Abstract
Abstract
Background:
Mesh reinforcement as part of open ventral hernia repair (OVHR) has become the standard of care. However, there is no consensus on the ideal type of mesh to use. In many clinical situations, surgeons are reluctant to use synthetic mesh. Options in these complicated OVHRs include suture repair or the use of biologic mesh such as porcine acellular dermal matrix (PADM). There has been a paucity of controlled studies reporting long-term outcomes with biologic meshes. We hypothesized that compared with synthetic mesh in OVHR, PADM is associated with fewer surgical site infections (SSI) but more seromas and recurrences. Additionally, compared with suture repair, we hypothesized that PADM is associated with fewer recurrences but more SSIs and seromas.
Methods:
A retrospective study was performed of all complicated OVHRs performed at a single institution from 2000–2011. All data were captured from the electronic medical records of the service network. Data were compared in two ways. First, patients who had OVHR with PADM were case-matched with patients having synthetic mesh repairs on the basis of incision class, Ventral Hernia Working Group (VHWG) grade, hernia size, American Society of Anesthesiologists (ASA) class, and emergency status. The PADM cases were also matched with suture repairs on the basis of incision class, hernia grade, duration of the operation, ASA class, and emergency status. Second, we developed a propensity score-adjusted multi-variable logistic regression model utilizing internal resampling to identify predictors of primary outcomes of the overall cohort. The U.S. Centers for Disease Control and Prevention (CDC) definition of SSI was utilized; seromas and recurrences were defined and tracked similarly for all patients. Data were analyzed using the McNemar, X2, paired two-tailed Student t, or Mann-Whitney U test as appropriate.
Results:
A total of 449 complicated OVHR cases were reviewed for a median follow up of 61 mos (range 1–143 mos): 94 patients had PADM repairs, whereas 154 patients underwent synthetic mesh repairs, and 201 had suture repairs. The 40 PADM repairs were matched to synthetic repairs and 59 were matched to suture repairs. The PADM repairs that could not be well matched (n=54 unmatched for synthetic repairs, 35 unmatched for suture repairs) were characterized generally by larger hernias, VHWG grades of 3 or 4, and incision class 3 or 4 with longer operative durations and more ASA class 4 cases. The patients were well matched. Comparing PADM with synthetic mesh, there was no difference in SSI (20% vs. 35%; p=0.29), seromas (32.5% vs. 15%; p=0.17), mesh explantations (5% vs. 15%, p=0.28), readmissions within 90 d (37.5% vs. 45%; p=1.00), or recurrence (8.5% vs. 22.5%; p=0.15). Compared with suture repair, patients with PADM had fewer recurrences (11.9% vs. 33.9%; p<0.01) and more seromas (32.2% vs. 10.2%; p=0.02), but a similar number of SSIs (23.7% vs. 39.0%; p=0.19) and 90-d readmissions (35.6% vs. 39.0%; p=0.88). Propensity score-adjusted multi-variable logistic regression of the entire cohort corroborated the results of the case-matched patients.
Conclusions:
The PADM repair of complicated OVHR resulted in fewer recurrences, more seromas, and no difference in SSI compared with suture repair. Although no reduction in SSI was identified with the use of PADM rather than synthetic mesh or suture for OVHR, the meaning of this finding is unclear, as this case-controlled study was underpowered and limited by selection bias. According to our data, 280 patients would have been needed to identify a clinically significant difference in the primary outcome of SSI as well as secondary outcomes of mesh explantation and recurrence (α=0.05; β=0.20). A randomized trial is warranted to compare PADM with synthetic mesh in complicated OVHR.
E
There has been growing interest in the use of biologic meshes in open ventral hernia repair (OVHR). In particular, high-risk patients and high-risk lesions have been targeted. The Ventral Hernia Working Group (VHWG) developed a grading system to stratify patients into four risk groups (low risk, comorbid, potentially contaminated, and infected) for the consideration of biologic mesh use in high-risk patients [4]. However, this stratification system is not based on direct patient data, instead relying on expert opinion and extrapolation from the literature.
