet available.¹⁸ The second International Breast Cancer Intervention Study is an ongoing study with a similar design.¹⁹ These two studies will help determine whether aromatase inhibitors can be considered, in addition to tamoxifen, in the management of postmenopausal patients with DCIS. Finally, given the recent publication of the NSABP P-2 trial results,²⁰ the second-generation selective ER modulator raloxifene may be considered an alternative for lowering breast cancer risk in postmenopausal women at high risk for breast cancer, including patients with DCIS.

In conclusion, between July 1997 and December 2003, we found that 41% of patients who underwent surgery for DCIS received adjuvant tamoxifen at NCCN centers. The proportion of patients who received tamoxifen increased rapidly around the time of the initial presentation of the NSABP B-24 results in December 1998, and then reached a plateau at approximately 50%. Several patient- and treatment-related factors seem to be associated with tamoxifen use. However, given the significant variation in tamoxifen use by NCCN institution, it appears that physician beliefs and preferences regarding the reduction in breast cancer risk and potential adverse effects of tamoxifen in this patient population play a substantial role in the decision-making process regarding tamoxifen as adjuvant therapy for women with DCIS. These findings highlight the need for additional studies in DCIS patients to define clearly the role of adjuvant tamoxifen in relation to hormone receptor status, and to define subgroups of patients for whom the benefits of tamoxifen may truly outweigh its risks (perhaps younger women with ER-positive DCIS who undergo BCS). Furthermore, our findings confirm the importance of the NSABP B-35 and second International Breast Cancer Intervention Study trials, as well as other future studies evaluating alternative adjuvant and chemopreventive strategies for patients with DCIS.

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Although all authors completed the disclosure declaration, the following authors or their immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.

Employment: N/A Leadership: N/A Consultant: N/A Stock: N/A Honoraria: Tina W.F. Yen, Society of Surgical Oncology and Health Science Center for Continuing Medical Education Research Funds: N/A Testimony: N/A Other: N/A

AUTHOR CONTRIBUTIONS

Conception and design: Tina W.F. Yen, Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Stephen B. Edge, Richard L. Theriault

Financial support: Henry M. Kuerer

Administrative support: Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Joyce C. Niland, Richard L. Theriault, Jane C. Weeks

Provision of study materials or patients: Henry M. Kuerer, Joyce C. Niland, Stephen B. Edge, Richard L. Theriault, Jane C. Weeks

Collection and assembly of data: Tina W.F. Yen, Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Joyce C. Niland, Stephen B. Edge, Richard L. Theriault, Jane C. Weeks

Data analysis and interpretation: Tina W.F. Yen, Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Joyce C. Niland, Stephen B. Edge, Jane C. Weeks

Manuscript writing: Tina W.F. Yen, Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Joyce C. Niland, Stephen B. Edge, Richard L. Theriault, Jane C. Weeks

Final approval of manuscript: Tina W.F. Yen, Henry M. Kuerer, Rebecca A. Ottesen, Layla Rouse, Joyce C. Niland, Stephen B. Edge, Richard L. Theriault, Jane C. Weeks