PATIENT SUBSETS WITH T1-T2, NODE-NEGATIVE BREAST CANCER AT HIGH LOCOREGIONAL RECURRENCE RISK AFTER MASTECTOMY

Transcription

1 doi: /j.ijrobp Int. J. Radiation Oncology Biol. Phys., Vol. 62, No. 1, pp , 2005 Copyright 2005 Elsevier Inc. Printed in the USA. All rights reserved /05/$ see front matter CLINICAL INVESTIGATION Breast PATIENT SUBSETS WITH T1-T2, NODE-NEGATIVE BREAST CANCER AT HIGH LOCOREGIONAL RECURRENCE RISK AFTER MASTECTOMY PAULINE T. TRUONG, M.D., C.M.,* MARY LESPERANCE, PH.D., AYDIN CULHACI, B.SC., HOSAM A. KADER, M.B., B.S.* CAROLINE H. SPEERS, B.A., AND IVO. A.OLIVOTTO, M.D.* *Radiation Therapy Program and Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver Island Centre, University of British Columbia, Victoria, BC, Canada; Department of Mathematics and Statistics, University of Victoria, Victoria, BC, Canada Purpose: To identify patient subsets with T1-T2N0 breast cancer at high risk of locoregional recurrence (LRR) who may warrant consideration for postmastectomy radiotherapy. Methods and Materials: Data were analyzed for 1505 women referred between 1989 and 1999 with pathologic T1-T2N0M0 breast cancer treated with mastectomy with clear margins and no adjuvant radiotherapy. Logistic regression analysis was performed to identify statistically significant factors associated with LRR. Recursive partitioning was used to develop a classification tree model for LRR given the prognostic variables. Results: The median follow-up was 7.0 years. The 10-year Kaplan-Meier LRR rate was 7.8%. On logistic regression analysis, the statistically significant factors predicting LRR were histologic grade (p <0.0001), lymphovascular invasion (LVI) (p <0.0001), T stage (p 0.05), and systemic therapy use (p 0.01). In the recursive partitioning model, the first split in the classification tree was histologic grade. For 972 patients without high-grade histologic features, the 10-year Kaplan-Meier LRR rate was 5.5%. For 533 patients with Grade 3 disease (LRR rate 12.1%), the concomitant presence of LVI was associated with a LRR rate of 21.2% (n 126). In patients with Grade 3 disease without LVI, T2 tumors conferred a LRR rate of 13.4% (n 194), which increased to 23.2% for patients who did not receive systemic therapy (n 63). Conclusion: Women with pt1-t2n0 breast cancer experienced a LRR risk of approximately 20% in the presence of Grade 3 disease with LVI or Grade 3 disease, T2 tumors, and no systemic therapy. These subsets of node-negative patients warrant consideration of for postmastectomy radiotherapy Elsevier Inc. Breast cancer, Mastectomy, Locoregional recurrence, Node-negative, Prognostic factors, Outcomes. INTRODUCTION Randomized controlled trials and meta-analyses have established that postmastectomy radiotherapy (PMRT) improves locoregional control in patients with invasive breast cancer (1 6). Three randomized controlled trials (3 5) and a metaanalysis (6) that demonstrated survival benefits with PMRT in women with predominantly node-positive breast cancer receiving systemic therapy sparked debate and prompted efforts to reevaluate the role of PMRT in modern breast cancer management (7 12). A primary tumor size of 5 cm, skin or deep fascia invasion, and axillary node-positive disease have been clearly documented to confer a high risk of locoregional recurrence (LRR) after mastectomy, warranting consideration of PMRT (3 5), but the role of other pathologic features in specifying additional indications for PMRT is less well defined. The most recent meta-analysis (12) from the Early Breast Reprint requests to: Pauline T. Truong, M.D., C.M., British Columbia Cancer Agency, Radiation Therapy Program, Vancouver Island Centre, 2410 Lee Ave., Victoria, BC V8R 6V5 Canada. Tel: (250) ; Fax: (250) ; ptruong@bccancer. bc.ca Cancer Trialists Collaborative Group of randomized controlled trials, comparing local therapy in 20,000 women with early breast cancer, demonstrated that although RT reduced LRR in node-negative patients after mastectomy, the baseline risk of LRR was relatively low without PMRT (9.2% without PMRT compared with 2.7% with PMRT, 2p ). The question of whether subsets exist of pt1- T2, node-negative patients with specific prognostic factors that confer greater baseline LRR risks is unclear. This study reports the results of a recursive partitioning analysis to identify subsets of patients with pt1-t2n0 breast cancer with LRR risk of 20% who may warrant consideration for PMRT. METHODS AND MATERIALS The Breast Cancer Outcomes Unit Database prospectively records tumor, treatment, and outcome data on all breast cancer patients Oral Presentation at the American Society for Therapeutic Radiology and Oncology, 46th Annual Meeting, October 3 7, 2004, Atlanta, GA. Received Jul 8, 2004, and in revised form Sep 13, Accepted for publication Sep 15,

