Primary fascial closure with biologic mesh reinforcement results in lesser complication and recurrence rates than bridged biologic mesh repair for abdominal wall reconstruction: A propensity score analysis

ARTICLE IN PRESS

Primary fascial closure with biologic mesh reinforcement results in lesser complication and recurrence rates than bridged biologic mesh repair for abdominal wall reconstruction: A propensity score analysis Salvatore Giordano, MD, PhD, Patrick B. Garvey, MD, FACS, Donald P. Baumann, MD, FACS, Jun Liu, PhD, and Charles E. Butler, MD, FACS, Houston, TX

Background. Previous studies suggest that bridged mesh repair for abdominal wall reconstruction may result in worse outcomes than mesh-reinforced, primary fascial closure, particularly when acellular dermal matrix is used. We compared our outcomes of bridged versus reinforced repair using ADM in abdominal wall reconstruction procedures. Methods. This retrospective study included 535 consecutive patients at our cancer center who underwent abdominal wall reconstruction either for an incisional hernia or for abdominal wall defects left after excision of malignancies involving the abdominal wall with underlay mesh. A total of 484 (90%) patients underwent mesh-reinforced abdominal wall reconstruction and 51 (10%) underwent bridged repair abdominal wall reconstruction. Acellular dermal matrix was used, respectively, in 98% of bridged and 96% of reinforced repairs. We compared outcomes between these 2 groups using propensity score analysis for risk-adjustment in multivariate analysis and for 1-to-1 matching. Results. Bridged repairs had a greater hernia recurrence rate (33.3% vs 6.2%, P < .001), a greater overall complication rate (59% vs 30%, P = .001), and worse freedom from hernia recurrence (log-rank P <.001) than reinforced repairs. Bridged repairs also had a greater rate of wound dehiscence (26% vs 14%, P = .034) and mesh exposure (10% vs 1%, P = .003) than mesh-reinforced abdominal wall reconstruction. When the treatment method was adjusted for propensity score in the propensity-score– matched pairs (n = 100), we found that the rates of hernia recurrence (32% vs 6%, P = .002), overall complications (32% vs 6%, P = .002), and freedom from hernia recurrence (68% vs 32%, P = .001) rates were worse after bridged repair. We did not observe differences in wound healing and mesh complications between the 2 groups. Conclusion. In our population of primarily cancer patients at MD Anderson Cancer Center bridged repair for abdominal wall reconstruction is associated with worse outcomes than mesh-reinforced abdominal wall reconstruction. Particularly when employing acellular dermal matrix, reinforced repairs should be used for abdominal wall reconstruction whenever possible. (Surgery 2016;j:j-j.) From the Department of Plastic Surgery, The University of Texas M.D. Anderson Cancer Center, Houston, TX

Supported in part by the National Institutes of Health through M.D. Anderson’s Cancer Center Support Grant CA016672. Dr Garvey is a consultant for Acelity Corporation. No other author has a relevant financial conflict with the material presented in this manuscript. Presented at the European Association of Plastic Surgeons (Euraps) Research Council, May 25–26, 2016, Bruxelles, Belgium.

Accepted for publication August 10, 2016. Reprint requests: Charles E. Butler, MD, FACS, The University of Texas M.D. Anderson Cancer Center, 1400 Pressler St, Unit 1488, Houston, TX 77030. E-mail: [email protected]. 0039-6060/$ - see front matter Ó 2016 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.surg.2016.08.009

SURGERY 1

ARTICLE IN PRESS 2 Giordano et al

THE INCIDENCE OF ABDOMINAL WALL RECONSTRUCTION (AWR), including ventral hernia repair, has been increasing annually in the United States over the past years, with over 350,000 such operations performed in 2014.1 AWR can be technically challenging, especially when large defects are encountered. Previous studies suggest that primary fascial closure with mesh reinforcement improves outcomes, resulting in decreased rates of hernia recurrence.2-6 Nevertheless, surgeons still commonly employ techniques of bridged mesh repairs, owing to limited definitive evidence demonstrating the benefits of primary fascial closure to decrease rates of hernia recurrence.7 To date, few studies have compared bridged and mesh-reinforced repairs, and the data available are confounded by substantial heterogeneity within studies, particularly regarding defect size.2-4,7 In general, larger abdominal wall defects are more likely to be repaired with a bridged mesh technique than with complete fascial closure, especially with laparoscopic repairs,7 while for the smaller defects, abdominal fascia can be reapproximated and reinforced with mesh.2-4 Indeed, the nature of the abdominal defect is different in patients who undergo bridged repair versus primary fascial closure. In addition, in the available studies, the comorbidities, types of operative defects, and treatment factors vary, which can affect the operative approach and, consequently, the outcomes. Furthermore, the duration of follow-up is very heterogeneous among the published studies, further complicating interpretation of the data.8 Thus, the comparison of these 2 repair techniques in randomized controlled trials is difficult due to problems with determining comparable groups and associated confounding.9 To overcome this issue, we analyzed our longterm results of bridged and mesh-reinforced AWR by adjusting the baseline differences between the 2 groups using propensity score analysis.10 We hypothesized that primary fascial closure with underlay mesh placement decreases hernia recurrence rates in comparison to bridged fascial repairs. METHODS We evaluated retrospectively all consecutive patients who underwent midline AWR with underlay (preperitoneal or intraperitoneal) mesh due to a ventral hernia or oncologic resection defect. Over 95% of repairs were performed with a biologic, acellular dermal matrix (ADM) mesh. The study was performed at The University of

