BREAST Using Propensity Score Analysis to Compare Major Complications between DIEP and Free Muscle-Sparing TRAM Flap Breast Reconstructions Toni Zhong, M.D., M.H.S. Christine B. Novak, Ph.D. Shaghayegh Bagher, M.Sc. Saskia W. M. C. Maass, M.D. Jing Zhang, M.D. Udi Arad, M.D. Anne C. O’Neill, M.D., Ph.D. Kelly A. Metcalfe, R.N., Ph.D. Stefan O. P. Hofer, M.D., Ph.D. Toronto, Ontario, Canada
Background: Previous studies comparing muscle-sparing transverse rectus abdominis myocutaneous (TRAM) versus deep inferior epigastric artery perforator (DIEP) free flaps have not considered procedure selection bias. Propensity score analysis provides a statistical approach to consider preoperative factors in flap selection, and was used to compare major complications (breast and abdominal) between these microsurgical breast reconstruction (free musclesparing TRAM versus DIEP). Methods: This study evaluated major breast and abdominal complications in 292 consecutive patients (428 free abdominal flaps). Propensity scores were calculated for patient differences affecting flap selection (DIEP versus free muscle-sparing TRAM). Multivariate logistic models using selected covariates separately analyzed breast and abdominal complications between flap methods. Results: There were 83 major complications (28 percent): breast, 20 percent; abdomen, 8 percent. Using propensity scores, the adjusted odds of abdominal complications were significantly higher in free muscle-sparing TRAM than in DIEP flaps (OR, 2.73; 95 percent CI, 1.01 to 7.07). With prior chemotherapy, body mass index significantly increased abdominal complications (OR, 1.16; 95 percent CI, 1.01 to 1.34). Using propensity scores, there was no significant association between reconstruction method and breast complications; diabetics had significantly increased breast complications (OR, 4.19; 95 percent CI, 1.14 to 15.98). Previous abdominal operations (OR, 1.77; 95 percent CI, 0.96 to 3.30) and immediate reconstruction (OR, 1.86; 95 percent CI, 0.94 to 3.71) approached significance. Conclusions: Propensity score analysis indicated significantly higher abdominal complications in free muscle-sparing TRAM compared with DIEP flaps. This study highlights the importance of separately evaluating recipient breast and donor abdominal complications and use of propensity scores to minimize procedure selection bias. (Plast. Reconstr. Surg. 133: 774, 2014.) CLINICAL QUESTION/LEVEL OF EVIDENCE: Therapeutic, III.
T
he goal of autologous breast reconstruction is to provide an aesthetically pleasing natural breast, using a reliable flap that imposes minimal compromise to abdominal wall integrity. The muscle-sparing transverse rectus From the Division of Plastic and Reconstructive Surgery, Department of Surgery and Surgical Oncology, University Health Network; and the Division of Plastic and Reconstructive Surgery and the Lawrence S. Bloomberg Faculty of Nursing, University of Toronto. Received for publication May 6, 2013; accepted October 18, 2013. Copyright © 2014 by the American Society of Plastic Surgeons DOI: 10.1097/PRS.0000000000000024
774
Disclosure: Dr. Zhong is supported by an American Society of Clinical Oncology, Conquer Cancer Foundation Career Development Award, and Dr. Hofer is supported by the Wharton Chair in Plastic and Reconstructive Surgery at the University Health Network. Dr. Hofer, Dr. Metcalfe, and Dr. Zhong are supported by grant funding from the Canadian Breast Cancer Foundation and the Canadian Institutes of Health Research. The other authors do not have a financial interest to declare. None of the authors has a financial interest in any of the products or devices mentioned in this article.
www.PRSJournal.com
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications abdominis myocutaneous (TRAM) and the deep inferior epigastric perforator (DIEP) free flap are well-established techniques for autologous breast reconstruction. Debate continues regarding the merits of muscle-preserving techniques to minimize abdominal wall morbidity and maintain flap reliability.1–6 Although several retrospective studies have reported decreased abdominal complications in DIEP flaps compared with free muscle-sparing TRAM flaps, other studies have not found significant differences in abdominal wall morbidity.1,4,7–9 The trade-offs to the theoretical abdominal wall benefits from DIEP flaps, however, are reported higher rates of partial flap necrosis, fat necrosis, and venous congestion in the breast, which are related to decreased flap vascularity.10–12 A meta-analysis published in 2009 reported increased breast-related complications associated with DIEP flaps compared with free TRAM flap, whereas the free muscle-sparing TRAM flap was associated with greater donor-site morbidity.2 A weakness intrinsic to these observational studies relates to the concept of “confounding by indication,” which can threaten the validity of the findings.13 Specifically, the preoperative confounding variables related to the selection of the flap have not been considered in the previous studies. Factors such as the choice of performing a DIEP versus a free TRAM flap are not random (i.e., an intended DIEP flap may be converted to a free muscle-sparing TRAM flap with inadequate intraoperative perforator anatomy). In patients who do not have a single dominant perforator, a multiperforator DIEP flap or free muscle-sparing TRAM flap may be used, and this decision is often influenced by preoperative patient characteristics. This is termed “confounding by indication” or “indication bias.”14 The most common indication biases toward the selection of a free muscle-sparing TRAM flap over a DIEP flap when a single dominant perforator is not present are in smokers or diabetics, to minimize the risk of recipient fat necrosis.2,3 Also, most microsurgeons will strive to perform multiperforator DIEP flap surgery over free muscle-sparing TRAM flap surgery in younger patients and bilateral breast reconstructions.15,16 Conventional methods such as restriction and multivariable regression adjust for outcome confounding factors but do not address confounding by indication or treatment allocation. The purpose of this study was to evaluate DIEP flaps versus muscle-sparing TRAM free flaps in terms of breast complications and abdominal morbidity
following breast reconstruction while controlling for indication bias through propensity score analyses17: Propensity score is a statistical approach to control the a priori probability of receiving one treatment over another, attempting to equate it with randomization in circumstances where this is not possible.18 The propensity score for an individual is the conditional probability of receiving a particular treatment, based on observed pretreatment covariates.17,18 In this study, the propensity score was the probability that the surgeon will convert from an intended DIEP flap to a free muscle-sparing TRAM flap based on known preoperative patient characteristics. This score is included with other baseline covariates and known confounders in the final multivariable regression analysis to reduce confounding by indication bias.19 In the absence of randomized controlled trials, the use of propensity scores in this study enabled a retrospective comparison between free muscle-sparing TRAM and DIEP flaps on abdominal and breast complications while providing the most balance between the two surgical treatment groups.
