Long-term outcomes in breast cancer patients undergoing immediate 2-stage expander/implant reconstruction and postmastectomy radiation




Breast reconstruction with tissue expander (TE)/permanent implant (PI) followed by postmastectomy radiation (PMRT) is an increasingly popular treatment for breast cancer patients. The long-term rates of permanent implant removal or replacement (PIRR) and clinical outcomes in patients treated with a uniform reconstructive surgery and radiation regimen were evaluated.


Between 1996 and 2006, 1639 patients with stage II-III breast cancer received modified radical mastectomy (MRM) at Memorial Sloan-Kettering Cancer Center. A total of 751 received TE placement at the time of mastectomy. Of these, 151 patients went on to receive chemotherapy and exchange of the TE for a permanent implant, followed by PMRT. Clinical outcomes and PIRR-free rates were estimated by Kaplan-Meier methods. Cox regression model was used to examine patient, disease, and treatment characteristics associated with PIRR.


Median follow-up was 86 months (range, 11-161 months). The 7-year PIRR-free rate was 71% (38 PIRRs in 35 patients). The 7-year rate of PI replacement was 17.1% (21), and removal was 13.3% (17). Reasons for PIRR included infection (15); implant extrusion, shift, leak, or rupture (4); patient request (1), or multifactorial (17). On univariate analysis, no factor was significantly associated with PIRR. Two patients experienced local recurrence in the chest wall, both after 7 years. The 7-year distant metastasis–free survival rate was 81% and overall survival 93%.


Favorable 7-year PIRR rates and clinical outcomes were achieved in a sizable cohort of patients treated with homogeneous sequencing, radiation, and reconstructive surgery and lengthy follow-up. Factors predictive for high risk of PIRR were not identifiable in this population. Cancer 2012. © 2011 American Cancer Society.


Postmastectomy radiation in patients with 4 or more positive lymph nodes is widely accepted as an essential element of treatment for locally advanced breast cancer. An update of the seminal Early Breast Cancer Collaborative Trialist Group meta-analysis demonstrated that the survival benefit of postmastectomy radiation therapy (PMRT) was also applicable to a subset with 1 to 3 positive lymph nodes.1 Consequently, increasing use of PMRT in breast cancer patients with lymph node-positive disease is anticipated.

Enhancing quality of life is also an important goal for breast cancer survivors. Breast reconstruction offers aesthetic and psychological advantages that contribute significantly to patient satisfaction and quality of life.2 Many mastectomy patients are opting to receive implant-based breast reconstruction; a 2-stage tissue expander (TE)/permanent implant (PI) approach constituted 2/3 of the breast-reconstructive surgeries performed in the United States between 2008 and 2009.3 Knowledge of long-term oncologic outcomes and surgical complication rates is therefore relevant to thousands of women who are faced with the decision to receive implant reconstruction prior to radiation.

The interpretation of current literature on radiation and breast reconstruction is limited by small patient populations and heterogeneous treatment regimens, which has resulted in the reporting of conflicting outcomes. At our institution, patients who choose immediate implant reconstruction and require PMRT undergo the following treatment algorithm: 1) modified radical mastectomy with immediate placement of TE(s); 2) initiation of chemotherapy with expansion of TE(s) performed throughout treatment; 3) exchange of TE(s) for permanent implant after completion of chemotherapy; 4) initiation of postmastectomy radiation (Fig. 1).

Figure 1.

The Memorial Sloan-Kettering Cancer Center treatment algorithm for integration of immediate 2-stage tissue expander-permanent implant reconstruction and postmastectomy radiation is shown. ALND, axillary lymph node dissection; RT, radiation therapy.

Five-year clinical outcomes in a cohort of patients treated with this treatment algorithm has been previously detailed.4 The current study aimed to achieve the following objectives in a larger group of patients with prolonged follow-up: 1) evaluate the incidence of permanent implant removal or replacement (PIRR) in a cohort of patients treated at a single academic institution with a uniform reconstructive surgery and radiation regimen; 2) assess long-term clinical outcomes; 3) identify any clinical or pathologic characteristics associated with PIRR.