Surgery j 2016

Texas M.D. Anderson Cancer Center and included patients who underwent AWR both for incisional hernias (about two thirds) and for abdominal wall defects created by radical resections for cancer involving the abdominal wall (about one third of patients) between March 2005 and October 2015. We followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines for observational cohort studies.11 The clinical investigation for the present study was conducted in accordance with the ethical principles of the World Medical Association Declaration of Helsinki and the laws of the United States. The Institutional Review Board approved this study, and individual informed consent was waived, because the source data were deidentified. The outcomes of patients who underwent a bridged fascial closure with mesh (experimental group) were compared with those of patients who underwent a mesh-reinforced, primary fascial closure (control group). Patients with defects that did not involve the midline (lateral defects), primary closure of their abdominal wall fascia without mesh, onlay mesh reconstructions, or defects reconstructed or bridged with tissue from free or local musculocutaneous or fasciocutaneous flaps or fascial grafts were excluded from the study. Patients were followed up with physical examination and computed tomography (CT) imaging (88.8% of cases had a postoperative CT at followup). The AWR follow-up was robust, because our institution serves an oncologic population; both clinical and radiologic surveillance were accomplished according to the tumor protocol individually of each patient, typically quarterly for the first year and then annually thereafter. Data were collected both from a prospectively maintained, departmental database and from electronic medical records. Patient, treatment, and defect characteristics were analyzed, and operative outcomes were directly compared between the experimental and control groups. Wounds were considered contaminated if they met the definition of the American College of Surgeon National Surgical Quality Improvement Program of contaminated or infected (class 3–4).12 We defined obesity as a body mass index (BMI) $30 kg/m2.13 Patients who smoked tobacco within 1 month of operation were considered to be active smokers. The primary outcome measure was development of a hernia after AWR. Secondary outcome measures included the incidences of the following postoperative complications: bulging or laxity of the abdominal wall and wound healing

ARTICLE IN PRESS Surgery Volume j, Number j

complications (skin dehiscence, skin necrosis, fat necrosis, cellulitis, abscess, hematoma, and seroma). Development of a hernia was defined as a contour abnormality associated with a fascial defect detected by physical examination and/or CT, while bulging was a contour abnormality without a fascial defect. Hernia and bulge were considered mutually exclusive conditions. Wound dehiscence was defined as a skin breakdown with full-thickness skin separation extending >2 cm with or without infection, while skin necrosis involved clearly demarcated necrotic skin edges >1 cm in width. Fat necrosis was a palpable firmness 1 cm or greater in diameter that persisted beyond 3 months postoperatively. Infection was an infectious process (cellulitis/abscess) requiring treatment with intravenous or oral antibiotics with or without an operation. Hematoma and seroma were subcutaneous collections of blood or serous fluid, respectively, requiring percutaneous or operative drainage. Operative technique. The choice of the type of mesh was left to the surgeon’s discretion, but, in this study, bioprosthetic mesh materials were used in over 95% of patients in contrast to synthetic prosthetics. The indications for use of bioprosthetic mesh included bacterial contamination, unavoidable direct placement of mesh over viscera, high-risk patients with multiple comorbidities, and compromised soft tissue coverage over the AWR with increased risk of wound complications. The utilization of bioprosthetic mesh allowed for a potential decrease in mesh-related infection, adhesions, enterocutaneous fistulae, and risk of inadvertent enterotomy at subsequent reoperation. A multidisciplinary approach was used in performing AWR, with a similar general technique for all cases. The surgical oncologist performed the laparotomy, adhesiolysis, and tumor resection (if required).14 The reconstructive surgeon defined the defect, including excision of the hernia sac and debridement of devitalized tissue and fascia, and decided whether lateral anterior fascial release was necessary in order to facilitate the medialization of rectus muscles. When deemed necessary, anterior open or minimally invasive15,16 component separation, including release of the external oblique aponeurosis from the pubis to above the costal margin, was performed to provide lateral release and to decrease tension from the midline fascial closure.16 The indication for a component separation in AWR was an inability to approximate the fascial edges without excessive tension that might place the repair at risk of failure.

Giordano et al 3

An underlay or acellular dermal matrix (ADM) or occasionally a synthetic mesh (<5%) reinforced the midline fascial repair with 3 to 5 cm of abdominal wall overlap, fixed circumferentially with interrupted #1 polypropylene sutures, typically spaced approximately 1.5 to 2 cm apart, followed by midline primary fascial closure over the mesh with interrupted #1 polypropylene sutures.17 When the fascial defect could not be entirely closed primarily over the mesh, the prosthesis was left in place as a bridge at the point of maximal tension to span the residual defect using an underlay technique secured with circumferential sutures15,16 and a dual circumferential inlay technique.18 Decisions concerning the operative techniques and whether to perform a mesh-reinforced primary fascial closure or bridged repair were at the discretion of the reconstructive surgeon. When the defect could not be reinforced and necessitated bridging, a component separation was still generally performed in order to decrease the size of the bridged portion of the closure. Scarred, nonviable, and/or redundant skin was resected and subcutaneous drains were placed in attempt to decrease the risk of seroma formation.15,16 Statistical analysis. Continuous variables are reported as the mean ± standard deviation. The Pearson v2 test, Fisher exact test, and the MannWhitney test were used for univariate analysis. Logistic regression with backward selection was performed to calculate the likelihood of patients to be included in either the bridged repair or fascial closure group. Hosmer-Lemeshow test was used to assess the regression model fit. Variables included in the regression model are shown in Table I. Analysis using a receiver operating characteristic (ROC) curve was used to estimate the area under the curve of the model predicting the probability of being included in the bridged repair or fascial closure group. The calculated propensity score was employed for 1-to-1 matching as well as to adjust for other variables in estimating their impact on the postoperative outcomes.10 One-to-one propensity score matching between study groups was done using the nearest neighbor method and a caliper of 0.2 of the standard deviation of the logit of the propensity score.19 Continuous outcomes were compared in the propensity score–matched groups using a paired t test or Wilcoxon signed rank test as appropriate; differences in proportions were compared using the McNemar or binominal test as appropriate.