2 176 I. J. Radiation Oncology Biology Physics Volume 62, Number 1, 2005 referred to the British Columbia Cancer Agency (BCCA). The database identified 7782 women diagnosed between January 1, 1989 and December 31, 1999 referred to the BCCA with pt1-t2n0m0 breast cancer. Mastectomy was performed in 2461 patients (31.6%) and breast-conserving surgery was performed in 5321 (68.4%). Because the objective was to evaluate the baseline LRR risk after mastectomy, patients who received PMRT (n 185) were excluded. Of 2276 unirradiated mastectomy patients, those with positive margins (n 59) and those with unknown information on margin status (n 226), histologic grade (n 188), lymphovascular invasion (LVI) status (n 113), or estrogen receptor (ER) status (n 309) were also excluded. The remaining 1505 patients with pt1-t2n0 breast cancer with complete pathologic information treated with mastectomy with clear surgical margins and no adjuvant RT formed the study cohort for this analysis. The institutional ethics review board approved the anonymous reporting of patient data. Data were extracted on each subject s date of pathologic diagnosis, dates and sites of relapse, and date of death as of December 1, Patients were followed at the BCCA and in the community. For patients who were discharged from the BCCA and followed by their primary care physicians, relapse information was obtained from the community physician on an annual basis. Death information was obtained from the Department of Vital Statistics on a monthly basis. Patients were censored 6 months before the date of data retrieval, thus ensuring that most relapse and death information to that point had been received. The tumor factors analyzed were histologic features (ductal, lobular, other), T stage (T1, T2), histologic grade (1, 2, 3); LVI status (positive, negative), ER status (positive, negative), and number of axillary nodes removed ( 5, 6 10, 11 15, 16; and 10 vs. 10). The primary outcomes examined were LRR, defined as the first site of tumor recurrence involving the ipsilateral chest wall (local) and/or axillary, supra- or infraclavicular, and internal mammary nodes (regional). Patients with LRR that occurred simultaneously with distant recurrence were scored as having an LRR event. Patients with LRR events that occurred after distant recurrence were censored at the time of distant recurrence. Secondary outcomes were distant recurrence, breast cancer-specific survival, and overall survival. Statistical analysis The clinical characteristics of those with and without LRR were compared using chi-square tests. The 10-year Kaplan-Meier local, regional, and combined LRR probability estimates and the associated standard errors (SEs) were computed for the entire cohort and for subgroups of patient, tumor, and treatment characteristics. Log rank tests of the equality of survival curves for the subgroups were performed (13). The indicator for LRR was modeled using logistic regression analysis, with the prognostic and treatment factors as independent variables. Statistical tests were two-tailed and were performed using the Statistical Package for Social Sciences, version 11.0 (SPSS, Chicago, IL). Recursive partitioning analysis was used to develop a classification tree model for LRR (14). Recursive partitioning methods have an advantage over traditional models in that complex pathways leading to high-order and/or nonlinear interactions can be readily characterized. Cross-validation was used to assist in tree model selection. With cross-validation, the data set was divided into mutually exclusive sets. For each set, a