Between May 1996 and August 2006, 1639 patients with American Joint Committee on Cancer (AJCC) 6th edition stage II-III breast cancer underwent total mastectomy and axillary lymph node dissection at our institution. Of these, 751 received TE placement at the time of mastectomy. A subset of 260 patients who subsequently underwent PMRT was identified. We excluded 80 patients who received PMRT or chemotherapy at an outside institution, 5 patients who received PMRT subsequent to exchange for a permanent implant, 19 patients who underwent neoadjuvant chemotherapy, 1 patient with a prior history of breast malignancy, 1 patient with a history of prior thoracic radiation, and 3 patients in whom the TE was removed without subsequent replacement prior to radiation. This left 151 patients who comprised our study population. No patients were lost to follow-up. All patients received all components of their care at our institution under the treatment algorithm outlined above.


All patients underwent skin-sparing mastectomy of the affected breast with lymph node dissection. None received nipple-sparing mastectomies. Twelve patients of the entire cohort also had prophylactic mastectomy and expander/implant reconstruction of the contralateral breast. Sixty-three patients had sentinel node biopsies performed prior to axillary dissection. At the time of mastectomy, expanders were routinely placed in the subpectoral position. TE expansion was initiated 1 to 2 weeks after surgery and continued throughout the chemotherapy. Approximately 4 weeks after chemotherapy, patients received exchange of the TE for a PI. A full circumferential capsulotomy was performed in 69% of patients at the time of exchange. Reconstructive surgeries were performed by 3 plastic surgeons within our institution.

Systemic Therapy

All patients received adjuvant chemotherapy, 95% of which was anthracycline and taxane based. Other chemotherapy regimens included cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) or a combination of regimens. One hundred twenty-two patients (81%) received hormonal therapy, which was initiated after the completion of radiation.


Radiation treatment was initiated at 8 weeks from the date of completion of chemotherapy. All patients received 50 Gy in 25 to 28 fractions to the chest wall, implant, and regional lymph nodes. Internal mammary lymph nodes were not routinely included. No patients received a boost to the chest wall. Radiation was delivered to the chest wall and implant with 2 tangential photon beams matched to an anterior oblique field encompassing the supraclavicular fossa. Either 6 MV or 15 MV photons were used, depending on the medial-lateral separation of the patient. A 0.5-cm or 1.0-cm bolus over the chest wall was used on a daily basis.

Outcomes Analysis

The incidence of PIRR was retrospectively assessed through review of the patients' medical records, including annual follow-up that included physical examination by the plastic surgeon. Removal was defined as explantation of the implant. Replacement was defined as replacement of the implant for another implant, autologous breast reconstruction, or a combination. Reasons necessitating removal or replacement of the implant were captured from medical records and then individually confirmed with the treating plastic surgeon. Univariate analysis was used to examine the following characteristics and its potential association with PIRR: age (≤45 years vs >45 years), menopausal status (pre- or peri- vs postmenopausal), smoking history, T stage, N stage, extranodal extension, tumor histology (invasive ductal vs other), presence of vascular and perineural invasion, multifocal/multicentric tumor, margin status, estrogen receptor (ER) status, progesterone receptor (PR) status, her-2-neu status, chemotherapy regimen, use of hormone therapy, laterality of irradiated implant (right vs left), and implant type (saline vs silicone). Time to PIRR was calculated from end of radiation therapy to date of replacement or removal, whichever occurred first. PIRR-free rate at 7 years was estimated by Kaplan-Meier methods.