PATIENTS AND METHODS After approval from the institutional ethics review board, a consecutive series of women undergoing breast reconstruction from January 1, 2009, to August 30, 2012, at the University Health Network, Toronto, Canada, were identified using a prospectively maintained institutional breast reconstruction database. Demographic, oncologic treatment, and surgical data were obtained from this database. Demographic data included age at diagnosis, tobacco use, body mass index, diabetes, previous pregnancies, and previous abdominal operations. Oncologic data included previous irradiation and chemotherapy treatment. Surgical data included date of reconstruction, type of flap (DIEP flap versus free muscle-sparing TRAM flap), laterality (unilateral versus bilateral), and timing of reconstruction (immediate versus delayed). Surgery Our selection criteria for timing and method of abdominally based breast reconstruction are consistent with previously published guidelines.20–22 Immediate reconstruction was usually offered to women with a strong family history or gene positivity for breast cancer, in situ breast cancer, or stage I or IIa breast cancer when postmastectomy radiation therapy was unlikely. In general, autologous tissue reconstruction was our first option
775
Plastic and Reconstructive Surgery • April 2014 for women who had had previous breast or chest wall irradiation, nonpliable chest wall soft tissues, or redundant abdominal tissues, and for those undergoing delayed reconstruction or requesting this type of reconstruction. All autologous tissue reconstructions were microsurgically anastomosed to the internal mammary vessels in the third or fourth interspaces under the operating microscope by two plastic surgeons (T.Z. and S.O.P.H.). Selection of Free Muscle-Sparing TRAM versus DIEP Flap: Intraoperative Decision Making In all abdominally based breast reconstructions, DIEP flaps were the primary choice. If the intraoperative perforator anatomy was not ideal, these cases were converted to a free m uscle-sparing TRAM flap with a fascia-sparing technique, harvesting only a small cuff of fascia around each perforator (Fig. 1). When assessing perforator anatomy intraoperatively, our main determining factor to convert a DIEP flap to a free musclesparing TRAM flap was an absent perforator vein (1 cm in diameter >6 months after surgery) and breast hematoma that required evacuation in the operating room. Major abdominal complications included detection of a bulge (abdominal wall laxity without an associated fascial defect) in the standing position or hernia (with an associated fascial defect) elicited with the Valsalva maneuver. These are detected during routine follow-up and are prospectively documented in our standard postoperative clinic forms. Only those bulges that are symptomatic to the patient are corrected in a revision surgery by exploration and fascial reinforcement using a polypropylene mesh (Prolene) suture. Statistical Analysis Demographic, clinical, and surgical data were summarized and compared between the DIEP and free muscle-sparing TRAM flap groups using Fisher’s exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. The overall incidence of breast and abdominal complications was compared between flap groups using Fisher’s exact test. Univariate and multivariable logistic regression were used to examine the risk factors associated with major complications. Factors included in the univariate analyses were age; body mass index; diabetes; smoking status; previous breast irradiation, chemotherapy, and abdominal surgery; and timing, laterality, and type of reconstruction (DIEP versus free muscle-sparing TRAM flap). The propensity score was defined as the probability of conversion from a planned DIEP flap
to a free muscle-sparing TRAM flap intraoperatively based on preoperative clinical characteristics including smoking status, age, and laterality of breast reconstruction.2,15,16,24–26 To estimate this probability, an initial logistic regression analysis was performed with patient clinical characteristics as the independent variables and free muscle-sparing TRAM flap (yes/no) as the dependent variable. Two final multivariable models were constructed as outlined by Harrell, which prioritizes the inclusion of key confounders of a primary predictor (reconstruction) and outcome (major complication) in the final regression model in addition to the propensity score estimate.28 The first model used abdominal complication as the dependent variable and included the following independent variables: flap reconstruction method, propensity score, body mass index, and previous chemotherapy. The second model used major breast complications as the dependent variable and the following independent variables: flap reconstruction method, propensity score, body mass index, diabetes, timing of reconstruction, previous irradiation, and previous abdominal surgery. Multivariable logistic regression models were built using stepwise backward elimination (based on the lowest Akaike information criterion value). Second-order interactions and test of collinearity were performed in all models. The area under the receiver operating characteristic curve (c-statistic) was 0.7 for the propensity score model, indicating good discrimination between patients who received free musclesparing TRAM versus DIEP flaps. Assessment of model fit using H osmer-Lemeshow goodness of fit and analysis of variance were performed in all models.29 A two-sided value of p ≤ 0.05 was considered statistically significant. Statistical analyses were performed using the R statistical package version 2.14.
RESULTS There were 292 women who underwent a total of 428 abdominally based free flap breast reconstructions from January 1, 2009, to August 30, 2012, performed by two surgeons. The mean age of the patients was 50.1 ± 8.6 years, and the mean body mass index was 27.8 ± 4.7. The majority of the women (84 percent) had DIEP flap reconstructions. Following surgery, 28 percent of the patients had one or more major complications; 20 percent had breast complications and 8 percent had abdominal complications, including three total flap losses (1 percent). Unadjusted analysis revealed no significant difference in occurrence of breast complications between the two methods of breast reconstruction;
777
Plastic and Reconstructive Surgery • April 2014 however, the unadjusted odds of abdominal complication were significantly higher in free muscle-sparing TRAM compared with DIEP flaps (OR, 3.04; 95 percent CI, 1.04 to 8.24) (Table 1). Univariate analyses examined the association of demographic, clinical, and surgical factors with overall complications (Table 2). The occurrence of major complications was not significantly different between DIEP and free muscle-sparing TRAM flaps. Bilateral reconstructions had a 1.7-times higher odds of a major complication compared with a unilateral reconstruction (OR, 1.7; 95 percent CI, 1.0 to 3.0). The odds of a major complication occurrence were 3.6 times higher in patients with diabetes (OR, 3.6; 95 percent CI, 1.1 to 12.9) compared with a nondiabetic, and for every 1-year increase in patient age, the odds of any major complication increased by 4 percent (OR, 1.04; 95 percent CI, 1.0 to 1.1). The first final multivariable model compared abdominal complications between DIEP versus free muscle-sparing TRAM flaps after controlling for obesity and prior chemotherapy treatment (Table 3). Because previous breast irradiation, abdominal surgery, and timing of reconstruction were not significantly related to abdominal complications and no diabetic patients had abdominal complications, these variables were omitted from this analysis. Previous chemotherapy and obesity had a significant interaction effect on abdominal complications; thus, an interaction variable was created. The adjusted odds of abdominal complications were 2.73 times higher in free muscle-sparing TRAM compared with DIEP flap breast reconstruction (OR, 2.73; 95 percent CI, 1.01 to 7.07). In the subgroup of patients who had prior chemotherapy, for every one-unit increase in body mass index, the odds of an abdominal complication increased by 16 percent (OR, 1.16; 95 percent CI, 1.01 to 1.34). The effect of obesity on abdominal complications was mediated by the presence of previous
chemotherapy (Fig. 2); abdominal complications were increased only in obese patients who had previous chemotherapy, whereas complications remained fairly low for those who did not have previous chemotherapy, regardless of body mass index. The second final multivariable model compared breast complications between DIEP flaps and free muscle-sparing TRAM flaps after controlling for body mass index, diabetes, previous abdominal operations, timing of breast reconstruction, and previous breast irradiation (Table 4). Previous chemotherapy and breast irradiation were collinear and therefore we included only breast irradiation in the regression model. The adjusted analysis using the propensity score showed no significant association between method of breast reconstruction and breast complications. However, diabetic patients had a 4.19-fold increased odds of breast complications (OR, 4.19; 95 percent CI, 1.14 to 15.98) compared with nondiabetic patients. Both previous abdominal operations (OR, 1.77; 95 percent CI, 0.96 to 3.30) and immediate breast reconstruction (OR, 1.86; 95 percent CI, 0.94 to 3.71) approached statistical significance as risk factors for major breast complications.
DISCUSSION This is the first study to use propensity score analysis to separately evaluate postoperative complications related to the recipient site (breast) and donor site (abdomen) following muscle-sparing TRAM and DIEP free flap breast reconstruction while accounting for the confounding factors that contribute to selection bias with the use of a propensity score analysis. It is important to analyze the breast and abdominal complications separately, because one flap method may be superior for achieving breast outcomes but inferior in abdominal outcomes in the same patient. Our initial analysis indicated no difference in overall
Table 1. Major Complications following Breast Reconstruction with DIEP or Muscle-Sparing TRAM Free Flaps No. of patients Complications Breast complications Total flap loss Partial flap loss Fat necrosis Hematoma in breast Abdominal complications Bulge Hernia MS, muscle-sparing.
