Potential conflict of interest: Nothing to report.
See Editorial on Page 814
Infectious complications after orthotopic liver transplantation (OLT) are a major clinical problem. The lectin pathway of complement activation is liver-derived and a crucial effector of the innate immune defense against pathogens. Polymorphisms in lectin pathway genes determine their functional activity. We assessed the relationship between these polymorphic genes and clinically significant bacterial infections, i.e., sepsis, pneumonia, and intra-abdominal infection, and mortality within the first year after OLT, in relation to major risk factors in two cohorts from different transplant centers. Single-nucleotide polymorphisms in the mannose-binding lectin gene (MBL2), the ficolin-2 gene (FCN2), and the MBL-associated serine protease gene (MASP2) of recipients and donors were determined. Recipients receiving a donor liver in the principal cohort with polymorphisms in all three components i.e., MBL2 (XA/O; O/O), FCN2+6359T, and MASP2+371A, had a cumulative risk of an infection of 75% as compared to 18% with wild-type donor livers (P = 0.002), an observation confirmed in the second cohort (P = 0.04). In addition, a genetic (mis)match between donor and recipient conferred a two-fold higher infection risk for each separate gene. Multivariate Cox analysis revealed a stepwise increase in infection risk with the lectin pathway gene profile of the donor (hazard ratio = 4.52; P = 8.1 × 10−6) and the donor-recipient (mis)match genotype (hazard ratio = 6.41; P = 1.9 × 10−7), independent from the other risk factors sex and antibiotic prophylaxis (hazard ratio > 1.7 and P < 0.02). Moreover, patients with a lectin pathway gene polymorphism and infection had a six-fold higher mortality (P = 0.9 × 10−8), of which 80% was infection-related. Conclusion: Donor and recipient gene polymorphisms in the lectin complement pathway are major determinants of the risk of clinically significant bacterial infection and mortality after OLT. (HEPATOLOGY 2010;)
The occurrence of infectious complications is a major clinical problem after orthotopic liver transplantation (OLT).1 Immunosuppressive agents that prevent graft rejection interfere with the adaptive immune response and thereby increase the susceptibility to infections. These drugs do not affect, however, the innate immune system that is crucial for the first line of immunological defense.
Lectins, humoral pattern recognition molecules of the innate immune system, recognize pathogen-associated carbohydrate motifs on microorganisms and elicit activation of multiple processes of innate immunity. In order to execute the elimination of microorganisms, these lectins, such as mannose-binding lectin (MBL) and ficolins, cooperate with phagocytes and other humoral factors, including complement. Upon pathogen binding, both lectins activate the complement system via MBL-associated serine proteases (MASPs), leading to C3b-mediated opsonization of the microorganism followed by phagocytosis and the formation of a complement membrane attack complex that directly kills the pathogen.2, 3
Common single-nucleotide polymorphisms (SNPs) in the genes coding for these three members of the lectin complement pathway result in liver-derived protein variants with important functional implications. Structural variants in the exon 1 region of the MBL gene (MBL2) interfere with the oligomerization of the protein and polymorphisms in the promoter regions alter the rate of synthesis of the protein, leading to changes in level, avidity, and pattern recognition of the lectin.4, 5 These polymorphisms are known to be associated with increased susceptibility to infections in conditions accompanied by an immature or compromised adaptive immune system.6-9 In a proof-of-concept study, we previously showed that gene polymorphisms of MBL from the donor liver are associated with the risk of a clinically significant infection after OLT, an observation that recently has been confirmed independently.10, 11
Ficolin-2 has similarities in structure and function to MBL and its preferential binding target is N-acetylglucosamine,12 a constituent of bacterial peptidoglycans and a major component of their cell wall.13 Polymorphisms in the promoter region of the ficolin-2 (FCN2) gene are associated with differences in ficolin-2 serum levels. Structural amino acid substituting polymorphisms within the carbohydrate-recognition domain encoding region of the FNC2 gene are associated with altered ligand binding of ficolin-2.14 MASP-2 is the serine protease associated with MBL and ficolin-2 that is essential for activation of the complement cascade.15 Two polymorphisms in the MASP2 gene that change the amino acid sequence are known to lead to a functional defect in the protease that prevents its interaction with the lectins.16 One SNP leads to the inability to activate complement,17, 18 and the other SNP is located in the complement control protein domain 2 of MASP2, which is important in stabilizing the structure of the serine protease domain19 and is essential for effective cleavage of complement C4.20
The risk of infection after transplantation of a solid organ is the combined effect of all of the factors that contribute to a patient's susceptibility to infection, i.e., the net state of immunodeficiency,21 in which not only the immunosuppressive therapy but also the genetic predisposition of recipient and donor organ are likely to play a role. Given the fact that MBL, as well as ficolin-2 and MASP-2, are almost exclusively synthesized in the liver10, 22 and that the studied gene polymorphisms are quite common in the Caucasian population, there is a realistic chance that a patient in need of liver transplantation will receive a liver from a donor with one or more genetic alterations in the components of the lectin complement pathway.