There has been a paucity of evidence on the role of biologic meshes in OVHR [5–7]. Although reported outcomes have been encouraging, the quality of the evidence has been low, with mainly case reports and case series and no controlled studies. A handful of reviews recently were published reporting pooled data [5–7]. Only level IV (uncontrolled case series) evidence is available in these reports on the utility of biologic meshes.
The present study evaluated outcomes of porcine acellular dermal matrix (PADM) compared with synthetic mesh and suture repair of OVHR. We hypothesized that compared with synthetic mesh, PADM is associated with fewer surgical site infections (SSIs) but more seromas and recurrences and that compared with suture repair, PADM has fewer recurrences but more SSIs and seromas.
Patients and Methods
This was a retrospective study of all complicated OVHR procedures performed from January 2000 to December 2010 at a single Veterans Affairs Medical Center. Complicated OVHR was defined as any concomitant, emergency, or incisional procedure. All data were captured from the electronic medical records of the entire service network. Patients having PADM were compared separately with those undergoing synthetic mesh repair and those having suture repair. The primary outcomes tracked were overall SSI and hernia recurrence. Secondary outcomes were seroma formation, mesh explantation, and hospital readmission. Patient demographics (age, gender, ethnicity), comorbidities (American Society of Anesthesiologists score, body mass index, prostate disease, abdominal aortic aneurysm, chronic obstructive pulmonary disease, corticosteroid use, immunosuppression, diabetes mellitus, current smoking, and alcohol use disorder), hernia data, and surgical history (hernia type, hernia area, VHWG grade, prior SSI, prior ventral hernia repair, and number of prior ventral hernia repairs), operative data (emergency status, incidental repair, incision class, operative duration, fascial release procedure, skin flaps, bridged repair, and drain use), outcomes (intensive care unit admission, peak blood glucose concentration 24 h after surgery, SSI, fascial dehiscence, seroma, hematoma, urinary tract infection, urinary retention, pulmonary infection, central venous catheter-related infection, all infections, ileus, surgical site occurrence, length of hospital stay, 30-d hospital readmission, reoperation, mesh explantation, small bowel obstruction, fistula, recurrence, and followup) were recorded.
Immunosuppression was defined as corticosteroid use, autoimmune disorder syndrome, or chemotherapy within 30 days. Alcohol use disorder was defined as >2 alcoholic drinks per day. Hernia type was defined by the European Hernia Society classification system for ventral hernias [8]. Hernia grade was defined by the VHWG classification system [4]. Concomitant repair was defined as any repair during performance of another abdominal operation (such as colectomy with ventral hernia repair). Fascial releasing procedures included any component separation, fascial rotational flap (e.g., Chevrel procedure), or relaxing fascial incision. Bridged repair was defined as any repair where the abdominal wall fascia was not completely re-approximated. An SSI was defined by the U.S. Centers for Disease Control and Prevention (CDC) description of superficial, deep, and organ/space infection [9]. Seroma was defined as any clinically important fluid collection anterior to the implanted mesh. Ileus was recorded for any patient in whom bowel function was not regained by post-operative day three or if bowel function was present previously and gastrointestinal function was lost with no signs of small bowel obstruction. Surgical site occurrence was defined as any SSI, seroma, hematoma, dehiscence, enterocutaneous fistula, or site complication. Recurrence was captured by clinical examination on re-operation or radiographic follow up.
Patients with PADM repair were compared separately with those having both synthetic mesh and suture repairs. Data were analyzed by two methods: Case-control and propensity score-adjusted multivariable logistic regression.
Case–control series
Patients who had OVHR with PADM were matched with patients in the synthetic mesh group and the suture repair group on the basis of incision class, VHWG grade, operative duration, hernia size, ASA class, and emergency status. Exact matching was attempted for categorical variable, whereas nearest-neighbor matching was used for continuous variables. Data were analyzed using the McNemar, X2, paired two-tailed Student t, or Mann Whitney U test, as appropriate.