778
DIEP Flap (%)
Free MS-TRAM Flap (%)
244
48
50 (20.5) 3 (1.2) 6 (2.5) 25 (10.2) 17 (7.0) 15 (6) 11 (4.5) 4 (1.5)
10 (20.8) 0 0 7 (14.6) 3 (6.2) 8 (17) 7 (14) 1 (2)
OR (95% CI) 1.02 (0.42–2.27) 1.41 (0.07–27.6) 2.64 (0.15–47.47) 0.67 (0.26–1.96) 1.12 (0.31–6.22) 3.04 (1.04–8.24) 0.27 (0.09–0.89) 1.39 (0.17–63.86)
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications Table 2. Univariate Analysis of Major Overall Complications Major Complications Variable Categorical Method DIEP MS-TRAM Laterality Unilateral Bilateral Timing Immediate Delayed Diabetes No Yes Smoker Yes No Previous abdominal operations Yes No Previous breast irradiation Yes No Previous chemotherapy Yes No Continuous Age, yr BMI
Total (%)
No (%)
Yes (%)
OR (95% CI)
244 (84) 48 (16)
184 (75) 32 (67)
60 (25) 16 (33)
1.5 (0.7–3.1)
156 (53) 136 (47)
123 (79) 93 (68)
33 (21) 43 (32)
1.7 (1.0–3.0)
134 (46) 158 (54)
91 (68) 125 (79)
43 (32) 33 (21)
0.56 (0.3–0.9)
281 (96) 11 (4)
211 (75) 5 (45)
70 (25) 6 (55)
3.6 (1.1–12.9)
25 (9) 267 (91)
17 (68) 199 (75)
8 (32) 68 (25)
0.7 (0.3–1.8)
135 (46) 157 (54)
94 (70) 122 (78)
41 (30) 35 (22)
0.7 (0.4–1.2)
146 (50) 146 (50)
112 (77) 104 (71)
34 (23) 42 (29)
1.3 (0.2–1.7)
163 (56) 129 (44)
124 (76) 92 (71)
39 (24) 37 (29)
1.3 (0.7–2.2)
50.1 ± 8.6 27.8 ± 4.7
49.4 ± 8.5 27.5 ± 4.5
52.2 ± 8.6 28.5 ± 5.1
1.04 (1.0–1.1 1.05 (0.9–1.1
MS, muscle-sparing; BMI, body mass index.
major complications following breast reconstruction using DIEP or muscle-sparing TRAM free flaps (Table 2). However, when the donor and recipient sites were analyzed separately, abdominal complications were significantly higher in muscle-sparing TRAM flaps compared with DIEP flaps (Table 3). The unique advantage of using a propensity score adjustment is that comparability between the comparison groups is improved by creating pseudorandomization of measured confounders in clinical situations where a randomized controlled trial is not feasible.18 Analyses with propensity scores revealed a strong association between breast complications and diabetes Table 3. Multivariate Logistic Regression Model with Propensity Score for Abdominal Complications Variable Categorical Method DIEP Free MS-TRAM Continuous BMI With previous chemotherapy Without previous chemotherapy
OR (95% CI)
p
* 2.73 (1.01–7.07)
0.04
1.16 (1.01–1.34)
0.02
0.93 (0.81–1.06)
0.28
MS, muscle-sparing; BMI, body mass index. *Reference category.
but no difference in type of flap. Also, increased abdominal complications were associated with previous chemotherapy in those with a higher body mass index. The DIEP flap was developed as a derivative of the muscle-sparing free flap to minimize morbidity related to abdominal wall integrity.30 In two recent meta-analyses, the authors independently found the pooled relative risk of abdominal bulge or hernia to be lower in DIEP flaps compared with muscle-sparing TRAM free flaps.2,31 Similarly, we found an adjusted OR of 2.73 (95 percent CI, 1.01 to 7.07) for bulge/hernia formation in free muscle-sparing TRAM compared with DIEP flaps. Because we found a significant interaction between body mass index and previous chemotherapy, our conclusions were based on analyzing the effect of obesity on abdominal complications stratified by previous chemotherapy. The stratum-specific estimates provided a more valid representation of the effect modification observed in this study, and it was likely the combination of obesity and the detrimental effects of chemotherapy on wound healing that increased the risk of abdominal complications in this subgroup. It is believed that the large and heavy abdominal flap stretches and attenuates the musculocutaneous perforators in
779
Plastic and Reconstructive Surgery • April 2014
Fig. 2. Effect modification of previous chemotherapy on body mass index (BMI) and the association with abdominal complications.
an obese patient, leading to an abdominal flap that is compromised in its blood supply. This combined with the detrimental effect of chemotherapeutic agents on the macrophages and fibroblasts that are intimately involved in wound healing can result in poor healing within the important layers of the abdominal wall, including the fascial Table 4. Multivariate Logistic Regression Model with Propensity Score for Breast Complications Variable Categorical Method DIEP Free MS-TRAM Diabetes No Yes Timing Delayed Immediate Previous abdominal operations No Yes Previous breast irradiation No Yes Continuous BMI
OR (95% CI)
p
* 0.98 (0.40–2.14)
0.95
* 4.19 (1.14–15.98)
0.02
* 1.86 (0.94–3.71)
0.07
* 1.77 (0.96–3.30)
0.07
* 1.23 (0.67–2.29)
0.51
1.02 (0.96–1.09)
0.48
*Reference category. MS, muscle-sparing; BMI, body mass index.
780
repair, leading to problems with bulges and hernias. Our finding that DIEP flaps likely represent the preferred autologous breast reconstruction technique for obese patients to minimize their abdominal morbidity corroborates the findings reported by Garvey et al.25 We did not observe a significant difference in major breast complications between DIEP and muscle-sparing TRAM free flaps. This is contrary to reports by Sailon et al.3 and Man et al.,2 who reported a higher frequency of breast complications in DIEP flaps compared with muscle-sparing TRAM flaps. However, addition of the propensity score analysis indicated that the presence of diabetes was a significant predictor of major breast complications. Similar to previous studies, we also identified immediate reconstruction and previous abdominal operations to be risk factors for breast complications that approached statistical significance.20 Our finding that previous abdominal surgical scars may predispose a patient to increased breast flap complications is in contradiction to a recent study that reported similar flap complications between patients with and without Pfannenstiel scars.32 These authors found the number of larger-caliber skin perforators to be significantly higher in DIEP flaps that have Pfannenstiel scars compared with the virginal abdomens using
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications computed tomographic angiography.32 Although low transverse abdominal incisions, such as Pfannenstiel incisions, do not generally impact the concentration of larger perforators around the periumbilical region, it is possible that other incisions located more centrally in the abdomen may affect perfusion of the DIEP flap, as may be the case for our study. Because this subgroup analysis may have been underpowered, it is plausible that both immediate reconstruction and previous abdominal operations may pose “real” threats to breast reconstruction outcomes in a larger sample. The multivariable regression using propensity score analysis allowed the opportunity to control for selection bias and confounding by indication. This is particularly important when there is a relatively large number of confounders compared with the number of major complication outcomes. This unique approach of using propensity scores allowed us to consider not only the factors that affected outcome but also the factors that contributed to selection of the reconstruction procedure. The limitations to this study include an observational study with a retrospective design and a small sample of muscle-sparing TRAM flaps. In addition, our study was undertaken at a cancer center with a high volume of reconstructive procedures, and the complications reported may not be representative of all institutions. The challenge of comparing a series of patients with DIEP and muscle-sparing TRAM free flap reconstruction is recognition of the differences between surgeons regarding surgical approaches, techniques, and postoperative evaluation mechanisms. In our study, all operations were performed by two experienced microsurgeons who shared the same intraoperative approach and postoperative regimens. Although the use of the propensity score added another degree of reliability to our results, the interpretation of our conclusions remains limited to a retrospective study of patients who were operated on by two surgeons who followed the intraoperative flap selection guide as outlined in Figure 1. Another caveat is that Figure 1 is intended to reflect the clinical decision-making process for the two surgeons when selecting a DIEP versus a free muscle-sparing TRAM flap for our series, and is not intended to be an instructional algorithm to be followed by all surgeons. As such, the preoperative clinical variable “younger age” does not reflect a specific age cut-off; rather, it only represented the surgeon’s gestalt. In addition, although the two surgeons routinely use preoperative computed tomographic angiography to map out the perforator anatomy of the abdominal
wall to expedite perforator dissection, the final decision to use a DIEP or muscle-sparing TRAM flap nevertheless is ultimately based on the intraoperative assessment of the perforators and the key patient characteristics as outlined in Figure 1. Based on our study findings, we will continue to use the DIEP flap as our first choice in patients with adequate single-perforator anatomy and strive to dissect multiperforator DIEP flaps in patients with insufficient single-perforator anatomy as much as possible to achieve the best abdominal outcome. Lastly, our study also highlights the need for improved methods of abdominal fascial closure in patients who have undergone free muscle-sparing TRAM flap surgery, and our current approach to always close fascia primarily after a fascia-sparing elevation of the free muscle-sparing TRAM flap is clearly an oversimplified solution to the complex dynamics of the abdominal musculature. The ideal method of abdominal wall closure following muscle-sparing TRAM flap surgery is an important area of research that warrants a separate study of its own.