We evaluated our unchallenged hypothesis that an intergenic interaction between MBL2, FCN2, and MASP2 genes, representing the liver-specific lectin complement cascade, from the donor and recipient contributes to the susceptibility for bacterial infections and associated mortality in OLT recipients.
All patients who received OLT at the Leiden University Medical Center in The Netherlands were taken into consideration for the principal study. Genomic DNA was extracted routinely from peripheral blood and/or tissue samples, when possible, without given preference to any explicit clinical variables. For this study, 202 patients were identified who underwent OLT between 1992 and 2005, of whom we were able to unselectively retrieve 148 patients whose DNA was available from both donor and recipient. From these patients, 143 were finally included who had at least 7 days of follow-up after liver transplantation, excluding perioperative complication morbidity and mortality. The confirmation study consisted of patients who received OLT at the University Medical Center Groningen between 2000 and 2005. From the 212 available patients, 178 unselected patients could be retrieved for whom we had DNA from both recipient and donor, and 167 had at least 7 days of follow-up after transplantation.
The study was performed with informed consent from the patients according to the guidelines of the Medical Ethics Committee of the Leiden University Medical Center and according to the guidelines of the Medical Ethics Committee of the University Medical Center Groningen and in compliance with the Helsinki Declaration.
All patients in the principal study received standard immunosuppressive therapy consisting of corticosteroids, a calcineurin inhibitor (i.e., cyclosporine or tacrolimus) with or without mycophenolate mofetil or azathioprine and/or basiliximab. Patients in the confirmation study received standard immunosuppressive therapy consisting of basiliximab combined with a calcineurin inhibitor with or without corticosteriods and/or mycophenolate mofetil. With respect to the immunosuppressive therapy, azathioprine was used until 2001, and thereafter mycophenolate mofetil was given in case of impaired renal function. From 2001, basiliximab was also used on days 0 and 4. In addition, all patients received 24 hours of prophylactic antibiotics intravenously: gentamycin, cefuroxim, penicillin G, and metronidazol in the principal study; amoxicillin-clavulanate and ciprofloxacin in the confirmation study. The patients in the principal study also received 3 weeks of selective digestive tract decontamination (polymyxin/neomycin, norfloxacin, and amfotericin B) after OLT. After surgery, all patients were intensively monitored according to standardized protocols for any infection, rejection, or poor function of the new liver. After hospital discharge, frequent regular visits and standard procedures in case of suspicion of an infection after OLT, including additional visits upon febrile temperature, are operational in both transplant centers.
The general patient database, original patient reports, transplantation databases, and microbiology records were evaluated to identify episodes of clinical and laboratory-confirmed bacterial infections within the first year after transplantation, without knowledge of the genotypes. The indentified infections were considered clinically significant bacterial infections (CSI) when they complied with the Centers for Disease Control and Prevention criteria23 for diagnosing infection. All infections found could be categorized into sepsis, including symptomatic urinary tract infection (urosepsis); pneumonia; and intra-abdominal infections, i.e., cholangitis and peritonitis.
Demographic and clinicopathological characteristics of the recipient at the time of OLT (age, sex, indication for liver transplantation, cytomegalovirus serostatus, Child-Pugh classification, and laboratory Model for End-Stage Liver Disease [MELD] score), donor information (age, sex, cytomegalovirus serostatus, and donor type), and posttransplant follow-up data (immunosuppressive regimen, acute cellular rejection according to the Banff scheme24) were also collected from the transplantation databases.
