Mycophenolate mofetil (MMF) and sirolimus impair wound healing. We compared sirolimus vs. MMF to determine the relative impact on surgical complications and wound healing in adult kidney transplant recipients. This retrospective, single center study of 235 adult kidney transplants performed between 1 January 2000 and 31 January 2002 identified 158 adult, kidney-only recipients treated with tacrolimus and prednisone, from which two groups were defined: group 1 (n = 84) received MMF, group 2 (n = 74) received sirolimus. The incidence of fluid collections, wound problems, dehiscence, and urine leak were compared. A multivariate stepwise logistical regression analysis was performed to identify risk factors. The overall incidence of complications was 21.5%, with rates significantly lower in group 1 (2.4%) vs. group 2 (43.2%, p < 0.0001). Regression analysis showed only sirolimus (p < 0.001) and hypo-albuminemia (p = 0.006) to independently correlate with complication occurrence. In subanalyses, lymphoceles correlated only with sirolimus (p = 0.003), while other wound problems also correlated with higher body mass index (p = 0.067). The use of sirolimus, tacrolimus and prednisone was associated with a greater incidence of lymphoceles, non-lymphocele perinephric fluid collections and other consequences of poor wound healing, as compared to contemporary patients treated with MMF, tacrolimus and prednisone.
Anti-proliferative drugs are commonly used in the immunosuppressive regimens administered to kidney transplant recipients. A recent report indicates that immunosuppressive regimens including mycophenolate mofetil (MMF) cause a greater impairment in healing than previously used drug combinations (1). Sirolimus is a recently introduced inhibitor of cellular proliferation and cell-signal transduction, with a favorable impact on rejection rates and a toxicity profile that differs from the calcineurin inhibitors. Sirolimus is a potent anti-proliferative agent that has proven to be efficacious in preventing rejection, but also results in greater numbers of lymphoceles (2). Sirolimus impairs fibroblasts (3,4) and inhibits angiogenesis under certain conditions (5). Therefore, a negative effect on wound healing should also be expected. To evaluate the impact of sirolimus compared to MMF on wound healing, we retrospectively reviewed our recent experience in patients receiving sirolimus-based immunosuppression after kidney transplantation, and compared it to a group of contemporary patients receiving MMF, but otherwise identical immunosuppression.
Materials and Methods
Between 1 January 2000 and 31 January 2002, 236 kidney transplants were performed at the University Hospitals of Cleveland. Pediatric recipients, those receiving immunosuppression other than sirolimus or MMF in combination with tacrolimus and prednisone, and recipients of multiple organ transplants were excluded. Enrollment in sirolimus and MMF treatment arms was consistent in proportion over the entire study period. Two patients whose grafts were removed on postoperative days 4 and 5 were excluded due to the short duration of drug therapy. The first patient had venous thrombosis and acute rejection; the other had ischemia/necrosis associated with an anti-endothelial antibody. Both were initially treated with sirolimus and neither suffered any wound complications. A retrospective chart review allowed analysis of the outcomes of the remaining 158 patients. Demographic and clinical characteristics evaluated included age, sex, cadaver or live donor source, body mass index (BMI), presence of diabetes, incidence of delayed graft function (DGF) and rejection, and immediate preoperative (day 0) serum albumin level. Delayed graft function was defined as the need for dialysis during the first week after transplantation. Ethnicity was not evaluated as a study variable because African American patients, for the most part, were specifically chosen to receive sirolimus therapy by study protocol. Fewer non-African American patients enrolled in sirolimus regimens, which included tacrolimus and steroids, allowing inclusion in this study. Ethnicity variations between the two groups were as follows: MMF-treated patients were 89% Caucasian, 5% African American (and 6% other races) vs. sirolimus-treated patients, who were 12% Caucasian and 88% African American. Therefore, ethnicity was considered a selection criterion and was not deemed appropriate for formal analysis. However, when ethnicity was informally included in a multiple regression analysis, it did not correlate with the occurrence of wound problems, with p = 0.48 (using a p < 0.05 for initial correlations and p < 0.05 for significance of stepwise comparisons, see statistical methods below).