CONCLUSIONS Using the propensity score analysis, abdominal complications were significantly higher in free muscle-sparing TRAM flaps compared with DIEP flaps. The effect of obesity on abdominal complications was mediated by the presence of previous chemotherapy. There was no significant association between type of flap and breast complications; however, diabetes was a significant risk factor. This study highlights the importance of evaluating recipient breast complications separately from donor abdominal complications and the use of propensity scores to account for preoperative patient factors to reduce confounding by indication. Toni Zhong, M.D., M.H.S. University Health Network 8NU-871 200 Elizabeth Street Toronto, Ontario M5G 2C4, Canada [email protected]
REFERENCES 1. Bajaj AK, Chevray PM, Chang DW. Comparison of d onor-site complications and functional outcomes in free muscle-sparing TRAM flap and free DIEP flap breast reconstruction. Plast Reconstr Surg. 2006;117:737–746; discussion 747. 2. Man LX, Selber JC, Serletti JM. Abdominal wall following free TRAM or DIEP flap reconstruction: A meta-analysis and critical review. Plast Reconstr Surg. 2009;124:752–764.
781
Plastic and Reconstructive Surgery • April 2014 3. Sailon AM, Schachar JS, Levine JP. Free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps for breast reconstruction: A systematic review of flap complication rates and donor-site morbidity. Ann Plast Surg. 2009;62:560–563. 4. Chen CM, Halvorson EG, Disa JJ, et al. Immediate postoperative complications in DIEP versus free/muscle-sparing TRAM flaps. Plast Reconstr Surg. 2007;120:1477–1482. 5. Nahabedian MY, Dooley W, Singh N, Manson PN. Contour abnormalities of the abdomen after breast reconstruction with abdominal flaps: The role of muscle preservation. Plast Reconstr Surg. 2002;109:91–101. 6. Momoh AO, Colakoglu S, Westvik TS, et al. Analysis of complications and patient satisfaction in pedicled transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flap breast reconstruction. Ann Plast Surg. 2012;69:19–23. 7. Nahabedian MY, Tsangaris T, Momen B. Breast reconstruction with the DIEP flap or the muscle-sparing (MS-2) free TRAM flap: Is there a difference? Plast Reconstr Surg. 2005;115:436–444; discussion 445–436. 8. Garvey PB, Salavati S, Feng L, Butler CE. Perfusion-related complications are similar for DIEP and muscle-sparing free TRAM flaps harvested on medial or lateral deep inferior epigastric artery branch perforators for breast reconstruction. Plast Reconstr Surg. 2011;128:581e–589e. 9. Nelson JA, Guo Y, Sonnad SS, et al. A comparison between DIEP and muscle-sparing free TRAM flaps in breast reconstruction: A single surgeon’s recent experience. Plast Reconstr Surg. 2010;126:1428–1435. 10. Nahabedian MY, Momen B, Galdino G, et al. Breast reconstruction with the free TRAM or DIEP flap: Patient selection, choice of flap, and outcome. Plast Reconstr Surg. 2002;110:466–475; discussion 476–467. 11. Kroll SS. Fat necrosis in free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps. Plast Reconstr Surg. 2000;106:576–583. 12. Blondeel PN, Arnstein M, Verstraete K, et al. Venous congestion and blood flow in free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps. Plast Reconstr Surg. 2000;106:1295–1299. 13. Giordano SH, Kuo YF, Duan Z, et al. Limits of observational data in determining outcomes from cancer therapy. Cancer 2008;112:2456–2466. 14. Salas M, Hofman A, Stricker BH. Confounding by indication: An example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981–983. 15. Blondeel N, Vanderstraeten GG, Monstrey SJ, et al. The donor site morbidity of free DIEP flaps and free TRAM flaps for breast reconstruction. Br J Plast Surg. 1997;50: 322–330. 16. Nahabedian MY, Manson PN. Contour abnormalities of the abdomen after transverse rectus abdominis muscle flap breast reconstruction: A multifactorial analysis. Plast Reconstr Surg. 2002;109:81–87; discussion 88–90.
782
17. Austin PC. The performance of different propensity score methods for estimating marginal odds ratios. Stat Med. 2007;26:3078–3094. 18. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. 19. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17:2265–2281. 20. Zhong T, Hofer SO, McCready DR, et al. A comparison of surgical complications between immediate breast reconstruction and mastectomy: The impact on delivery of chemotherapy. An analysis of 391 procedures. Ann Surg Oncol. 2012;19:560–566. 21. Styblo TM, Lewis MM, Carlson GW, et al. Immediate breast reconstruction for stage III breast cancer using transverse rectus abdominis musculocutaneous (TRAM) flap. Ann Surg Oncol. 1996;3:375–380. 22. Buchholz TA, Kronowitz SJ, Kuerer HM. Immediate breast reconstruction after skin-sparing mastectomy for the treatment of advanced breast cancer: Radiation oncology considerations. Ann Surg Oncol. 2002;9:820–821. 23. Disa JJ, Pusic AL, Zhong T. Muscle-sparing and free TRAM flap breast reconstruction. In: Hall-Findlay EJ, Evans GRD, eds. Aesthetic and Reconstructive Surgery of the Breast. St. Louis: Saunders Elsevier; 2010:93–116. 24. Chang DW, Reece GP, Wang B, et al. Effect of smoking on complications in patients undergoing free TRAM flap breast reconstruction. Plast Reconstr Surg. 2000;105:2374–2380. 25. Garvey PB, Buchel EW, Pockaj BA, et al. The deep inferior epigastric perforator flap for breast reconstruction in overweight and obese patients. Plast Reconstr Surg. 2005;115:447–457. 26. Scheer AS, Novak CB, Neligan PC, Lipa JE. Complications associated with breast reconstruction using a perforator flap compared with a free TRAM flap. Ann Plast Surg. 2006;56:355–358. 27. Nahabedian MY, Momen B, Manson PN. Factors associated with anastomotic failure after microvascular reconstruction of the breast. Plast Reconstr Surg. 2004;114:74–82. 28. Harrell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer-Verlag; 2001. 29. Sturmer T, Joshi M, Glynn RJ, et al. A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol. 2006;59:437–447. 30. Allen RJ, Treece P. Deep inferior epigastric perforator flap for breast reconstruction. Ann Plast Surg. 1994;32:32–38. 31. Egeberg A, Rasmussen MK, Sørensen JA. Comparing the donor-site morbidity using DIEP, SIEA or MS-TRAM flaps for breast reconstructive surgery: A meta-analysis. J Plast Reconstr Aesthet Surg. 2012;65:1474–1480. 32. Mahajan AL, Zeltzer A, Claes KE, et al. Are Pfannenstiel scars a boon or a curse for DIEP flap breast reconstructions? Plast Reconstr Surg. 2012;129:797–805.