We genotyped a total of 13 SNPs in the MBL2, FCN2, and MASP2 genes, with known functional implications on protein level or function, which are common in the Caucasian population,4, 5, 14, 16 with the use of high-resolution DNA melting assays with the oligonucleotide primers as indicated in Supporting Table 1.25-27 In brief, high-resolution melting analysis of polymerase chain reaction products amplified in the presence of a saturating double-stranded DNA dye (LCGreenPlus, Idaho Technology, Inc., Salt Lake City, UT) and a 3′-blocked probe identified both heterozygous and homozygous sequence variants. Heterozygotes were identified by a change in melting curve shape, and different homozygotes are distinguished by a change in melting temperature. In each experiment, sequence-verified control donors for each genotype were used.
Genotypic MBL studies have shown that each of the three exon 1 variants (B, C, and D, which are collectively called O, whereas the wild-type is called A) is in strong linkage disequilibrium with a different promoter haplotype. The association between MBL genotype and phenotype is very strong: sufficient MBL levels are associated with YA/YA, YA/XA, XA/XA, and YA/O genotypes, and insufficient/deficient MBL levels are associated with O/O and XA/O genotypes.28, 29
Associations between baseline characteristics of the liver transplant recipients, donors, and posttransplant follow-up data and CSI were analyzed by using the log-rank and two-tailed Student t tests.
The probability of clinically significant infection within the first year after transplantation according to MBL2, FCN2, and MASP2 gene variants was determined with cumulative incidence curves using Kaplan-Meier analysis, and the differences between groups were assessed by log-rank test and Cox regression analysis. Patients were censored at the date of the last follow-up, death, or liver retransplantation.
The multivariate Cox proportional hazards regression analysis was used to evaluate the independence of the MBL2, FCN2, and MASP2 SNPs or the quantity of gene polymorphisms. The forced entry method, including all variables, as well as the backward elimination regression method (Wald statistic) was applied. Results were considered statistically significant when P values were <0.05. Bonferroni correction for multiple comparison tests was not performed because SNPs were selected on the basis of a deducible hypothesis. All analyses were performed with the SPSS statistical software package (version 16.02; SPSS, Inc., Chicago, IL).
The principal study consisted of 143 patients who received OLT (Table 1) of which 59 (41%) encountered a CSI within the first year after transplantation. The MBL2, FCN2, and MASP2 genotype distribution of recipients and donors were analyzed in relation to the cumulative incidence of CSI in the first year after OLT (Supporting Table 2).
Table 1. Characteristics of Orthotopic Liver Transplant Recipients and Donors
Patients receiving a liver from an MBL-deficient donor (XA/O or O/O) had an increased cumulative incidence of CSI compared to those receiving a wild-type liver (Table 2). In addition, patients receiving a donor liver with at least one copy of the minor T-allele of FCN2 SNP rs17549193 (+6359C→T) also had an increased cumulative CSI incidence. Interestingly, the absence of the minor C-allele (homozygosity for the major A-allele) of MASP2 SNP rs12711521 (+371A→C) in the donor liver was also accompanied with an increased incidence of CSI.
Table 2. Univariate Analysis of the Association of Liver Donor Polymorphisms with the Cumulative Incidence of Clinically Significant Bacterial Infection in the Principal Study (N = 143)
Donor Variant Genotype
Cumulative Incidence of Clinically Significant Infection
Five single-nucleotide polymorphisms in MBL2 were genotyped: MBL2 −550 (H/L), rs11003125; MBL2 −221 (X/Y), rs7096206; MBL2 codon 52 (D), rs5030737; MBL2 codon 54 (B), rs1800450; and MBL2 codon 57 (C), rs1800451. Structural variants D, B, and C are collectively called O; A is considered the wild type; Secretor haplotypes were constructed: A/A and YA/O versus XA/O and O/O.
The presence of multiple intergenic variants within the individual were clustered as the lectin pathway gene profile.
P Values were calculated with the use of the log-rank test.
P Values calculated with the use of the log-rank test compared to variant absent.
The joint genetic effect of risk-conferring variants of MBL2, FCN2, and/or MASP2 present within the donor liver were clustered as the lectin pathway gene profile (Table 2). This profile, including the number of risk-conferring gene variants, showed an ever-increasing cumulative risk for developing CSI with increasing numbers of variants: 18% CSI with no genetic variant, 33% in those with one, 50% in those with two, and 75% in those with three variants (P = 0.002; Table 2 and Fig. 1).
