- Email: [email protected]
Postmastectomy radiotherapy: Patterns of recurrence and long-term disease control using electrons
Int. J. Radiation Oncology Biol. Phys., Vol. 56, No. 3, pp. 716 –725, 2003 Copyright © 2003 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/03/$–see front matter
doi:10.1016/S0360-3016(03)00112-3
CLINICAL INVESTIGATION
Breast
POSTMASTECTOMY RADIOTHERAPY: PATTERNS OF RECURRENCE AND LONG-TERM DISEASE CONTROL USING ELECTRONS STEVEN J. FEIGENBERG, M.D., NANCY PRICE MENDENHALL, M.D., RASHMI K. BENDA, M.D., CHRISTOPHER G. MORRIS, M.S.
AND
Department of Radiation Oncology, University of Florida College of Medicine, Gainesville, FL Purpose: To determine the patterns of failure and prognostic factors for locoregional recurrence after postmastectomy radiotherapy (RT), using a specific electron beam technique. Methods and Materials: A uniform electron beam was used in 323 patients with invasive breast cancer at the University of Florida Health Science Center. The patterns of disease recurrence, prognostic factors, and overall outcome were studied. Results: At 10 years, the freedom from locoregional recurrence, disease-free survival, and absolute survival rate was 90%, 62%, and 55%, respectively. The 10-year disease-free survival rate for patients with 0, 1–3, and >3 positive lymph nodes was 73%, 75%, and 47%, respectively. On multivariate analysis, the three factors significantly associated with locoregional recurrence were T stage, number of involved nodes, and RT fields. Full axillary fields appeared to be beneficial (p ⴝ 0.02). Patients with positive surgical margins appeared to benefit from a mastectomy incision boost to >65 Gy. Finally, patients with T2N0 disease had a substantial risk of chest wall recurrence without chest wall RT. Conclusion: Findings include a low rate of clinically detectable locoregional recurrence. The data suggest benefits for the addition of full axillary RT in node-positive patients and chest wall RT in patients with T2N0 disease. © 2003 Elsevier Inc. Breast neoplasms, Radiotherapy, Adjuvant, Electrons, Combined modality therapy, Adverse effects.
INTRODUCTION
with the treatment of postmastectomy patients with an en face electron-beam technique.
Recently, three prospective randomized trials (1–3) showed improved local control, disease-free survival, and overall survival with the addition of radiotherapy (RT) to systemic therapy for postmastectomy breast cancer patients. Previous trials (4) of postmastectomy RT showed significant improvements in clinically detectable locoregional recurrence (LRR), but no convincing evidence of a survival benefit. One reason for the lack of survival benefit from RT was an increase in non– cancer-related deaths, particularly cardiac deaths in patients with left-sided breast cancer (4 –7). The increased risk was related to the RT technique, specifically tangential photon fields to the left chest wall and en face photon fields to the internal mammary nodes that included a large volume of cardiac tissue (8 –11). Since 1978, electrons have been used to treat postmastectomy patients at the Shands Cancer Center at the University of Florida in an effort to cover the target volume accurately while minimizing the exposure to normal heart and lung tissue. We report the long-term patterns of failure and outcomes associated
Between January 1978 (when electrons became available in the Shands Cancer Center at the University of Florida) and June 1998, 323 postmastectomy breast cancer patients were treated with RT using an electron beam technique for treatment of the chest wall and internal mammary nodes. During this time, postmastectomy patients were treated almost exclusively with electrons. Forty-two women treated with neoadjuvant chemotherapy and 5 men were not included in the study. Eight patients were lost to follow-up without evidence of cancer ⬍2 years after treatment, leaving 268 patients. Three patients with metachronous or synchronous contralateral breast cancer underwent bilateral postmastectomy RT, so that 271 treatment courses in 268 patients were analyzed. Seventy-nine women (29%) were premenopausal, 173 (65%) were postmenopausal, 14 (5%) were perimenopausal,
Reprint requests to: Nancy Price Mendenhall, M.D., Department of Radiation Oncology, University of Florida Health Science Center, 2000 SW Archer Rd., P.O. Box 100385, Gainesville, FL 32610-0385. Tel: (352) 265-0287; Fax: (352) 265-0759; E-mail: [email protected]
Poster presentation at the 86th Scientific Assembly and Annual Meeting of the Radiological Society of North America, Chicago, IL, 2000. Received Sep 10, 2002, and in revised form Jan 2, 2003. Accepted for publication Jan 10, 2003.
METHODS AND MATERIALS
716
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
717
Table 1. Patient distribution according to stage and axillary status Positive nodes T stage
0
1–3
4–9
ⱖ10
Unknown (⬎0)
Total cases* (n)
T1 T2 T3 T4 Unknown Total
23 26 10 8 1 68
17 33 11 11 1 73
18 28 5 5 5 61
12 22 14 15 0 63
2 1 0 0 0 3
72 110 40 39 7 268
Data presented as the number of patients. * Excludes 2 patients with no axillary lymph nodes found in pathologic specimen and 1 with 4 positive lymph nodes but no tumor at primary site.
and 2 (1%) were of unknown menopausal status. The median age was 57 years (range 27– 84). The left side was treated in 143 cases and the right side in 128 cases. All pathology specimens, including outside biopsy or mastectomy specimens, were reviewed at Shands Teaching Hospital. Close margins were defined as ⱕ2 mm from the inked margin. Close (n ⫽ 33) or positive (n ⫽ 16) deep chest wall margins were present in 49 patients. The 1997 American Joint Committee on Cancer (12) guidelines were used for staging. The distribution by stage for the 271 cancers was as follows: Stage I, 23 (8%); Stage II, 154 (57%); and Stage III, 86 (32%). The stage could not be determined in 8 cases (3%). The distribution by T stage and number of involved lymph nodes are shown in Table 1. Of the 268 patients, 70 had 0, 73 had 1–3, 62 had 4 –9, and 63 patients had ⱖ10 positive lymph nodes. Three additional patients had positive lymph nodes, although the exact number of involved and retrieved lymph nodes was not stated in the pathology report. Follow-up The median follow-up among living patients was 8 years (range 2–21). Sixty-two percent of the patients had ⬎5 years of follow-up, and 30% had ⬎10 years of follow-up. The data were obtained from the RT chart, the hospital chart, on-line medical records (when available), and interviews of patients and/or family members. Surgery Most patients were treated with either a modified radical mastectomy (including Level I and II axillary node dissection) or a simple mastectomy after attempted breast-conserving surgery and axillary dissection. Radical mastectomy was performed in 5 patients. The median number of lymph nodes identified by the pathologist was 18, ranging from 0 (2 patients) to 60 (1 patient). Of 268 patients, 29 (10%) had ⬍10 lymph nodes in the surgical specimen. Chemotherapy Chemotherapy was administered to 159 patients. Ninetytwo patients received a doxorubicin-based regimen with (51
patients) or without (41 patients) additional treatment with cyclophosphamide, methotrexate, and fluorouracil (CMF). Of the 51 patients who received doxorubicin in addition to CMF, 42 had RT delivered concurrently with CMF, in contrast to 40 of 41 patients treated with doxorubicin regimens alone whose RT followed the chemotherapy sequentially. Only 1 patient received concomitant doxorubicin and RT. Fifty-nine patients received CMF alone (concomitant with RT in 37 patients, sequential in 22 patients). Overall, RT was delivered sequentially in 78 patients and concomitantly (during CMF in all cases except for 1) in the remaining 81 patients. Details regarding the exact systemic regimen were unavailable in 7 patients. Hormonal therapy Hormonal therapy was given to 102 patients, including tamoxifen (101 patients) and megestrol acetate (1 patient). Forty-six patients were treated with tamoxifen alone; 56 received chemotherapy in addition to hormonal therapy. Hormonal therapy was generally started after RT completion. Radiotherapy Indications. The indications used for field selection in postmastectomy RT during this time (13) are shown in Table 2. The treatment of the internal mammary chain (IMC) nodes and supraclavicular fossa (SCF), referred to as peripheral lymphatic irradiation (PLI), was used for patients with medial T1–2N0 or T1–2N1bi cancers with 1–3 involved lymph nodes before 1994. Patients with larger tumors (T3–T4) and/or ⬎3 positive axillary lymph nodes received chest wall RT as well. After 1994, all patients with any positive lymph nodes received chest wall, SCF, and IMC node RT. The entire axilla was treated with an anterior SCF field extended laterally and inferiorly and a posterior axillary field in patients with axillary lymph nodes ⬎2 cm, extensive extracapsular extension, clinical matted adenopathy, or an incomplete axillary dissection. The chest wall was treated in all patients with close or positive margins. Field design. The fields were designed on the basis of surface anatomy. CT treatment planning was performed in
718
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
Table 2. Indications for postmastectomy radiotherapy Stage
N0
N1bi
N1bii
N1biii-iv or N2
T1 medial T2 medial T3 T4
IMC PLI PLI ⫹ CW PLI ⫹ CW
PLI PLI PLI ⫹ CW PLI ⫹ CW
PLI ⫹ CW PLI ⫹ CW PLI ⫹ CW PLI ⫹ CW
PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX
Abbreviations: IMC ⫽ internal mammary chain; PLI ⫽ peripheral lymphatic irradiation (internal mammary chain and supraclavicular fossa); CW ⫽ chest wall; AX ⫽ axillary. Modified, with permission, from Mendenhall et al. (13).