Background: Previous studies comparing muscle-sparing transverse rectus abdominis myocutaneous (TRAM) versus deep inferior epigastric artery perforator (DIEP) free flaps have not considered procedure selection bias. Propensity score analysis provides a statistical approach to consider preoperative factors in flap selection, and was used to compare major complications (breast and abdominal) between these microsurgical breast reconstruction (free musclesparing TRAM versus DIEP). Methods: This study evaluated major breast and abdominal complications in 292 consecutive patients (428 free abdominal flaps). Propensity scores were calculated for patient differences affecting flap selection (DIEP versus free muscle-sparing TRAM). Multivariate logistic models using selected covariates separately analyzed breast and abdominal complications between flap methods. Results: There were 83 major complications (28 percent): breast, 20 percent; abdomen, 8 percent. Using propensity scores, the adjusted odds of abdominal complications were significantly higher in free muscle-sparing TRAM than in DIEP flaps (OR, 2.73; 95 percent CI, 1.01 to 7.07). With prior chemotherapy, body mass index significantly increased abdominal complications (OR, 1.16; 95 percent CI, 1.01 to 1.34). Using propensity scores, there was no significant association between reconstruction method and breast complications; diabetics had significantly increased breast complications (OR, 4.19; 95 percent CI, 1.14 to 15.98). Previous abdominal operations (OR, 1.77; 95 percent CI, 0.96 to 3.30) and immediate reconstruction (OR, 1.86; 95 percent CI, 0.94 to 3.71) approached significance. Conclusions: Propensity score analysis indicated significantly higher abdominal complications in free muscle-sparing TRAM compared with DIEP flaps. This study highlights the importance of separately evaluating recipient breast and donor abdominal complications and use of propensity scores to minimize procedure selection bias. (Plast. Reconstr. Surg. 133: 774, 2014.) CLINICAL QUESTION/LEVEL OF EVIDENCE: Therapeutic, III.
T
he goal of autologous breast reconstruction is to provide an aesthetically pleasing natural breast, using a reliable flap that imposes minimal compromise to abdominal wall integrity. The muscle-sparing transverse rectus From the Division of Plastic and Reconstructive Surgery, Department of Surgery and Surgical Oncology, University Health Network; and the Division of Plastic and Reconstructive Surgery and the Lawrence S. Bloomberg Faculty of Nursing, University of Toronto. Received for publication May 6, 2013; accepted October 18, 2013. Copyright © 2014 by the American Society of Plastic Surgeons DOI: 10.1097/PRS.0000000000000024
774
Disclosure: Dr. Zhong is supported by an American Society of Clinical Oncology, Conquer Cancer Foundation Career Development Award, and Dr. Hofer is supported by the Wharton Chair in Plastic and Reconstructive Surgery at the University Health Network. Dr. Hofer, Dr. Metcalfe, and Dr. Zhong are supported by grant funding from the Canadian Breast Cancer Foundation and the Canadian Institutes of Health Research. The other authors do not have a financial interest to declare. None of the authors has a financial interest in any of the products or devices mentioned in this article.
www.PRSJournal.com
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications abdominis myocutaneous (TRAM) and the deep inferior epigastric perforator (DIEP) free flap are well-established techniques for autologous breast reconstruction. Debate continues regarding the merits of muscle-preserving techniques to minimize abdominal wall morbidity and maintain flap reliability.1–6 Although several retrospective studies have reported decreased abdominal complications in DIEP flaps compared with free muscle-sparing TRAM flaps, other studies have not found significant differences in abdominal wall morbidity.1,4,7–9 The trade-offs to the theoretical abdominal wall benefits from DIEP flaps, however, are reported higher rates of partial flap necrosis, fat necrosis, and venous congestion in the breast, which are related to decreased flap vascularity.10–12 A meta-analysis published in 2009 reported increased breast-related complications associated with DIEP flaps compared with free TRAM flap, whereas the free muscle-sparing TRAM flap was associated with greater donor-site morbidity.2 A weakness intrinsic to these observational studies relates to the concept of “confounding by indication,” which can threaten the validity of the findings.13 Specifically, the preoperative confounding variables related to the selection of the flap have not been considered in the previous studies. Factors such as the choice of performing a DIEP versus a free TRAM flap are not random (i.e., an intended DIEP flap may be converted to a free muscle-sparing TRAM flap with inadequate intraoperative perforator anatomy). In patients who do not have a single dominant perforator, a multiperforator DIEP flap or free muscle-sparing TRAM flap may be used, and this decision is often influenced by preoperative patient characteristics. This is termed “confounding by indication” or “indication bias.”14 The most common indication biases toward the selection of a free muscle-sparing TRAM flap over a DIEP flap when a single dominant perforator is not present are in smokers or diabetics, to minimize the risk of recipient fat necrosis.2,3 Also, most microsurgeons will strive to perform multiperforator DIEP flap surgery over free muscle-sparing TRAM flap surgery in younger patients and bilateral breast reconstructions.15,16 Conventional methods such as restriction and multivariable regression adjust for outcome confounding factors but do not address confounding by indication or treatment allocation. The purpose of this study was to evaluate DIEP flaps versus muscle-sparing TRAM free flaps in terms of breast complications and abdominal morbidity
following breast reconstruction while controlling for indication bias through propensity score analyses17: Propensity score is a statistical approach to control the a priori probability of receiving one treatment over another, attempting to equate it with randomization in circumstances where this is not possible.18 The propensity score for an individual is the conditional probability of receiving a particular treatment, based on observed pretreatment covariates.17,18 In this study, the propensity score was the probability that the surgeon will convert from an intended DIEP flap to a free muscle-sparing TRAM flap based on known preoperative patient characteristics. This score is included with other baseline covariates and known confounders in the final multivariable regression analysis to reduce confounding by indication bias.19 In the absence of randomized controlled trials, the use of propensity scores in this study enabled a retrospective comparison between free muscle-sparing TRAM and DIEP flaps on abdominal and breast complications while providing the most balance between the two surgical treatment groups.