Clinical outcomes, including local, regional, and distant disease control, as well as survival, were examined. The length of follow-up was defined as the time from mastectomy to the date of local failure or last follow-up. Locoregional failure was defined as biopsy-proven tumor in the ipsilateral chest wall or axillary, supraclavicular, or internal mammary lymph nodes. Time to biopsy-proven distant metastasis was calculated from date of mastectomy to date of distant metastasis or last follow-up. Patients who died without a distant metastasis (DM) were censored (n = 3). Overall survival (OS) time was calculated from date of mastectomy to date of death or last follow-up. Time to local failure, DM-free survival, and OS were assessed by Kaplan-Meier methods.


Median follow-up, defined as date of mastectomy to death or last follow-up, was 86 months (range, 11-161 months). Median survival was not reached in this cohort. The median follow-up from end of radiation to last follow-up for those patients who did not experience PIRR was 73.5 months (range, 2-145 months).

Table 1 outlines patient and tumor characteristics. The median age of the patients was 44 years (range, 26-74 years). A positive smoking history was defined by patient self-reporting of smoking, whether active at the time of diagnosis or in the past (defined as >1 year from diagnosis). Among the 61 smokers, 40 were past and 21 were active smokers. The majority (75%) of patients were pre- or peri-menopausal. Thirty-one percent had AJCC 6th edition stage II breast cancer, and 69% had stage III. There were no cases of inflammatory breast cancer or patients with chest-wall invasion. Lymphovascular invasion was present in 60% of patients. Seventy-four percent of patients had poorly differentiated tumors. The median number of lymph nodes dissected was 25 (range, 4-50). Forty-four percent of patients had extranodal extension. Eighty percent of patients were estrogen receptor and/or progesterone receptor positive. Twenty-three (15%) patients had her-2-neu–amplified breast cancer, as defined by 3+ immunohistochemistry staining and/or amplification by fluorescent in situ hybridization. Ninety-four percent of patients had negative or close (<2 mm) surgical margins.

Table 1. Patient and Tumor Characteristics
  No.(%)   No.(%)
  1. ER, estrogen receptor; PR, progesterone receptor.

LateralityRight79(52) T StageT133(22)
 Left72(48)  T281(54)
Age at diagnosis≤4586(57)  T329(19)
 >4565(43)  T48(5)
Menopausal statusPre/peri114(75) N stageN05(3)
 Post37(25)  N153(35)
RaceCaucasian130(86)  N254(36)
 Black/Asian/other21(14)  N339(26)
     Histologic gradeWell1(1)
History of smokingYes61(40)  Moderate11(7)
 No90(60)  Poor112(74)
HistologyInvasive ductal107(71)  Unspecified27(18)
 Invasive lobular26(17) Nuclear gradeLow1(1)
 Mixed/other18(12)  Intermediate42(28)
Lymphovascular invasionPresent90(60)  High78(52)
 Absent58(38)  Unspecified30(20)
 Unspecified3(2) ER statusPositive120(79)
Perineural invasionPresent13(9)  Negative31(21)
 Absent138(91) PR statusPositive92(39)
Extranodal extensionPresent66(44)  Negative59(61)
 Absent85(56) Her-2-neu statusPositive23(15)
Multifocal/multicentricPresent72(48)  Negative111(74)
 Absent79(52)  Unknown17(11)
Margin statusPositive9(6)     
 Close (<2 mm)15(10)     

Table 2 describes radiation and systemic treatment characteristics. The internal mammary lymph nodes were intentionally targeted in 3% of patients. A posterior axillary boost was utilized in 15% of patients—all cases in which microscopic residual tumor in the axilla was suspected. Ninety-five percent (143) of patients received an anthracycline and taxane–based chemotherapy regimen. Five patients received CMF, and the remaining 3 patients received a combination of CMF and other regimens. All 120 ER+ patients and 2 ER− patients received hormonal therapy; 42 (28%) received tamoxifen alone, 17 (11%) aromatase inhibitors alone, and 63 (42%) a sequence of Nolvadex and an aromatase inhibitor. Seventeen of the 23 her-2-neu–positive patients received trastuzumab, which was continued throughout radiation for a total duration of 1 year. The six her-2-neu–positive patients who did not receive trastuzumab were treated in the early years of the study in which trastuzumab was not routinely administered for her-2-neu–positive patients.