Propensity score matching
The overall cohort was analyzed using propensity score-adjusted multi-variable logistic regression. Patients were propensity-matched on the basis of incision class, VHWG grade, operative duration, hernia size, ASA class, and emergency status. A multi-variable logistic regression model was built to assess the effect of a given predictor on each of the primary and secondary outcomes. All pre-operative and intraoperative variables, as well as the duration of followup, were included. Model selection with internal validation and propensity score adjustment was performed using a re-sampling technique (boot-strapping) with the RMS package in R. Diagnostics of the model were assessed and validation was performed using a ten-fold cross-validation. All statistical analysis was performed on the software program R (R Foundation for Statistical Computing, Vienna, Austria).
Results
A total of 449 complicated OVHR cases were reviewed for a median follow up of 61 mos (range 1–143 mos): 94 patients had PADM repairs, whereas 154 patients had synthetic mesh repairs, and 201 had suture repairs.
PADM vs. synthetic mesh
Forty PADM repairs were case-matched to 40 synthetic mesh repairs. The PADM repairs that could not be well matched (n=54) had larger hernia defects and were more likely to have lesions in VHWG grade 3 or 4, incision class 3 or 4 and longer operative duration and to be in ASA class 4.
The patients were well-matched, with no differences in demographics, comorbidities, hernia characteristics, or operative details except for operative duration (Table 1). There were no differences in outcomes (Table 2). Patients with PADM were followed for a median of 34 (range 1–54) mos, whereas patients with synthetic mesh were followed for a median of 84 (range 1–140) mos.
No patients were in ASA class 1.
ASA=American Society of Anesthesiologists; SD=standard deviation; SSI=surgical site infection; VHWG=Ventral Hernia Working Group.
SD=standard deviation.
On propensity score matching, the full cohort (PADM n=94 and synthetic mesh n=154) demonstrated that mesh type was not an independent predictor of SSI or recurrence (Table 3).
For continuous variables and multicategory discrete variables, odds ratios represent degree of risk per unit of change. For body mass index, increase in odds per 1 kg/m2; for follow up duration, increase in odds per 1 mo of follow up; for incision class and American Society of Anesthesiologists class, increase in odds per class; for hernia area, increase in odds per 1 cm2.
VHWG=Ventral Hernia Working Group.
PADM vs. suture repair
Fifty-nine PADM repairs were matched with 59 suture repairs. The PADM repairs that could not be well matched (n=35) involved larger hernias and were more likely to be VHWG grade 3 or 4 and incision class 3 or 4, and to have a longer operative duration and be in ASA class 4.
Although patients were well-matched, with no demographic or comorbidity differences, the PADM patients had larger hernias, longer operations, more complex repairs, and more drains placed (Table 4). The PADM repairs were more inclined to develop seromas post-operatively. However, hernia recurrence was more common with suture repair (Table 5). Patients with PADM were followed for a median of 41 (range 1–57) mo, whereas patients with suture repair were followed for a median of 55 (range 1–142) months. On propensity score matching of the full cohort (PADM n=94 and suture repair n=201), identified mesh use was not correlated with SSI, but suture repair was correlated with hernia recurrence (see Table 3).
ASA=American Society of Anesthesiologists; SD=standard deviation; SSI=surgical site infection.
No patients were in ASA class 1.
Discussion
Complicated OVHR remains a challenging problem, with no clear consensus on the best type of repair or the best mesh to use. In this study, we demonstrate that with complex OVHR, PADM has superior long-term outcomes compared with suture repair, with a lower recurrence rate; however, when compared with synthetic mesh repair, there were no statistically significant outcome differences. In addition, we note that in a substantial number of complex cases, particularly the most complicated ones, surgeons tended to use PADM rather than suture or synthetic mesh repairs, indicating a strong selection bias for the use of PADM in complex cases.