only 16 patients. Fluoroscopy and simulation films were used to confirm placement of the IMC field over the first three rib interspaces, inclusion of the coracoid process and/or surgical clips within the SCF and axillary field border, and inclusion of the clavicle and a margin of lung inside the rib cage within the axillary field borders. The IMC nodes were treated with an en face electron field. The medial border of the IMC field was placed on a line drawn from 1 cm across the midline superiorly to the midline inferiorly (Fig. 1). The lateral border of the field was 6 cm lateral to the midline. The IMC field was thus 7 cm wide superiorly and 6 cm wide inferiorly. With this design, the sternal– costal junction and the parasternal tissue that harbors the IMC nodes were always centered in the IMC field. The inferior border was placed on the fourth rib to encompass the first three interspaces. All six interspaces were included if the primary breast lesion was located in the inferior and medial quadrants. The superior border of the IMC field abutted the SCF. The inferior border of the SCF included the first rib for PLI and the second rib when the whole axilla was treated. The lateral border of the SCF field included any clips the surgeon placed at the top of the axillary dissection and coracoid process. If treatment was planned to the whole axilla (Fig. 2), the lateral border of the SCF was extended to the mid humerus. The medial border of the SCF field on the skin extended from 1 cm across the midline inferiorly to the cricothyroid ligament superiorly. The lateral superior border flashed across the trapezius muscle in most cases. The gantry on the SCF field was angled 15° to reduce esophageal and spinal cord exposure. The chest wall fields abutted the IMC and SCF fields on the skin and included a minimum of a 5-cm margin below the mastectomy scar with at least 2-cm margins medial and lateral to the mastectomy incision. The chest wall field was divided into medial and lateral fields on the basis of the contour of the chest wall. The chest wall fields were matched on the skin surface, producing an area of inhomogeneity with increased dosage at the skin surface. Junctions of abutting electron fields were changed once during treatment to reduce the dose inhomogeneity. Dose and dose specification. The usual dose was 50 Gy within 5 weeks at 2 Gy/fraction given 5 d/wk. Before 1982, 17 patients received 50 Gy within 4 weeks (2.5 Gy/fraction). An additional boost of 10 Gy in 5 fractions with 6- or
8-MeV electrons was given to the mastectomy incision; the boost dose was increased to 15 or 20 Gy in patients with close or positive mastectomy margins. The depth of the inner table of the sternum was presumed
Fig. 1. Standard field setup using electron beam technique. Reprinted with permission from Mendenhall et al. (13).
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
719
Fig. 2. Standard extension of SCF field used to treat entire axilla and low neck in a patient with more than N1bi axillary disease (opposed posterior axillary boost field not shown.)
to reflect the maximal potential depth of the internal mammary nodes. This distance was determined by placing a radiopaque marker on the skin surface and measuring the distance between the skin and inner table of the sternum on a cross-table lateral film. The electron beam energy for the IMC field was then selected to include the maximal depth of the inner table of the sternum within the 90% isodose volume and to provide a steep dose gradient to underlying tissues to minimize the dose to the heart and/or lung. In most patients, the depth of the inner table of the sternum was 2– 4 cm. Consequently, the IMC field in most patients was treated with 10 –14-MeV electrons. When ⱖ14 MeV electrons were used, no bolus was applied to the skin because of the increased skin dose with the higher energy electrons; in patients with internal mammary node depths of ⱖ4 cm, 20% of the dose was occasionally delivered with 6-MV photons to increase the coverage at depth and decrease the skin dose. The target volume for the chest wall fields was considered the surgical flap, not the entire thickness of the chest wall. In most cases, the chest wall flap was estimated to be between 1 and 2 cm thick. The determination of electron energy was based on the clinical assessment. Tissue-equivalent bolus material (0.5 cm) was applied to the chest wall to increase the skin dose to 90% of the prescription dose. Consequently, 6-MeV electrons were selected for most slender patients, which produced 90% isodose coverage at a 1.8 cm depth (including the bolus), thus covering chest wall flaps up to 1.3 cm thick. Larger patients were treated with 8-MeV electrons, which produced 90% isodose coverage at approximately 2.3 cm depth (including the bolus), thus covering chest wall flaps up to 1.8 cm thick within the 90% isodose shell. Higher electron energies were used occasion-
ally according to physician preference to treat the entire chest wall thickness. Only 4 patients had chest wall treatment with ⬎10 MeV electrons: 12 MeV (n ⫽ 1) and 14 MeV (n ⫽ 3). The dose to the anterior SCF field was delivered with photons (frequently with a 20% electron component). A dose of 50 Gy was prescribed to the depth of the maximal dose deposition, resulting in an actual target dose of 45–50 Gy. When the entire axilla was judged to be at risk, the SCF field was extended laterally and inferiorly, and a posterior field was added to increase the dose at the midplane of the axilla to 45–50 Gy. The posterior field was usually treated with 20-MV photons. This meant that the highest risk tissues (including the axillary apex), which were anterior to the midplane of the axilla and included in the posterior field, actually received ⬎45–50 Gy, in some cases up to 60 Gy. The inferior border of the SCF and posterior axillary boost fields abutted the superior borders of the medial and lateral chest wall fields. The region of low axillary nodes (which were resected) was frequently included in superior portion of the lateral chest wall field, rather than the SCF field, if the electron energy was judged adequate for coverage (Fig. 2).
Statistical analysis All statistical analyses were performed using SAS software (14). The end points of the study included LRR, any relapse, death, and other events that were possibly related to treatment. LRR was calculated directly as the crude number of events over the number of patients at risk, as well as the rate projected over time by the product-limit method (15). LRR was defined as the appearance of tumor on the chest wall or in the ipsilateral supraclavicular, axillary, or internal
720
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
Table 3. Ten-year freedom from locoregional recurrence, freedom from relapse, and absolute survival rates according to T stage, number of positive axillary nodes, and overall stage Variable T stage (n ⫽ 266) Tx–T1 T2 T3 T4 p Positive axillary nodes (n ⫽ 268) 0 1–3 ⬎3 p Overall stage (n ⫽ 260) I IIa IIb IIIa IIIb p
Locoregional control (%)
Freedom from relapse (%)
Absolute survival (%)
97 86 87 88 0.2351
75 60 47 54 0.0115
67 52 49 43 0.0029
91 95 86 0.1520
73 75 45 0.0006
63 69 41 0.0050
100 94 88 89 88 0.7380
95 66 63 43 53 0.0003
89 56 59 32 41 0.0001
mammary lymph nodes before, simultaneous with, or after the development of distant metastases. We considered the occurrence of distant metastasis and LRR within 1 month of each other as simultaneous events. For disease-free survival, any relapse was considered an event and patients were censored from analysis if they died without relapse or evidence of disease. The only event counted for absolute survival was death. At the time of death, patients with no LRR were censored from the actuarial calculation of LRR probability. The follow-up time was calculated from the time of mastectomy. Multivariate analyses of potential prognostic variables for chest wall recurrence, locoregional control, disease-free survival, and absolute survival were performed using a Cox regression backward selection procedure (16). Potential prognostic variables in the analysis of chest wall recurrences included the number of lymph nodes involved with metastatic disease (0, 1–3, or ⬎3), pathologic T stage, margin status (negative, close, or microscopically positive), and electron energy used to treat the chest wall (none, 6 MeV, 8 MeV, 10 MeV, or other). Factors with potential prognostic value in the analysis of locoregional control included the number of lymph nodes involved with metastatic disease, pathologic overall stage, pathologic T stage, fields treated (PLI, PLI ⫹ chest wall, or PLI ⫹ chest wall ⫹ axillary), systemic therapy (concurrent chemotherapy, sequential chemotherapy, tamoxifen alone, or no systemic therapy), menopausal status (pre-, peri-, or postmenopausal), and margin status. Factors assessed for potential prognostic value in predicting disease-free survival and absolute survival included the number of lymph nodes involved with metastatic disease, pathologic stage, pathologic T stage, fields treated, addition of chemotherapy, menopausal status, and age (⬎50 or ⬍50 years).
