Comparison of survival and regional failure between accelerated partial breast irradiation and whole breast irradiation

Brachytherapy 11 (2012) 311e315

Comparison of survival and regional failure between accelerated partial breast irradiation and whole breast irradiation Chirag Shah1, J. Ben Wilkinson1, Maureen Lyden2, Alfred Mbah2, Martin Keisch3, Peter Beitsch4, Frank A. Vicini1,* 1

Department of Radiation Oncology, Beaumont Cancer Institute, Oakland University William Beaumont School of Medicine, Royal Oak, MI 2 Biostat Inc., Tampa, FL 3 Cancer HealthCare Associates, University of Miami Hospital, Miami, FL 4 Dallas Surgical Group, Dallas, TX

ABSTRACT

PURPOSE: To compare rates of regional recurrence (RR) and overall survival (OS) between a pooled set of 1400 patients treated on the American Society of Breast Surgeons MammoSite (Hologic, Inc., Bedford, MA) Registry Trial to a cohort of 3600 patients treated with whole breast irradiation (WBI). METHODS AND MATERIALS: A total of 1440 women underwent accelerated partial breast irradiation (APBI) between 2002 and 2004 as part of the American Society of Breast Surgeons Registry Trial and a total of 3593 patients who received WBI were evaluated from the Surveillance Epidemiology and End Results database with treatment received between 1980 and 2009. A matched-pair analysis was performed based on age, receipt of hormonal therapy, chemotherapy, nodal status, and tumor size (1051 patients per arm). Rates of RR and OS were then analyzed for each group. RESULTS: After the match, no differences in patient characteristics were noted when tumor size was evaluated as a continuous variable. Rates of RR and OS were similar between the WBI and APBI groups. A Cox regression model found no difference between WBI and APBI with regard to RR; however, OS was improved in the APBI cohort (hazard ratio 0.008, p!0.0001). CONCLUSIONS: With one of the largest patient populations to date comparing WBI and APBI, no difference in RR or OS was noted between WBI and APBI treatment. Until the publication of prospective Phase III trials, these data support the continued use of APBI on protocol and off protocol in appropriately selected patients. Ó 2012 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.

Keywords:

Partial breast irradiation; Breast cancer; Breast conservation therapy; APBI; Whole breast irradiation

Introduction With the publication of multiple prospective Phase III trials, breast conserving therapy with whole breast irradiation (WBI) has been found to provide equivalent outcome to

Received 6 December 2011; received in revised form 29 December 2011; accepted 9 January 2012. Disclosures: The MammoSite Breast Brachytherapy Registry Trial was supported in part by a grant from Hologic, Inc. to the American Society for Breast Surgeons and BioStat International, Inc. There are no conflicts of interest to declare except Martin Keisch, MD: Consultant/Advisory Board, Hologic, Inc., Bedford, MA. * Corresponding author. Michigan Healthcare Professionals/21st Century Oncology, 28595 Orchard Lake Road, Farmington Hills, MI 48331. Tel.: þ1-248-994-0632; fax: þ1-248-553-7674. E-mail address: [email protected] (F.A. Vicini).

mastectomy alone in terms of both local tumor control and survival. However, over the past two decades, novel strategies for delivering adjuvant radiation therapy after breast conserving surgery have emerged because of the protracted 6e7-week course of WBI required and the fact that up to 20% of patients do not receive any form of adjuvant radiotherapy (1, 2). One of the techniques that reduces the length of the radiotherapy course while delivering a biologically equivalent dose of radiation is accelerated partial breast irradiation (APBI). This technique limits the radiation target to the volume of tissue immediately surrounding the surgical cavity with a variable margin and therefore decreases treatment time to 1 week or less. As the techniques for delivering APBI have been modified and improved, APBI has been shown not only to be feasible but associated with excellent clinical outcomes with

1538-4721/$ - see front matter Ó 2012 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.brachy.2012.01.008