ARTICLE IN PRESS 4 Giordano et al

Surgery j 2016

Table I. Baseline characteristics of patients who underwent bridged repair or reinforced fascia closure for abdominal wall reconstruction Overall series

Age (y)* Women BMI* (kg/m2) Comorbidity Coronary artery disease Diabetes* Hypertension Pulmonary disease Renal disease ASA status* Smoking* Preoperative abdominal wall XRT Postoperative abdominal wall XRT Preoperative chemotherapy Postoperative chemotherapy Indication for AWR Complex hernia Defect width,* (cm) Defect area,* (cm2) Contamination grades 3 or 4* Prior abdominal operation History of hernia repair* Follow-up (mo)

Propensity score–matched pairs

Bridged

Reinforced

Bridged

Reinforced

n = 51

n = 484

n = 50

n = 50

P value

59 ± 12 24 (47.1%) 32 ± 7 44 (86.3%) 9 (17.6%) 9 (17.6%) 29 (56.9%) 6 (11.8%) 10 (19.6%) 3.0 ± 0.4 12 (23.5%) 10 (32.3%)

59.2 ± 12.3 254 (52.5%) 31.0 ± 6.9 400 (90.1%) 56 (10.7%) 84 (17.4%) 260 (53.7%) 43 (8.9%) 51 (10.5%) 2.9 ± 0.4 132 (27.3%) 126 (26.1%)

.870 .461 .581 .512 .140 .958 .668 .498 .053 .190 .566 .450

59 ± 12 23 (46%) 32 ± 7 43 (86%) 9 (18%) 9 (18%) 28 (56%) 6 (12%) 10 (20%) 3.0 ± 0.4 12 (24%) 10 (20%)

56.7 ± 12.8 22 (44.0%) 31.5 ± 7.1 41 (82.0%) 8 (16.0%) 8 (16.0%) 25 (50.0%) 6 (12.0%) 3 (6.0%) 3.0 ± 0.5 15 (30.0%) 18 (36.0%)

.247 1.000 .905 .791 1.000 1.000 .664 1.000 .092 .670 .453 .896

3 (5.9%)

21 (4.2%)

.813

3 (6%)

2 (4.0%)

1.000

26 (51.0%) 13 (25.5%)

295 (61.0%) 68 (14.1%)

.167 .031

25 (50%) 13 (26%)

28 (56.0%) 8 (16.0%)

.690 .249

20 (39.2%) 14.1 ± 5.0 337 ± 196 22 (43.1%) 48 (94.1%) 17 (33.3%) 28 ± 24

264 (54.7%) 11.0 ± 5.3 216.4 ± 183.7 79 (16.3%) 465 (96.1%) 113 (23.3%) 30.8 ± 22.2

.107 <.001 <.001 <.001 .503 .114 .350

19 (38%) 14.2 ± 5.1 338 ± 197 21 (42%) 47 (94%) 17 (34%) 27.9 ± 24.2

23 (46.0%) 14.3 ± 6.9 308.5 ± 246.4 21 (42.0%) 48 (96.0%) 16 (30.0%) 30.0 ± 23.3

.571 .770 .461 1.000 1.000 1.000 .431

P value

*Variable included into regression model for estimation of the propensity score. Data are x ± SD or percentage. Statistically significant comparisons are indicated in italics. AWR, Abdominal wall reconstruction; XRT, radiotherapy.

The outcome of development of a hernia postoperatively was evaluated by Kaplan-Meier methods in the overall series and in the propensitymatched pairs. Univariate and multivariate logistic regression models were used to determine the associations between the development of a hernia and patient and operative variables. A stepwise model selection method was used to fit a multivariate regression model. A post hoc statistical power calculation was performed for primary end point variables before and after propensity score matching.20 All analyses were carried out using SPSS statistical software (version 23; IBM SPSS Statistics, Armonk, NY). Data are presented as the mean ± standard deviation (SD). RESULTS We identified 535 consecutive patients who underwent AWR with underlay mesh (51 bridged;

484 mesh reinforced) during the period of the study. The mean follow-up was 30.5 ± 22.3 months. Among the bridged reconstructions, the mean (±SD) bridge width was 6.7 ± 3.2 cm and the mean bridge area was 79 ± 65 cm2. Our surgeons employed ADM rather than synthetic mesh in 96.3% of the cases. Results of bridged versus mesh-reinforced AWR in the overall series. Baseline characteristics are outlined in Table I, showing significant difference in abdominal defect width, defect area, and degree of contamination between the 2 groups, which were greater in the bridged cases. Mean follow-up was similar in both groups: 28 months vs 30.8 months for bridged and reinforced repairs, respectively. Our surgeons employed synthetic mesh in only 3.3% of reinforced cases and never in bridged repair (Table II). The most frequently used bioprosthetic matrices included non-cross–linked porcine ADM (47.1%; Strattice; Acelity Corp, Bridgewater, NJ) followed by non-cross–linked