The genotype distribution in OLT recipients (Supporting Table 2) showed no significant association with the occurrence of CSI either for the independent SNPs or for the number of risk-conferring variants (not shown). However, some remarkable interactions between the genotype of the donor and the recipient with the occurrence of CSI were found. Recipients with an MBL-sufficient or wild-type MBL genotype (A/A and YA/O) receiving an MBL-insufficient (XA/O and O/O) donor liver developed significantly more CSI than the other patients (61% [17/28] versus 37% [42/115], respectively, P < 0.006). For the SNPs in FCN2 and MASP2 genes, we found that absence of the minor allele in FCN2 SNP rs17549193 and the absence of homozygosity for the major allele in MASP2 SNP rs12711521 in both recipient and donor showed a clear trend toward less CSI (FCN2: 28% [11/39] versus 46% [48/104] and MASP2: 22% [5/23] versus 45% [54/120], both P = 0.06).
The demographic characteristics of the patients in the confirmation study were comparable to those of the patients in the principal study (Table 1), even though there was a difference in selective digestive tract decontamination, type of intravenous antibiotic prophylaxis, and immunosuppressive therapy. The frequencies for the various SNPs of the recipients were similar compared to the principal group (Supporting Table 2). The cumulative incidence of CSI within the first year was significantly lower (22% [36/167] versus 41% [59/143], P = 0.007) and the percentage of transplanted donor livers with an MBL-deficient genotype was significantly lower in this confirmation group compared to the principal study (13% [22/167] versus 22% [31/143], P = 0.05; Supporting Table 2). Nevertheless, the lectin pathway gene profile of the donor liver in this confirmation group showed a similar significant association with the cumulative incidence of CSI (56% [5/9] with three variants, 26% [15/57] with two variants, 15% [12/81] with one variant, and 20% [4/20] when genetic variants were absent, log-rank = 8.2; P = 0.04). Furthermore, the effect of the donor-recipient genotypic match was also confirmed. MBL mismatch, i.e., a sufficient recipient and an insufficient donor liver, conferred a significantly increased risk for developing clinically significant infection compared to the other MBL combinations (40% [8/20] versus 19% [28/147], P = 0.03), whereas again a lower risk of CSI was associated with absence of the minor T-allele in FCN2 SNP rs17549193 (10% [5/50] versus 27% [31/117], P < 0.03) and the absence of homozygosity for the major A-allele in MASP2 SNP rs12711521 (9% [2/23] versus 24% [34/144], P = 0.11) in both recipient and donor.
In the univariate regression models, a significant association was found for the separate donor gene polymorphisms with CSI of the combined data from both cohorts, in particular for MBL2 (XA/O and O/O) and FCN2 (rs17549193), and less so for MASP2 (rs12711521) (Table 3). In addition, the lectin pathway gene profile of the donor liver showed a significant stepwise association with CSI. In the presence of three variants, 67% CSI was found; 38% CSI was found in the case of two variants, 23% CSI in the case of one variant, and 19% CSI was found when genetic variants in the lectin pathway were absent (P < 0.001). The only other factors associated with the infection risk were found to be male sex of the donor and recipient, the antibiotic prophylactic regimen used, and acute cellular rejection.
Table 3. Univariate and Multivariate Analysis for the Association of Risk Factors of Clinically Significant Bacterial Infection in all OLT Patients (N = 310)
Multivariate analysis Cox regression analyses and the backward elimination procedure, taking all demographic and clinicopathological characteristics into account, indicated that the three individual lectin pathway genes of the donor were independently associated with the infection risk, next to sex of the donor-recipient combination and the prophylactic antibiotic regimen, whereas acute cellular rejection lost its significance (Table 3).
A final adjusting multivariate model was derived using the lectin pathway gene profile and the backward elimination procedure, which indicated that recipients had an even higher CSI risk if they received a donor liver with two or three genetic variants up to an adjusted hazard ratio (HR) of 4.52 (confidence interval [CI] = 1.81-11.31), again independent from sex and antibiotic prophylaxis, which were also found to have significant and independent HRs of more than 2.21.