312

C. Shah et al. / Brachytherapy 11 (2012) 311e315

5e12 years of followup (3e6). However, data are limited comparing APBI to WBI. Currently, multiple Phase III trials are underway comparing adjuvant WBI with APBI, with one of the largest being the National Surgical Adjuvant Breast and Bowel Project B-39 (7). Previously, data from William Beaumont Hospital compared outcomes between 200 patients undergoing interstitial APBI and standard WBI and showed no difference in clinical outcomes at 12 years, albeit with limited numbers and therefore, a reduced power to detect differences (4). Prospective randomized data from Polgar et al. (3) compared WBI and interstitial APBI or partial breast with electron fields. At 5 years, no difference in local recurrence, disease free survival, or overall survival (OS) was noted with improved cosmesis in the high-dose-rate partial breast cohort. The partial breast arm, however, was mixed as the interstitial cohort received APBI, whereas the electron cohort received partial breast treatment over a standard length of treatment. More recently, intraoperative radiation has been used to deliver APBI with the results of a prospective randomized trial comparing this technique to WBI finding no difference in outcomes at 4 years, albeit with only 18.8% of patients having followup of at least 4 years (8). Therefore, the purpose of this analysis was to perform a comparison of APBI and WBI using the data from the American Society of Breast Surgeons (ASBS) MammoSite Registry Trial and the Surveillance Epidemiology and End Results (SEER) database and to control for differences in patient characteristics via a matched-pair analysis.

Methods and materials ASBS patient population The ASBS MammoSite Registry Trial consisted of 97 institutions treating 1449 patients with the original MammoSite Radiation Therapy System (Hologic, Inc., Bedford, MA) between May 4, 2002 and July 30, 2004. All centers that were provided training in the use of the MammoSite device were offered participation in the registry trial. Information on enrollment criteria, data collection, and data management has previously been published (9). Of note, patients could be enrolled in the trial at any time during their treatment (before, during, or after), but pretreatment enrollment was encouraged. Since the inception of the trial, two full-service, independent contract research organizations, Synergos, Inc. (The Woodlands, TX) and Biostat International, Inc. (Tampa, FL) have provided data management services and statistical analyses for the ASBS Registry Trial. Followup was complete through July, 2011. SEER patient population The Metropolitan Detroit SEER Registry was queried for patients with invasive breast cancer between 1980 and 2008.

Data queried included tumor size, stage, age, nodal status, radiation treatment, hormonal therapy, chemotherapy, recurrence information, and survival. The initial query identified 25,863 patients. Only patients with primary breast tumors and initial primaries were included (n 5 22,801). Only patients who had undergone postoperative radiation treatment (n 5 21,521) were included with the exclusion of patients with preoperative, intraoperative, or unknown sequencing of radiation. Only patients receiving breast conserving surgery were included (n 5 16,130). Patients were required to have known chemotherapy and hormonal treatment information (n 5 16,128). The only histologies included in this analysis were ductal, lobular, and medullary carcinomas (n 5 16,053). SEER database patients were required to have initial staging and followup data available, which yielded a final SEER database cohort size of 3593 patients available for the matching process. Local recurrence data is not consistently maintained within the SEER database and was therefore not evaluated in this study. Statistical analyses Patient characteristics for each cohort were analyzed including age, tumor size/stage, estrogen receptor status, margin status, lymph node status, and receipt of hormonal therapy. Clinical outcomes that were analyzed included regional recurrence (RR) and OS. RR was defined as failure within the regional lymphatics (axillary, supraclavicular, or internal mammary nodes). OS was defined as an absence of death secondary to any cause. In matching invasive patients who received APBI to those who received WBI with respect to their baseline variables (e.g., age, chemotherapy, hormonal therapy, and nodal status), we used the propensity scoring algorithm (10). Therefore, for each patient treated with APBI or WBI, we calculated the propensity score using the logistic regression model with the baseline characteristics (age, chemotherapy, hormonal therapy, and nodal status) as explanatory variables. For each computed propensity score for patients with APBI (n 5 1440), we selected a match from the WBI group (n 5 3593) based on the closest absolute propensity score, that is, the ‘‘nearest neighbor’’ (11). We then conducted the selection process without replacement so that there was a 1:1 match (11, 12). After the propensity score matching, the distribution of the covariates between patients with APBI and patients with WBI were expected to be the same, leading to nonsignificant differences in the covariates across the two groups of patients (13). The matched-pair process yielded a final cohort size of 1051 women in each treatment group (WBI and APBI) with a total patient population of 2102 women in this study. The Chi-square test was used to assess differences in baseline characteristics (age, chemotherapy, hormonal therapy, and nodal status). We compared the outcome variables (RR and OS) between APBI and WBI patients using KaplaneMeier productelimit estimator, which calculated

C. Shah et al. / Brachytherapy 11 (2012) 311e315 Table 1 Patient characteristics prior to matched-pair analysis

at 5%. SAS version 9.1 (SAS Institute, Cary, NC) was used to perform all analyses.