ARTICLE IN PRESS Giordano et al 5

Surgery Volume j, Number j

Table II. Reconstruction characteristics and perioperative data for patients who underwent bridged repair or reinforced fascia closure for abdominal wall reconstruction Overall series

Operative time (min) Estimated blood loss (L) Component separation Minimally invasive Mesh type Synthetic Bioprosthetic Both Synthetic mesh type Polypropylene Polygalactin 910 ePTFE Seri scaffold Bioprosthetic mesh type HADM PADM BADM Bovine pericardium Hospital stay (days)

Bridged

Reinforced

n = 51

n = 484

429 ± 193 0.51 ± 0.78 30 (59%) 15 (50%)

Propensity score–matched pairs Bridged

Reinforced

P value

n = 50

n = 50

P value

368.7 ± 193.8 0.50 ± 1.14 339 (70.0%) 129 (37.4%)

.035 .949 .100 .173

427 ± 194 0.50 ± 0.8 29 (58%) 15 (23%)

414.5 ± 220.7 0.94 ± 2.2 36 (72.0%) 14 (21.5%)

.768 .357 .210 1.000

0 (0%) 50 (98%) 1

16 (3.3%) 465 (96.1%) 2

.187 .482

0 49 (98%) 1 (2%)

1 (2.0%) 49 (98.0%) 0 (0.0%)

1.000 1.000 1.000

1 (2%) 0 0 0

16 (3.3%) 1 1 4

.602

1 (2%) 0 0 0

1 (2.0%) 0 0 0

1.000

4 28 (55%) 19 (37%) 0 (0%) 14 ± 7

22 224 (46.3%) 204 (42.1%) 1 (0.2%) 10.0 ± 19.1

4 (8%) 28 (56%) 18 (36%) 0 (0%) 14 ± 7

3 (6.0%) 23 (46.0%) 23 (46.0%) 0 (0.0%) 9.9 ± 10.0

1.000 .405 .383 1.000 .023

1.000 .001

Data are x ± SD or percentage. Statistically significant comparisons are indicated in italics. BADM, Bovine acellular dermal matrix; ePTFE, expanded polytetrafluoroethylene; HADM, human acellular dermal matrix; PADM, porcine acellular dermal matrix.

bovine ADM (41.7%; SurgiMend; TEI Biosciences, Inc, Boston, MA). Component separation was performed in 68.9% of cases, with similar frequency between the 2 comparison groups. Patients undergoing a bridged repair had operations of greater duration (429 minutes vs 369 minutes, P = .035) and longer hospital stays (14 days vs 10 days, P = .001, Table II) compared to the reinforcedrepair patients. The bridged repair group experienced a >7fold greater rate of hernia development (33% vs 6%) than the reinforced repair patients (odds ratio [OR] 7.6, 95% confidence interval [CI] 3.8–15.1; P = .001, Table III). Compared to the reinforced repair group, the bridged repair group also demonstrated a greater overall complication rate (59% vs 30.0%; OR 3.3, 95% CI 1.8–6.0; P = .001), skin dehiscence rate (26% vs 14.3%; OR 2.0, 95% CI 1.0–4.0; P = .034), and mesh exposure rate (10% vs 1.4%; OR 7.4, 95% CI 2.3–24.3; P = .003). Kaplan-Meier curves showed a significant difference in freedom from development of a hernia between the 2 groups (Fig 1). Hernia recurrences were repaired operatively in 49% of cases, performing a component separation if not previously performed and using a new biologic or synthetic mesh reinforcement in both groups. Of

note, bridging repairs that failed were found to have separation at the musculofascial-mesh interface in all cases. Univariate analysis showed bridged repair, overall complications, skin dehiscence, defect width, mesh exposure, infection, bulging/laxity, and use of human cadaveric ADM to be associated with development of a hernia (Table IV). These factors were then analyzed with multivariate logistic regression, which confirmed bridged repair (OR 6.3, 95% CI 2.8–14.2; P < .001), overall complications (OR 4.5, 95% CI 2.0–9.8; P < .001), and the use of human cadaveric ADM (OR 3.9, 95% CI 1.2–12.2; P = .020) to be independent predictors of hernia recurrence (Table IV). Results of the propensity score analysis. Age (beta-coefficient –0.020), BMI (beta-coefficient –0.013), diabetes (beta-coefficient 0.173), American Society of Anesthesiologists (ASA) status (beta-coefficient 0.241), defect width (beta-coefficient 0.057), defect area (beta-coefficient 0.001), contamination grade 3 or 4 (beta-coefficient –1.356), and prior hernia repair (beta-coefficient –0.347) were used as independent predictors for assigning patients to the bridged repair or reinforced group (constant beta-coefficient –1.412, Hosmer-Lemeshow test; P = .115) (Table I).