Two groups, based on immunosuppression, were defined as follows: Group 1 (n = 84) received MMF, Group 2 (n = 74) received sirolimus. Both groups received corticosteroids according to our standard protocol. Methylprednisolone, 250 mg, was given intraoperatively, followed by a rapid taper to an initial maintenance dose of prednisone 30 mg/day. Steroids dosage was further reduced to 10 mg/day by 3 months. All patients received tacrolimus at an initial dose of 0.05 mg/kg, with subsequent adjustments to maintain a target blood level of 8–12 ng/mL in group 1 and 5–8 ng/mL in group 2. Patients in group 1 were given 1 g of MMF every 12 h. MMF dosage was adjusted downward if toxicity was observed. Patients in group 2 were given a loading dose of 15 mg of sirolimus orally shortly before or after surgery, followed by 5 mg/day. Sirolimus dosage was adjusted to maintain a trough level at 10–20 ng/mL, but dosage reductions and levels below 10 ng/mL were targeted if neutropenia or other toxicity unrelated to healing was observed. Immunosuppressive drug dosing and exposure was examined, including actual achieved steroid and MMF doses, tacrolimus doses and trough levels, and sirolimus doses and trough levels. Comparisons were made between doses and levels, both at the mean time to complications and between values averaged over the median time to onset of complications (i.e. the first 30 days following transplantation).
All transplants were performed by one of four general surgeons trained in American Society of Transplant Surgeons approved multi-organ abdominal transplant programs. Assistant surgical support was provided by senior surgical residents or by a second attending. The four surgeons involved (labeled surgeons 1 through 4 in the data) have 20, 10, 3 and 8 years of post-fellowship experience and have worked in this program for 15, 5, 2, and 1 years, respectively. All surgeons used similar technique, including lateral pelvic incision, extra-peritoneal approach, standard vascular technique, extravesical ureteroneocystostomy over double-J ureteral stent (in all cases), and layered fascial closure using no. 1 monofilament, nonabsorbable suture followed by skin staples.
All wound-healing complications were noted and analyzed regardless of whether they were symptomatic or not. All fluid collections were diagnosed by either ultrasound examination or computed tomography (CT) and were categorized as being either superficial or deep (in relationship to the fascia) and by their composition as being hematomas, seromas, lymphoceles, or as undesignated. A collection was designated as a lymphocele if cell counts from direct aspirates proved lymphatic content. A wound dehiscence was defined as any spontaneous separation in the skin and/or fascia of sufficient size to require surgical intervention or application of dressing changes. Urine leaks were defined by the presence of a high concentration of creatinine relative to plasma, in perinephric aspirate or drainage fluid. Each complication was reviewed with respect to the presence or absence of surgical site infection. Culture documentation leading to treatment with specific antibiotics was considered proof of an infected site.
Statistical analysis was performed using SPSS® standard software (Chicago, IL, USA). Since ethnicity co-varied with sirolimus use as described above, it was not included in the risk analysis. Univariate analyses between the groups were determined using a 2-tailed, independent samples, Student's t-test assuming unequal variances with significance taken at p = 0.05. Risk factors for outcomes were defined using multivariate stepwise logistical regression analysis, using initial inclusion criteria meeting p < 0.05 defining significant correlations, followed by stepwise analysis with significance taken both at a p = 0.05 for analysis of the whole population, and at p = 0.10 to detect broader trends, especially in analysis of the subpopulation of patients in the sirolimus group. Continuous variables such as serum albumin and BMI were entered into the model as such, identifying significant positive correlations as ‘high’ levels (e.g. higher BMI) and significant negative correlations as ‘low’ levels (e.g. hypoalbuminemia). Univariate differences between the four individual surgeons were determined using analysis of variance (anova) taking significance at p = 0.05, while regression analysis of the surgeon effect included ‘surgeon’ as a variable in the model.
Demographics and selected clinical parameters for the study population and the subgroups are presented in Table 1. The two study groups were not different with regard to age, gender, incidence of pretransplant diabetes, donor source, serum albumin on admission, or the incidence of delayed graft function (DGF). The whole study population consisted of 42.4% women with an average age of 47.0 ± 13.9 years (range 21–75 years). Overall, 63% of patients received their grafts from cadaveric donors. Group 2 was significantly heavier than group 1 with a BMI of 28.1 vs. 25.5 (p = 0.002). Interestingly, the incidence of acute rejection episodes was not statistically different between groups (12.2% for group 1 vs. 8.3% for group 2, p = 0.43). The individual surgeons performed 51 vs. 28 (surgeon 1), 14 vs. 11 (surgeon 2), 16 vs. 25 (surgeon 3) and 3 vs. 10 (surgeon 4) transplants in group 1 vs. group 2. The percentage of transplants assigned to group 2 for each surgeon was significantly different – 35%, 44%, 61% and 77% for surgeons 1 through 4, respectively (p = 0.004 by anova) – while an insignificant trend towards higher BMI was also noted for surgeons 1 through 4, respectively (p = 0.06).
Table 1. Demographics and clinical data
|Age (years)||47.0 ± 13.9||48.9 ± 15.0||44.9 ± 12.4||0.073|
|BMI** (kg/m2)||26.69 ± 5.2||25.5 ± 4.5||28.1 ± 5.7||0.002|
|Albumin (g/dL)||3.8 ± 0.6||3.8 ± 0.5||3.7 ± 0.6||0.187|
|Time†† (days)||68.5 ± 130||26 ± 14||75 ± 138||0.39|
There were 42 wound complications observed in 34 patients, yielding an incidence of 21.5% for the population overall. These included 4 urine leaks, 4 wound episodes of wound dehiscence (2 superficial and 2 fascial), 9 superficial wound collections (5 seromas, 2 abscesses and one hematoma), and 25 deep collections (9 lymphoceles, 7 seromas, and 9 that were undesignated). Eight of the 34 patients suffered more than one complication. The time in days to diagnosis of a wound complication was not different, but varied widely at 26 ± 14 days (range 1–38 days) for group 1 vs. 75 ± 139 days (range 5–762 days for group 2. The median number of days to diagnosis was more similar between the groups at 32 and 30 days for groups 1 and 2, respectively. The incidence of wound complications was statistically different for patients receiving MMF compared to sirolimus: 2.4% for group 1 vs. 43.2% for group 2 (p < 0.0001). In a univariate analysis of the entire study population, those with wound problems had lower serum albumin than those without (3.47 ± 0.6 vs. 3.83 ± 0.5 mg/dL, p = 0.009). The BMI was greater in those with wound complications (28.3 ± 6.3 vs. 26.3 ± 4.9, p = 0.045), and further analysis showed 38% of those with a BMI above 30 suffered wound complications vs. only 20% of those with a BMI less than 30 (p = 0.029). Interestingly, there was no difference in the incidence of pretransplant diabetes mellitus when those with and without complications were compared (32% vs. 35%, respectively).
A further breakdown of the wound complications identified in groups 1 and 2 is summarized in Table 2. Problems occurred in two patients in group 1 (with 3 separately identified complications) and in 31 patients in group 2 (with 39 complications). Deep fluid collections occurred in 2.4% of patients in group 1, compared to 22 patients in group 2 (30%). There were no wound separations, superficial fluid collections or urine leaks in MMF-treated patients, whereas 4, 9, and 4 occurrences, respectively, were seen in sirolimus-treated patients. Comparing surgeons 1 through 4, statistical differences were seen between the independent surgeons' rates of superficial wound problems (p = 0.013) but not for lymphoceles, any deep fluid collection, fascial dehiscence or urine leak.
Table 2. Wound complication rates
|Deep fluid collections|
| Lymphocele||9 (5.7%)||2 (2.3%)||7 (9.5%)||0.066|
| Seroma||7 (4.4%)||0||7 (9.5%)||0.007|
| Not specified||9 (5.7%)||1 (1.2%)||8 (10.8%)||0.014|
| Subcutaneous||2 (1.2%)||0||2 (2.7%)||0.16|
| Fascial||2 (1.2%)||0||2 (2.7%)||0.16|
|Superficial wound*||9 (5.7%)||0||9 (12.2%)||0.002|
|Urine leak||4 (2.5%)||0||4 (5.4%)||0.045|
|Infection**||10 (6.3%)||0||10 (13.5%)||0.001|
|Number with complications||34 (21.5%)||2 (2.4%)||32 (43.2%)||<0.0001|
Logistic regression analyses
Since group 2 (sirolimus-treated) patients were more prone to developing wound complications, the demographic, clinical, and physiologic variables that may have contributed to this difference were analyzed not only for the entire study population, but also for the sirolimus-treated patients alone. Further subgroup analysis of lymphoceles (as a specific and well-defined complication) was performed, as well as for other ‘non-lymphocele’ wound complications. Urine leaks and other narrowly defined complications were not encountered in sufficient numbers to allow a reliable analysis of independent risk factors. Results of the multiple logistical regression analysis performed on the entire study population are shown in Table 3 for variables meeting a p-value less than 0.1. Using any wound complication as the dependent variable, only the use of sirolimus and the level of serum albumin were found to be independent correlates. Setting a regression analysis cutoff of p = 0.05, both sirolimus and lower serum albumin retained significance, with p < 0001 and p = 0.008, respectively. Sirolimus therapy alone predicted the formation of lymphoceles regardless of analysis stringency with a p-value of 0.003, while sirolimus use, hypoalbuminemia, and higher BMI were found to be independent risk factors for non-lymphocele wound problems. Complications were not related to age, gender, pretransplant diabetes, cadaveric donor source, re-transplantation, delayed graft function or acute rejection. It must be noted that only 3 of the 16 episodes of rejection occurred in patients with wound complications prior to or concurrent with the complication, making analysis of the potentially important effects of steroid bolus or anti-rejection therapy impossible in this small study. Regression analysis showed no surgeon effect on wound complications overall, and the effect on superficial wound problems in particular was lost when BMI and sirolimus use were taken into account. As expected, analysis within the sirolimus-treated subgroup (group 2) showed lower serum albumin to be the only independent risk factor for wound complications overall, while hypoalbuminemia and higher BMI were independent correlates of non-lymphocele wound problems (see Table 4). BMI weakly correlated with absence of lymphocele formation in this group (odds ratio 0.91; p = 0.087), while none of the other clinical variables showed a significant correlation.
Table 3. Analyses of risk factors for subsets of complications
| ||Sirolimus|| 3.34||28.1||6.2–127||<0.001|
| ||Sirolimus|| 2.35||10.5||2.2–49.8|| 0.003|
| ||Albumin||−1.9||0.15||0.05–0.45|| 0.008|
| ||BMI|| 0.11||1.11||0.99–1.26|| 0.067|
| ||Sirolimus|| 2.1||7.98||1.62–39.3|| 0.107|
Table 4. Analysis of risk factors in sirolimus-treated patients
| ||BMI:|| 1.3||1.14||0.99–1.30||0.58|
Immunosuppressive drug dosing and levels
No significant differences were found between groups 1 and 2 with respect to the actual steroid dose achieved at 30 days (22.9 ± 2.9 mg vs. 22.8 ± 3.0 mg, p = 0.53) or the mean steroid dosing over the first 30 days following transplantation (26.4 ± 2.0 mg vs. 26.5 ± 1.6 mg, p = 0.75). As anticipated by protocol, the tacrolimus dose administered was higher in group 1 vs. group 2 (7.0 ± 3.5 mg/day vs. 5.0 ± 2.5 mg/day, p = 0.05), while tacrolimus levels were not significantly different 9.9 ± 3.2 vs. 7.7 ± 2.6 ng/mL. The actual MMF dose achieved in group 1 at 30 days was 1.75 ± 0.5 g/day, while the actual sirolimus does achieved in group 2 at 30 days was 5.2 ± 0.6 mg/day with an average trough level of 13.5 ± 6.0 ng/mL. Comparison of sirolimus dosing, sirolimus levels, steroid doses, tacrolimus doses and tacrolimus levels between those with and those without wound complications showed no significant differences. Multiple logistic regression analysis also showed no correlation between any drug parameter and the occurrence of lymphoceles or wound complications overall.
Superficial dehiscence and other superficial wound complications were treated variably with either surgical intervention (drainage or debridement, with or without closure), or expectant management to secondary healing according to surgeon preference. One such patient failed to heal following a procedure for closure of dehisced subcutaneous tissues and a smaller wound developed. Both fascial breakdowns were returned to the operating room for repair. Urine leaks were all returned to the operating room for repair utilizing the standard techniques of ureteroureterostomy, ureteroneocystostomy, or creation of a bladder flap. Two patients required second surgical repairs. All symptomatic fluid collections were initially drained percutaneously, with good results being achieved in most cases. Two patients with deep lymphoceles required surgical creation of a peritoneal window for relief of symptoms. One of these was performed laparoscopically, the other with an open procedure. Both recovered uneventfully.
Sirolimus is a potent anti-proliferative drug that reduces early rejection rates following kidney transplantation, even among high-risk groups (6,7). One recent report highlights the potential impact of anti-proliferative drugs on healing and found the use of MMF to be an independent risk factor for these wound complications (1). Tacrolimus, steroids and MMF vs. tacrolimus, steroids and sirolimus regimens have been shown to have similar 2-year actuarial patient and graft survivals (89% and 85% vs. 97 and 89%, respectively) in our hands (6), making the comparison of sirolimus to MMF in terms of wound complications relevant. Wound healing relies on specific growth factors to promote cellular proliferation and angiogenesis in the presence of adverse conditions (such as relative hypoxia). Such responses appear to be dependent upon target of rapamycin (TOR)-related pathways (5), making sirolimus a specific inhibitor of angiogenesis, and broadening the range of its potential impact on healing. Furthermore, sirolimus may be included on a list of new, more potent agents, which may now contribute to greater complications of immunosuppression.
Using modern immunosuppressive agents, the cumulative risk of serious wound complications reported in the literature has been less than 10% for loss of fascial continuity, wound infection or both (1). Lymphocele formation occurs in 6–22% (8,9), urine leak in 2–3% and urologic complications in general occur in 5–6%(10–12). We performed a retrospective chart review of patients treated with MMF or sirolimus, and while there is a potential for bias in any retrospective analysis, the overall complication rates are similar to those reported in the literature (wound and/or fascial dehiscence in 5.6%, combined wound infection and dehiscence rate 10%, lymphocele: 6.3%, and urine leak 2.5%). While our MMF-treated patients had a 2.4% complication rate overall, somewhat lower than the long-term rates of 4.8% wound infection and 3.6% dehiscence reported in the literature, Humar et al. did not report on lymphoceles or minor wound problems, and had longer follow-up in which to report late events such as hernia formation (1).
Symptomatic lymphocele formation occurred in 2.3% of MMF-treated patients compared to 9.5% of those treated with sirolimus during the same period. Analysis of risk factors showed only the use of sirolimus to predict the increase in lymphoceles and deep fluid collections overall. The presumed mechanism of lymphocele formation is a technical failure to seal perivascular lymphatic channels divided during surgical exposure of the iliac vessels or lymphatic leak from the allograft itself. Good evidence suggests that acute rejection is associated with lymphocele formation (8). A recent review found that lymphoceles were more common with sirolimus use compared to historical controls, but all perinephric fluid collections were defined as lymphoceles (2). Of interest, the overall rate of deep fluid collections in patients treated with sirolimus (38.1%) was very similar to our findings. We found an increase in lymphoceles and seromas in those treated with sirolimus, but no increase in rejection rates, pointing to a new mechanism involved in the pathogenesis of this complication. Certainly, cellular processes that allow healing and fluid re-absorption are required. Nevertheless, the causative mechanism(s) of TOR inhibition in the formation of lymphoceles and other fluid collections remains speculative.
Urologic complications cause significant morbidity following renal transplantation. In our population urine leaks did not occur in MMF-treated patients, but were found in 5.4% with the use of sirolimus. While this difference was not statistically significant (p = 0.2), it had a strong clinical impact. Leakage from the ureteral anastomosis is best attributed to technical issues. All patients underwent an extravesicle ureteroneocystostomy. Most of the leaks in the sirolimus group were noted after successful removal of bladder and perinephric drains, despite the presence of double-J stents (which are routinely used in all patient transplants in our center). Half of the leaks were discovered several weeks following initial discharge. Notably, the incidence of delayed graft function or acute cellular rejection, potential risk factors for distal ureteral ischemia, were not different between the two groups and in none of these grafts were ureteral issues noted at procurement or implantation.
Finally, we evaluated the impact of sirolimus on local wound problems and dehiscence. All patients were transplanted through lateral, lower-quadrant incisions and given a prophylactic dose of a first-generation cephalosporin or equivalent antibiotic. One of the most important risk factors for any surgical complication is the surgeon himself, and a significant trend was noted in the incidence of superficial wound problems comparing surgeon 1 (lowest risk) through surgeon 4 (highest risk), but this difference was lost in the multivariate analysis due to the strong trends towards increasing sirolimus use and BMI. Thus, an independent surgeon effect could not be shown in this study. The incidence of superficial wound problems or fascial dehiscence in MMF-treated patients was 0%, while 17.5% of sirolimus-treated patients suffered from these problems. Analysis using multivariate techniques showed sirolimus to be an independent risk factor for complications overall, and for superficial healing problems. Similar healing difficulties are noted with MMF for both abdominal wall (1) and visceral sites (13). Preoperative serum albumin and BMI also have an impact on the occurrence of wound-healing problems, but only for non-lymphocele-related complications. Other, more specific markers of protein stores were not measured in a sufficient number of cases to allow further analysis of the impact of nutrition in this study. Among the known risks for dehiscence, elevated BMI, infection, steroids and other factors common in renal transplantation remain significant (14). Nevertheless, obesity (BMI above 30) has been variably reported to have an adverse impact on outcomes in renal transplantation (15,16). Recent reports continue to highlight the controversy over BMI as a risk factor (17), but it does appear than wound separations are elevated in recipients with a BMI above 30 kg/m2 (18). Many of the superficial wound problems associated with sirolimus caused ongoing drainage, another risk factor for poor outcome (19).
Infection is a significant predictor of morbidity and increased cost in transplantation (20). Surgical site infections were not noted in the absence of other documented complications, and were found only in those treated with sirolimus, but did occasionally predate the documentation of further complications. All infections were documented by culture results, confirmed by surgeon examination, and eventually responsive to antibiotic therapy and local measures. Surgical site infections occurred at all phases of care, i.e. before, during and after diagnosis and initial therapy for wound complications. There were no deaths or graft losses caused by infection or sepsis.
Once opened, wounds in sirolimus-treated patients were variably characterized by: recurrent bleeding from dermal edges, a lack of granulation and slowed wound contracture. Some wounds required 6–12 months of local dressing care before resolution. Open wounds remained susceptible to recurrent bacterial infections causing surrounding soft-tissue infection. In one case, late (9 months) re-operation in a sirolimus-treated patient with a clinically healed wound showed minimal degradation of absorbable suture material in the subcutaneous tissues, and a lack of mechanical strength at the fascial level, allowing spontaneous separation following removal of fascial sutures.
Whether the use of sirolimus alone or the combination of sirolimus with tacrolimus and steroids used in this population caused the increase in wound-healing difficulties cannot be determined from our data. The use of an initial loading dose of sirolimus is not a uniform practice, which may contribute to the toxicity of this drug. Nevertheless, a 15% incidence of lymphoceles was noted in the multicenter US trials using doses of 2 or 5 mg sirolimus daily without a loading dose (21). Also of interest is the lack of any effect attributable to pretransplant diabetes, and the distinct difference between risk factors for lymphoceles vs. other complications. The influence of ethnicity could not be fully addressed, since grouping criteria are not appropriate outcome variables for analysis by multiple regression analysis. Could genetic influences tracking with race be important? Our informal analysis did not show race to correlate with poor healing, but the specific effect of ethnicity on healing has already been addressed elsewhere (22). In a study of incisional hernias following renal transplantation, Mazzucchi et al. found 13 hernias in 221 ‘white’ patients (5.8%) vs. 1 hernia in 129 ‘black’ patients (0.8%). Furthermore, the sentinel study on the effect of MMF on wound infection, dehiscence or incisional hernia following kidney transplantation did not include race or ethnicity as a risk factor (1). While the ethnic make-up of the groups fails to explain the differences in healing seen, this difference must still be emphasized and taken as a limitation of this study. Careful patient selection based on preoperative nutritional status and BMI, and the use of alternative agents or dosing regimens in the immediate postoperative period are possible approaches suggested by this data to limit wound problems associated with sirolimus following renal transplantation.
The current logistic regression analysis of over 150 patients receiving either sirolimus or MMF demonstrates that, apart from certain previously described risk factors, the use of sirolimus has an independent effect on wound complications in general, and on lymphocele formation in particular. Depending upon results at other centers, sirolimus may be added to the list of the newer agents responsible for greater wound-healing problems in renal transplantation.