Fax: (902) 494-2519
Immediate postmastectomy reconstruction is associated with improved breast cancer-specific survival
Evidence and new challenges from the Surveillance, Epidemiology, and End Results database
Article first published online: 24 JUL 2009
Copyright © 2009 American Cancer Society
Volume 115, Issue 20, pages 4648–4654, 15 October 2009
How to Cite
Bezuhly, M., Temple, C., Sigurdson, L. J., Davis, R. B., Flowerdew, G. and Cook, E. F. (2009), Immediate postmastectomy reconstruction is associated with improved breast cancer-specific survival. Cancer, 115: 4648–4654. doi: 10.1002/cncr.24511
- Issue published online: 5 OCT 2009
- Article first published online: 24 JUL 2009
- Manuscript Accepted: 24 FEB 2009
- Manuscript Revised: 17 FEB 2009
- Manuscript Received: 21 NOV 2008
- breast cancer;
- breast reconstruction;
Although immediate breast reconstruction is increasingly offered as part of postmastectomy psychosocial rehabilitation, concerns remain that it may delay adjuvant therapy or impair detection of local recurrence. No single population-based study has examined the relationship between immediate breast reconstruction and breast cancer-specific survival.
By using data from the US National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) registries, breast cancer-specific survival was compared for female unilateral mastectomy patients who did or did not undergo immediate breast reconstruction. Cox proportional hazards models were fitted, adjusting for known demographic and disease severity variables and stratifying on reconstruction type (implant or autologous) and age.
Improved breast cancer-specific survival was observed among all immediate breast reconstruction patients compared with patients who underwent mastectomy alone (hazard ratio [HR] = 0.74; 95% confidence interval [CI], 0.68 to 0.80). Implant reconstruction patients below 50 years of age demonstrated the greatest apparent survival benefit (HR = 0.47; 95% CI 0.28 to 0.80). Similarly, autologous reconstruction was associated with improved cancer-specific survival among patients below the age of 50 (HR = 0.58; 95% CI, 0.42 to 0.80) and between ages 50 to 69 (HR = 0.61; 95% CI, 0.43 to 0.85).
Immediate breast reconstruction is associated with decreased breast cancer-specific mortality, particularly among younger women. We believe this association is more likely attributable to imbalances in socioeconomic factors and access to care than to inadequate adjustment for tumor characteristics and disease severity. Further research is needed to identify additional prognostic factors responsible for the improved cancer survival among women undergoing immediate postmastectomy reconstruction. Cancer 2009. © 2009 American Cancer Society.
Breast reconstruction, particularly when performed at the time of mastectomy, reduces depression and improves well-being in breast cancer survivors.1, 2 With advances in both implant and autologous (tissue remodelling) techniques, an esthetically acceptable breast can be achieved in most patients.3 Although several studies suggest that reconstruction neither impairs the detection of locoregional recurrence4-6 nor delays adjuvant therapy,7, 8 many surgeons and patients remain unconvinced of the oncological safety of breast reconstruction.9, 10 To date, 2 limited population-based studies have demonstrated an association between implant reconstruction and improved breast cancer-specific survival.11, 12 However, no population-based study has examined the relationship between autologous reconstruction and cancer-specific mortality. Here, we compare breast cancer-specific survival between patients who underwent unilateral mastectomy alone or immediate breast reconstruction using population-based data from US cancer registries participating in the US National Cancer Institute's (NCI) Surveillance, Epidemiology, and End Results (SEER) program.
MATERIALS AND METHODS
Data were obtained from all 17 US cancer registries participating in the SEER program using SEER*Stat version 6.2.3 (http://seer.cancer.gov/seerstat). As reconstruction was not recorded before 1998,13 to ensure sufficient follow-up time for meaningful interpretation, analyses were restricted to cases diagnosed from 1998 to 2003 (with follow-up through 2005 inclusive). The analysis included all women who underwent unilateral simple, modified radical, radical or extended radical mastectomy for histologically confirmed invasive breast cancer without metachronous disease or concomitant surgery on the contralateral breast or axillary lymph node basin. Cases whose reporting sources were only from autopsy records or death certificates were excluded.