Table 2. Treatment Characteristics
  1. AI, aromatase inhibitors; CMF, cyclophosphamide, methotrexate, and 5-fluorouracil; IMN, internal mammary node; PAB, posterior axillary boost.

Treatment of IMNsYes4(3)
Posterior axillary boostYes23(15)
Chemotherapy regimenAnthracycline and taxane based143(95)
Endocrine therapyTamoxifen42(28)
 Aromatase inhibitor17(11)
 Tamoxifen and aromatase inhibitor63(42)

The total number of DMs was 29. Three patients died without documented DM and were censored at date of death. The 7-year distant disease-free survival (DDFS) rate was 81%, and the 7-year overall survival rate was 93%. The 7-year local and regional control was 100% (Fig. 2A). Two patients experienced a local failure in the chest wall, both of which occurred >7 years after mastectomy. One developed DM at 3.7 years and a subsequent local failure in the chest wall and brachial plexus 7.2 years after mastectomy. The second patient developed local failure in the chest wall 7.7 years after mastectomy and did not have DM at last follow-up, which was 9.6 years after mastectomy.

Figure 2.

(A) A Kaplan-Meier curve demonstrates 7-year locoregional control, distant metastasis–free survival, and overall survival in study patients. Median follow-up from time of mastectomy was 86 months. (B) A Kaplan-Meier curve demonstrates time to permanent implant removal or replacement (PIRR). Median time to PIRR was not reached.

Time to PIRR was defined as time from the completion of radiation to the date of either replacement or removal of the implant, whichever occurred first. The 7-year PIRR-free rate was 71%, with 38 PIRRs (21 replacements and 17 removals) in 35 patients (Fig. 2B). Three patients underwent both removal and replacement of the implant. The interval between the 2 procedures ranged from 3 weeks to 9 months after radiation. Two patients had implant removal secondary to infection, followed by replacement with a latisimus flap reconstruction. One patient had the implant replaced with another implant due to Baker's grade 4 capsular contracture, followed by permanent removal secondary to implant extrusion. Two patients had 2 consecutive implant replacements: the first replacement with an implant, followed by the second replacement with an autologous reconstruction. None of the PIRRs were associated with the 2 patients who experienced locoregional recurrences. The 7-year rates of implant replacement and removal were 17.1% and 13.3%, respectively.

Table 3 delineates the time intervals at which the PIRRs occurred. The 2-year PIRR rate was 15%, whereas the 7-year PIRR rate was 29%, indicating that half of the PIRR events occurred within the first 2 years after the completion of radiation. The 2-year and 7-year rate of PI replacement was 8% and 17.1%, respectively, whereas the 2-year and 7-year rate of PI removal was 9% and 13.3%, respectively. Reasons for the 38 PIRR events are summarized as follows: 47% (17) of the PIRRs were attributable to multifactorial etiologies, defined as patient or physician dissatisfaction, suboptimal cosmesis, grade 3 or 4 capsular contracture, or a combination of these factors. The second most common cause of PIRR was infection (14), followed by implant extrusion (3). Mechanical issues with the implant such as leak (1), rupture (1), or shift (1) were less common reasons for PIRR. One patient requested removal of her implant secondary to desired symmetry after undergoing contralateral mastectomy without reconstruction. When reasons for the 17 removals and 21 replacements were examined separately (Fig. 3), multifactorial etiologies (17 of 21) was the most common reason for implant replacement, whereas infection (12 of 17) was the dominant etiology of implant removal. Of the 17 patients whose implants were replaced due to multifactorial etiologies, 7 had Baker's grade 3 and 10 had grade 4 capsular contracture. Sixty-two percent (13 of 21) of the implants were replaced by either another implant or combination of a latissimus flap and implant, whereas the remaining 38% consisted of autologous reconstructions (6 transverse rectus abdominis muscle, 1 latissimus, 1 deep inferior epigastric artery perforator flap).