PATIENTS AND METHODS After approval from the institutional ethics review board, a consecutive series of women undergoing breast reconstruction from January 1, 2009, to August 30, 2012, at the University Health Network, Toronto, Canada, were identified using a prospectively maintained institutional breast reconstruction database. Demographic, oncologic treatment, and surgical data were obtained from this database. Demographic data included age at diagnosis, tobacco use, body mass index, diabetes, previous pregnancies, and previous abdominal operations. Oncologic data included previous irradiation and chemotherapy treatment. Surgical data included date of reconstruction, type of flap (DIEP flap versus free muscle-sparing TRAM flap), laterality (unilateral versus bilateral), and timing of reconstruction (immediate versus delayed). Surgery Our selection criteria for timing and method of abdominally based breast reconstruction are consistent with previously published guidelines.20–22 Immediate reconstruction was usually offered to women with a strong family history or gene positivity for breast cancer, in situ breast cancer, or stage I or IIa breast cancer when postmastectomy radiation therapy was unlikely. In general, autologous tissue reconstruction was our first option
775
Plastic and Reconstructive Surgery • April 2014 for women who had had previous breast or chest wall irradiation, nonpliable chest wall soft tissues, or redundant abdominal tissues, and for those undergoing delayed reconstruction or requesting this type of reconstruction. All autologous tissue reconstructions were microsurgically anastomosed to the internal mammary vessels in the third or fourth interspaces under the operating microscope by two plastic surgeons (T.Z. and S.O.P.H.). Selection of Free Muscle-Sparing TRAM versus DIEP Flap: Intraoperative Decision Making In all abdominally based breast reconstructions, DIEP flaps were the primary choice. If the intraoperative perforator anatomy was not ideal, these cases were converted to a free m uscle-sparing TRAM flap with a fascia-sparing technique, harvesting only a small cuff of fascia around each perforator (Fig. 1). When assessing perforator anatomy intraoperatively, our main determining factor to convert a DIEP flap to a free musclesparing TRAM flap was an absent perforator vein (1 cm in diameter >6 months after surgery) and breast hematoma that required evacuation in the operating room. Major abdominal complications included detection of a bulge (abdominal wall laxity without an associated fascial defect) in the standing position or hernia (with an associated fascial defect) elicited with the Valsalva maneuver. These are detected during routine follow-up and are prospectively documented in our standard postoperative clinic forms. Only those bulges that are symptomatic to the patient are corrected in a revision surgery by exploration and fascial reinforcement using a polypropylene mesh (Prolene) suture. Statistical Analysis Demographic, clinical, and surgical data were summarized and compared between the DIEP and free muscle-sparing TRAM flap groups using Fisher’s exact test for categorical variables and the Wilcoxon rank sum test for continuous variables. The overall incidence of breast and abdominal complications was compared between flap groups using Fisher’s exact test. Univariate and multivariable logistic regression were used to examine the risk factors associated with major complications. Factors included in the univariate analyses were age; body mass index; diabetes; smoking status; previous breast irradiation, chemotherapy, and abdominal surgery; and timing, laterality, and type of reconstruction (DIEP versus free muscle-sparing TRAM flap). The propensity score was defined as the probability of conversion from a planned DIEP flap
to a free muscle-sparing TRAM flap intraoperatively based on preoperative clinical characteristics including smoking status, age, and laterality of breast reconstruction.2,15,16,24–26 To estimate this probability, an initial logistic regression analysis was performed with patient clinical characteristics as the independent variables and free muscle-sparing TRAM flap (yes/no) as the dependent variable. Two final multivariable models were constructed as outlined by Harrell, which prioritizes the inclusion of key confounders of a primary predictor (reconstruction) and outcome (major complication) in the final regression model in addition to the propensity score estimate.28 The first model used abdominal complication as the dependent variable and included the following independent variables: flap reconstruction method, propensity score, body mass index, and previous chemotherapy. The second model used major breast complications as the dependent variable and the following independent variables: flap reconstruction method, propensity score, body mass index, diabetes, timing of reconstruction, previous irradiation, and previous abdominal surgery. Multivariable logistic regression models were built using stepwise backward elimination (based on the lowest Akaike information criterion value). Second-order interactions and test of collinearity were performed in all models. The area under the receiver operating characteristic curve (c-statistic) was 0.7 for the propensity score model, indicating good discrimination between patients who received free musclesparing TRAM versus DIEP flaps. Assessment of model fit using H osmer-Lemeshow goodness of fit and analysis of variance were performed in all models.29 A two-sided value of p ≤ 0.05 was considered statistically significant. Statistical analyses were performed using the R statistical package version 2.14.
RESULTS There were 292 women who underwent a total of 428 abdominally based free flap breast reconstructions from January 1, 2009, to August 30, 2012, performed by two surgeons. The mean age of the patients was 50.1 ± 8.6 years, and the mean body mass index was 27.8 ± 4.7. The majority of the women (84 percent) had DIEP flap reconstructions. Following surgery, 28 percent of the patients had one or more major complications; 20 percent had breast complications and 8 percent had abdominal complications, including three total flap losses (1 percent). Unadjusted analysis revealed no significant difference in occurrence of breast complications between the two methods of breast reconstruction;
777
Plastic and Reconstructive Surgery • April 2014 however, the unadjusted odds of abdominal complication were significantly higher in free muscle-sparing TRAM compared with DIEP flaps (OR, 3.04; 95 percent CI, 1.04 to 8.24) (Table 1). Univariate analyses examined the association of demographic, clinical, and surgical factors with overall complications (Table 2). The occurrence of major complications was not significantly different between DIEP and free muscle-sparing TRAM flaps. Bilateral reconstructions had a 1.7-times higher odds of a major complication compared with a unilateral reconstruction (OR, 1.7; 95 percent CI, 1.0 to 3.0). The odds of a major complication occurrence were 3.6 times higher in patients with diabetes (OR, 3.6; 95 percent CI, 1.1 to 12.9) compared with a nondiabetic, and for every 1-year increase in patient age, the odds of any major complication increased by 4 percent (OR, 1.04; 95 percent CI, 1.0 to 1.1). The first final multivariable model compared abdominal complications between DIEP versus free muscle-sparing TRAM flaps after controlling for obesity and prior chemotherapy treatment (Table 3). Because previous breast irradiation, abdominal surgery, and timing of reconstruction were not significantly related to abdominal complications and no diabetic patients had abdominal complications, these variables were omitted from this analysis. Previous chemotherapy and obesity had a significant interaction effect on abdominal complications; thus, an interaction variable was created. The adjusted odds of abdominal complications were 2.73 times higher in free muscle-sparing TRAM compared with DIEP flap breast reconstruction (OR, 2.73; 95 percent CI, 1.01 to 7.07). In the subgroup of patients who had prior chemotherapy, for every one-unit increase in body mass index, the odds of an abdominal complication increased by 16 percent (OR, 1.16; 95 percent CI, 1.01 to 1.34). The effect of obesity on abdominal complications was mediated by the presence of previous
chemotherapy (Fig. 2); abdominal complications were increased only in obese patients who had previous chemotherapy, whereas complications remained fairly low for those who did not have previous chemotherapy, regardless of body mass index. The second final multivariable model compared breast complications between DIEP flaps and free muscle-sparing TRAM flaps after controlling for body mass index, diabetes, previous abdominal operations, timing of breast reconstruction, and previous breast irradiation (Table 4). Previous chemotherapy and breast irradiation were collinear and therefore we included only breast irradiation in the regression model. The adjusted analysis using the propensity score showed no significant association between method of breast reconstruction and breast complications. However, diabetic patients had a 4.19-fold increased odds of breast complications (OR, 4.19; 95 percent CI, 1.14 to 15.98) compared with nondiabetic patients. Both previous abdominal operations (OR, 1.77; 95 percent CI, 0.96 to 3.30) and immediate breast reconstruction (OR, 1.86; 95 percent CI, 0.94 to 3.71) approached statistical significance as risk factors for major breast complications.
DISCUSSION This is the first study to use propensity score analysis to separately evaluate postoperative complications related to the recipient site (breast) and donor site (abdomen) following muscle-sparing TRAM and DIEP free flap breast reconstruction while accounting for the confounding factors that contribute to selection bias with the use of a propensity score analysis. It is important to analyze the breast and abdominal complications separately, because one flap method may be superior for achieving breast outcomes but inferior in abdominal outcomes in the same patient. Our initial analysis indicated no difference in overall
Table 1. Major Complications following Breast Reconstruction with DIEP or Muscle-Sparing TRAM Free Flaps No. of patients Complications Breast complications Total flap loss Partial flap loss Fat necrosis Hematoma in breast Abdominal complications Bulge Hernia MS, muscle-sparing.