Descriptive statistics of demographic and tumor characteristics were generated for women who underwent mastectomy alone (control cases) or who underwent the first course of breast reconstruction immediately at the time of their mastectomy (index cases). In addition, descriptive statistics were computed for implant only and autologous only (rectus abdominis flap, latissimus dorsi flap, or reconstruction with flap not otherwise specified) reconstruction subgroups. Women in whom reconstruction occurred but was not specified or in whom a combination of implant and autologous techniques was used were included among the all immediate breast reconstruction group. Stages I to IV were defined by SEER according to the American Joint Committee on Cancer staging scheme third edition.14 Histologic grade 1 comprised well-differentiated tumors, grade 2 were moderately differentiated tumors, grade 3 were poorly differentiated tumors, and grade 4 were undifferentiated or anaplastic. Fifty-six histology codes were collapsed into 6 categories as previously described15: ductal (defined using ICD-0 code 8500), lobular (8520), ductal/lobular (8522), mucinous (8480), inflammatory (8530), and other, comprising tubular (8211), comedocarcinoma (8501), medullary (8510), and papillary (8050 and 8503). Patients with missing, borderline or unknown data on progesterone receptor (PR) status or estrogen receptor (ER) status were placed in a separate category or combined with ER positive patients, respectively, as previously described.16, 17 The Markov chain Monte Carlo multiple imputation method assuming multivariate normality18 was used to correct for missing data on marital status and stage.
Differences in patient characteristics between control and index cases and reconstructive methods were evaluated using chi-square tests. Kaplan-Meier plots and log-rank tests compared unadjusted cancer-specific survival between control and index cases, as well as between control cases and each of the 2 reconstruction subgroups.
Cox proportional hazards models fitting demographic (age, marital status, race, year of diagnosis, and registry) and disease characteristics (stage, grade, histology, ER and progesterone receptor status, and radiotherapy) were used to estimate hazard ratios and 95% confidence intervals [CI] for breast cancer-specific mortality. Cause-specific survival was measured from the time of diagnosis to breast cancer-specific death, date of last follow-up or death from another cause, or December 2005. The assumption of proportional hazards was evaluated by visual inspection of hazard function and residual plots. Failures of the proportional hazards assumptions were addressed by the inclusion of time-covariate interaction terms or, in the case of age, by stratification.
All statistical tests were 2-sided, and P values less than .05 were considered statistically significant. All analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, NC).
Demographic and tumor characteristics of the 54,660 women included in this analysis are shown in Table 1. Regional differences in breast reconstruction techniques and rates were observed, with a significantly greater proportion of patients from the New Jersey, Detroit, and Atlanta registries receiving immediate reconstruction (data not shown). Women who underwent immediate breast reconstruction were more likely to be younger, married, and white, and to be diagnosed with early stage, lymph node-negative disease. They were also less likely to receive radiotherapy. Compared with control cases, tumors among the immediate reconstruction group tended to be estrogen and progesterone receptor positive, grade 2 or 3, and of ductal histologic type. Greater similarity in patient and tumor characteristics was observed between the reconstruction subgroups. The unadjusted relationship between immediate breast reconstruction, including implant and autologous methods, and mastectomy alone is illustrated in Figure 1. Median follow-up was 59 months. Women with immediate breast reconstruction of any type, or implant and autologous reconstruction specifically, had improved overall breast cancer-specific mortality compared with women who underwent mastectomy alone.