ARTICLE IN PRESS 6 Giordano et al

Surgery j 2016

Table III. Postoperative outcomes of patients who underwent bridged repair or reinforced fascia closure for abdominal wall reconstruction Overall series

Recurrent hernia Time to recurrence (mo) Bulging/laxity Overall complications Abdominal wall complications Wound healing complications Skin dehiscence Fat necrosis Infection cellulitis Infection abscess Hematoma Seroma Mesh, exposure Mesh, infected Mesh, removal Reoperation

Propensity score–matched pairs

Bridged

Reinforced

n = 51

n = 484

17 (33%) 18 ± 13 4 (8%) 30 (59%) 15 (29%)

30 (6.2%) 19.0 ± 12.4 17 (3.5%) 145 (30.0%) 111 (22.9%)

13 3 3 4

(26%) (6%) (6%) (8%) 1 3 (6%) 5 (10%) 1 0 5 (10%)

69 33 32 32 6 20 7 4 6 43

(14.3%) (6.8%) (6.6%) (6.6%) (1.2%) (4.1%) (1.4%) (0.8%) (1.2%) (8.9%)

Bridged

Reinforced

P value

n = 50

n = 50

P value

<.001 .797 .131 .001 .300

16 (32%) 18 ± 14 4 (8%) 30 (68%) 15 (30%)

3 (6.0%) 16.2 ± 9.1 0 14 (31.8%) 12 (24.0%)

.002 .849 .125 .005 .607

.034 .800 .841 .738 .506 .558 .003 .423 .424 .827

13 3 3 4

(26%) (6%) (6%) (8%) 1 3 (6%) 5 (10%) 1 0 5 (10%)

9 4 3 6 2 1 2 1 1 5

(18.0%) (8.0%) (6.0%) (12.0%) (4.0%) (2.0%) (4.0%) (2.0%) (2.0%) (10.0%)

.454 1.000 1.000 .687 1.000 .625 .375 1.000 1.000 1.000

Statistically significant comparisons are indicated in italics.

Fig 1. Kaplan-Meier curves of hernia recurrence for patients who underwent bridged repair or reinforced fascia closure for abdominal wall reconstruction. (Color version of this figure is available online.)

The obtained propensity score had an area under the ROC curve of 0.773 (95% CI 0.715– 0.832; P < .0001). After the propensity score analysis, the baseline characteristics of abdominal defect width, defect area, contamination rate,

and postoperative chemotherapy that were significant for the overall series did not differ between the 2 groups. In addition, mean follow-up, operative times, the use of component separation, and durations of hospitalization were similar for both groups after matching. Nevertheless, after matching the groups according to their baseline characteristics, we found that the bridged repair group still experienced a 7-fold greater rate of hernia development than the reinforced repair group (32.0% vs 6.0%; OR 7.4, 95% CI 2.0–27.3; P = .002; Table III). Furthermore, compared to the reinforced repair group, the bridged group also experienced a greater overall complication rate (68% vs 31.8%; OR 3.8, 95% CI 1.7–8.9; P = .002), but after propensity matching, however, we found there no longer to be significant differences in skin dehiscence, mesh exposure, or bulging/laxity between the 2 comparison groups. Kaplan-Meier curves showed a significant difference in freedom from development of a hernia between the 2 matched paired groups (Fig 2). A post hoc statistical power of 86.6% was calculated for the hypothesis of the discrete primary end point variables after propensity score matching. DISCUSSION These data support our hypothesis that a bridged repair in AWR is associated with a greater

ARTICLE IN PRESS Giordano et al 7

Surgery Volume j, Number j

Table IV. Significant univariate variables put into a multivariate model for hernia recurrence Odds ratio*

Bridged repair Overall complications Skin dehiscence Defect width Mesh, exposure Mesh, infected Bulging Human ADM

Univariate analysis

95% CI

P value

7.6 5.7

3.4–15.1 3.0–11.0

<.001 <.001

2.0 2.0 5.6 7.2 5.9 5.3

1.0–4.1 0.4–3.6 1.6–19.3 1.2–44.1 2.2–15.5 2.2–13.1

.042 .014 .002 .013 <.001 .001

6.3 4.5

2.8–14.2 2.0–9.8

<.001 <.001

3.9

1.2–12.2

.020

Multivariate analysis Bridged repair Overall complications Human ADM

*For continuous variables and discrete variables, odds ratios represent degree of risk per unit of change. ADM, Acellular dermal matrix.

rate of development of a hernia compared to a mesh-reinforced repair on long-term follow-up. This increased risk of developing a hernia persists even after propensity score analysis of characteristics known to be predictive of increased risk of complications. We believe that these findings support favoring a mesh-reinforced fascial closure and avoiding a bridged repair for AWR whenever possible, particularly in cases that employ a biologic material, such as ADM. AWR can be performed by closing the fascia or by bridging the fascial defect with or without a component separation. Despite data suggesting the importance of primary fascial coaptation in complex AWR, bridged repair with mesh remains a very commonly employed technique by many surgeons, especially among laparoscopic surgeons who generally bridge defects with laparoscopically placed, intraperitoneal (or rarely preperitoneal) synthetic mesh prostheses.7,21,22 Increasing evidence has demonstrated superior outcomes with primary fascial closure, with or without a component separation, over traditional bridged repair with respect to lesser rates of hernia recurrence (8–19% vs 37.5–57%) and overall complication rates (32–50% vs 52.5–74%).2,5,8,23,24 These differences in hernia recurrence are also seen with laparoscopic ventral hernia repairs,24 which have lesser rates of recurrence (0–5.7% vs

Fig 2. Kaplan-Meier curves of hernia recurrence for patients who underwent bridged repair or reinforced fascia closure for abdominal wall reconstruction after propensity score matching. (Color version of this figure is available online.)