Table 3. Time Intervals to First Permanent Implant Removal or Replacement
Permanent Implant EventYearTotal
  1. PIRR, permanent implant removal or replacement.

Figure 3.

Reasons for implant removal versus replacement are depicted.

On univariate analysis (Table 4), multifocal/multicentric tumor reached borderline significant association with PIRR (P = .058). Given the lack of statistical significance achieved by other examined clinical, pathologic, or treatment factors, multivariate analysis was not performed.

Table 4. Univariate Analysis of Clinicopathologic Features and Incidence of Permanent Implant Removal or Replacement
Patient CharacteristicsPTumor CharacteristicsPTreatment CharacteristicsP
  1. ER, estrogen receptor; LN, lymph node; LVI, lymphovascular invasion; PAB, posterior axillary boost; PR, progesterone receptor.

Age at diagnosis.39Histology.302Laterality.533
Menopausal status.287Histologic grade.136Implant type.911
Ethnicity.263LVI.164Photon energy.687
Smoking history.48T-size.772Receipt of PAB.58
  No. positive LN.09Chemo type.385
  ER.123Hormone therapy.67


The compatibility of PMRT and immediate breast reconstruction is a subject of controversy among breast cancer physicians. Some have hypothesized that immediate reconstruction can complicate the technical delivery of PMRT and oncologic outcomes in patients receiving radiation.5, 6 Radiation in patients with implant-based reconstructions has been associated with the development of increased surgical complications compared with unirradiated controls.7, 8

Existing studies of radiation and immediate reconstruction are confounded by small groups of patients with irradiated implants treated with heterogeneous radiotherapy and surgical sequencing and techniques.9-15 In addition, the median follow-ups of these studies were relatively short, and end points were variably defined. These differences led to a wide spectrum of reported complication rates ranging from 5% to 48%, underscoring the need for reliable data with lengthy follow-up in homogeneously treated patients who received immediate implant-based reconstruction and radiation. Our study examined clinical outcomes and PIRR rates in a sizable group of patients who were consecutively treated with uniform sequencing, radiation, and surgery techniques. The median follow-up of 86 months is the longest reported follow-up to date.

With the exception of 1 study that reported 1 locoregional recurrence in 93 patients,16 these studies of PMRT and immediate breast reconstruction have focused largely on cosmetic outcomes and complication rates, rather than disease control. Our results demonstrated excellent long-term local and regional control after PMRT, with 2 reported chest-wall failures, each of which occurred after 7 years. This mitigates the concern that immediate reconstruction may compromise clinical outcomes by impairing the effective delivery of radiation to the target volume.

The 7-year PIRR rate was 29% in our study. It is important to consider PIRR rates in context of the longevity of implants. Breast prostheses are not permanent devices and may require revision over the life course of patients, even in women who receive implants primarily for augmentation purposes. A study from the University of Michigan17 showed that implants provide less stable aesthetic satisfaction in the long term compared with autologous reconstructions. We viewed this incidence of PIRR as an acceptable rate of corrective surgery in patients who received PMRT after immediate reconstruction. This rate is similar to the 23% reconstruction failure rate reported by the recently published multicenter prospective French trial that included 141 patients and defined reconstruction failure as removal or replacement of the breast prosthesis.18 In contrast to our study, these patients received PMRT to the TE, with exchange for the permanent implant occurring after the completion of radiotherapy. Details that could impact surgical complications and contracture rates, such as the timing of exchange for a permanent implant and the use of capsulotomy at the time of exchange, were not reported.