|Variable||Mastectomy Only||Immediate Breast Reconstruction||P|
|All||Implant Only||Autologous Only|
|No. of Patients||%||No. of Patients||%||No. of Patients||%||No. of Patients||%|
|Other (comedo, tubular, medullary, papillary)||1241||2.7||295||3.5||74||3.8||121||3.3|
|Estrogen receptor status||<.0001|
|Progesterone receptor status||<.0001|
|Positive lymph nodes||<.0001|
|1 to 5||11524||25.0||2055||24.2||471||24.2||940||26.0|
Table 2 shows the results of the Cox proportional hazards model including dichotomized immediate breast reconstruction. Immediate breast reconstruction was associated with a 26% (P < .0001) decrease in the risk of breast cancer death. Other factors associated with reduced risk of cancer death in this model included younger age, being married, non-African American race, noninflammatory histology, ER positive status, PR positive status, and absence of radiotherapy. The results of Cox proportional hazards models fitting each of the reconstruction methods and stratifying by age are presented in Table 3. Immediate breast reconstruction was associated with improved breast cancer-specific survival among women in all age groups, with those 70 years of age and older having the lowest risk of death among all those undergoing reconstruction. Among women who received implant reconstruction, those younger than 50 years of age experienced the lowest mortality risk. All other age groups in the implant subgroup failed to demonstrate reduced cause-specific mortality. Autologous reconstruction was associated with a survival advantage among women below the age of 70.
|Immediate breast reconstruction|
|Yes||.74||.68 to .80||<.0001|
|50-69||1.12||1.01 to 1.24||.046|
|≥70||2.65||2.36 to 2.98||<.0001|
|Yes||.71||.65 to .77||<.0001|
|Black||1.37||1.21 to 1.55||<.0001|
|Other||.84||.76 to .92||.0002|
|II||3.55||2.97 to 4.25||<.0001|
|III||11.97||9.85 to 14.54||<.0001|
|IV||23.74||15.96 to 34.65||<.0001|
|2||1.76||1.37 to 2.25||<.0001|
|3||3.78||2.96 to 4.84||<.0001|
|4||4.46||3.25 to 6.11||<.0001|
|Lobular||.71||.58 to .87||.0009|
|Ductal/lobular||.77||0.65 to 0.91||.0023|
|Mucinous||.75||.56 to 1.02||.063|
|Inflammatory||1.35||1.17 to 1.56||<.0001|
|Other (comedocarcinoma, tubular, medullary, papillary)||.61||.49 to .77||<.0001|
|Estrogen receptor status|
|Positive/unknown||.30||.27 to .35||<.0001|
|Progesterone receptor status|
|Negative||2.33||2.05 to 2.65||<.0001|
|Unknown||1.47||1.29 to 1.69||<.0001|
|Yes||1.27||1.15 to 1.40||<.0001|
|Age, y||All Immediate Breast Reconstruction||Implant Only||Autologous Only|
|HR||95% CI||P||HR||95% CI||P||HR||95% CI||P|
|<50||.73||.65 to .82||<.0001||.47||.28 to .80||.0053||.58||.42 to .80||.0011|
|50-69||.75||.66 to .85||<.0001||.78||.43 to 1.40||.40||.61||.43 to .85||.0036|
|≥70||.66||.51 to .86||.0017||.25||.06 to 1.01||.05||1.01||.52 to 1.97||.99|
Our finding that immediate postmastectomy reconstruction is associated with improved breast cancer-specific survival is in keeping with previous population-based studies that demonstrated a relationship between implant reconstruction and reduced breast cancer mortality.11, 12 In a study of the Danish Cooperative Group registry comparing 580 delayed implant reconstruction patients and 1158 matched controls, women who had reconstruction displayed improved disease-free survival (HR, 0.78 [0.64-0.95]).11 Similarly, a US study demonstrated reduced risk of breast cancer mortality (HR, 0.54 [0.43-0.67]) for mastectomy patients who underwent implant reconstruction.12 Our study is the first population-based study to show improved breast cancer-specific mortality for all reconstruction types, and autologous reconstruction specifically. Data was drawn from all US cancer registries participating in the SEER program making it a larger study and our findings more generalizable and less susceptible to variations in regional practice patterns.