There have been no controlled trials evaluating the outcomes of PADM or other biologic meshes in OVHR [5–7]. A number of case reports and case series have reported promising results; however, recent pooled analyses of biologic mesh outcomes suggested a paucity of data on this subject and a need for caution about the use of biologic meshes in complicated cases [5–7]. The present article represents the first controlled and risk-adjusted study on outcomes with biologic mesh vs. synthetic or suture repair in complex OVHR.
The challenge of interpreting these pooled reports remains the stark differences between different biologic meshes. One of the key areas of difference is the method of processing the meshes and the formation of additional cross-links. Multiple studies have demonstrated that non-cross-linked mesh demonstrated earlier cell infiltration, extracellular matrix deposition, scaffold degradation, and neovascularization than cross-linked materials [10]. Whereas some studies demonstrated that cross-linked meshes simply shift the interval over which tissue in-growth occurs [10], others have demonstrated that these meshes become encapsulated, preventing further integration [11]. In short, cross-linked mesh may function more like synthetic mesh in the short term and ultimately be degraded by the body with little tissue in-growth.
In addition, non-cross-linked meshes may be more resistant to bacterial colonization [10–12]. Although in our study, PADM did not demonstrate any difference in overall, superficial incisional, deep incisional, or organ/space SSI, PADM was three times less likely to be explanted than synthetic mesh (5% vs. 15%).
Our study is also the first controlled trial comparing the outcomes of non-cross-linked biologic mesh with synthetic mesh and suture repair in OVHR. Compared with suture repair, there was no difference in SSI rates, but PADM was associated with more seromas. Long term, PADM outperformed suture repair, with one-third the recurrence rate (11.9% vs. 33.9%; p=0.02). When PADM was compared with synthetic mesh repairs, there was no difference in seroma (32.5% vs. 15%; p=0.17), SSI (20% vs. 35%; p=0.29), or recurrence (7.5% vs. 22.5%; p=0.15). However, there was a trend toward more seromas with biologic mesh than with synthetic mesh and fewer hernia recurrences and mesh explantation. Because of the small sample, this study did not have the power to confirm these differences.
This study has a number of limitations. First, as a retrospective study, it is subject to selection bias and limitations in identifying adverse events accurately. We attempted to normalize for selection by analyzing the data by two risk-adjusted methods: Case-matching and propensity score multi-variable analysis. In addition, with trained reviewers utilizing strict definitions of primary and secondary outcomes along with established definitions such as the CDC definition of SSI, we attempted to normalize data collection. Second, our study was performed in a Veterans Affairs Medical Center, where patients are largely high-risk older males. Application of these data to other patient populations should be approached with caution. Third, because of the sample size, we were not able to demonstrate differences in seroma formation, hernia recurrence, or mesh explantation between biologic vs. synthetic mesh (likely a type II error). Assuming similar results in a larger trial with an α=0.05 and β=0.20, 280 patients would be needed to power a randomized trial comparing PADM with synthetic mesh repair, and 112 patients would be needed to power such a trial comparing PADM with suture repair in moderate-risk OVHR. Finally, because of the differences in the length of followup of the various treatment groups, the data on late outcomes such as hernia recurrence and mesh explantation must be interpreted with caution. The multi-variable analysis did adjust for follow up duration, taking into account the differences in the two cohorts.
In conclusion, we present the first controlled and risk-adjusted study comparing the outcomes of biologic mesh vs. synthetic mesh and suture repair in complicated OVHR. Our results demonstrate that PADM has superior long-term outcomes compared with suture repair and may be preferable to simple suture repair in complicated ventral hernias. Although we noted no difference in outcomes of PADM vs. synthetic mesh, there was a trend toward fewer hernia recurrences and mesh explantations with PADM. Part of the challenge remains that in extremely complicated cases, PADM was used preferentially over suture or synthetic mesh repair, complicating interpretation of the outcome data. Further randomized study comparing PADM with synthetic mesh repair is warranted.
Footnotes
Author Disclosure Statement
No competing financial interests exist.