Number (percentage of cohort)

WBI (n 5 3593)

APBI (n 5 1255)

p-Value

Nodal status N1 N0 Nx

853 (23.7%) 2724 (75.8%) 16 (0.5%)

38 (3.0%) 1152 (91.8%) 65 (5.2%)

!0.01

Chemotherapy No Yes

1893 (52.7%) 1700 (47.3%)

1089 (86.8%) 166 (13.2%)

!0.01

Hormonal therapy Yes No

1705 (47.5%) 1888 (52.5%)

785 (62.6%) 470 (37.4%)

!0.01

ER status Negative Positive Unknown

774 (21.5%) 2707 (75.3%) 112 (3.1%)

127 (10.1%) 873 (69.6%) 255 (20.3%)

!0.01

Age (yr) Tumor size (cm, continuous)

59.5 1.67

313

!0.0001 !0.0001

66.0 1.13

APBI 5 accelerated partial breast irradiation; WBI 5 whole breast irradiation; ER 5 estrogen receptor.

the cumulative probability of the outcomes for each group of patients. To determine whether the difference in each outcome between groups was significant, we applied the Log-rank statistic. Cox Proportional Hazards Regression models were used to generate risk estimates. All tests of hypothesis were two tailed with a Type 1 error rate fixed

Results Prematch patient characteristics are presented in Table 1. Before the matching process, significant differences were noted in all patient characteristics between the WBI and APBI patients. Patients in the WBI cohort were more likely to be node positive, to receive chemotherapy, to be estrogen receptor negative, and to have larger tumors. APBI patients were younger and more likely to have received hormonal therapy. When matched by the factors previously noted, no statistical differences in patient characteristics between the WBI and APBI cohorts were noted when tumor size was evaluated as a continuous variable (Table 2). When tumor size was evaluated with a 2 cm cutoff, no differences were seen with the exception of age, with WBI patients being significantly older. Clinical outcomes after matching are presented in Table 3 with tumor size evaluated continuously and using a cutoff of 2 cm. No differences in RR (0.1% vs. 0.7%, p 5 0.12) or OS (94.6% vs. 91.3%, p 5 0.23) were noted between WBI (n 5 1051) and APBI (n 5 1051) when evaluating tumor size continuously. Further, no differences in RR (0.2% vs. 0.7%, p 5 0.09) or OS (94.7% vs. 93.2%, p 5 0.14) were noted between WBI (n 5 1052) and APBI (n 5 1052) when evaluating tumor size using a 2 cm cutoff. Cox regression analysis was performed after matching the WBI and APBI cohorts. No difference in RR

Table 2 Patient characteristics after matched-pair analysis Tumor size ! vs. $2 cm

Tumor size continuous Number (percentage of cohort)

WBI (n 5 1051)

APBI (n 5 1051)

p-Value

WBI (n 5 1052)

APBI (n 5 1052)

p-Value

Nodal status N1 N0 Nx

31 (3.0%) 1009 (96.0%) 11 (1.0%)

38 (3.6%) 1001 (95.2%) 12 (1.1%)

0.68

47 (4.5%) 991 (94.2%) 14 (1.3%)

38 (3.6%) 999 (95.0%) 15 (1.4%)

0.60

Chemotherapy No Yes

907 (86.3%) 144 (13.7%)

890 (84.7%) 161 (15.3%)

0.29

899 (85.5%) 153 (14.5%)

889 (84.5%) 163 (15.5%)

0.54

Hormonal therapy Yes No

673 (64.0%) 378 (36.0%)

672 (63.9%) 379 (36.1%)

0.96

652 (62.0%) 400 (38.0%)

673 (64.0%) 379 (36.0%)

0.34

ER status Negative Positive Unknown

131 (12.5%) 845 (80.4%) 75 (7.1%)

122 (11.6%) 837 (79.6%) 92 (8.8%)

0.35

143 (13.6%) 830 (78.9%) 79 (7.5%)

124 (11.8%) 842 (80.0%) 86 (8.2%)

0.42

0.83 0.63 d

66.6 d

65.4 d

0.02 d

923 (87.7%) 122 (11.6%) 7 (0.7%)

933 (88.7%) 115 (10.9%) 4 (0.4%)

0.58

Age (yr) Tumor size (cm, continuous) Tumor size (cm) !2 cm $2 cm Unknown

65.37 1.14 d

65.26 1.15 d

APBI 5 accelerated partial breast irradiation; WBI 5 whole breast irradiation; ER 5 estrogen receptor.