The incidence of PI replacement and removal were also examined separately, as the ability to replace an implant with another reconstruction may not necessarily be deemed an implant “failure.” The 7-year replacement and removal rate was 17.1% and 13.3%, respectively. The implant removal rate is consistent with the 11% PI removal rate reported in a prior institutional study by Cordeiro et al, who studied outcomes in a population of 81 patients who received PMRT after TE/PI reconstruction with a mean follow-up period of 33 months.19 Although some of these patients are represented in our current study population, there is not complete overlap, as the prior report included patients who received radiation at other institutions, whereas our study did not.

Late complications have been found to be common in patients undergoing immediate implant-based reconstruction, whereas patients with delayed autologous tissue reconstruction experience more acute complications, such as infections, seromas, and impaired wound healing secondary to the detrimental effect of prior radiation.20 Capsular contracture is a well-recognized late complication of implant-based reconstruction and can take 5 to 7 years to develop.7 Although capsular contracture can occur in the absence of radiation in patients with implant reconstructions, radiotherapy is a known risk factor for increased capsular contracture rates.19, 21 The retrospective nature of our study limited the ability to distinguish between radiation-induced capsular contracture and other subjective reasons for PIRR that may impact the decision for implant replacement. Therefore, the category “multifactorial” was used also to describe etiologies for implant replacement other than capsular contracture, such as suboptimal cosmesis or patient or physician dissatisfaction, many of which are often interrelated.

The timing of the development of PIRRs in relation to radiotherapy warrants discussion (Table 3), with the caveat that the number of observed events was small. Approximately 1/3 of the PIRRs (13 of 38) occurred within the first year after the completion of PMRT. Seventy-seven percent (10 of 13) of these year 1 events consisted of implant removals secondary to infection. This is in accordance with the observation that infections are relatively early complications of breast prostheses,8 and when they occur can necessitate removal of the prosthesis. Half of the implant replacements (10 of 20) occurred by the end of year 2 after PMRT, all of which were attributable to contracture/other. Replacements due to multifactorial etiologies occurred steadily through year 8, which is consistent with the finding that the incidence of capsular contracture and patient dissatisfaction with cosmesis after implant reconstruction can increase with follow-up.13, 17, 18

Data regarding the impact of clinical factors on surgical complication rates have been variable. Some have described the influence on complication rates of treatment factors such as the use of systemic therapy and type of bolus, as well as patient and tumor-related factors such as smoking, positive lymph node status, and T3/T4 primary tumor size.10, 12-14, 18 Another study found no association between these factors and surgical outcomes.16 We were unable to identify any patient or treatment-related predictors of PIRR. Chemotherapy was received by all patients as part of the treatment algorithm, and the use of a daily 0.5-cm bolus over the chest wall has been standard practice for all mastectomy patients treated in our department during the years of the study. Therefore, these factors were not assessed as independent risk factors for reconstruction failure. Smoking has been cited as a predictor for postoperative complications after implant reconstruction in multiple studies.13, 18, 22 However, misclassification errors due to inconsistent patient reporting made retrospective evaluation difficult.

Several limitations of our study should be noted. Revisional surgeries for correction of fibrosis, pain, rippling, seroma, skin necrosis, or prophylaxis before impending implant exposure using regenerative tissue matrix to cover the implant were not captured, as implant removal or replacement were the most objective end points that could be assessed. Nevertheless, these details could be important, as the total number of revisional surgeries can also impact cost and quality of life. Furthermore, our results, which were generated from a single cancer center, may not be generalizable to other populations at large. They are likely not applicable to patients who receive neoadjuvant chemotherapy and require expedient initiation of radiation after mastectomy, leaving inadequate time for expansion and exchange of the expander for the permanent implant before radiation therapy.

Despite these shortcomings, we believe that this study documents reliable long-term outcomes in a large group of breast cancer patients accumulated over a decade, who received 2-stage expander/implant reconstruction with anthracycline and taxane–based chemotherapy followed by PMRT. These data have important implications for both the breast cancer physicians who counsel patients on the efficacy and complications of this approach, as well as the patients who choose to receive the treatment.


No specific funding was disclosed.


The authors made no disclosures.