tree sequence was generated using the remaining sets. Next, the held-out data set was used to evaluate the goodness-of-fit of the tree sequence. Trees with small goodness-of-fit measures were preferred. Calculations were performed using tree and prune tree and cv.tree in S-PLUS, version 6.2. RESULTS The median follow-up time was 7.0 years. The median age at diagnosis was 62 years (range, years). Tumor and treatment characteristics The tumor and treatment characteristics of the study cohort are presented in Table 1 All patients underwent modified radical mastectomy with clear surgical margins, as well as axillary nodal staging. The median number of axillary nodes removed was 10 (range, 1 37). None of the study subjects underwent PMRT. Adjuvant systemic therapy, delivered according to the BCCA guidelines (15), was used in 50.7% of patients (14.1% chemotherapy alone, 29.9% hormonal therapy alone, and 6.7% both chemotherapy and hormonal therapy). Comparisons of characteristics of cohorts with and without LRR Table 1 presents the comparisons of the clinical characteristics of cohorts with and without LRR. Patients who experienced LRR had greater proportions of T2 stage, Grade 3 histologic features, LVI, and ER-negative disease (all p 0.05). The distributions of age at diagnosis, histologic type, number of nodes removed, and systemic therapy use were comparable between the cohorts with and without LRR. Recurrence and survival outcomes Local recurrence occurred in 61 patients (4.1%), regional recurrence in 42 patients (2.8%) and combined LRR in 89 patients (5.9%). The 10-year Kaplan-Meier local, regional, and combined LRR estimates were 5.7% (SE 0.8), 3.6% (SE 0.6), and 7.8% (SE 0.9), respectively. Of the 42 patients with regional recurrence, 22 (52%) had involved axillary nodes, 17 (40%) had involved clavicular nodes, and 3 (7%) had involved internal mammary nodes. Distant recurrence occurred in 176 patients (11.7%). Patients with LRR had statistically significantly greater risks of distant recurrence (10-year Kaplan-Meier rate 51.5%) compared with patients without LRR (10-year Kaplan- Meier rate 11.1%; p ). In the entire cohort, the 10-year Kaplan-Meier breast cancer-specific survival and overall survival rates were 88.4% (SE 1.0) and 74.7% (SE 1.4), respectively. Compared with patients without LRR, patients with LRR experienced poorer 10-year Kaplan-Meier breast cancer-specific survival (56.0% vs. 90.8%, p ). Overall survival at 10 years was also significantly lower in patients with LRR compared with the patients without LRR (53.4% vs. 76.1%, p ). Table 2 presents the 10-year Kaplan-Meier local, regional, and combined LRR estimates for the entire cohort and patient subgroups stratified by age, tumor characteris-

3 Postmastectomy recurrence in T1-T2N0 breast cancer P. T. TRUONG et al. 177 Table 1. Patient, tumor, and treatment characteristics of entire cohort and comparisons of patients with and without locoregional recurrence LRR (%) Characteristic Entire cohort (n 1505) No (n 1416) Yes (n 89) p Age (y) (117) (1388) (401) (1104) Menopausal status 0.05* Premenopausal 26.0 (391) Postmenopausal 74.0 (1114) Histologic type 0.40 Ductal 88.9 (1388) Lobular 9.6 (144) Others 1.5 (23) T stage 0.01* T (913) T (592) Grade * (230) (742) (533) (972) * (533) Lymphovascular invasion * No 81.2 (1222) Yes 18.8 (283) Estrogen receptor status 0.009* Positive 75.6 (1138) Negative 24.4 (367) Nodes removed (n) (238) (572) (456) (239) (810) (695) Systemic therapy 0.62 No 49.3 (763) Yes 50.7 (742) Abbreviation: LRR locoregional recurrence. Data presented as the percentage of patients, with the number in parentheses. * Statistically significant. tics, and systemic therapy use. The risk of LRR increased significantly in the presence of one of the following factors: Grade 3 histology, LVI, T2 Stage, and ER-negative disease (all p 0.05). As individual factors, Grade 3 histology, LVI, and T2 Stage were associated with LRR risks of 10%. No single factor was associated with a 10-year LRR risk 15%. Logistic regression analysis The logistic regression analysis (Table 3) confirmed the histologic grade (p ), LVI (p ), T stage (p 0.05), and systemic therapy use (p 0.01) to be statistically significant independent predictors of LRR. Age at diagnosis, menopausal status, histologic type, ER status, and number of nodes removed were statistically not significant. Recursive partitioning analysis Figure 1 depicts the classification tree of LRR risk based on combinations of the prognostic variables. Six terminal nodes were demonstrated. The first split was histologic grade. The 10-year Kaplan-Meier LRR rate was 5.5% (SE 1.1) for 972 patients with Grade 1 2 histologic features compared with 12.1% (SE 1.7) for 533 patients with Grade 3 disease (p ; Fig. 2). The concomitant presence of