778
DIEP Flap (%)
Free MS-TRAM Flap (%)
244
48
50 (20.5) 3 (1.2) 6 (2.5) 25 (10.2) 17 (7.0) 15 (6) 11 (4.5) 4 (1.5)
10 (20.8) 0 0 7 (14.6) 3 (6.2) 8 (17) 7 (14) 1 (2)
OR (95% CI) 1.02 (0.42–2.27) 1.41 (0.07–27.6) 2.64 (0.15–47.47) 0.67 (0.26–1.96) 1.12 (0.31–6.22) 3.04 (1.04–8.24) 0.27 (0.09–0.89) 1.39 (0.17–63.86)
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications Table 2. Univariate Analysis of Major Overall Complications Major Complications Variable Categorical Method DIEP MS-TRAM Laterality Unilateral Bilateral Timing Immediate Delayed Diabetes No Yes Smoker Yes No Previous abdominal operations Yes No Previous breast irradiation Yes No Previous chemotherapy Yes No Continuous Age, yr BMI
Total (%)
No (%)
Yes (%)
OR (95% CI)
244 (84) 48 (16)
184 (75) 32 (67)
60 (25) 16 (33)
1.5 (0.7–3.1)
156 (53) 136 (47)
123 (79) 93 (68)
33 (21) 43 (32)
1.7 (1.0–3.0)
134 (46) 158 (54)
91 (68) 125 (79)
43 (32) 33 (21)
0.56 (0.3–0.9)
281 (96) 11 (4)
211 (75) 5 (45)
70 (25) 6 (55)
3.6 (1.1–12.9)
25 (9) 267 (91)
17 (68) 199 (75)
8 (32) 68 (25)
0.7 (0.3–1.8)
135 (46) 157 (54)
94 (70) 122 (78)
41 (30) 35 (22)
0.7 (0.4–1.2)
146 (50) 146 (50)
112 (77) 104 (71)
34 (23) 42 (29)
1.3 (0.2–1.7)
163 (56) 129 (44)
124 (76) 92 (71)
39 (24) 37 (29)
1.3 (0.7–2.2)
50.1 ± 8.6 27.8 ± 4.7
49.4 ± 8.5 27.5 ± 4.5
52.2 ± 8.6 28.5 ± 5.1
1.04 (1.0–1.1 1.05 (0.9–1.1
MS, muscle-sparing; BMI, body mass index.
major complications following breast reconstruction using DIEP or muscle-sparing TRAM free flaps (Table 2). However, when the donor and recipient sites were analyzed separately, abdominal complications were significantly higher in muscle-sparing TRAM flaps compared with DIEP flaps (Table 3). The unique advantage of using a propensity score adjustment is that comparability between the comparison groups is improved by creating pseudorandomization of measured confounders in clinical situations where a randomized controlled trial is not feasible.18 Analyses with propensity scores revealed a strong association between breast complications and diabetes Table 3. Multivariate Logistic Regression Model with Propensity Score for Abdominal Complications Variable Categorical Method DIEP Free MS-TRAM Continuous BMI With previous chemotherapy Without previous chemotherapy
OR (95% CI)
p
* 2.73 (1.01–7.07)
0.04
1.16 (1.01–1.34)
0.02
0.93 (0.81–1.06)
0.28
MS, muscle-sparing; BMI, body mass index. *Reference category.
but no difference in type of flap. Also, increased abdominal complications were associated with previous chemotherapy in those with a higher body mass index. The DIEP flap was developed as a derivative of the muscle-sparing free flap to minimize morbidity related to abdominal wall integrity.30 In two recent meta-analyses, the authors independently found the pooled relative risk of abdominal bulge or hernia to be lower in DIEP flaps compared with muscle-sparing TRAM free flaps.2,31 Similarly, we found an adjusted OR of 2.73 (95 percent CI, 1.01 to 7.07) for bulge/hernia formation in free muscle-sparing TRAM compared with DIEP flaps. Because we found a significant interaction between body mass index and previous chemotherapy, our conclusions were based on analyzing the effect of obesity on abdominal complications stratified by previous chemotherapy. The stratum-specific estimates provided a more valid representation of the effect modification observed in this study, and it was likely the combination of obesity and the detrimental effects of chemotherapy on wound healing that increased the risk of abdominal complications in this subgroup. It is believed that the large and heavy abdominal flap stretches and attenuates the musculocutaneous perforators in
779
Plastic and Reconstructive Surgery • April 2014
Fig. 2. Effect modification of previous chemotherapy on body mass index (BMI) and the association with abdominal complications.
an obese patient, leading to an abdominal flap that is compromised in its blood supply. This combined with the detrimental effect of chemotherapeutic agents on the macrophages and fibroblasts that are intimately involved in wound healing can result in poor healing within the important layers of the abdominal wall, including the fascial Table 4. Multivariate Logistic Regression Model with Propensity Score for Breast Complications Variable Categorical Method DIEP Free MS-TRAM Diabetes No Yes Timing Delayed Immediate Previous abdominal operations No Yes Previous breast irradiation No Yes Continuous BMI
OR (95% CI)
p
* 0.98 (0.40–2.14)
0.95
* 4.19 (1.14–15.98)
0.02
* 1.86 (0.94–3.71)
0.07
* 1.77 (0.96–3.30)
0.07
* 1.23 (0.67–2.29)
0.51
1.02 (0.96–1.09)
0.48
*Reference category. MS, muscle-sparing; BMI, body mass index.
780
repair, leading to problems with bulges and hernias. Our finding that DIEP flaps likely represent the preferred autologous breast reconstruction technique for obese patients to minimize their abdominal morbidity corroborates the findings reported by Garvey et al.25 We did not observe a significant difference in major breast complications between DIEP and muscle-sparing TRAM free flaps. This is contrary to reports by Sailon et al.3 and Man et al.,2 who reported a higher frequency of breast complications in DIEP flaps compared with muscle-sparing TRAM flaps. However, addition of the propensity score analysis indicated that the presence of diabetes was a significant predictor of major breast complications. Similar to previous studies, we also identified immediate reconstruction and previous abdominal operations to be risk factors for breast complications that approached statistical significance.20 Our finding that previous abdominal surgical scars may predispose a patient to increased breast flap complications is in contradiction to a recent study that reported similar flap complications between patients with and without Pfannenstiel scars.32 These authors found the number of larger-caliber skin perforators to be significantly higher in DIEP flaps that have Pfannenstiel scars compared with the virginal abdomens using
Volume 133, Number 4 • DIEP and Free TRAM Flap Complications computed tomographic angiography.32 Although low transverse abdominal incisions, such as Pfannenstiel incisions, do not generally impact the concentration of larger perforators around the periumbilical region, it is possible that other incisions located more centrally in the abdomen may affect perfusion of the DIEP flap, as may be the case for our study. Because this subgroup analysis may have been underpowered, it is plausible that both immediate reconstruction and previous abdominal operations may pose “real” threats to breast reconstruction outcomes in a larger sample. The multivariable regression using propensity score analysis allowed the opportunity to control for selection bias and confounding by indication. This is particularly important when there is a relatively large number of confounders compared with the number of major complication outcomes. This unique approach of using propensity scores allowed us to consider not only the factors that affected outcome but also the factors that contributed to selection of the reconstruction procedure. The limitations to this study include an observational study with a retrospective design and a small sample of muscle-sparing TRAM flaps. In addition, our study was undertaken at a cancer center with a high volume of reconstructive procedures, and the complications reported may not be representative of all institutions. The challenge of comparing a series of patients with DIEP and muscle-sparing TRAM free flap reconstruction is recognition of the differences between surgeons regarding surgical approaches, techniques, and postoperative evaluation mechanisms. In our study, all operations were performed by two experienced microsurgeons who shared the same intraoperative approach and postoperative regimens. Although the use of the propensity score added another degree of reliability to our results, the interpretation of our conclusions remains limited to a retrospective study of patients who were operated on by two surgeons who followed the intraoperative flap selection guide as outlined in Figure 1. Another caveat is that Figure 1 is intended to reflect the clinical decision-making process for the two surgeons when selecting a DIEP versus a free muscle-sparing TRAM flap for our series, and is not intended to be an instructional algorithm to be followed by all surgeons. As such, the preoperative clinical variable “younger age” does not reflect a specific age cut-off; rather, it only represented the surgeon’s gestalt. In addition, although the two surgeons routinely use preoperative computed tomographic angiography to map out the perforator anatomy of the abdominal
wall to expedite perforator dissection, the final decision to use a DIEP or muscle-sparing TRAM flap nevertheless is ultimately based on the intraoperative assessment of the perforators and the key patient characteristics as outlined in Figure 1. Based on our study findings, we will continue to use the DIEP flap as our first choice in patients with adequate single-perforator anatomy and strive to dissect multiperforator DIEP flaps in patients with insufficient single-perforator anatomy as much as possible to achieve the best abdominal outcome. Lastly, our study also highlights the need for improved methods of abdominal fascial closure in patients who have undergone free muscle-sparing TRAM flap surgery, and our current approach to always close fascia primarily after a fascia-sparing elevation of the free muscle-sparing TRAM flap is clearly an oversimplified solution to the complex dynamics of the abdominal musculature. The ideal method of abdominal wall closure following muscle-sparing TRAM flap surgery is an important area of research that warrants a separate study of its own.