4.8–16.7%) and seroma formation (5.6–11.4% vs 4.3–27.8%) when the fascia is coapted rather than bridged. Only one recent study, a retrospective, multi-institutional study based on 196 patients, found a greater rate of hernia recurrence for primary fascial closure versus bridged repair (20% vs 15%), however, that study excluded patients who had undergone concomitant component separation, bringing the validity of its findings into question, because the repairs were likely performed under excessive fascial tension in the absence of a component separation.7 Most other studies comparing fascial reinforcement to bridging have been underpowered or have had short follow-up times, ranging from 1 month7 to 33 months.5 Furthermore, such studies were not able to compare these 2 techniques directly because of lack of control for patient variables, such as comorbidity, BMI, previous operations, and, in particular, defect size.2,5,7,8 Patients with larger abdominal defects, with extensive scarring due to multiple procedures, or with important comorbidities are more likely to undergo a bridged mesh repair technique. In contrast, patients with smaller defects, less scarring, and less comorbidity are more likely to undergo procedures involving fascia reapproximation.8 A recent meta-analysis on this topic demonstrated that bridged repair was associated with

ARTICLE IN PRESS 8 Giordano et al

more surgical site occurrences (SSOs) than primary fascial closure, irrespective of whether a synthetic or biologic mesh was employed.8 Another issue is that of bulge/laxity. Although the incidence of bulging/laxity was low in both the reinforced and bridged groups in our study both before and after propensity score analysis, abdominal wall bulging and asymmetry can be the source of considerable consternation and anxiety related to body self-image for patients. This study is the first to control for important variables including defect size using propensity score analysis to obtain similar treatment groups in attempt to decrease heterogeneity, and to use multivariate logistic regression. Our study supports strongly the superiority of fascial reinforcement over bridging in terms of hernia recurrence and overall complication rates. With a mean patient follow-up of >2.5 years, our study ranks as one of the adequately powered studies of its kind with sufficient follow-up to address this question of bridging versus reinforced fascial closures using a sublay biologic mesh technique. Furthermore, because of the large size of our study sample, we were able to show the difference between these 2 techniques with adequate power: 98.3% in the comparative analysis and 86.6% after propensity score matching.20 Component separation was performed consistently in the majority of cases, and there was no difference in its use between groups. Even when bridging was unavoidable due to excessive tension on the midline, a component separation was performed to decrease the amount of fascial tension and the size of the fascial defect. Before propensity score matching, patients who underwent bridged repairs had significantly larger abdominal wall defects and greater rates of grades 3–4 contamination, factors shown to affect operative outcomes.8,25 To overcome the differences in fascial defect size, we adjusted the baseline differences between the 2 groups using propensity score analysis in order to obtain 2 comparable groups. Mesh choice is another important issue in these procedures. The use of bioprosthetic mesh has been shown to be associated with a low incidence of mesh-related infection, adhesion, and enterocutaneous fistulae25; lesser rates of infection, mesh exposure leading to explantation, and reconstruction failure26-31; and fewer infectious wound complications.26,27 In addition, biologic mesh in one series had a similar rate of hernia recurrence compared to synthetic mesh27 and is more expensive.26-31 Some surgeons have advanced the opinion that ADM may not be the best option when bridging is

Surgery j 2016

necessary, because very high recurrence rates have been observed in some studies evaluating the outcomes of complex AWR with ADM.26-31 These studies often detail the experiences of surgeons who employ ADM selectively for only the most challenging cases, such as cases with gross contamination or active infection. The less complex hernia patients receive synthetic mesh and thus, not surprisingly, may experience better outcomes. In contrast, our practice has been to employ ADM more commonly than synthetic materials due to the high rates of medical comorbidities, previous radiotherapy or chemotherapy, nutritional deficiency, adhesion risk, and multiple previous operations in our populations of oncology patients. Therefore, our data represent a more accurate appraisal of the performance of ADM in complex AWR, because selection bias for ADM is more controlled based on our practice pattern. In this cohort of AWR with ADM, the rate of development of a hernia was 8.8% overall, and this outcome was optimized when bridged repairs were excluded, obtaining a rate of hernia development of 6.2% at 2.5 years of follow-up. A recent meta-analysis demonstrated that bridged repair was associated with greater SSO rates, potentially due to the fact that the mesh lies between subcutaneous tissue and the intra-abdominal contents, leaving it more susceptible to infectious complications.26 When abdominal fascia is approximated over the mesh, this approach creates a dynamic abdominal wall and isolates the mesh from the less vascular subcutaneous adipose tissue, which likely decreases the risk of seroma over the mesh and translocation of bacteria from an overlying compromised wound. In our study, we found that mesh exposure seemed to be greater in the bridged group; however, this difference was not significant after propensity score matching. The fact that ADM was placed in an underlay position to facilitate mesh revascularization from the posterior sheath and because we excluded onlay reinforcements may explain why ADM exposure did not appear to lead to serious sequelae, such as mesh explantation. This placement allows for conservative wound management, like negative pressure therapy, avoids further procedures in the operating room, such as new mesh implantation, and, therefore, decreases morbidity and potential overall costs.32,33 One finding that deserves mention is the fact that the use of human ADM was significantly associated with development of a hernia. In light of this, we have abandoned the use of human allografts in favor of xenograft ADM in complex AWR.