4 178 I. J. Radiation Oncology Biology Physics Volume 62, Number 1, 2005 Table 2. Ten-year Kaplan-Meier local, regional, and combined locoregional recurrence according to patient, tumor, and treatment characteristics Characteristic Local recurrence (%) Regional recurrence (%) Locoregional recurrence (%) All patients 5.7 (0.8) 3.6 (0.6) 7.8 (0.9) Age (y) (2.2) 5.3 (2.4) 9.1 (3.0) (0.9) 3.4 (0.7) 7.7 (1.0) p (1.7) 5.4 (1.6) 9.3 (1.8) (0.9) 2.8 (0.6) 7.3 (1.1) p Menopausal status Premenopausal 5.7 (1.4) 4.9 (1.4) 8.6 (2.0) Postmenopausal 5.7 (1.0) 3.1 (0.7) 7.6 (1.1) p Histologic type Ductal 5.8 (0.9) 3.8 (0.7) 8.0 (1.0) Lobular 5.3 (2.5) 2.0 (1.5) 7.3 (2.9) Other p T stage T1 4.5 (0.9) 2.7 (0.8) 6.2 (1.1) T2 7.5 (1.6) 4.9 (1.1) 10.3 (1.7) p Grade (3.6) 5.4 (3.4) 7.9 (3.6) (0.9) 1.8 (0.6) 5.0 (1.0) (1.5) 5.8 (1.1) 12.1 (1.7) p p Lymphovascular invasion Absent 5.0 (0.9) 2.9 (0.7) 6.4 (1.0) Present 8.6 (2.1) 6.7 (1.6) 13.5 (2.4) p Data in parentheses are standard errors. Grade 3 histologic type and LVI was associated with a 10-year Kaplan-Meier LRR estimate of 21.2% (SE 4.2) compared with 9.0% (SE 1.8) for those with Grade 3 disease without LVI (p ; Fig. 3). In 407 patients with Grade 3 disease without LVI, T2 tumors conferred a 10-year Kaplan-Meier LRR risk of 13.4% (SE 3.6), which was increased to 23.2% (SE 5.7) for patients who did not receive systemic therapy (n 63). Among patients without highgrade disease, the LRR risks were 8.8% and 4.5% in subgroups aged 50 and 50 years, respectively (p 0.02). DISCUSSION The present study has identified Grade 3 histology, LVI, T2 tumor stage, and lack of systemic therapy as factors associated with a greater LRR risk among pt1-t2n0 patients. When these factors were examined individually, the 10-year Kaplan-Meier LRR risk associated with any single factor was 15%, which arguably is not sufficiently high to indicate routine PMRT. The current recursive partitioning analysis has demonstrated that the combination of Grade 3 and LVI-positive disease, present in 8% of women in this node-negative cohort, conferred a LRR risk of 21%. The magnitude of this risk was comparable to that of patients with one to three positive nodes (16, 17). The lack of consensus regarding the role of PMRT in patients with one to three positive nodes relates to the controversy regarding baseline LRR risk in these patients. In the Danish trials, a 10-year LRR rate of approximately 30% was observed in the absence of PMRT (3 5). This rate was substantially greater than the 10-year actuarial LRR rate of approximately 15% reported in other trials with long-term follow-up (16 18). Although limited axillary dissection has been implicated as a potential reason for the greater LRR rates in the Danish trials, the prognostic significance of the number of nodes removed has been unclear. In the present study s analysis, the number of negative nodes removed was not a statistically significant predictor of LRR. Table 3. Logistic regression analysis of postmastectomy locoregional recurrence Variable Local recurrence (p, overall test) Regional recurrence (p, overall test) Locoregional recurrence (p, overall test) Age ( 50 vs. 50 y) Menopausal status (pre- vs postmenopausal) Histologic type (ductal vs lobular) Tumor stage (T2 vs. T1) * (1.61, ) Grade (3 vs. 1 2) 0.009* (2.17; ) 0.003* (3.03; ) * (2.61; ) Lymphovascular invasion 0.03* (1.93; ) 0.001* (3.07; ) * (2.51; ) (present vs. absent) Estrogen receptor status (negative vs. positive) Nodes removed ( 10 vs. 10) Systemic therapy (no vs. yes) (1.87; ) Data in parentheses are hazard ratio followed by 95% confidence interval.