There are several inherent limitations to the use of retrospective registry data in the current study. As reconstruction was recorded only after 1998, follow-up was limited to a median of 59 months, meaning that many cases had follow-up intervals far shorter than 5 years. In addition, the SEER database included only the first round of reconstruction. Some patients may have completed their reconstruction only after follow-up time was computed. In addition, some patients may have received reconstruction subsequent to their initial diagnosis and mastectomy. Such delayed reconstruction may have had an impact on breast cancer-specific mortality. Another limitation is that while we controlled for all available disease severity measures and key tumor characteristics, we could not adjust for every biological prognostic factor. For example, although we adjusted for PR/ER status, information on the proportion of index and control cases receiving hormonal therapy or adjuvant chemotherapy is not available in the SEER database. Similarly, the SEER database does not capture information on comorbidities and operative risk factors such as cardiovascular disease, obesity, and smoking status. Baseline health status would tend to distort the relationship between reconstruction and survival, with sicker patients less likely to be offered or to request additional surgery in the form of reconstruction. To address this potential source of confounding, we stratified by age. Cardiovascular disease and concomitant illnesses are less common in younger women, and, thus, they are less likely to have distorted the association between reconstruction and cancer-specific survival in this age group. We specifically examined young women in the implant subgroup. Implant reconstruction is an operative option that carries minimal additional surgical risk when performed at the time of mastectomy. Although a staged approach is often required, many secondary procedures can be performed under local anesthesia and are of minimal risk.19, 20 If a patient is eligible to undergo mastectomy, there is little to preclude her medically from being offered implant reconstruction. Tellingly, the estimated apparent survival advantage of reconstruction was particularly strong within this subgroup (HR = 0.47; 95% CI, 0.28 to 0.80).
Although the observed association between reconstruction and improved survival may, as suggested by others, be due, in part, to direct physiologic or immunologic effects,21 a more likely hypothesis is that reconstruction is a surrogate marker for breast cancer patients destined to do well on the basis of socioeconomic factors. Women that select breast reconstruction after mastectomy have been shown to differ from women without breast reconstruction with regards to socioeconomic status.22, 23 Similarly, Morrow et al have shown that patients with a family income of $40,000 or more were more likely than patients with less than $40,000 to undergo reconstruction.24 Several studies demonstrate that general surgeons are more likely to discuss reconstructive options with more educated women.25, 26 In addition, general surgeons with a higher propensity to refer breast cancer patients to plastic surgeons before mastectomy tend to work in cancer centers, suggesting that women who undergo reconstruction may also have greater access to medical surveillance.10 Women with higher incomes may experience improved survival after breast cancer not only because of better access to cancer care and treatment but also as a result of healthy lifestyle behaviors (diet, exercise). Although we captured dimensions of socioeconomic status by adjusting for marital status, race, and registry location, we were unable to adjust more specifically for disparities in socioeconomic status between the index and control groups.
In the absence of information on comorbidities and socioeconomic factors, the data available from the registries participating in the SEER program remains inadequate to fully assess survival patterns associated with immediate reconstruction. Such patient-level information may be obtained in the future through linkages between participating SEER site registries and state health insurance datasets, as well as through the use of patient questionnaires and medical chart reviews. Although we have suggested that imbalances in socioeconomic status may be contributing to the observed improvement in breast cancer-specific survival, it is possible that other unmeasured confounders are involved. These additional factors could be identified using qualitative techniques, including semistructured interviews of surgeons and patients after mastectomy.
A randomized trial to control for imbalances in all prognostic factors will never be feasible, as women interested in reconstruction are unlikely to consent to randomization. In the absence of prospective evidence, our current retrospective study represents an important step in examining the interaction of immediate breast reconstruction and breast cancer-specific survival. In addition, it points to the need to capture patient-level information on socioeconomic status and other potential prognostic factors involved in the complex pathway that leads a medically eligible patient to choose reconstruction.
Conflict of Interest Disclosures
The authors made no disclosures.
- 13Summary staging guide for the cancer Surveillance, Epidemiology, and End Results reporting program. In: FritzA, RiesL, eds. SEER Program Code Manual. 3rd ed. Bethesda, MD: National Cancer Institute Available at: http://seer.cancer.gov/manuals/AppendC.pdf, , , et al.
- 14SEER: Breast AJCC/UICC Stage. SEER Program: Comparative Staging Guide for Cancer, Major Cancer Sites, version 1.1; 2006 Available at: http://seer.cancer.gov/manuals/historic/comp_stage1.1.pdf
- 18Analysis of Incomplete Multivariate Data. New York: Chapman and Hall, 1997..