The combined genotypes of the donor and the recipient showed even stronger association with CSI than the donor genes alone. Although CSI risk is related to the donor MBL genotype, the risk is even higher when the recipient genotype is taken into account. Thus, receiving an MBL-insufficient liver when having previously had an MBL-sufficient liver almost doubles the risk of CSI as compared to the other donor/recipient MBL combinations (52% [25/48] versus 27% [70/262], respectively; P < 0.0001). Similarly increased infection risks were found for the FCN2 and MASP2 donor-recipient combinations, as described in Table 4 and Supporting Table 3. The different genotypic donor-recipient combinations also gave rise to (mis)match genotypes associated with increasing infection risk scores from 0% in those without a variant to 65% in those with three variants within the lectin pathway gene profile (Table 4 and Fig. 2). Because the multivariate model revealed that the individual (mis)matches were independently associated with the infection risk, all donor-recipient (mis)match variant genotypes were included in the final multivariate model, which showed an even higher infection risk profile for two or three variants as compared to one or no variant, with adjusted HRs of 2.74 (CI = 1.56-4.82) and 6.41 (CI = 3.19-12.89), respectively, than that for the donor gene profile alone.
Table 4. Univariate and Multivariate Analysis for the Association of Risk Factors of Clinically Significant Bacterial Infection in all OLT Patients Stratified According Donor- Recipient Genotypes (N = 310)
The all-cause mortality rate in the first year after OLT for recipients who received a donor liver with one or more variants in the lectin complement pathway was significantly higher in patients who encountered a CSI (28% [25/88] versus 4% [8/185] in those without a CSI; Fig. 3). In the absence of a genetic variant in the lectin pathway of the donor liver (n = 37), none of the recipients died in the first year of follow-up, despite a CSI rate of 19%. These differences in CSI-associated mortality persisted after adjustment for the D-MELD score,30 the product of donor age and preoperative laboratory MELD score (unadjusted HR = 7.34; 95% CI = 3.31-16.29), whereas the HR adjusted for D-MELD > 1600 was 7.35 (95% CI = 3.31-16.32). A similar association with mortality was found in the patients with a (mis)match in the MBL2, FCN2, and MASP2 genes between donor and recipient. Presence of one or more (mis)matches was accompanied by a 26% (25 of 95 patients) mortality rate, 80% (20 of 25 patients) of which was infection-related, in case of a clinically significant infection when compared to only a 4% (8 of 209 patients) mortality rate in patients without an infection (unadjusted HR = 7.71 [95% CI = 3.47-17.10]; adjusted for D-MELD > 1600, HR = 7.81 [95% CI = 3.52-17.33], both P < 0.001). None of the remaining six patients without a genetic (mis)match had died during follow-up.
The presence of common functional gene polymorphisms in MBL2, FCN2, and MASP2, which affect the composition, structure, and function of the respective proteins, was found to confer an increased risk of CSI after liver transplantation. Thus, the multifactorial antimicrobial lectin complement activation pathway is of eminent importance to the risk of bacterial infections such as sepsis, peritonitis, and pneumonia, after OLT.
Earlier studies already indicated that MBL deficiency of the donor liver is accompanied by an increased risk of infections after liver transplantation.10, 11 We now showed that the minor T-allele of FCN2 SNP rs17549193 (+6359C→T) and homozygosity for the major A-allele, or the absence of the minor allele, of MASP2 SNP rs12711521 (+371A→C), which are the other main components of the lectin complement activation pathway, also have a significant impact on this infection risk. Diverse combined SNPs in the MBL2 gene, in conjunction with SNPs in the FCN2 and MASP2 genes of the donor liver, constitute a genetic profile of the lectin complement activation pathway which carry a gene dose-dependent risk for bacterial infection in the first year after OLT, as demonstrated and confirmed in the two separate cohorts.
The recipient lectin complement pathway gene profile seemed not to convey a major clinical risk itself. However, MBL-sufficient recipients receiving an MBL-insufficient donor liver were found to be at high risk for these infections. In addition, combined donor and recipient FCN2 and MASP2 genotype analyses showed that when there is no match in the allele associated with reduced infection, the relative risk of CSI is also highly increased.
The essential components of the lectin pathway of complement activation that we studied are mainly produced in the liver.10, 22 After liver transplantation, the adaptive immunity of the recipient is reduced by immunosuppression and the recipient will, to a major extent, be dependent on the lectin complement activation pathway of the donor liver. The functional SNPs in these polymorphic genes may thus lead to reduced complement activation and opsonization, which results in increased susceptibility to infections in patients with an immature or compromised adaptive immune system. Our study is the first to show that the interplay between the genotype of three members of the lectin complement pathway in both donor and recipient has a major impact on the risk of developing infections and on related death in immunocompromised OLT recipients.
Apart from the lectin pathway polymorphisms, univariate and multivariate analyses showed that the male-male donor-recipient sex combination and the type of antimicrobial prophylaxis were consistently found to be accompanied by an increased infection risk. The regimen with selective digestive tract decontamination showed significantly more infections after OLT, as reported before in several other studies.31-34 In the multivariate analyses, the lectin pathway gene profile was found to convey the risk of infection independent from the prophylactic antibiotic regimen. Also noteworthy was the association of the male-male donor-recipient sex combination as an independent risk factor for CSI after liver transplantation. Sex differences in terms of infection and sepsis have been observed in several clinical and epidemiological studies with a predominance of risk in male patients, leading to lower proinflammatory innate immune responses and a worse prognosis with sepsis.35, 36 These findings indicate that male patients receiving a male donor liver should be monitored more intensively and perhaps receive more preemptive antibiotic treatment because of the increased infection risk.
The present study further revealed an important contribution of the MBL2 gene donor-recipient mismatch in the occurrence of CSI. The impact of the MBL2 gene on the increased infection risk was particularly seen in MBL-sufficient recipients whose liver was replaced by an MBL-insufficient donor liver. This raises the question as to whether MBL supplementation might be beneficial. However, the MBL-insufficient recipient does not seem to profit from MBL supplementation, i.e., transplantation of an MBL-sufficient liver. This is in line with the observation that MBL protein substitution in other conditions seems to be ineffective; for example, neutropenic MBL-deficient children who were treated with MBL substitution still encountered neutropenic fever and sepsis.37 Substitution only appeared to be beneficial in some case reports and preclinical studies in knockout mice.38, 39
Finally, the high mortality risk in the first year after OLT in patients with one or more gene polymorphisms in the lectin complement pathway who encountered an infectious event as opposed to those without infection illustrates the major clinical impact of these polymorphisms, in particular because of the high percentage of infection-related deaths. A similar association with survival was recently reported in a small group of patients with only the MBL2 exon 1 gene mutations of the donor liver.40
Our findings account for up to 84% (80/95) but not for all infections observed in the patients who underwent OLT. This might arise for several reasons. For example, other low-allele-frequency SNPs in the lectin pathway genes might also have an impact, but these can only be examined in a considerably larger study population. Furthermore, the lectin pathway is not the only innate immune response to bacterial infections in immunocompromised patients. Toll-like receptors (TLRs) and Nod-like receptors, for instance, also play a role in the immune response to bacterial infections,41 but their function and gene polymorphisms have not been studied extensively in OLT. Interestingly, MBL is able to interact with TLR2 in the phagosome to initiate proinflammatory signaling,42 which thereby might also play a role in infection after OLT.
Gene association studies have several potential limitations which should be taken into consideration when interpreting the results. One is that selection bias may arise from the fact that not all patients were included (patients were excluded because DNA was absent or because of perioperative morbidity or mortality within the first 7 days after transplantation). However, frequencies for the studied SNPs in recipients were comparable in both cohorts. Another limitation is that the study may suffer from bias due to population stratification. In our study, however, a similar association was observed in a second independent cohort, despite differences in treatment regimes and donor genotype frequencies. An additional theoretical limitation is the possibility that the evaluated polymorphisms may not be directly associated with CSI, but instead may be associated with other factors that influence that clinical endpoint. However, the multivariate analyses identify each of the separate SNPs, the number of risk-conferring SNPs, sex, and antimicrobial prophylaxis as independent risk factors for infection.
In conclusion, the genetic profile of the lectin complement activation pathway has a major impact on bacterial infection after liver transplantation. These observations also confirm the importance of the liver as primary source of the lectin complement pathway constituents: MBL, FCN2, and MASP2. Further studies on these genetic risk factors in liver transplantation could contribute to novel infection prevention strategies and improvement of postoperative outcome. This should be evaluated in prospective intervention studies. Such an approach based on lectin complement pathway genes might in time lead to more personalized treatment protocols and improved survival after OLT.
We thank Rolf Vossen and Willem Verduyn for technical assistance, and Dr. James Hardwick for his advice regarding the final text.