5 Postmastectomy recurrence in T1-T2N0 breast cancer P. T. TRUONG et al. 179 Fig. 1. Recursive partitioning analysis of postmastectomy locoregional recurrence (LRR) in women with pt1-t2n0 breast cancer. Because a randomized controlled trial designed to address the role of PMRT in patients with one to three positive nodes was recently closed owing to lack of accrual (19), the controversy over the absolute LRR risk reductions with PMRT are unlikely to be resolved definitively by prospective data. Recognizing that clinicians vary in their manage- Fig. 2. Kaplan-Meier locoregional recurrence (LRR) rate in patients with pt1-t2n0 breast cancer stratified by histologic grade.

6 180 I. J. Radiation Oncology Biology Physics Volume 62, Number 1, 2005 Fig. 3. Kaplan-Meier locoregional recurrence (LRR) in patients with Grade 3 histologic findings stratified by lymphovascular invasion (LVI) status. ment approaches for women with one to three positive nodes, the similar baseline LRR risks identified in this analysis suggest that high-risk node-negative patients warrant treatment considerations similar to those with one to three positive nodes. Such management should include radiation oncology consultation for evaluation and discussion of RT options and the potential benefits and risks. In patients with Grade 3 disease without LVI, the finding that T2 tumors conferred LRR risk estimates of 13%, which was increased to 23% in the absence of systemic therapy, suggest that these are additional situations in which PMRT may be considered. With respect to low-risk subsets, in the absence of highgrade histologic features and LVI, combinations of other patient and tumor factors were associated with a LRR risk of 10%. These patients, which comprised two-thirds of the pt1-t2n0 cohort, may be spared PMRT when balancing the small absolute locoregional control benefits against the risk of normal tissue toxicities. Reservations about the adverse effects of PMRT have stemmed from older meta-analyses of PMRT trials in which reductions in breast cancer-related deaths were offset by significantly increased cardiovascular-related deaths (1, 2). PMRT reduced LRR in node-negative patients, but the baseline risk was relatively low even in the absence of PMRT (LRR 9.2% without PMRT vs 2.7% with PMRT, 2p ) (2). In the 25-year update of the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-04 study, local recurrence occurred in 5% of patients treated with radical mastectomy, 7% of patients treated with total mastectomy, and 1% of patients treated with total mastectomy plus PMRT. The corresponding regional recurrence rates were 4%, 6%, and 4%. Although PMRT significantly reduced LRR (p 0.002), the absolute improvements were small, and no survival benefit was demonstrated (20). These early trials provided important information regarding recurrence risk in early-stage breast cancer patients but lacked subgroup analyses of prognostic factors other than primary tumor stage and the number of axillary nodes involved. Contemporary data on the postmastectomy relapse risks associated with histopathologic and treatment characteristics are sparse. The current study s demonstration of LVI presence and T2 stage as factors associated with greater postmastectomy LRR risk has been corroborated by an analysis conducted by the International Breast Cancer Study Group of 1275 women with node-negative breast cancer treated with modified radical mastectomy without PMRT enrolled in a systemic therapy trial (21). In that trial, one-third of patients received no adjuvant systemic therapy and two-thirds received a single cycle of perioperative cyclophosphamide, methotrexate, and fluorouracil chemotherapy (21). Among 692 premenopausal patients, the concomitant presence of vascular invasion and tumor size 2 cm(n 122) was associated with a 10-year LRR risk of 15%. Among 583 postmenopausal node-negative patients, vas-