RESULTS Absolute survival The absolute survival rate at 5, 10, and 15 years was 72%, 55%, and 40%, respectively. Multivariate analysis of factors potentially predictive of absolute survival revealed the following to be strongly correlated with survival: pathologic T stage (p ⬍0.0001), addition of chemotherapy (p ⫽ 0.0013), and number of lymph nodes involved with metastatic disease (p ⫽ 0.007). The 10-year survival rates as a function of T stage, overall stage, and number of lymph nodes involved are shown in Table 3. Of the 268 patients, 127 have died— 85 of breast cancer, 42 of intercurrent disease, and 0 of treatment complications. Of the 13 patients who developed subsequent cardiac disease, no difference was found in the incidence between patients with right or left-sided breast cancer (p ⫽ 0.1777). Disease-free survival The disease-free survival rate at 5, 10, and 15 years was 71%, 62%, and 59%, respectively. Multivariate analysis of potentially predictive variables for disease-free survival revealed the following to be significantly associated with disease-free survival: number of lymph nodes involved with metastatic disease (p ⫽ 0.0008) and pathologic T stage (p ⫽ 0.002). The relationship of 10-year freedom from locoregional recurrence, disease-free survival, and overall survival with T stage, overall stage, and number of involved lymph nodes is shown in Table 3. Locoregional control The rate of locoregional disease control at 5, 10, and 15 years was 92%, 90%, and 90%, respectively. Patterns of failure. LRR developed in 25 patients, includ-
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
721
Table 4. Sites of locoregional recurrence by treatment field Recurrence site
CW only (n ⫽ 4)
PLI only (n ⫽ 27)
CW ⫹ PLI (n ⫽ 171)
CW ⫹ PLI ⫹ AX (n ⫽ 66)
Local—in boost field Local—marginal to boost field Local—regional outside RT field Regional—axillary Regional—supraclavicular Regional—internal mammary Regional—other Multiple sites Total
0 0 0 0 0 0 0 0 0 (0)
0 0 3 0 0 0 0 0 3 (11)
6 0 3 1 3 0 1 3* 17 (10)
0 0 3 1 1 0 0 1† 6 (9)
Abbreviations as in Table 2. Numbers in parentheses are percentages. * One local (in boost field) and regional (internal mammary); one local (other) and regional (axillary); and one local (other) and regional (other). † One local (in boost field) and regional (internal mammary).
ing isolated LRR (14 patients), simultaneous LRR and distant metastasis (8 patients), and LRR after distant metastasis (3 patients). Thirteen of the 14 isolated LRRs were confined to either the chest wall alone (6 patients) or the regional lymph nodes (7 patients); one LRR involved both the chest wall and regional lymph nodes simultaneously. Two-thirds of LRRs (16 of 25) were detected within 2 years, and almost 88% (22 of 25) were detected within 3 years. The relationship between the patterns of LRR and fields selected for RT is shown in Table 4. Chest wall recurrences. There were 19 recurrences on the chest wall, including 3 in the untreated chest walls of patients receiving PLI alone. In most cases (194 treatment courses), the chest wall was treated with electron energy to cover only the surgical flap (i.e., the skin and subcutaneous tissue). By physician preference, higher energy electrons were used to treat the entire chest wall thickness in 50 cases. No correlation was found between the electron energy used and the T stage, number of positive nodes, or pathologic stage, indicating physician preference rather than disease extent as the motivation to treat the entire chest wall thickness rather than the surgical flap. The infield LRR rate at 10 years was 4%, 5%, and 15% for patients treated with 6, 8, and 10 MeV electrons, respectively. Multivariate analysis of factors potentially predictive of chest wall recurrences identified no significant correlations: energy used to treat the chest wall (p ⫽ 0.50), pathologic T stage (p ⫽ 0.18), number of lymph nodes involved with metastatic disease (p 0.63), and surgical margin status (p ⫽ 0.48). Because patients with positive margins, extensive nodal disease, and/or large tumor were at a greater risk of chest wall recurrence, the lack of correlation suggests efficacy of the therapeutic policy. No evidence of greater chest wall control rates with the use of higher energy electrons was seen, suggesting that coverage of the full thickness of the chest wall was not generally necessary. Only three chest wall failures occurred in the 49 patients with close or positive margins (Table 5). No chest wall
failures occurred in the 8 patients who received a boost of 15–20 Gy to the incision dose. Chest wall recurrence did occur in 2 of 8 patients with close margins not treated with a boost to the surgical scar. Both recurrences were located in the scar and would have been covered in a boost field. One chest wall failure outside the boost field occurred in the 33 patients who received a boost of 10 Gy to the surgical scar. Regional lymph node recurrence. Regional lymph node recurrences were detected in 12 patients. The SCF was the most common site of clinically detected regional node treatment failure (7 patients) followed by axilla (4 patients), and the IMC region (4 patients). Three of four IMC recurrences were detected either simultaneously with SCF recurrence (2 patients) or after LRR (1 patient). All occurred after treatment of the IMC field to 50 Gy in 25 fractions with 12-, 14-, or 15-MeV electrons or a combination of 15-MeV electrons and 20-MV photons (weighted 4:1 in favor of electrons). No significant difference was seen in the supraclavicular or axillary lymph node recurrence rate as a function of the extent of the axillary dissection. The SCF recurrences occurred in patients with a median of 9 positive nodes (range 0 –17) at mastectomy. One patient who developed an SCF recurrence had no axillary lymph nodes identified in the axillary dissection. The SCF recurrences were located primarily in the medial part of the SCF. A review of the SCF port films showed that all failures were located within the
Table 5. Relationship between chest wall recurrence and radiation boost dose (after 50 Gy to chest wall) to mastectomy incision in patients with close or positive mastectomy margins Boost dose to incision (Gy) 0 10 15–20
Chest wall recurrences/Treated (n) 2/8 1/33 0/8
722
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
wall recurrence than patients additionally treated with a chest wall field. All these patients had T2N0 disease and developed LRR on the chest wall outside the PLI fields. Twenty-two of 244 patients treated to PLI and chest wall with or without axillary fields had a LRR, 19 of whom had recurrence within the treatment volume. Two patients had recurrence in the chest wall outside the treatment fields. One patient, who had indications for full axillary treatment (lymph node ⬎2 cm), had a recurrence in the untreated axilla. The overall rates of freedom from LRR at 10 years according to the extent of RT were as follows: PLI only, 88%; PLI and chest wall, 89%; and PLI plus chest wall plus axillary, 94%. Because only 4 cases were in the chest wall-only treatment group, this group was not included in this stratification. Prognostic factors for LRR. Multivariate analysis of factors potentially influencing the probability of locoregional disease control revealed the following factors to be significantly associated with overall LRR: T stage (p ⫽ 0.03), number of lymph nodes involved with metastatic disease (p ⫽ 0.03), and fields treated (p ⫽ 0.02). Factors not significantly associated with LRR in this study included pathologic stage (p ⫽ 0.08), addition of chemotherapy (p ⫽ 0.90), menopausal status (p ⫽ 0.90), and margin status (p ⫽ 0.32).
Fig. 3. Example of recurrence in surgical scar on unirradiated chest wall.
irradiated field. Both patients who had no nodal tissue identified in the axillary dissection had a recurrence in the regional lymph nodes, in the axilla in 1 patient and in the SCF in 1 patient. The number of lymph nodes recovered in the axillary dissection did not correlate with rate of supraclavicular or axillary failure (1 of 35 in patients with 1–9 nodes removed, 3 of 141 in patients with 10 –19 nodes removed, 5 of 132 in patients with ⱖ19 nodes removed, and 1 of 8 in patients with an unknown number of nodes removed). Adjuvant therapy. The actuarial incidence of LRR at 10 years as a function of the use and sequence of adjuvant therapy was as follows: RT alone (16%); RT followed by tamoxifen (5%); concomitant RT and chemotherapy (8%); and sequential RT and chemotherapy (11%). Field selection. The 10-year actuarial LRR rate was 12% for patients who received only PLI compared with 10% for patients who received chest wall RT and PLI. The three instances of LRR in the group receiving PLI alone (27 total) occurred in the untreated chest wall (Fig. 3). Patients treated with PLI alone were considered to be at lower risk of chest
Morbidity Lymphedema developed in 55 (21%) of 268 patients. Most (n ⫽ 33, 60%) had mild or asymptomatic edema, which was recorded if the patient complained of swelling or if visible edema was present that measured ⱕ2 cm in difference between the two sides. Nine (3%) of 268 patients developed symptomatic radiation pneumonitis. The incidence of pneumonitis appeared to correlate with treatment of an SCF field and with the use of higher electron energies to treat the chest wall field. When the SCF was treated, the incidence of symptomatic pneumonitis was 3.6% (9 of 247 patients) compared with 0% (0 of 24 patients) when the SCF was not treated. The incidence of pneumonitis increased with the use of higher energies on the chest wall: 1.6% with 6 MeV, 2.2% with 8 MeV, and 8% with 10-MeV electrons. One patient in our series developed brachial plexus injury (for an incidence of 1 of 268). The SCF field in this patient was treated to 50 Gy in 20 fractions at 2.5 Gy/fraction. An axillary field was used to increase the dose to the midplane to 50 Gy. The actual dose and dose rate to the brachial plexus was estimated to be 15–20% higher than the prescription dose. Asymptomatic telangiectasia was noted along the incision boost field and along the matchline between the medial and lateral chest wall fields in approximately one-third of patients. Telangiectasia outside the matchline or boost field was rarely noted.
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
723
Fig. 4. CT scans of 32-year-old woman with a 4.5-cm upper inner quadrant breast cancer with 3 of 16 nodes positive at modified radical mastectomy. Two months after mastectomy before beginning chemotherapy, the new presence of a parasternal mass was noted. (a) CT confirmed the internal mammary adenopathy, as well as (b) Rotter’s nodes and axillary apical nodes seen on the scan but not clinically detected.