ARTICLE IN PRESS Giordano et al 9

Surgery Volume j, Number j

The comparison of bridged repair versus mesh-reinforced fascial closure in abdominal wall reconstruction is difficult, because there can be considerable heterogeneity among patients who undergo AWR. The etiology of the abdominal defect, the chronicity of patient comorbidities, and the specific characteristics of the abdominal wall defect differ extensively among these patients. In comparative settings, patient groups tend to differ in terms of risk factors as well as the size of the abdominal wall defect requiring treatment. Thus, in prior nonrandomized observational studies, there has been no ability to control assignment for treatment. This issue can be avoided partially if the covariates or risk factors are integrated into the study design. Propensity score is a measure of the likelihood that a patient would have been treated using their covariate scores and is used in an attempt to decrease bias and to improve precision by adjusting for risk factors prior to or while calculating the effect of the treatment.10,19 We adjusted the treatment method for the most relevant variables in the 1-to-1, propensity-matched pairs analysis in order to obtain similar comorbidities, defect sizes, and grades of wound contamination among the 2 groups to decrease confounding factors. In contrast, most prior comparative studies have had substantial differences among the study population characteristics, especially concerning defect/ hernia size and duration of follow-up. The strengths of this study include its large sample size, consistent operative technique often using a component separation, adequate statistical power, long-term follow-up, and comparable groups in terms of comorbidities and, more importantly, defect/hernia size. Also, the routine protocols of tumor recurrence surveillance of our center, with both clinical examination and CT (in 88.8% of cases), confer very high sensitivity and specificity for the detection of development of a hernia that is unusual in the currently available medical literature.34 Several limitations of this study should be acknowledged. This is a retrospective study, and the lack of randomization introduces the possibility of selection bias. To compensate for the potential selection bias and evident preoperative baseline imbalance between the study groups, propensity score matching was performed and provided a well-balanced study population. Confounders not taken into account in this analysis may have still biased the results after propensity score matching, and selection bias may have been present.10

The sample size for bridged repair is relatively small, which decreased the power of our comparative analyses. In an effort to mitigate this potential study design bias with respect to the uneven distribution of bridged and reinforced repairs, we analyzed a large cohort of patients that provided a greater pool of candidates for propensity score matching. Additionally, we included a large number of variables previously demonstrated to be predictors of risk in AWR. Other possible patient characteristics, such as wound healing in pre- or postmenopausal women related to hormone replacement therapy and concomitant steroids, might have potentially influenced the outcomes of this study; yet, we did not include these factors in our analyses, as they were not tracked consistently in our database or patients’ medical records. Another limitation may be the illness severity and complexity of the patients in our unique oncologic practice setting, which exceeds those usually encountered in AWR practices. Indeed, the complexity of our oncologic patients has led our surgeons to employ ADM far more readily than synthetic mesh, which differs from many other practitioners, where synthetic mesh is used primarily due to lesser patient complexity and initial cost pressures. For this reason, our high utilization of ADM does not represent most common operative practices dealing with incisional hernias. We acknowledge that large pore synthetic and coated meshes can be placed against viscera with satisfactory outcomes, but, as a cancer center, every effort is made to avoid postoperative wound complications that might delay the provision of adjuvant chemotherapy or radiation therapy and thus compromise cancer survival. Finally, because of the 10-year study period, the effect of each surgeon on the outcomes has not been possible to take into account, as 37 different surgeons contributed to this study cohort; however, the operative technique employed on all cases had been established by the senior author (CEB) at the beginning of the study period, and the techniques were similar during the study period. Further randomized studies on this topic are warranted and, whenever possible, comparable groups in terms of comorbidities and, more importantly, in defect/hernia size should be obtained. REFERENCES 1. Poulose BK, Shelton J, Phillips S, Moore D, Nealon W, Penson D, et al. Epidemiology and cost of ventral hernia repair: making the case for hernia research. Hernia 2012; 16:179-83.