7 Postmastectomy recurrence in T1-T2N0 breast cancer P. T. TRUONG et al. 181 cular invasion (n 225) was the only factor that significantly predicted LRR, with a 10-year LRR risk of 14% (21). The survival implication of PMRT in patients with pt1-t2 tumors and negative nodes is unclear. A retrospective study from Germany reported reduced LRR but no difference in overall survival among 114 pt1-t2, node-negative patients treated with PMRT before 1979 compared with 804 patients who did not undergo PMRT (22). A comparison of that study with the current analysis and its generalizability to contemporary practice is difficult because none of the subjects received adjuvant systemic therapy, a modality that has been demonstrated to affect local control (23, 24) and is currently used extensively, including among node-negative patients (25, 26). In a study from Denmark, 388 patients who underwent PMRT at a single institution were compared with 32,000 patients in the Surveillance, Epidemiology, and End Results (SEER) database treated without PMRT. The 10-year actuarial overall survival rates in the small cohort treated with mastectomy and RT were 79.9% for pt1 and 70.4% for pt2 disease. The overall survival rates in the large SEER cohort treated without PMRT were 73.8% for pt1 and 63.5% for pt2 disease. The authors suggested that the greater survival rates observed in the 388 patients treated with PMRT compared with the SEER patients treated without PMRT argued for a survival benefit from adjuvant RT (27). This argument, however, was not substantiated by disease recurrence data nor a demonstration that the survival differences between the two cohorts were statistically significant. The available data thus are insufficient to establish whether PMRT confers a survival benefit in women with node- negative breast cancer. Despite consensus that adjuvant RT reduces LRR risk (1, 2), the question of what baseline risk and associated absolute risk reduction constitute the threshold for routine RT is controversial. Information on patients expectations and perceptions of benefits and risks pertaining to adjuvant locoregional therapy is lacking. In a survey of 318 women treated with adjuvant chemotherapy, when asked what degree of absolute benefit patients would find acceptable to choose to undergo systemic therapy, the median acceptable reduction in recurrence risk among the surveyed population was only % (28). Acknowledging that treatment selection and decision-making are highly complex and individualized processes, this survey raised the suggestion that patients with breast cancer may often accept very small degrees of net benefit in their adjuvant therapy choices. Thus, the question of whether patients with an estimated baseline LRR risk of approximately 20% would be motivated to undergo PMRT to improve locoregional control warrants discussion, with careful consideration of the individual s goals and preferences. CONCLUSION This population-based analysis identified subsets of women with pt1-t2n0 breast cancer who had a postmastectomy LRR risk of approximately 20%. The presence of Grade 3 disease with LVI or Grade 3 disease, T2 tumors, and no systemic therapy identified patient subgroups who may achieve sufficient reductions in LRR that PMRT could be considered. Prospective research is required to confirm the role of PMRT on locoregional control and survival in these node-negative patients. REFERENCES 1. Early Breast Cancer Trialists Collaborative Group. Effects of radiotherapy and surgery in early breast cancer: An overview of the randomized trials. N Engl J Med 1995;333: Early Breast Cancer Trialists Collaborative Group. Favourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer: An overview of the randomised trials. Lancet 2000;355: Overgaard M, Hansen PS, Overgaard J, etal.postoperative radiotherapy in high- risk premenopausal women with breast cancer who receive adjuvant chemotherapy. Danish Breast Cancer Cooperative Group 82b trial. N Engl J Med 1997;337: Ragaz J, Jackson SM, Le N, etal.adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 