DISCUSSION The recent postmastectomy RT trials (1–3) have secured the role for postmastectomy RT in patients at risk of residual subclinical disease. Several questions remain to be answered, however, including which subsets of patients have insufficient risks to warrant RT, which fields must be treated to obtain the efficacy documented in the recent landmark trials, and what are the optimal doses and techniques to achieve the best therapeutic ratio. It was the policy in the present study to use postmastectomy RT in all node-positive patients, all patients with T3 primaries, all patients with medial or central T2 primaries, and all patients with close or positive chest wall margins. The treatment fields varied according to the indication for postmastectomy RT. Before 1994, patients with T2N0 medial or central tumors or 1–3 positive axillary lymph nodes received RT to the IMC and SCF; after 1994, these patients received chest wall, IMC, and SCF RT. Any patient with a close or positive surgical margin received chest wall RT. Any patients with either a T3 tumor or ⬎3 positive lymph nodes received chest wall, SCF, and IMC nodal RT. Most patients with axillary lymph nodes that were matted or ⬎2 cm, had extracapsular extension, or had incomplete dissections also received full axillary RT with an extended supraclavicular field and a supplemental posterior axillary boost. This study of patterns of recurrence showed a very low rate of clinically detectable LRR. Several observations may be made from this experience that shed light on the optimal field selection, treatment technique, and dose prescription. It is currently generally accepted that the chest wall is at risk in patients with ⱖ4 positive nodes. Treatment is warranted because the most common site of clinically detected LRR after mastectomy is on the chest wall. It is unclear whether the chest wall requires treatment when few or no axillary lymph nodes are involved and the tumor size is ⬍5
cm. In this study, 27 patients were treated with IMC and SCF RT only, including 12 with T2N0 medial tumors. The overall chest recurrence rate after IMC and SCF RT alone was 12%, but all three chest wall recurrences were among the 12 T2N0 patients. These patients were all operated on by experienced surgical oncologists, and no particular concern was indicated by the surgeon regarding a risk of recurrence in the chest wall. None of these patients had close or positive chest wall margins. The numbers were small, but the risk for chest wall recurrence even with T2N0 breast cancer does appear to be substantial. Another controversial issue is whether the entire axilla needs RT. The question is of more than academic interest, because it is well known that the extent of both surgery and postmastectomy RT contribute to the risk of arm edema. In this study, the full axilla was treated only in the setting of indications listed above. Occasionally, concern about arm edema precluded delivery of axillary RT even with these indications. On multivariate analysis of the clinical and treatment factors predicting a higher rate of LRR, patients who received full axillary RT had a lower rate of clinically detected LRR than those who did not receive full axillary RT (p ⫽ 0.02). This is, at first, a curious finding in that there were only four clinically detected axillary recurrences in this study. The axilla, however, can be a clinically silent area. Uncontrolled disease in the axilla may precede clinically detectable LRR in other sites such as the SCF, chest wall, or even internal mammary nodes (Fig. 4). Because patients did not routinely have follow-up CT imaging in this study, uncontrolled disease in the axilla could easily have been overlooked. We suspect that recurrences in the apex of the axilla were underestimated in this and other studies and that the finding that axillary RT is associated with better disease control is real. Given the finding that whole axillary RT appeared to be associated with better locoregional disease control, the method of dose prescription used in this study is of interest.
724
I. J. Radiation Oncology
● Biology ● Physics
After 50 Gy was delivered to the anterior extended supraclavicular field, prescribed at the maximal depth of dose deposition for whatever beam was used, the actual dose delivered at the midplane of the axilla was calculated and then supplemented with a posterior axillary boost field. The supplemental posterior axillary boost field exited through the axillary apical tissues anterior to the midplane. These anterior tissues had already received a higher dose of RT from the anterior extended supraclavicular field than the tissues at the midplane. Retrospective dose analyses suggested that those tissues anterior to the midplane generally received 55– 60 Gy when the entire axilla was treated and a posterior field was used to bring the midplane axillary dose to 50 Gy. The apex of the axilla after mastectomy may be the nodal area at highest risk of residual subclinical disease in node-positive patients, both because of the orderly fashion in which breast cancer spreads to the lymphatics and because lymphatic flow through the axillary lymphatics is more common than through the IMC. Other studies have demonstrated a dose–response relationship for subclinical disease in breast cancer (17). The European Organization for the Research and Treatment of Cancer demonstrated in patients undergoing breast-conserving therapy that even after lumpectomy with negative surgical margins, the addition of a boost to the tumor bed to a total dose of 66 Gy resulted in a significantly lower local recurrence rate than did a total dose of 50 Gy (17). It is conceivable that treatment of the entire axilla was a critically important part of the postmastectomy RT techniques used in the recent landmark trials (1–3) that finally demonstrated a survival benefit associated with postmastectomy RT. It is also conceivable that the actual dose associated with successful RT of subclinical disease in this very high-risk site is 10 –15% greater than that actually prescribed and reported. The final finding of interest is the pattern of recurrence in the 49 patients with close or positive mastectomy margins. As
Volume 56, Number 3, 2003
noted previously, no recurrences developed in 8 such patients when a boost dose of 15–20 Gy was given to the mastectomy incision after the initial 50 Gy to the chest wall. One recurrence developed outside the boost field in the 33 patients who received a 10-Gy boost, and two recurrences developed in the mastectomy incision among the 8 patients in whom no boost was given after the initial 50 Gy. Although the numbers were small, there was a suggestion that when the mastectomy margins are close or positive, areas on the chest wall, including the mastectomy incision, are at very high risk of residual disease and would best be treated with doses ⬎50 Gy. With 10 –20 years of follow-up, no apparent excess in cardiac deaths occurred in patients with left-sided breast cancers in this series, suggesting that the goal of minimizing radiation exposure to the heart may have been achieved with this technique. Additional follow-up is necessary, however, to ensure no long-term impact on cardiac morbidity. CONCLUSION This report documented a very low rate of clinically detected LRR associated with the electron beam technique in the guidelines used for postmastectomy RT at the University of Florida. The findings in this study of the patterns of recurrence suggest several possible conclusions. First, chest wall treatment may be beneficial even with T2N0 tumors. Second, full axillary RT may be beneficial, at least in all patients with more than N1Bi disease. Third, the optimal dose to very high-risk areas, such as a mastectomy incision in patients with close or positive mastectomy margins and the apex of the axilla, may be ⬎50 Gy. Finally, we recommend that CT staging be done before delivery of adjuvant chemotherapy and RT and at the time of suspected treatment failure to better understand the true pattern of LRR and its relationship to distant metastases. This was a retrospective study, and all findings and possible conclusions suggested here should be confirmed in a larger controlled study.
REFERENCES 1. Overgaard M, Hansen PS, Overgaard J, et al. Danish Breast Cancer Cooperative Group 82b trial: Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. N Engl J Med 1997;337: 949–955. 2. Overgaard M, Jensen MB, Overgaard J, et al. Postoperative radiotherapy in high-risk postmenopausal breast-cancer patients given adjuvant tamoxifen: Danish Breast Cancer Cooperative Group DBCG 82c randomised trial. Lancet 1999;353: 1641–1648. 3. Ragaz J, Jackson SM, Le N, et al. Adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 1997;337:956–962. 4. 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;333:1444–1455. 5. Cuzick J, Stewart H, Rutqvist L, et al. Cause-specific mortality in long-term survivors of breast cancer who participated in trials of radiotherapy. J Clin Oncol 1994;12:447–453.
6. Rutqvist LE, Johansson H. Mortality by laterality of the primary tumour among 55,000 breast cancer patients from the Swedish Cancer Registry. Br J Cancer 1990;61:866–868. 7. Paszat LF, Mackillop WJ, Groome PA, et al. Mortality from myocardial infarction after adjuvant radiotherapy for breast cancer in the Surveillance, Epidemiology, and End-Results cancer registries. J Clin Oncol 1998;16:2625–2631. 8. Host H, Brennhovd IO, Loeb M. Postoperative radiotherapy in breast cancer—Long-term results from the Oslo study. Int J Radiat Oncol Biol Phys 1986;12:727–732. 9. Rutqvist LE, Lax I, Fornander T, et al. Cardiovascular mortality in a randomized trial of adjuvant radiation therapy versus surgery alone in primary breast cancer. Int J Radiat Oncol Biol Phys 1992;22:887–896. 10. Gagliardi G, Lax I, Ottolenghi A, et al. Long-term cardiac mortality after radiotherapy of breast cancer—Application of the relative seriality model. Br J Radiol 1996;69:839–846. 11. Gyenes G, Fornander T, Carlens P, et al. Morbidity of ischemic heart disease in early breast cancer 15–20 years after
Postmastectomy RT using electrons
adjuvant radiotherapy. Int J Radiat Oncol Biol Phys 1994;28: 1235–1241. 12. American Joint Committee on Cancer. AJCC cancer staging manual. 5th ed. Philadelphia: Lippincott Williams & Wilkins; 1997. 13. Mendenhall NP, Fletcher GH, Million RR. Adjuvant radiation therapy following modified radical or radical mastectomy. In: Bland KI, Copeland EM, III, editors. The breast: Comprehensive management of benign and malignant diseases, 1st ed. Philadelphia: WB Saunders; 1991. p, 770–778.
● S. J. FEIGENBERG et al.
725
14. SAS Institute Inc. SAS OnlineDoc(R), Version 8. Cary, NC: SAS Institute Inc.; 1999. 15. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–481. 16. Cox DR. Regression models and life tables—Series B (Methodological). J R Stat Soc 1972;34:187–220. 17. Bartelink H, Poortmans P, et al. Recurrence rates after treatment of breast cancer with standard radiotherapy with or without additional radiation [Abstract]. N Engl J Med 2001; 345:1378–1387.
doi:10.1016/S0360-3016(03)00112-3
CLINICAL INVESTIGATION
Breast
POSTMASTECTOMY RADIOTHERAPY: PATTERNS OF RECURRENCE AND LONG-TERM DISEASE CONTROL USING ELECTRONS STEVEN J. FEIGENBERG, M.D., NANCY PRICE MENDENHALL, M.D., RASHMI K. BENDA, M.D., CHRISTOPHER G. MORRIS, M.S.