314

C. Shah et al. / Brachytherapy 11 (2012) 311e315

Table 3 Five-year clinical outcomes after matching Tumor size 2! vs. $2 cm

Tumor size continuous Outcome

WBI

APBI

p-Value

WBI

APBI

p-Value

Regional recurrence Overall survival

1 (0.1%) 994 (94.6%)

7 (0.7%) 960 (91.3%)

0.12 0.23

2 (0.2%) 996 (94.7%)

7 (0.7%) 980 (93.2%)

0.09 0.14

APBI 5 accelerated partial breast irradiation; WBI 5 whole breast irradiation.

( p 5 0.24) was noted between WBI and APBI patients; however, a statistically significant decrease in mortality was noted in APBI patients (hazard ratio 5 0.008, p ! 0.0001) using a tumor size cutoff of 2 cm. When evaluating tumor size continuously, similar findings were seen with regard to RR ( p 5 0.10) and improved OS (hazard ratio 0.006, p !0.001) in the APBI arm.

Discussion The results of our analysis demonstrate that no difference in RR or OS is noted between comparably matched patients undergoing WBI or APBI. These data further confirm previously published series that compared WBI and APBI, and represents one of the largest series to date comparing the two treatment modalities. One of the few published randomized trials comparing WBI and APBI comes from Hungary; at 5 years, there was no difference in local recurrence, disease-free survival, or OS (3). However, this study was limited to 258 patients and over 30% of the partial breast patients were treated using a nonhypofractionated treatment schedule, which does not represent a comparable technique to current APBI methodology. A more recent prospective trial from Vaidya et al. (8) evaluated WBI and intraoperative partial breast irradiation with no difference in clinical outcomes, including regional failures, noted at 4 years. Limitations of this study include the noninferiority design of the trial, unknown margin status at the time of treatment, limited followup, and the use of an intraoperative technique, which is not a commonly used method to deliver APBI at this time. Long-term retrospective data for the interstitial technique is available, with a matched-pair analysis from William Beaumont Hospital demonstrating no difference in clinical outcomes between patients treated with interstitial APBI or WBI at 12 years (4). With the increased use of APBI over the past decade, questions have arisen regarding the potential for increased regional failure, as treatment with APBI does not incorporate the regional lymphatics. However, our analysis did not find any difference in the rates of regional failure with the application of APBI when compared with WBI. These data are supported by data from William Beaumont Hospital, which evaluated over 500 patients treated with APBI and had a 5-year axillary failure rate of 0.19% and a 5-year regional nodal failure rate of 0.37% (14). Further, data from the

ASBS MammoSite Registry Trial demonstrated a 5-year axillary failure rate of 0.8% with histologic high grade as the only associated risk factor (15). Dosimetric analysis of WBI has determined that coverage of Level IeII lymph nodes with traditional tangents is only 55% of the target volumes (16). This continues to raise the question of whether WBI, without the addition of intentional regional irradiation, is capable of reducing the rate of axillary and regional failure because it delivers incomplete coverage and nonprescription doses to the Level I and II lymph nodes (16). There are limitations to our analysis. Our study was not randomized or prospective and while we attempted to alleviate some patient differences between groups by matching patients, the potential for other factors that were not matched has the potential to impact our outcomes such as Her-2 and margin status. Further, patients treated with WBI were treated between 1980 and 2008 and therefore, presenting characteristics may be different for patients treated in the early 1980s compared with the APBI patients who were treated between 2002 and 2004. Treatment with WBI in the 1980s may have also included treatment of internal mammary lymph nodes, which are not addressed by ABPI. Potentially, toxicities from older WBI treatment techniques could partially explain the detriment in survival noted on regression analysis between WBI and APBI. As the WBI patients were accrued over 22 years, there exists the potential for bias as outcomes have improved over this time period. Another limitation of our study is that we did not evaluate differences in rates of local recurrence. This was intentional, as the local SEER data queried did not provide sufficient local recurrence data to perform this analysis (i.e., new primaries vs. ipsilateral breast tumor recurrence are not included in the recurrence data). Also, SEER data are based on reporting from participating institutions; therefore, the data queried are only as accurate as the institution has available, and the potential for patients lost to followup is high, which could lead to underscoring of failure rates. Of note, our analysis demonstrated a significant improvement in OS with APBI using the Cox regression analysis, which was not noted with the KaplaneMeier values. This may be because of the semiparametric method for Cox analysis compared with the nonparametric assumptions of the KaplaneMeier analysis despite the use of the same tumor size parameters. Finally, this analysis did not examine cosmetic outcomes or toxicity associated with treatment. Going forward, these factors must also be considered when comparing WBI and APBI.