CONCLUSIONS Using the propensity score analysis, abdominal complications were significantly higher in free muscle-sparing TRAM flaps compared with DIEP flaps. The effect of obesity on abdominal complications was mediated by the presence of previous chemotherapy. There was no significant association between type of flap and breast complications; however, diabetes was a significant risk factor. This study highlights the importance of evaluating recipient breast complications separately from donor abdominal complications and the use of propensity scores to account for preoperative patient factors to reduce confounding by indication. Toni Zhong, M.D., M.H.S. University Health Network 8NU-871 200 Elizabeth Street Toronto, Ontario M5G 2C4, Canada [email protected]
REFERENCES 1. Bajaj AK, Chevray PM, Chang DW. Comparison of d onor-site complications and functional outcomes in free muscle-sparing TRAM flap and free DIEP flap breast reconstruction. Plast Reconstr Surg. 2006;117:737–746; discussion 747. 2. Man LX, Selber JC, Serletti JM. Abdominal wall following free TRAM or DIEP flap reconstruction: A meta-analysis and critical review. Plast Reconstr Surg. 2009;124:752–764.
781
Plastic and Reconstructive Surgery • April 2014 3. Sailon AM, Schachar JS, Levine JP. Free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps for breast reconstruction: A systematic review of flap complication rates and donor-site morbidity. Ann Plast Surg. 2009;62:560–563. 4. Chen CM, Halvorson EG, Disa JJ, et al. Immediate postoperative complications in DIEP versus free/muscle-sparing TRAM flaps. Plast Reconstr Surg. 2007;120:1477–1482. 5. Nahabedian MY, Dooley W, Singh N, Manson PN. Contour abnormalities of the abdomen after breast reconstruction with abdominal flaps: The role of muscle preservation. Plast Reconstr Surg. 2002;109:91–101. 6. Momoh AO, Colakoglu S, Westvik TS, et al. Analysis of complications and patient satisfaction in pedicled transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flap breast reconstruction. Ann Plast Surg. 2012;69:19–23. 7. Nahabedian MY, Tsangaris T, Momen B. Breast reconstruction with the DIEP flap or the muscle-sparing (MS-2) free TRAM flap: Is there a difference? Plast Reconstr Surg. 2005;115:436–444; discussion 445–436. 8. Garvey PB, Salavati S, Feng L, Butler CE. Perfusion-related complications are similar for DIEP and muscle-sparing free TRAM flaps harvested on medial or lateral deep inferior epigastric artery branch perforators for breast reconstruction. Plast Reconstr Surg. 2011;128:581e–589e. 9. Nelson JA, Guo Y, Sonnad SS, et al. A comparison between DIEP and muscle-sparing free TRAM flaps in breast reconstruction: A single surgeon’s recent experience. Plast Reconstr Surg. 2010;126:1428–1435. 10. Nahabedian MY, Momen B, Galdino G, et al. Breast reconstruction with the free TRAM or DIEP flap: Patient selection, choice of flap, and outcome. Plast Reconstr Surg. 2002;110:466–475; discussion 476–467. 11. Kroll SS. Fat necrosis in free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps. Plast Reconstr Surg. 2000;106:576–583. 12. Blondeel PN, Arnstein M, Verstraete K, et al. Venous congestion and blood flow in free transverse rectus abdominis myocutaneous and deep inferior epigastric perforator flaps. Plast Reconstr Surg. 2000;106:1295–1299. 13. Giordano SH, Kuo YF, Duan Z, et al. Limits of observational data in determining outcomes from cancer therapy. Cancer 2008;112:2456–2466. 14. Salas M, Hofman A, Stricker BH. Confounding by indication: An example of variation in the use of epidemiologic terminology. Am J Epidemiol. 1999;149:981–983. 15. Blondeel N, Vanderstraeten GG, Monstrey SJ, et al. The donor site morbidity of free DIEP flaps and free TRAM flaps for breast reconstruction. Br J Plast Surg. 1997;50: 322–330. 16. Nahabedian MY, Manson PN. Contour abnormalities of the abdomen after transverse rectus abdominis muscle flap breast reconstruction: A multifactorial analysis. Plast Reconstr Surg. 2002;109:81–87; discussion 88–90.
782
17. Austin PC. The performance of different propensity score methods for estimating marginal odds ratios. Stat Med. 2007;26:3078–3094. 18. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:41–55. 19. D’Agostino RB Jr. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med. 1998;17:2265–2281. 20. Zhong T, Hofer SO, McCready DR, et al. A comparison of surgical complications between immediate breast reconstruction and mastectomy: The impact on delivery of chemotherapy. An analysis of 391 procedures. Ann Surg Oncol. 2012;19:560–566. 21. Styblo TM, Lewis MM, Carlson GW, et al. Immediate breast reconstruction for stage III breast cancer using transverse rectus abdominis musculocutaneous (TRAM) flap. Ann Surg Oncol. 1996;3:375–380. 22. Buchholz TA, Kronowitz SJ, Kuerer HM. Immediate breast reconstruction after skin-sparing mastectomy for the treatment of advanced breast cancer: Radiation oncology considerations. Ann Surg Oncol. 2002;9:820–821. 23. Disa JJ, Pusic AL, Zhong T. Muscle-sparing and free TRAM flap breast reconstruction. In: Hall-Findlay EJ, Evans GRD, eds. Aesthetic and Reconstructive Surgery of the Breast. St. Louis: Saunders Elsevier; 2010:93–116. 24. Chang DW, Reece GP, Wang B, et al. Effect of smoking on complications in patients undergoing free TRAM flap breast reconstruction. Plast Reconstr Surg. 2000;105:2374–2380. 25. Garvey PB, Buchel EW, Pockaj BA, et al. The deep inferior epigastric perforator flap for breast reconstruction in overweight and obese patients. Plast Reconstr Surg. 2005;115:447–457. 26. Scheer AS, Novak CB, Neligan PC, Lipa JE. Complications associated with breast reconstruction using a perforator flap compared with a free TRAM flap. Ann Plast Surg. 2006;56:355–358. 27. Nahabedian MY, Momen B, Manson PN. Factors associated with anastomotic failure after microvascular reconstruction of the breast. Plast Reconstr Surg. 2004;114:74–82. 28. Harrell FE. Regression Modeling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. New York: Springer-Verlag; 2001. 29. Sturmer T, Joshi M, Glynn RJ, et al. A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol. 2006;59:437–447. 30. Allen RJ, Treece P. Deep inferior epigastric perforator flap for breast reconstruction. Ann Plast Surg. 1994;32:32–38. 31. Egeberg A, Rasmussen MK, Sørensen JA. Comparing the donor-site morbidity using DIEP, SIEA or MS-TRAM flaps for breast reconstructive surgery: A meta-analysis. J Plast Reconstr Aesthet Surg. 2012;65:1474–1480. 32. Mahajan AL, Zeltzer A, Claes KE, et al. Are Pfannenstiel scars a boon or a curse for DIEP flap breast reconstructions? Plast Reconstr Surg. 2012;129:797–805.