1997;337: Overgaard M, Jensen MB, Overgaard J, etal.postoperative radiotherapy in high- risk postmenopausal breast cancer patients given adjuvant tamoxifen: Danish Breast Cancer Cooperative Group (DBCG) 82c randomised trial. Lancet 1999; 353: Whelan TJ, Julian J, Wright J, etal.does locoregional radiotherapy improve survival in breast cancer? A meta-analysis. J Clin Oncol 2000;18: Hellman S. Stopping metastases at their source. N Engl J Med 1997;337: Recht A, Bartelink H, Fourquet A, et al. Postmastectomy radiotherapy: Questions for the twenty-first century. J Clin Oncol 1998;16: Harris JR, Halpin-Murphy P, McNeese M, et al. Consensus statement on postmastectomy radiation therapy. Int J Radiat Oncol Biol Phys 1999;44: Taylor ME, Haffty BG, Shank BM, et al., for the American College of Radiology. Postmastectomy radiotherapy: ACR Appropriateness Criteria. Radiology 2000;215(Suppl.): Recht A, Edge SB, Solin LJ, etal.postmastectomy radiotherapy: Guidelines of the American Society of Clinical Oncology. J Clin Oncol 2001;19: Pierce LJ. Treatment guidelines and techniques in delivery of postmastectomy radiotherapy in management of operable breast cancer. J Natl Cancer Inst Monogr 2001;30: Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. 2nd ed. New York: John Wiley; Zhang H, Singer B. Recursive partitioning in the health sciences. New York: Springer-Verlag, British Columbia Cancer Agency Breast Tumour Group. Cancer treatment policy manual. 5th ed. Vancouver, BC; British Columbia Cancer Agency; Recht A, Gray R, Davidson NE, etal.locoregional failure 10 years after mastectomy and adjuvant chemotherapy with or

8 182 I. J. Radiation Oncology Biology Physics Volume 62, Number 1, 2005 without tamoxifen: Experience of the Eastern Cooperative Oncology Group. J Clin Oncol 1999;17: Katz A, Strom E, Bucholz TA, etal.locoregional recurrence patterns after mastectomy and doxorubicin-based chemotherapy: Implications for postoperative irradiation. J Clin Oncol 2000;18: Katz A, Strom EA, Bucholz TA, et al. The influence of pathologic tumor characteristics on locoregional recurrence rates following mastectomy. Int J Radiat Oncol Biol Phys 2001;50: Pierce L. Postmastectomy radiotherapy: Future directions. Semin Radiat Oncol 1999;9: Fisher B, Jeong JH, Anderson S, et al. Twenty-five-year follow-up of a randomized trial comparing radical mastectomy, total mastectomy and total mastectomy followed by irradiation. N Engl J Med 2002;347: Wallgren A, Bonetti M, Gelber RD, etal.risk factors for locoregional recurrence among breast cancer patients: Results from International Breast Cancer Study Group trials I through VII. J Clin Oncol 2003;21: Janni W, Dimpfl T, Braun S, etal.radiotherapy of the chest wall following mastectomy for early stage breast cancer: Impact on local recurrence and overall survival. Int J Radiat Oncol Biol Phys 2000;48: Buchholz TA, Tucker SL, Erwin J, etal.impact of systemic treatment on local control for patients with lymph nodenegative breast cancer treated with breast- conservation therapy. J Clin Oncol 2001;19: Levine MN, Bramwell V, Abu-Zahra H, etal.the effect of systemic adjuvant chemotherapy on local breast cancer recurrence in node positive breast cancer patients treated by lumpectomy without radiation. Br J Cancer 1992;65: Early Breast Cancer Trialists Collaborative Group: Systemic treatment of early breast cancer by hormonal, cytotoxic or immune therapy: 133 randomised trials involving recurrences and deaths among women. Lancet 1992;339:1 15 and Early Breast Cancer Trialists Collaborative Group: Tamoxifen for early breast cancer: An overview of the randomised trials. Lancet 1998;351: Voordeckers M, Van de Steene J, Vinh-Hung V, etal.adjuvant radiotherapy after mastectomy for pt2-pt2 node negative (pn0) breast cancer: Is it worth the effort? Radiother Oncol 2003;68: Ravdin PM, Siminoff IA, Harvey JA. Survey of breast cancer patients concerning their knowledge and expectations of adjuvant therapy. J Clin Oncol 1998;16: