Effect of sirolimus on infection incidence in liver transplant recipients


  • Adrian Fisher,

    Corresponding author
    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    • University Hospital Room E 350, 150 Bergen Street, PO Box 1709, Newark, NJ 07103-1709
    Search for more papers by this author
    • Telephone: 973-972-7218; FAX: 973-972-6227

  • Joseph M. Seguel,

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author
  • Andrew N. de la Torre,

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author
  • Dorian Wilson,

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author
  • Anand Merchant,

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author
  • Rakesh K. Arora,

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author
  • Baburao Koneru

    1. Division of Transplant Surgery, Department of Surgery, New Jersey Medical School–University Hospital, Newark, NJ 07103
    Search for more papers by this author


Sirolimus is a new immunosuppressive agent that lacks the nephrotoxicity and neurotoxicity associated with calcineurin inhibitors.1–3 The addition of sirolimus to immunosuppressive protocols may thus allow sparing of calcineurin inhibitors and reduction or elimination of associated toxicities.1, 6 Between January 2000 and July 2001, sirolimus was administered to 55 of 116 consecutive liver recipients. The remaining 61 patients served as the comparison group in the retrospective analysis. In the sirolimus group, perioperative steroids were reduced, and calcineurin inhibitor initiation was delayed. All infectious episodes that occurred within 60 days of liver transplantation were evaluated but were limited to 1 per patient for statistical analysis of sepsis. Demographic variables were comparable between groups. Patients receiving sirolimus experienced more infection (47.2% vs. 18.03%, P<0.001), and this effect persisted across high and low dosage ranges and sirolimus levels. A trend toward increased length of stay was noted (P=0.07). No difference between groups was found in acute rejection rates (17.5% vs. 22.5%), 1-year graft (81% vs. 89%), patient survival (86% vs. 89%), or hepatic artery thrombosis. In conclusion, despite reduction of other immunosuppressants, patients receiving even low doses of sirolimus experienced increased sepsis rates. This agent may have greater usefulness for patients with threatened renal function or patients with chronic rejection after wound healing has occurred. (Liver Transpl 2004;10:193–198.)


CI, calcineurin inhibitor; MMF, mycophenolate mofetil; CTP, Child-Turcotte-Pugh; LTX, liver transplantation; CMV, cytomegalovirus; BMI, body mass index; UNOS, United Network for Organ Sharing; MELD, Model for End-Stage Liver Disease; LOS, length of stay; IL-2, interleukin-2; CIT, cold ischemia time; WIT, warm ischemia time; q.o.d., every other day.


The availability of new immunosuppressant agents over the last several years has allowed the consideration of calcineurin inhibitor (CI) and steroid sparing or avoidance in liver transplant recipients. Although CI's are very effective in preventing rejection, these agents continue to cause serious nephrotoxicity and neurotoxicity in many patients. Other CI adverse effects include diabetes, gastrointestinal toxicity, hyperlipidemia, and, occasionally, hepatotoxicity. Steroid complications are well described and include osteoporosis, hypertension, mood changes, diabetes, obesity, and inhibition of wound healing.

Sirolimus is a new immunosuppressive agent with a different mechanism of action than CI, steroids, mycophenolate mofetil (MMF), or azathioprine.1–3 Sirolimus lacks the neurotoxicity and nephrotoxicity associated with CI use.1, 2, 4, 5 It also appears to act synergistically with CI.1, 6 After sirolimus demonstrated efficacy in renal transplantation, we reasoned that introducing it immediately after liver transplantation (LTX) would allow delayed introduction of CI and reduced CI dosage when initiated and, therefore, result in less nephrotoxicity and neurotoxicity in our patient population, particularly in the immediate postoperative period. We also hypothesized that concomitant reduction in perioperative steroid use would be tolerated without increasing rates of allograft rejection. As our patient population becomes older and sicker and manifests more renal dysfunction, attempts to reduce CI and steroid induced toxicities appear eminently reasonable. Watson et al. reported use of sirolimus in conjunction with CI in 15 liver transplant patients and found acceptable rejection rates and toxicity profiles.7 Similarly, Trotter et al. demonstrated lower acute rejection rates and good tolerance of sirolimus in 39 liver transplant recipients.8

In early 2000, sirolimus was added to our immunosuppressive regimens, initially for patients at high risk of CI toxicity: elderly patients, high Child-Turcotte-Pugh (CTP) status patients (2A and very ill 2B patients), patients in fulminant hepatic failure, and those with pretransplant acute or chronic renal insufficiency. Its use was expanded over the succeeding 18 months to include routine patients. We provide here a retrospective analysis of our sirolimus use in the first 55 patients who received this agent in a consecutive cohort of 116 liver allograft recipients.


A retrospective review was performed of a consecutive cohort of 119 cadaveric LTXs in 116 patients at our center between January 1, 2000, and July 31, 2001. This group encompassed all 55 patients who received sirolimus within 60 days after LTX. The 61 patients who did not receive sirolimus during this period served as the comparison group. Sirolimus was initiated in the majority of patients within 24 hours after LTX.

Before the introduction of sirolimus use in our program, liver transplant patients received 1 of 2 immunosuppressive regimens:

  • 1Tacrolimus and steroids: 500 mg methylprednisolone intraoperatively, followed by daily taper from 200 mg to 20 mg prednisone over 5 days. Tacrolimus was initiated 12 hours after LTX at 0.03–0.05 mg/kg orally every 12 hours, with dose adjustment to reach target levels of 8–15 ng/mL.
  • 2Tacrolimus, basiliximab, and steroids: basiliximab 20 mg intravenously intraoperatively and on postoperative day 4; methylprednisolone 250 mg infused intraoperatively; daily taper of methylprednisolone was given as previously mentioned or on a one-half dose schedule. Tacrolimus was begun 24–48 hours after LTX and then started at 1 to 2 mg orally every 12 hours, with dose escalation to target tacrolimus levels of 5–10 ng/mL.

In either regimen, postoperative steroids were reduced further in patients judged to be particularly sick or debilitated. CTP Status 2A patients received an intraoperative methylprednisolone dose of 100–250 mg intravenously, and postoperative steroids were omitted for 72 hours.

Sirolimus was used initially for patients at high risk of CI toxicity: elderly patients, high CTP status patients (2A and very ill 2B patients), patients in fulminant hepatic failure, and those with pretransplant acute or chronic renal insufficiency. Its use was expanded over the succeeding 18 months to include routine patients at the transplant surgeon's preference. Nearly all sirolimus group patients received sirolimus within 24 hours after LTX; all received it within 72 hours. Sirolimus was added initially at a loading dose of 8–12 mg daily, followed by maintenance doses of 4–8 mg to achieve target levels of 10–15 ng/mL, based on available literature. Because of the difficulty in obtaining timely sirolimus levels in our early experience, the loading and maintenance dosages were reduced to 4–8 mg daily, and periodic levels were obtained to ensure that levels less than 15 ng/mL were maintained. After an increase in infections and in the severity of infected and poorly healing surgical wounds were identified, initial and maintenance doses were further reduced to 1–4 mg daily, and no attempt was made to achieve “therapeutic” levels.

When sirolimus was added to the immunosuppressive regimens, an intraoperative methylprednisolone dose of 100–250 mg was infused. Methylprednisolone was then tapered from 100 mg to 5–20 mg over the first 5 postoperative days. Frequently, the standard steroid taper was truncated or eliminated. Maintenance prednisone at 5–20 mg daily was then used, based on perceived degree of patient debilitation. Prednisone dose at patient discharge ranged from 5 mg to 20 mg daily. Prednisone taper to discontinuation occurred over 3–9 months. Tacrolimus was begun 24–48 hours after LTX when basiliximab was not used or 72 hours after LTX when basiliximab was used. No attempt was made to achieve tacrolimus levels greater than 5 ng/mL. MMF was used only when CI use was not feasible because of renal insufficiency or CI toxicity or when rejection could not be treated with escalation of CI dose. Therapeutic monitoring of tacrolimus was performed by measuring whole-blood, 12-hour trough levels using an enzyme immunoassay. For each patient, a mean tacrolimus level for the entire study period of 60 days was determined by calculating a mean of 9 individual weekly values (a median of daily values in each week for in-patients and a single value for out patients). Sirolimus levels in whole blood were measured by high-performance liquid chromatography at an outside laboratory much less frequently than tacrolimus. A mean of all available levels during the study period was calculated for each patient receiving sirolimus.

Standard antibacterial, antifungal, and antiviral protocols were used. Patients received 3 g ampicillin/sulbactam intravenously preoperatively and for 48 hours postoperatively. Patients allergic to Penicillin or cephalosporin received levofloxacin for 48 hours, usually in conjunction with a preoperative dose of vancomycin. Fluconazole 100 mg intravenous/oral was administered postoperatively until hospital discharge. Thereafter, nystatin oral suspension (500,000 units q.i.d.) was administered for 3 months. Ganciclovir 5 mg/kg/day was administered intravenously to all patients during intensive care unit stay. Cytomegalovirus (CMV) seronegative patients who received livers from seropositive donors received intravenous ganciclovir for 30 postoperative days, then oral therapy for an additional 60 days. All other patients received acyclovir 200 mg t.i.d. for 3 months. All patients received 80 mg of sulfamethoxazole /trimethoprim every other day indefinitely.

All bacterial, viral, and fungal infectious episodes occurring within 60 days of LTX were evaluated. No protocol cultures were performed. All cultures were performed on clinical suspicion of infection. Appropriate therapeutic agents were administered for all positive cultures in consultation with infectious disease specialists. Blood infection was defined as 2 positive cultures or positive catheter tip plus 1 positive blood culture. Urinary infection was defined by a culture positive for more than 105 bacteria. Pulmonary sepsis was defined by the presence of 2 of the following 3 findings: (1) positive sputum culture, (2) gram stain with a dominant organism and many polymorphonuclear leukocytes, or (3) radiologic infiltrate.

Superficial and deep wound complications were defined by any of the following 4 findings: (1) superficial abscess or focal fluid collection with positive culture or gram stain, (2) gross tissue necrosis, (3) positive peritoneal fluid culture or gram stain, or (4) fascial dehiscence in the presence of a positive culture or gram stain. CMV disease was defined by the presence of CMV immunoglobulin M antibodies in blood samples or the presence of viral inclusion bodies on liver biopsy specimens. CMV-DNA levels were measured in all such patients to confirm the diagnosis. Immunohistochemical staining was performed on suspicious biopsy specimens. Although all sepsis events were tallied, patients with multiple infectious events were credited with only 1 infectious episode for purposes of statistical comparison.

Recipient variables examined included age, gender, body mass index (BMI), pretransplant creatinine, graft warm and cold ischemia times, United Network for Organ Sharing (UNOS) status, and disease etiology. Model for End-Stage Liver Disease (MELD) scores were calculated for each patient to provide an additional method of comparison of recipient disease severity. Hepatocellular carcinoma incidence was compared between groups to assess comparability of MELD scores. Primary end points included infection, patient and graft survival, and acute rejection. Length of hospital stay and incidence of hepatic artery thrombosis constituted secondary end points. Rejection was, in most cases, diagnosed by liver biopsy. Presumptive acute rejection was defined as a 2-fold or 3-fold increase in liver function enzymes without other defined etiology, which responded to methylprednisolone bolus therapy and subsequent steroid taper in conjunction with increase in CI dosage or addition of a third agent. Isolated steroid bolus doses were recorded for comparison between groups but were not counted as rejection episodes.

The 26 patients who received sirolimus in the last 6 months of the study period were, in general, treated with 1–4 mg sirolimus daily. Sirolimus levels and infection incidence in this group were compared with those indices in the first 29 patients who received sirolimus in our program.

Statistical Analysis

Results were evaluated by Chi-square analysis for categorical variables. Student t test or analysis of variance were used for continuous variables. Logistic regression analysis was used to assess independence of multiple variables such as recipient age, MELD score, wound dehiscence with use of basiliximab, and incidence of infection. Six-month, 1-year and 2-year survival data were plotted using actuarial analysis. Trend log-rank test was used to test significance for differences in survival between the two groups.


One hundred sixteen patients received 119 liver allografts over a 19-month period. Fifty-five patients who received sirolimus underwent 57 liver transplants. Sixty-one patients who received 62 allografts were not treated with sirolimus. Demographic variables (Table 1) were similar between the two groups. Mean MELD scores did not differ between groups (21.6 vs. 21.9). There was no significant difference for occurrence of hepatocellular carcinoma between the groups (7/61 in the no sirolimus arm and 2/55 in the sirolimus arm; P=0.11). Pretransplant creatinine levels were slightly higher in patients who received sirolimus but did not reach significance (P=0.08). As shown in Fig. 1, a significant difference (P<0.001) in infection rate was found between patients who received sirolimus (26/55, 47%) and those who did not receive sirolimus (11/61, 18%). Analysis of UNOS status and infection (Table 2) revealed that more CTP status 2A patients received sirolimus (7 vs. 1) and that 57% of the sirolimus group experienced infection (4/7). Illness severity was counterbalanced in the sirolimus group patients, however, by a preponderance of Status 3 patients (15 vs. 3) in the sirolimus arm. Forty percent (6/15) of these patients suffered sepsis episodes. The BMI did not differ significantly between all patients with and without infections or within the sirolimus group between patients with and without infections. Eleven patients in the sirolimus group received no CI inhibitor at anytime during the study period. The mean tacrolimus trough level in 46 patients who received both sirolimus and tacrolimus was significantly lower (5.7±2.6 vs. 9.8±3.1 ng/mL, P<0.001).

Table 1. Demographic Variables in 116 Liver Recipients Who Did and Did Not Receive Sirolimus
 Sirolimus Group (N = 55)No Sirolimus Group (N = 61)
  1. Note: The continuous variables shown in the table represent mean ± SD.

  2. Abbreviations: MELD, Model for End-Stage Liver Disease; CIT, cold ischemia time; WIT, warm ischemia time.

Gender (M/F)34/ 2141/ 20
Age (yr)53.1 ± 13.449.4 ± 11.5
Body mass index (kg/m2)27.4 ± 5.329.8 ± 7.3
Pretransplant creatinine (mg/dL)1.4 ± 1.11.1 ± 0.7
MELD score21.6 ± 9.221.9 ± 8.7
CIT (min)453.9 ± 99.1452.0 ± 103.4
WIT (min)43.1 ± 9.942.4 ± 7.5
Hepatocellular carcinoma27
Figure 1.

Infection and rejection outcomes in 116 liver recipients who did and did not receive sirolimus (infection episodes limited to one per patient).

Table 2. Distribution of United Network for Organ Sharing (UNOS) Status and the Incidence of Infections in 116 Liver Recipients Who Did and Did Not Receive Sirolimus
UNOS StatusSirolimus (N = 55)No Sirolimus (N = 61)
  • Note: The values shown in the parentheses represent percentages of patients developing infections.

  • *

    P = 0.001.

12/6 (33)1/4 (25)
2a4/7 (57)0/1
2b14/27 (52)*9/53 (17)
36/15 (40)0/3
Total26/55 (47)11/61 (18)

The mean sirolimus level of the entire sirolimus group was (9.3±4.8 ng/mL).The mean sirolimus levels did not differ significantly between infected sirolimus patients (9.4±3.8 ng/mL) and noninfected sirolimus patients (9.2±6.1 ng/mL, P=0.92). Thirteen of the 26 patients who received sirolimus (1–4 mg daily) in the last 6 months of the study period suffered infection (50%). Mean sirolimus level of this group was 6.7±3.7ng/mL. In comparison, 12 of the first 29 patients to receive sirolimus became infected (41%). Mean sirolimus level in this group was (11.7±4.7 ng/mL), which was significantly different at p=0.002.

The distribution of infections and wound dehiscence is shown in Table 3. Blood and wound sepsis constituted the majority of infections. Superficial and deep wound infections associated with tissue necrosis or dehiscence accounted for 18/40 (45%) of all sepsis events. Although the wounds in patients receiving sirolimus took longer to heal than those not receiving sirolimus, the difference was not significant (78±42 days vs. 60±41 days, P=0.33). More frequent debridement and lavage were required. The trend toward increased length of hospital stay in this patient group supports these observations.

Table 3. Types of Infections in 55 Liver Recipients Who Received Both Sirolimus and Tacrolimus and 61 Recipients Who Did Not Receive Sirolimus
Site of InfectionSirolimus Group (N = 55)No Sirolimus Group (N = 61)
  1. Note: Includes infection episodes in patients with multiple infection sites.

Blood13 (24%)2 (3%)
Wound infection/dehiscence18 (33%)5 (8%)
Pneumonitis5 (9%)5 (8%)
Urinary tract2 (4%)1 (2%)
Cytomeglovirus disease2 (4%)4 (7%)

A standard variety of gram-positive and gram-negative organisms was cultured. Viral and fungal infections were rare. CMV (hepatitis and viremia) was the only viral infection encountered; it occurred in 4 patients in the control group and 2 patients in the sirolimus group patients. Fungal infections were limited to Candida albicans species. No primary fungal infections were identified. Fungal infections were encountered only after bacterial infection occurred first.

More sirolimus group patients received basiliximab. However, logistic likelihood ratio analysis of independent variables (recipient age, MELD score, wound dehiscence, and use of basiliximab) demonstrated no correlation with incidence of infection. Further, infection rate in our patient population did not increase in comparison to historical control when we began using basiliximab in our immunosuppressive protocols, well before introduction of sirolimus into our patient population.

Rejection rates did not differ significantly between the sirolimus (17.5%) and the no sirolimus (22.6%) groups (P=0.55) (Fig. 1). Steroid-resistant rejection was very rare and did not differ significantly between groups. Posttransplant creatinine values did not differ between groups at weeks 1, 4, or 8, despite the lower tacrolimus levels in the no sirolimus group. Despite the significant increase in infections in the sirolimus group, the graft and patient survival were not significantly different between the two groups (P=0.18) (Fig. 2A and B). Hospital length of stay (LOS) showed a trend toward significance between the study groups (P=0.07). Patients who did not receive sirolimus had a mean LOS of 9.6±7.2 days compared with 15.6±25.4 days in patients who received sirolimus. We detected no significant difference in hepatic artery thrombosis between the sirolimus (1/57) and the no sirolimus (3/62) groups in our cohort.

Figure 2.

Actuarial graft (A) and patient (B) survival curves in the sirolimus (N=55) and no sirolimus (N = 61) groups. There were no significant differences between the two groups.


The major finding in our study is the significantly greater incidence of infections in liver recipients given sirolimus immediately after transplantation. Although the increased infections did not result in greater graft loss and patient death, increased morbidity of patients in the sirolimus group was reflected by the increased LOS trend. In spite of the structural resemblance of sirolimus with tacrolimus, it does not act on calcineurin; instead, sirolimus inhibits interleukin-2 (IL-2)–mediated signal transduction.3, 5, 6, 9 The lack of vasomotor-induced renal injury demonstrated by CI suggests a role for sirolimus in solid organ transplantation via reduction or limitation of CI toxicities.1, 3 Several trials of sirolimus use in renal transplantation have shown a reduction of acute rejection rates. In several, pneumonia and herpes-derived aphthous ulcers were more common, especially when doses approaching 5 mg per day were administered.1, 9, 10 Kahan reported no increase in infection rates in a U.S. multicenter, randomized, sirolimus trial in renal transplantation.10 In a European randomized renal transplant trial, however, Groth et al. reported a higher incidence and severity of infections in sirolimus group patients.9

Several authors have reported the use of sirolimus in LTX, most in small numbers of patients. In some of these studies, infections were common, and wound dehiscence was reported.7, 11–13 In a randomized trial of sirolimus and tacrolimus immunotherapy in liver transplant recipients, Wiesner and the Rapamune Liver Transplant Study Group found a higher rate of wound infection and a reduced 1-year survival rate (with infection at or near the time of death) in patients who received sirolimus.14 Contrary to our experience, Kneteman et al., in a study similar to ours in sample size, managed 56 liver recipients with daclizumab induction, sirolimus, and low-dose delayed tacrolimus in a steroid-free protocol and found no increase in wound complications (dehiscence or diminished healing) or in viral or other infections.15 Very recently, Trotter et al.8 and Dunkelberg et al.16, in the largest series of liver recipients given sirolimus in the immediate post transplant period, reported a decrease in rejection but no increase in infections or wound complications. It is noteworthy that, unlike our practice, steroids were either avoided or eliminated early by these 2 groups. Data suggest that sirolimus inhibits growth factor elaboration in response to tissue injury.1 In light of the several mechanisms by which steroids interfere with wound healing and broadly compromise immune surveillance, the combination of sirolimus and steroids may potentiate both infection and delayed wound healing. This combination may have been important in the increased wound complications in our patients receiving sirolimus. Thus a steroid-free regimen may facilitate the use of sirolimus in the immediate post-LTX period.

A relative excess of immunosuppression may, in fact, explain our early high sepsis rates. The fact that our infection rate did not decline significantly despite reduction of sirolimus dosing to very low levels (confirmed by assays) is both troubling and perplexing. It is unclear what role the IL-2 receptor antagonist played in increasing the incidence of infections in our patients receiving sirolimus. Note also that use of an IL-2 receptor antibody in combination with sirolimus differentiates our series from all preceding reports except Kneteman et al.15 Kahan stated that the effects of sirolimus are “complementary to IL-2 receptor monoclonal antibodies.”1;p.1181 This combination may therefore result in more than merely additive immunosuppression.

The lack of statistical difference in acute rejection rates despite higher infection rates argues, in some measure, against simple excessive immunosuppression of our patients receiving sirolimus. However, it is also possible that any incremental increase of total immunosuppressive therapy from a relatively low baseline rejection rate of 22% favors increased sepsis without a statistically significant further reduction in allograft rejection rate. Alternatively, our patients may have been “incorrectly,” but not necessarily “over,” immunosuppressed. That is, our policy of delayed initiation of CI therapy may have allowed higher rejection rates than would have occurred if CI were initiated immediately after LTX but at very low dose. The rejection cascade might thus be more fully inhibited at multiple points early in the posttransplant period, while CI toxicity is still obviated. Earlier initiation of very low dose CI may allow rapid elimination of postoperative steroids, which may in turn reduce sepsis rates.

A role for routine use of low dose sirolimus in the immediate posttransplant phase potentially exists. Although our study is retrospective, and thus may suffer possible unidentified confounding issues, the fact that the infections increased significantly during a very short period of sirolimus use and declined after its discontinuation indicates that sirolimus should not be used in the manner described in our study. In patients with notable acute or chronic renal insufficiency, the advantage of improved renal function with sirolimus use may outweigh the risk of sepsis or delayed wound healing. We could not substantiate such an effect in our data, however. Other CI-sparing regimens, including MMF and IL-2 receptor antagonists, exist and may have advantages over sirolimus with regard to both efficacy and complications. It is possible that a steroid-free protocol may limit, or eliminate, the delay in wound healing as well as the increased infection rate, as shown in the experience of Trotter et al. and Kneteman et al. Alternatively, optimal use of sirolimus in LTX may occur after wound healing has been completed. It may then prove an ideal agent to which patients with threatened renal function may be converted. Elderly patients and others suffering neurotoxic side effects also may be good candidates for conversion to sirolimus after wound healing has occurred. Such an approach may allow substantial dose reduction, or even elimination of CI, and better overall patient recovery and function in the early (2–6 month) phase after LTX.


Addition of sirolimus to immediate after LTX immunosuppressive regimens, especially those with steroids, appears to increase rates of sepsis and to delay wound healing. If sirolimus is to be used during this period, an elimination or drastic reduction in steroids appears to be necessary. Conversion to sirolimus after wound healing and early postoperative recovery has occurred may be the optimal therapy for patients with persistently threatened renal function and patients manifesting CI toxicity.