ARTICLE IN PRESS 10 Giordano et al

2. Booth JH, Garvey PB, Baumann DP, Selber JC, Nguyen AT, Clemens MW, et al. Primary fascial closure with mesh reinforcement is superior to bridged mesh repair for abdominal wall reconstruction. J Am Coll Surg 2013;217: 999-1009. 3. Patel KM, Nahabedian MY, Albino F, Bhanot P. The use of porcine acellular dermal matrix in a bridge technique for complex abdominal wall reconstruction: an outcome analysis. Am J Surg 2013;205:209-12. 4. Basta MN, Fischer JP, Kovach SJ. Assessing complications and cost-utilization in ventral hernia repair utilizing biologic mesh in a bridged underlay technique. Am J Surg 2015;209:695-702. 5. Richmond B, Ubert A, Judhan R, King J, Harrah T, Dyer B, et al. Component separation with porcine acellular dermal reinforcement is superior to traditional bridged mesh repairs in the open repair of significant midline ventral hernia defects. Am J Surg 2014;80:725-31. 6. Henry CR, Bradburn E, Moyer KE. Complex abdominal wall reconstruction: an outcomes review. Ann Plast Surg 2013; 71:266-8. 7. Wennergren JE, Askenasy EP, Greenberg JA, Holihan J, Keith J, Liang MK, et al. Laparoscopic ventral hernia repair with primary fascial closure versus bridged repair: a risk-adjusted comparative study. Surg Endosc 2016;30: 3231-8. 8. Holihan JL, Askenasy EP, Greenberg JA, Keith JN, Martindale RG, Roth JS, et al. Component separation vs. bridged repair for large ventral hernias: a multiinstitutional risk-adjusted comparison, systematic review, and meta-analysis. Surg Infect (Larchmt) 2016;17:17-26. 9. Le D, Deveney CW, Reaven NL, Funk SE, McGaughey KJ, Martindale RG. Mesh choice in ventral hernia repair: so many choices, so little time. Am J Surg 2013;205:602-7. 10. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res 2011;46:399-424. 11. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. STROBE Initiative. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806-8. 12. American College of Surgeons National Surgical Quality Improvement Program. User guide for the 2011 Participant Use Data File. Chicago (IL): American College of Surgeons; 2016. Available from: http://site.acsnsqip.org/wp-content/ uploads/2012/03/2011-User-Guide_Final.pdf. 13. Garvey PB, Villa MT, Rozanski AT, Liu J, Robb GL, Beahm EK. The advantages of free abdominal-based flaps over implants for breast reconstruction in obese patients. Plast Reconstr Surg 2012;130:991-1000. 14. Garvey PB, Bailey CM, Baumann DP, Liu J, Butler CE. Violation of the rectus complex is not a contraindication to component separation for abdominal wall reconstruction. J Am Coll Surg 2012;214:131-9. 15. Butler CE, Campbell KT. Minimally invasive component separation with inlay bioprosthetic mesh (MICSIB) for complex abdominal wall reconstruction. Plast Reconstr Surg 2011;128:698-709. 16. Ghali S, Turza KC, Baumann DP, Butler CE. Minimally invasive component separation results in fewer wound-healing complications than open component separation for large ventral hernia repairs. J Am Coll Surg 2012;214:981-9. 17. Baumann DP, Butler CE. Bioprosthetic mesh in abdominal wall reconstruction. Semin Plast Surg 2012;26:18-24.

Surgery j 2016

18. Butler CE, Langstein HN, Kronowitz SJ. Pelvic, abdominal, and chest wall reconstruction with AlloDerm in patients at increased risk for mesh-related complications. Plast Reconstr Surg 2005;116:1263-75. 19. Austin PC. Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharm Stat 2011;10:150-61. 20. Srinivas TR, Ho B, Kang J, Kaplan B. Post hoc analyses: after the facts. Transplantation 2015;99:17-20. 21. Sauerland S, Walgenbach M, Habermalz B, Seiler CM, Miserez M. Laparoscopic versus open surgical techniques for ventral or incisional hernia repair. Cochrane Database Syst Rev 2011:CD007781. 22. Awaiz A, Rahman F, Hossain MB, Yunus RM, Khan S, Memon B, et al. Meta-analysis and systematic review of laparoscopic versus open mesh repair for elective incisional hernia. Hernia 2015;19:449-63. 23. Iacco A, Adeyemo A, Riggs T, Janczyk R. Single institutional experience using biological mesh for abdominal wall reconstruction. Am J Surg 2014;208:480-4. 24. Nguyen DH, Nguyen MT, Askenasy EP, Kao LS, Liang MK. Primary fascial closure with laparoscopic ventral hernia repair: systematic review. World J Surg 2014;38:3097-104. 25. Garvey PB, Martinez RA, Baumann DP, Liu J, Butler CE. Outcomes of abdominal wall reconstruction with acellular dermal matrix are not affected by wound contamination. J Am Coll Surg 2014;219:853-64. 26. Darehzereshki A, Goldfarb M, Zehetner J, Moazzez A, Lipham JC, Mason RJ, et al. Biologic versus nonbiologic mesh in ventral hernia repair: a systematic review and meta-analysis. World J Surg 2014;38:40-50. 27. Montgomery A, Kallinowski F, K€ ockerling F. Evidence for replacement of an infected synthetic by a biological mesh in abdominal wall hernia repair. Front Surg 2016;2:67. 28. Rosen MJ, Krpata DM, Ermlich B, Blatnik JA. A 5-year clinical experience with single-staged repairs of infected and contaminated abdominal wall defects utilizing biologic mesh. Ann Surg 2013;257:991-6. 29. Guerra O. Noncrosslinked porcine-derived acellular dermal matrix for single-stage complex abdominal wall herniorrhaphy after removal of infected synthetic mesh: a retrospective review. Am Surg 2014;80:489-95. 30. Albino FP, Patel KM, Nahabedian MY, Attinger CE, Bhanot P. Immediate, multistaged approach to infected synthetic mesh: outcomes after abdominal wall reconstruction with porcine acellular dermal matrix. Ann Plast Surg 2015; 75:629-33. 31. Breuing K, Butler CE, Ferzoco S, Franz M, Hultman CS, Kilbridge JF, et al. Incisional ventral hernias: review of the literature and recommendations regarding the grading and technique of repair. Surgery 2010;148:544-58. 32. Abdelfatah MM, Rostambeigi N, Podgaetz E, Sarr MG. Long term outcome (>5 year follow-up) with porcine acellular dermal matrix (PermacolÔ) in incisional hernias at risk for infection. Hernia 2015;19:135-40. 33. Campbell KT, Burns NK, Ensor J, Butler CE. Metrics of cellular and vascular infiltration of human acellular dermal matrix in ventral hernia repairs. Plast Reconstr Surg 2012; 129:888-96. 34. Baucom RB, Beck WC, Holzman MD, Sharp KW, Nealon WH, Poulose BK. Prospective evaluation of surgeon physical examination for detection of incisional hernias. J Am Coll Surg 2014;218:363-6.