AND
Department of Radiation Oncology, University of Florida College of Medicine, Gainesville, FL Purpose: To determine the patterns of failure and prognostic factors for locoregional recurrence after postmastectomy radiotherapy (RT), using a specific electron beam technique. Methods and Materials: A uniform electron beam was used in 323 patients with invasive breast cancer at the University of Florida Health Science Center. The patterns of disease recurrence, prognostic factors, and overall outcome were studied. Results: At 10 years, the freedom from locoregional recurrence, disease-free survival, and absolute survival rate was 90%, 62%, and 55%, respectively. The 10-year disease-free survival rate for patients with 0, 1–3, and >3 positive lymph nodes was 73%, 75%, and 47%, respectively. On multivariate analysis, the three factors significantly associated with locoregional recurrence were T stage, number of involved nodes, and RT fields. Full axillary fields appeared to be beneficial (p ⴝ 0.02). Patients with positive surgical margins appeared to benefit from a mastectomy incision boost to >65 Gy. Finally, patients with T2N0 disease had a substantial risk of chest wall recurrence without chest wall RT. Conclusion: Findings include a low rate of clinically detectable locoregional recurrence. The data suggest benefits for the addition of full axillary RT in node-positive patients and chest wall RT in patients with T2N0 disease. © 2003 Elsevier Inc. Breast neoplasms, Radiotherapy, Adjuvant, Electrons, Combined modality therapy, Adverse effects.
INTRODUCTION
with the treatment of postmastectomy patients with an en face electron-beam technique.
Recently, three prospective randomized trials (1–3) showed improved local control, disease-free survival, and overall survival with the addition of radiotherapy (RT) to systemic therapy for postmastectomy breast cancer patients. Previous trials (4) of postmastectomy RT showed significant improvements in clinically detectable locoregional recurrence (LRR), but no convincing evidence of a survival benefit. One reason for the lack of survival benefit from RT was an increase in non– cancer-related deaths, particularly cardiac deaths in patients with left-sided breast cancer (4 –7). The increased risk was related to the RT technique, specifically tangential photon fields to the left chest wall and en face photon fields to the internal mammary nodes that included a large volume of cardiac tissue (8 –11). Since 1978, electrons have been used to treat postmastectomy patients at the Shands Cancer Center at the University of Florida in an effort to cover the target volume accurately while minimizing the exposure to normal heart and lung tissue. We report the long-term patterns of failure and outcomes associated
Between January 1978 (when electrons became available in the Shands Cancer Center at the University of Florida) and June 1998, 323 postmastectomy breast cancer patients were treated with RT using an electron beam technique for treatment of the chest wall and internal mammary nodes. During this time, postmastectomy patients were treated almost exclusively with electrons. Forty-two women treated with neoadjuvant chemotherapy and 5 men were not included in the study. Eight patients were lost to follow-up without evidence of cancer ⬍2 years after treatment, leaving 268 patients. Three patients with metachronous or synchronous contralateral breast cancer underwent bilateral postmastectomy RT, so that 271 treatment courses in 268 patients were analyzed. Seventy-nine women (29%) were premenopausal, 173 (65%) were postmenopausal, 14 (5%) were perimenopausal,
Reprint requests to: Nancy Price Mendenhall, M.D., Department of Radiation Oncology, University of Florida Health Science Center, 2000 SW Archer Rd., P.O. Box 100385, Gainesville, FL 32610-0385. Tel: (352) 265-0287; Fax: (352) 265-0759; E-mail: [email protected]
Poster presentation at the 86th Scientific Assembly and Annual Meeting of the Radiological Society of North America, Chicago, IL, 2000. Received Sep 10, 2002, and in revised form Jan 2, 2003. Accepted for publication Jan 10, 2003.
METHODS AND MATERIALS
716
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
717
Table 1. Patient distribution according to stage and axillary status Positive nodes T stage
0
1–3
4–9
ⱖ10
Unknown (⬎0)
Total cases* (n)
T1 T2 T3 T4 Unknown Total
23 26 10 8 1 68
17 33 11 11 1 73
18 28 5 5 5 61
12 22 14 15 0 63
2 1 0 0 0 3
72 110 40 39 7 268
Data presented as the number of patients. * Excludes 2 patients with no axillary lymph nodes found in pathologic specimen and 1 with 4 positive lymph nodes but no tumor at primary site.
and 2 (1%) were of unknown menopausal status. The median age was 57 years (range 27– 84). The left side was treated in 143 cases and the right side in 128 cases. All pathology specimens, including outside biopsy or mastectomy specimens, were reviewed at Shands Teaching Hospital. Close margins were defined as ⱕ2 mm from the inked margin. Close (n ⫽ 33) or positive (n ⫽ 16) deep chest wall margins were present in 49 patients. The 1997 American Joint Committee on Cancer (12) guidelines were used for staging. The distribution by stage for the 271 cancers was as follows: Stage I, 23 (8%); Stage II, 154 (57%); and Stage III, 86 (32%). The stage could not be determined in 8 cases (3%). The distribution by T stage and number of involved lymph nodes are shown in Table 1. Of the 268 patients, 70 had 0, 73 had 1–3, 62 had 4 –9, and 63 patients had ⱖ10 positive lymph nodes. Three additional patients had positive lymph nodes, although the exact number of involved and retrieved lymph nodes was not stated in the pathology report. Follow-up The median follow-up among living patients was 8 years (range 2–21). Sixty-two percent of the patients had ⬎5 years of follow-up, and 30% had ⬎10 years of follow-up. The data were obtained from the RT chart, the hospital chart, on-line medical records (when available), and interviews of patients and/or family members. Surgery Most patients were treated with either a modified radical mastectomy (including Level I and II axillary node dissection) or a simple mastectomy after attempted breast-conserving surgery and axillary dissection. Radical mastectomy was performed in 5 patients. The median number of lymph nodes identified by the pathologist was 18, ranging from 0 (2 patients) to 60 (1 patient). Of 268 patients, 29 (10%) had ⬍10 lymph nodes in the surgical specimen. Chemotherapy Chemotherapy was administered to 159 patients. Ninetytwo patients received a doxorubicin-based regimen with (51
patients) or without (41 patients) additional treatment with cyclophosphamide, methotrexate, and fluorouracil (CMF). Of the 51 patients who received doxorubicin in addition to CMF, 42 had RT delivered concurrently with CMF, in contrast to 40 of 41 patients treated with doxorubicin regimens alone whose RT followed the chemotherapy sequentially. Only 1 patient received concomitant doxorubicin and RT. Fifty-nine patients received CMF alone (concomitant with RT in 37 patients, sequential in 22 patients). Overall, RT was delivered sequentially in 78 patients and concomitantly (during CMF in all cases except for 1) in the remaining 81 patients. Details regarding the exact systemic regimen were unavailable in 7 patients. Hormonal therapy Hormonal therapy was given to 102 patients, including tamoxifen (101 patients) and megestrol acetate (1 patient). Forty-six patients were treated with tamoxifen alone; 56 received chemotherapy in addition to hormonal therapy. Hormonal therapy was generally started after RT completion. Radiotherapy Indications. The indications used for field selection in postmastectomy RT during this time (13) are shown in Table 2. The treatment of the internal mammary chain (IMC) nodes and supraclavicular fossa (SCF), referred to as peripheral lymphatic irradiation (PLI), was used for patients with medial T1–2N0 or T1–2N1bi cancers with 1–3 involved lymph nodes before 1994. Patients with larger tumors (T3–T4) and/or ⬎3 positive axillary lymph nodes received chest wall RT as well. After 1994, all patients with any positive lymph nodes received chest wall, SCF, and IMC node RT. The entire axilla was treated with an anterior SCF field extended laterally and inferiorly and a posterior axillary field in patients with axillary lymph nodes ⬎2 cm, extensive extracapsular extension, clinical matted adenopathy, or an incomplete axillary dissection. The chest wall was treated in all patients with close or positive margins. Field design. The fields were designed on the basis of surface anatomy. CT treatment planning was performed in
718
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
Table 2. Indications for postmastectomy radiotherapy Stage
N0
N1bi
N1bii
N1biii-iv or N2
T1 medial T2 medial T3 T4
IMC PLI PLI ⫹ CW PLI ⫹ CW
PLI PLI PLI ⫹ CW PLI ⫹ CW
PLI ⫹ CW PLI ⫹ CW PLI ⫹ CW PLI ⫹ CW
PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX PLI ⫹ CW ⫹ AX
Abbreviations: IMC ⫽ internal mammary chain; PLI ⫽ peripheral lymphatic irradiation (internal mammary chain and supraclavicular fossa); CW ⫽ chest wall; AX ⫽ axillary. Modified, with permission, from Mendenhall et al. (13).
only 16 patients. Fluoroscopy and simulation films were used to confirm placement of the IMC field over the first three rib interspaces, inclusion of the coracoid process and/or surgical clips within the SCF and axillary field border, and inclusion of the clavicle and a margin of lung inside the rib cage within the axillary field borders. The IMC nodes were treated with an en face electron field. The medial border of the IMC field was placed on a line drawn from 1 cm across the midline superiorly to the midline inferiorly (Fig. 1). The lateral border of the field was 6 cm lateral to the midline. The IMC field was thus 7 cm wide superiorly and 6 cm wide inferiorly. With this design, the sternal– costal junction and the parasternal tissue that harbors the IMC nodes were always centered in the IMC field. The inferior border was placed on the fourth rib to encompass the first three interspaces. All six interspaces were included if the primary breast lesion was located in the inferior and medial quadrants. The superior border of the IMC field abutted the SCF. The inferior border of the SCF included the first rib for PLI and the second rib when the whole axilla was treated. The lateral border of the SCF field included any clips the surgeon placed at the top of the axillary dissection and coracoid process. If treatment was planned to the whole axilla (Fig. 2), the lateral border of the SCF was extended to the mid humerus. The medial border of the SCF field on the skin extended from 1 cm across the midline inferiorly to the cricothyroid ligament superiorly. The lateral superior border flashed across the trapezius muscle in most cases. The gantry on the SCF field was angled 15° to reduce esophageal and spinal cord exposure. The chest wall fields abutted the IMC and SCF fields on the skin and included a minimum of a 5-cm margin below the mastectomy scar with at least 2-cm margins medial and lateral to the mastectomy incision. The chest wall field was divided into medial and lateral fields on the basis of the contour of the chest wall. The chest wall fields were matched on the skin surface, producing an area of inhomogeneity with increased dosage at the skin surface. Junctions of abutting electron fields were changed once during treatment to reduce the dose inhomogeneity. Dose and dose specification. The usual dose was 50 Gy within 5 weeks at 2 Gy/fraction given 5 d/wk. Before 1982, 17 patients received 50 Gy within 4 weeks (2.5 Gy/fraction). An additional boost of 10 Gy in 5 fractions with 6- or
8-MeV electrons was given to the mastectomy incision; the boost dose was increased to 15 or 20 Gy in patients with close or positive mastectomy margins. The depth of the inner table of the sternum was presumed
Fig. 1. Standard field setup using electron beam technique. Reprinted with permission from Mendenhall et al. (13).
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
719
Fig. 2. Standard extension of SCF field used to treat entire axilla and low neck in a patient with more than N1bi axillary disease (opposed posterior axillary boost field not shown.)
to reflect the maximal potential depth of the internal mammary nodes. This distance was determined by placing a radiopaque marker on the skin surface and measuring the distance between the skin and inner table of the sternum on a cross-table lateral film. The electron beam energy for the IMC field was then selected to include the maximal depth of the inner table of the sternum within the 90% isodose volume and to provide a steep dose gradient to underlying tissues to minimize the dose to the heart and/or lung. In most patients, the depth of the inner table of the sternum was 2– 4 cm. Consequently, the IMC field in most patients was treated with 10 –14-MeV electrons. When ⱖ14 MeV electrons were used, no bolus was applied to the skin because of the increased skin dose with the higher energy electrons; in patients with internal mammary node depths of ⱖ4 cm, 20% of the dose was occasionally delivered with 6-MV photons to increase the coverage at depth and decrease the skin dose. The target volume for the chest wall fields was considered the surgical flap, not the entire thickness of the chest wall. In most cases, the chest wall flap was estimated to be between 1 and 2 cm thick. The determination of electron energy was based on the clinical assessment. Tissue-equivalent bolus material (0.5 cm) was applied to the chest wall to increase the skin dose to 90% of the prescription dose. Consequently, 6-MeV electrons were selected for most slender patients, which produced 90% isodose coverage at a 1.8 cm depth (including the bolus), thus covering chest wall flaps up to 1.3 cm thick. Larger patients were treated with 8-MeV electrons, which produced 90% isodose coverage at approximately 2.3 cm depth (including the bolus), thus covering chest wall flaps up to 1.8 cm thick within the 90% isodose shell. Higher electron energies were used occasion-
ally according to physician preference to treat the entire chest wall thickness. Only 4 patients had chest wall treatment with ⬎10 MeV electrons: 12 MeV (n ⫽ 1) and 14 MeV (n ⫽ 3). The dose to the anterior SCF field was delivered with photons (frequently with a 20% electron component). A dose of 50 Gy was prescribed to the depth of the maximal dose deposition, resulting in an actual target dose of 45–50 Gy. When the entire axilla was judged to be at risk, the SCF field was extended laterally and inferiorly, and a posterior field was added to increase the dose at the midplane of the axilla to 45–50 Gy. The posterior field was usually treated with 20-MV photons. This meant that the highest risk tissues (including the axillary apex), which were anterior to the midplane of the axilla and included in the posterior field, actually received ⬎45–50 Gy, in some cases up to 60 Gy. The inferior border of the SCF and posterior axillary boost fields abutted the superior borders of the medial and lateral chest wall fields. The region of low axillary nodes (which were resected) was frequently included in superior portion of the lateral chest wall field, rather than the SCF field, if the electron energy was judged adequate for coverage (Fig. 2).
Statistical analysis All statistical analyses were performed using SAS software (14). The end points of the study included LRR, any relapse, death, and other events that were possibly related to treatment. LRR was calculated directly as the crude number of events over the number of patients at risk, as well as the rate projected over time by the product-limit method (15). LRR was defined as the appearance of tumor on the chest wall or in the ipsilateral supraclavicular, axillary, or internal
720
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
Table 3. Ten-year freedom from locoregional recurrence, freedom from relapse, and absolute survival rates according to T stage, number of positive axillary nodes, and overall stage Variable T stage (n ⫽ 266) Tx–T1 T2 T3 T4 p Positive axillary nodes (n ⫽ 268) 0 1–3 ⬎3 p Overall stage (n ⫽ 260) I IIa IIb IIIa IIIb p
Locoregional control (%)
Freedom from relapse (%)
Absolute survival (%)
97 86 87 88 0.2351
75 60 47 54 0.0115
67 52 49 43 0.0029
91 95 86 0.1520
73 75 45 0.0006
63 69 41 0.0050
100 94 88 89 88 0.7380
95 66 63 43 53 0.0003
89 56 59 32 41 0.0001
mammary lymph nodes before, simultaneous with, or after the development of distant metastases. We considered the occurrence of distant metastasis and LRR within 1 month of each other as simultaneous events. For disease-free survival, any relapse was considered an event and patients were censored from analysis if they died without relapse or evidence of disease. The only event counted for absolute survival was death. At the time of death, patients with no LRR were censored from the actuarial calculation of LRR probability. The follow-up time was calculated from the time of mastectomy. Multivariate analyses of potential prognostic variables for chest wall recurrence, locoregional control, disease-free survival, and absolute survival were performed using a Cox regression backward selection procedure (16). Potential prognostic variables in the analysis of chest wall recurrences included the number of lymph nodes involved with metastatic disease (0, 1–3, or ⬎3), pathologic T stage, margin status (negative, close, or microscopically positive), and electron energy used to treat the chest wall (none, 6 MeV, 8 MeV, 10 MeV, or other). Factors with potential prognostic value in the analysis of locoregional control included the number of lymph nodes involved with metastatic disease, pathologic overall stage, pathologic T stage, fields treated (PLI, PLI ⫹ chest wall, or PLI ⫹ chest wall ⫹ axillary), systemic therapy (concurrent chemotherapy, sequential chemotherapy, tamoxifen alone, or no systemic therapy), menopausal status (pre-, peri-, or postmenopausal), and margin status. Factors assessed for potential prognostic value in predicting disease-free survival and absolute survival included the number of lymph nodes involved with metastatic disease, pathologic stage, pathologic T stage, fields treated, addition of chemotherapy, menopausal status, and age (⬎50 or ⬍50 years).
RESULTS Absolute survival The absolute survival rate at 5, 10, and 15 years was 72%, 55%, and 40%, respectively. Multivariate analysis of factors potentially predictive of absolute survival revealed the following to be strongly correlated with survival: pathologic T stage (p ⬍0.0001), addition of chemotherapy (p ⫽ 0.0013), and number of lymph nodes involved with metastatic disease (p ⫽ 0.007). The 10-year survival rates as a function of T stage, overall stage, and number of lymph nodes involved are shown in Table 3. Of the 268 patients, 127 have died— 85 of breast cancer, 42 of intercurrent disease, and 0 of treatment complications. Of the 13 patients who developed subsequent cardiac disease, no difference was found in the incidence between patients with right or left-sided breast cancer (p ⫽ 0.1777). Disease-free survival The disease-free survival rate at 5, 10, and 15 years was 71%, 62%, and 59%, respectively. Multivariate analysis of potentially predictive variables for disease-free survival revealed the following to be significantly associated with disease-free survival: number of lymph nodes involved with metastatic disease (p ⫽ 0.0008) and pathologic T stage (p ⫽ 0.002). The relationship of 10-year freedom from locoregional recurrence, disease-free survival, and overall survival with T stage, overall stage, and number of involved lymph nodes is shown in Table 3. Locoregional control The rate of locoregional disease control at 5, 10, and 15 years was 92%, 90%, and 90%, respectively. Patterns of failure. LRR developed in 25 patients, includ-
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
721
Table 4. Sites of locoregional recurrence by treatment field Recurrence site
CW only (n ⫽ 4)
PLI only (n ⫽ 27)
CW ⫹ PLI (n ⫽ 171)
CW ⫹ PLI ⫹ AX (n ⫽ 66)
Local—in boost field Local—marginal to boost field Local—regional outside RT field Regional—axillary Regional—supraclavicular Regional—internal mammary Regional—other Multiple sites Total
0 0 0 0 0 0 0 0 0 (0)
0 0 3 0 0 0 0 0 3 (11)
6 0 3 1 3 0 1 3* 17 (10)
0 0 3 1 1 0 0 1† 6 (9)
Abbreviations as in Table 2. Numbers in parentheses are percentages. * One local (in boost field) and regional (internal mammary); one local (other) and regional (axillary); and one local (other) and regional (other). † One local (in boost field) and regional (internal mammary).
ing isolated LRR (14 patients), simultaneous LRR and distant metastasis (8 patients), and LRR after distant metastasis (3 patients). Thirteen of the 14 isolated LRRs were confined to either the chest wall alone (6 patients) or the regional lymph nodes (7 patients); one LRR involved both the chest wall and regional lymph nodes simultaneously. Two-thirds of LRRs (16 of 25) were detected within 2 years, and almost 88% (22 of 25) were detected within 3 years. The relationship between the patterns of LRR and fields selected for RT is shown in Table 4. Chest wall recurrences. There were 19 recurrences on the chest wall, including 3 in the untreated chest walls of patients receiving PLI alone. In most cases (194 treatment courses), the chest wall was treated with electron energy to cover only the surgical flap (i.e., the skin and subcutaneous tissue). By physician preference, higher energy electrons were used to treat the entire chest wall thickness in 50 cases. No correlation was found between the electron energy used and the T stage, number of positive nodes, or pathologic stage, indicating physician preference rather than disease extent as the motivation to treat the entire chest wall thickness rather than the surgical flap. The infield LRR rate at 10 years was 4%, 5%, and 15% for patients treated with 6, 8, and 10 MeV electrons, respectively. Multivariate analysis of factors potentially predictive of chest wall recurrences identified no significant correlations: energy used to treat the chest wall (p ⫽ 0.50), pathologic T stage (p ⫽ 0.18), number of lymph nodes involved with metastatic disease (p 0.63), and surgical margin status (p ⫽ 0.48). Because patients with positive margins, extensive nodal disease, and/or large tumor were at a greater risk of chest wall recurrence, the lack of correlation suggests efficacy of the therapeutic policy. No evidence of greater chest wall control rates with the use of higher energy electrons was seen, suggesting that coverage of the full thickness of the chest wall was not generally necessary. Only three chest wall failures occurred in the 49 patients with close or positive margins (Table 5). No chest wall
failures occurred in the 8 patients who received a boost of 15–20 Gy to the incision dose. Chest wall recurrence did occur in 2 of 8 patients with close margins not treated with a boost to the surgical scar. Both recurrences were located in the scar and would have been covered in a boost field. One chest wall failure outside the boost field occurred in the 33 patients who received a boost of 10 Gy to the surgical scar. Regional lymph node recurrence. Regional lymph node recurrences were detected in 12 patients. The SCF was the most common site of clinically detected regional node treatment failure (7 patients) followed by axilla (4 patients), and the IMC region (4 patients). Three of four IMC recurrences were detected either simultaneously with SCF recurrence (2 patients) or after LRR (1 patient). All occurred after treatment of the IMC field to 50 Gy in 25 fractions with 12-, 14-, or 15-MeV electrons or a combination of 15-MeV electrons and 20-MV photons (weighted 4:1 in favor of electrons). No significant difference was seen in the supraclavicular or axillary lymph node recurrence rate as a function of the extent of the axillary dissection. The SCF recurrences occurred in patients with a median of 9 positive nodes (range 0 –17) at mastectomy. One patient who developed an SCF recurrence had no axillary lymph nodes identified in the axillary dissection. The SCF recurrences were located primarily in the medial part of the SCF. A review of the SCF port films showed that all failures were located within the
Table 5. Relationship between chest wall recurrence and radiation boost dose (after 50 Gy to chest wall) to mastectomy incision in patients with close or positive mastectomy margins Boost dose to incision (Gy) 0 10 15–20
Chest wall recurrences/Treated (n) 2/8 1/33 0/8
722
I. J. Radiation Oncology
● Biology ● Physics
Volume 56, Number 3, 2003
wall recurrence than patients additionally treated with a chest wall field. All these patients had T2N0 disease and developed LRR on the chest wall outside the PLI fields. Twenty-two of 244 patients treated to PLI and chest wall with or without axillary fields had a LRR, 19 of whom had recurrence within the treatment volume. Two patients had recurrence in the chest wall outside the treatment fields. One patient, who had indications for full axillary treatment (lymph node ⬎2 cm), had a recurrence in the untreated axilla. The overall rates of freedom from LRR at 10 years according to the extent of RT were as follows: PLI only, 88%; PLI and chest wall, 89%; and PLI plus chest wall plus axillary, 94%. Because only 4 cases were in the chest wall-only treatment group, this group was not included in this stratification. Prognostic factors for LRR. Multivariate analysis of factors potentially influencing the probability of locoregional disease control revealed the following factors to be significantly associated with overall LRR: T stage (p ⫽ 0.03), number of lymph nodes involved with metastatic disease (p ⫽ 0.03), and fields treated (p ⫽ 0.02). Factors not significantly associated with LRR in this study included pathologic stage (p ⫽ 0.08), addition of chemotherapy (p ⫽ 0.90), menopausal status (p ⫽ 0.90), and margin status (p ⫽ 0.32).
Fig. 3. Example of recurrence in surgical scar on unirradiated chest wall.
irradiated field. Both patients who had no nodal tissue identified in the axillary dissection had a recurrence in the regional lymph nodes, in the axilla in 1 patient and in the SCF in 1 patient. The number of lymph nodes recovered in the axillary dissection did not correlate with rate of supraclavicular or axillary failure (1 of 35 in patients with 1–9 nodes removed, 3 of 141 in patients with 10 –19 nodes removed, 5 of 132 in patients with ⱖ19 nodes removed, and 1 of 8 in patients with an unknown number of nodes removed). Adjuvant therapy. The actuarial incidence of LRR at 10 years as a function of the use and sequence of adjuvant therapy was as follows: RT alone (16%); RT followed by tamoxifen (5%); concomitant RT and chemotherapy (8%); and sequential RT and chemotherapy (11%). Field selection. The 10-year actuarial LRR rate was 12% for patients who received only PLI compared with 10% for patients who received chest wall RT and PLI. The three instances of LRR in the group receiving PLI alone (27 total) occurred in the untreated chest wall (Fig. 3). Patients treated with PLI alone were considered to be at lower risk of chest
Morbidity Lymphedema developed in 55 (21%) of 268 patients. Most (n ⫽ 33, 60%) had mild or asymptomatic edema, which was recorded if the patient complained of swelling or if visible edema was present that measured ⱕ2 cm in difference between the two sides. Nine (3%) of 268 patients developed symptomatic radiation pneumonitis. The incidence of pneumonitis appeared to correlate with treatment of an SCF field and with the use of higher electron energies to treat the chest wall field. When the SCF was treated, the incidence of symptomatic pneumonitis was 3.6% (9 of 247 patients) compared with 0% (0 of 24 patients) when the SCF was not treated. The incidence of pneumonitis increased with the use of higher energies on the chest wall: 1.6% with 6 MeV, 2.2% with 8 MeV, and 8% with 10-MeV electrons. One patient in our series developed brachial plexus injury (for an incidence of 1 of 268). The SCF field in this patient was treated to 50 Gy in 20 fractions at 2.5 Gy/fraction. An axillary field was used to increase the dose to the midplane to 50 Gy. The actual dose and dose rate to the brachial plexus was estimated to be 15–20% higher than the prescription dose. Asymptomatic telangiectasia was noted along the incision boost field and along the matchline between the medial and lateral chest wall fields in approximately one-third of patients. Telangiectasia outside the matchline or boost field was rarely noted.
Postmastectomy RT using electrons
● S. J. FEIGENBERG et al.
723
Fig. 4. CT scans of 32-year-old woman with a 4.5-cm upper inner quadrant breast cancer with 3 of 16 nodes positive at modified radical mastectomy. Two months after mastectomy before beginning chemotherapy, the new presence of a parasternal mass was noted. (a) CT confirmed the internal mammary adenopathy, as well as (b) Rotter’s nodes and axillary apical nodes seen on the scan but not clinically detected.
DISCUSSION The recent postmastectomy RT trials (1–3) have secured the role for postmastectomy RT in patients at risk of residual subclinical disease. Several questions remain to be answered, however, including which subsets of patients have insufficient risks to warrant RT, which fields must be treated to obtain the efficacy documented in the recent landmark trials, and what are the optimal doses and techniques to achieve the best therapeutic ratio. It was the policy in the present study to use postmastectomy RT in all node-positive patients, all patients with T3 primaries, all patients with medial or central T2 primaries, and all patients with close or positive chest wall margins. The treatment fields varied according to the indication for postmastectomy RT. Before 1994, patients with T2N0 medial or central tumors or 1–3 positive axillary lymph nodes received RT to the IMC and SCF; after 1994, these patients received chest wall, IMC, and SCF RT. Any patient with a close or positive surgical margin received chest wall RT. Any patients with either a T3 tumor or ⬎3 positive lymph nodes received chest wall, SCF, and IMC nodal RT. Most patients with axillary lymph nodes that were matted or ⬎2 cm, had extracapsular extension, or had incomplete dissections also received full axillary RT with an extended supraclavicular field and a supplemental posterior axillary boost. This study of patterns of recurrence showed a very low rate of clinically detectable LRR. Several observations may be made from this experience that shed light on the optimal field selection, treatment technique, and dose prescription. It is currently generally accepted that the chest wall is at risk in patients with ⱖ4 positive nodes. Treatment is warranted because the most common site of clinically detected LRR after mastectomy is on the chest wall. It is unclear whether the chest wall requires treatment when few or no axillary lymph nodes are involved and the tumor size is ⬍5
cm. In this study, 27 patients were treated with IMC and SCF RT only, including 12 with T2N0 medial tumors. The overall chest recurrence rate after IMC and SCF RT alone was 12%, but all three chest wall recurrences were among the 12 T2N0 patients. These patients were all operated on by experienced surgical oncologists, and no particular concern was indicated by the surgeon regarding a risk of recurrence in the chest wall. None of these patients had close or positive chest wall margins. The numbers were small, but the risk for chest wall recurrence even with T2N0 breast cancer does appear to be substantial. Another controversial issue is whether the entire axilla needs RT. The question is of more than academic interest, because it is well known that the extent of both surgery and postmastectomy RT contribute to the risk of arm edema. In this study, the full axilla was treated only in the setting of indications listed above. Occasionally, concern about arm edema precluded delivery of axillary RT even with these indications. On multivariate analysis of the clinical and treatment factors predicting a higher rate of LRR, patients who received full axillary RT had a lower rate of clinically detected LRR than those who did not receive full axillary RT (p ⫽ 0.02). This is, at first, a curious finding in that there were only four clinically detected axillary recurrences in this study. The axilla, however, can be a clinically silent area. Uncontrolled disease in the axilla may precede clinically detectable LRR in other sites such as the SCF, chest wall, or even internal mammary nodes (Fig. 4). Because patients did not routinely have follow-up CT imaging in this study, uncontrolled disease in the axilla could easily have been overlooked. We suspect that recurrences in the apex of the axilla were underestimated in this and other studies and that the finding that axillary RT is associated with better disease control is real. Given the finding that whole axillary RT appeared to be associated with better locoregional disease control, the method of dose prescription used in this study is of interest.
724
I. J. Radiation Oncology
● Biology ● Physics
After 50 Gy was delivered to the anterior extended supraclavicular field, prescribed at the maximal depth of dose deposition for whatever beam was used, the actual dose delivered at the midplane of the axilla was calculated and then supplemented with a posterior axillary boost field. The supplemental posterior axillary boost field exited through the axillary apical tissues anterior to the midplane. These anterior tissues had already received a higher dose of RT from the anterior extended supraclavicular field than the tissues at the midplane. Retrospective dose analyses suggested that those tissues anterior to the midplane generally received 55– 60 Gy when the entire axilla was treated and a posterior field was used to bring the midplane axillary dose to 50 Gy. The apex of the axilla after mastectomy may be the nodal area at highest risk of residual subclinical disease in node-positive patients, both because of the orderly fashion in which breast cancer spreads to the lymphatics and because lymphatic flow through the axillary lymphatics is more common than through the IMC. Other studies have demonstrated a dose–response relationship for subclinical disease in breast cancer (17). The European Organization for the Research and Treatment of Cancer demonstrated in patients undergoing breast-conserving therapy that even after lumpectomy with negative surgical margins, the addition of a boost to the tumor bed to a total dose of 66 Gy resulted in a significantly lower local recurrence rate than did a total dose of 50 Gy (17). It is conceivable that treatment of the entire axilla was a critically important part of the postmastectomy RT techniques used in the recent landmark trials (1–3) that finally demonstrated a survival benefit associated with postmastectomy RT. It is also conceivable that the actual dose associated with successful RT of subclinical disease in this very high-risk site is 10 –15% greater than that actually prescribed and reported. The final finding of interest is the pattern of recurrence in the 49 patients with close or positive mastectomy margins. As
Volume 56, Number 3, 2003
noted previously, no recurrences developed in 8 such patients when a boost dose of 15–20 Gy was given to the mastectomy incision after the initial 50 Gy to the chest wall. One recurrence developed outside the boost field in the 33 patients who received a 10-Gy boost, and two recurrences developed in the mastectomy incision among the 8 patients in whom no boost was given after the initial 50 Gy. Although the numbers were small, there was a suggestion that when the mastectomy margins are close or positive, areas on the chest wall, including the mastectomy incision, are at very high risk of residual disease and would best be treated with doses ⬎50 Gy. With 10 –20 years of follow-up, no apparent excess in cardiac deaths occurred in patients with left-sided breast cancers in this series, suggesting that the goal of minimizing radiation exposure to the heart may have been achieved with this technique. Additional follow-up is necessary, however, to ensure no long-term impact on cardiac morbidity. CONCLUSION This report documented a very low rate of clinically detected LRR associated with the electron beam technique in the guidelines used for postmastectomy RT at the University of Florida. The findings in this study of the patterns of recurrence suggest several possible conclusions. First, chest wall treatment may be beneficial even with T2N0 tumors. Second, full axillary RT may be beneficial, at least in all patients with more than N1Bi disease. Third, the optimal dose to very high-risk areas, such as a mastectomy incision in patients with close or positive mastectomy margins and the apex of the axilla, may be ⬎50 Gy. Finally, we recommend that CT staging be done before delivery of adjuvant chemotherapy and RT and at the time of suspected treatment failure to better understand the true pattern of LRR and its relationship to distant metastases. This was a retrospective study, and all findings and possible conclusions suggested here should be confirmed in a larger controlled study.
REFERENCES 1. Overgaard M, Hansen PS, Overgaard J, et al. Danish Breast Cancer Cooperative Group 82b trial: Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. N Engl J Med 1997;337: 949–955. 2. Overgaard M, Jensen MB, Overgaard J, et al. Postoperative radiotherapy in high-risk postmenopausal breast-cancer patients given adjuvant tamoxifen: Danish Breast Cancer Cooperative Group DBCG 82c randomised trial. Lancet 1999;353: 1641–1648. 3. Ragaz J, Jackson SM, Le N, et al. Adjuvant radiotherapy and chemotherapy in node-positive premenopausal women with breast cancer. N Engl J Med 1997;337:956–962. 4. 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;333:1444–1455. 5. Cuzick J, Stewart H, Rutqvist L, et al. Cause-specific mortality in long-term survivors of breast cancer who participated in trials of radiotherapy. J Clin Oncol 1994;12:447–453.
6. Rutqvist LE, Johansson H. Mortality by laterality of the primary tumour among 55,000 breast cancer patients from the Swedish Cancer Registry. Br J Cancer 1990;61:866–868. 7. Paszat LF, Mackillop WJ, Groome PA, et al. Mortality from myocardial infarction after adjuvant radiotherapy for breast cancer in the Surveillance, Epidemiology, and End-Results cancer registries. J Clin Oncol 1998;16:2625–2631. 8. Host H, Brennhovd IO, Loeb M. Postoperative radiotherapy in breast cancer—Long-term results from the Oslo study. Int J Radiat Oncol Biol Phys 1986;12:727–732. 9. Rutqvist LE, Lax I, Fornander T, et al. Cardiovascular mortality in a randomized trial of adjuvant radiation therapy versus surgery alone in primary breast cancer. Int J Radiat Oncol Biol Phys 1992;22:887–896. 10. Gagliardi G, Lax I, Ottolenghi A, et al. Long-term cardiac mortality after radiotherapy of breast cancer—Application of the relative seriality model. Br J Radiol 1996;69:839–846. 11. Gyenes G, Fornander T, Carlens P, et al. Morbidity of ischemic heart disease in early breast cancer 15–20 years after
Postmastectomy RT using electrons
adjuvant radiotherapy. Int J Radiat Oncol Biol Phys 1994;28: 1235–1241. 12. American Joint Committee on Cancer. AJCC cancer staging manual. 5th ed. Philadelphia: Lippincott Williams & Wilkins; 1997. 13. Mendenhall NP, Fletcher GH, Million RR. Adjuvant radiation therapy following modified radical or radical mastectomy. In: Bland KI, Copeland EM, III, editors. The breast: Comprehensive management of benign and malignant diseases, 1st ed. Philadelphia: WB Saunders; 1991. p, 770–778.
● S. J. FEIGENBERG et al.
725
14. SAS Institute Inc. SAS OnlineDoc(R), Version 8. Cary, NC: SAS Institute Inc.; 1999. 15. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457–481. 16. Cox DR. Regression models and life tables—Series B (Methodological). J R Stat Soc 1972;34:187–220. 17. Bartelink H, Poortmans P, et al. Recurrence rates after treatment of breast cancer with standard radiotherapy with or without additional radiation [Abstract]. N Engl J Med 2001; 345:1378–1387.