C. Shah et al. / Brachytherapy 11 (2012) 311e315

Our study, however, confirms the outcomes of singleinstitution, collaborative, and the few published prospective trials evaluating APBI, which have all showed excellent clinical outcomes using this technique. Until the presentation of data from the ongoing prospective Phase III trials, these data represent the largest series comparing rates of regional failure and survival between WBI and APBI.

Conclusions The results of our matched-pair analysis demonstrated no difference in OS or RR in women with early stage breast cancer treated with WBI or APBI. These findings support the continued accrual of eligible patients to the Phase III APBI clinical trials and, in the interim until these studies have been concluded, the treatment of patients off clinical protocols with appropriate selection criteria. References [1] Lazovich D, Solomon CC, Thomas DB, et al. Breast conservation therapy in the United States following the 1990 National Institutes of Health Consensus Development Conference on the treatment of patients with early stage invasive breast carcinoma. Cancer 1999;86:628e637. [2] Morrow M, White J, Moughan J, et al. Factors predicting the use of breast-conserving therapy in stage I and II breast carcinoma. J Clin Oncol 2001;19:2254e2262. [3] Polgar C, Fodor J, Major T, et al. Breast-conserving treatment with partial or whole breast irradiation for low-risk invasive breast carcinomad5-Year results of a randomized trial. Int J Radiat Oncol Biol Phys 2007;69:694e792. [4] Shah C, Antonucci JV, Wilkinson JB, et al. Twelve-year clinical outcomes and patterns of failure with accelerated partial breast irradiation versus whole-breast irradiation: Results of a matched-pair analysis. Radiother Oncol 2011;100:210e214.

315

[5] Vicini F, Beitsch PD, Quiet CA, et al. Five-Year analysis of treatment efficacy and cosmesis by the American Society of Breast Surgeons Mammosite Breast Brachytherapy Registry Trial in patients treated with accelerated partial breast irradiation. Int J Radiat Oncol Biol Phys 2011;79:808e817. [6] Vicini F, Winter K, Wong J, et al. Initial efficacy results of RTOG 0319: Three-dimensional conformal radiation therapy (3D-CRT) confined to the region of the lumpectomy cavity for stage I/II breast carcinoma. Int J Radiat Oncol Biol Phys 2010;77:1120e1127. [7] Radiation Therapy Oncology Group: RTOG 0413. Available at: http:// www.rtog.org/members/protocols/0413/0413.pdf Accessed June 1, 2011. [8] Vaidya JS, Joseph DJ, Tobias, et al. Targeted intraoperative radiotherapy versus whole breast radiotherapy for breast cancer (TARGITA trial): An international prospective, randomized, non-inferiority phase 3 trial. Lancet 2010;376:91e102. [9] Shaitelman SF, Vicini FA, Beitsch P, et al. Five-year outcome of patients classified using the American Society for Radiation Oncology consensus statement guidelines for the application of accelerated partial breast irradiation: An analysis of patients treated on the American Society of Breast Surgeons MammoSite Registry Trial. Cancer 2010;116:4677e4685. [10] Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41e55. [11] Dehejia RH, Wahba S. Propensity score matching methods for nonexperimental causal studies. Rev Econ Stat 2002;84:151e161. [12] Smith J, Todd P. Does matching overcome LaLonde’s critique of nonexperimental estimators? J Econom 2005;125:305e353. [13] Jiang M, Foster EM, Gibson-Davis CM. Breastfeeding and the child cognitive outcomes: A propensity score matching approach. Matern Child Health J 2011;15:1296e1307. [14] Dekhne N, Shah C, Wilkinson JB, et al. Axillary lymph node failure in patients treated with accelerated partial breast irradiation. Cancer 2012;118:38e43. [15] Aburabia M, Roses RE, Kuerer HM, et al. Axillary failure in patients treated with MammoSite accelerated partial breast irradiation. Ann Surg Oncol 2011;18:3415e3421. [16] Reed DR, Lindsley SK, Mann GN, et al. Axillary lymph node dose with tangential breast irradiation. Int J Radiat Oncol Biol Phys 2005;61:358e364.