Telephone: 0044-20-74726226; FAX: 0044-20-74726229
Combined HLA-DR and -DQ disparity is associated with a stable course of ulcerative colitis after liver transplantation for primary sclerosing cholangitis
Article first published online: 29 MAR 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 13, Issue 4, pages 552–557, April 2007
How to Cite
Cholongitas, E., Papatheodoridis, G. V., Zappoli, P., Giannakopoulos, A., Patch, D., Marelli, L., Shusang, V., Kalambokis, G., Shirling, G., Rolando, N. and Burroughs, A. K. (2007), Combined HLA-DR and -DQ disparity is associated with a stable course of ulcerative colitis after liver transplantation for primary sclerosing cholangitis. Liver Transpl, 13: 552–557. doi: 10.1002/lt.21077
- Issue published online: 29 MAR 2007
- Article first published online: 29 MAR 2007
- Manuscript Accepted: 25 NOV 2006
- Manuscript Received: 7 AUG 2006
Combined disparity of human leukocyte antigen (HLA)-DR and -DQ between mother and fetus is associated with less severe ulcerative colitis (UC) during pregnancy. We evaluated whether donor-recipient HLA disparity after liver transplantation (LT) affects UC in patients with primary sclerosing cholangitis (PSC). Sixty-nine consecutive patients with PSC underwent LT; all underwent colonoscopy before LT; 48 had UC before and 3 had de novo UC after LT. Clinical and laboratory data, activity and treatment of UC, post-LT cytomegalovirus infection, and disparity of HLA-A, -B, -DR, and -DQ for each donor-recipient pair were evaluated. Pre-LT quiescent UC was present in 26 patients. Post-LT UC activity was evaluated in 36 of 51 patients with UC who had not undergone pre-LT colectomy and who had >12 months' post-LT survival. Of these, 16 were stable, 17 had worsened, and 3 had de novo UC. Seven required colectomy (4 for dysplasia or cancer) after LT. Post-LT cytomegalovirus viremia was neither associated with worse UC activity (P = 0.58) nor de novo UC. Disparity with respect to HLA-A, -B, -DR, and -DQ was found in 58%, 27%, 44%, and 39% donor-recipient pairs, respectively. Post-LT UC course was similar with respect to single HLA disparity. However, disparity in none or only one HLA-DR or -DQ was significantly associated with worse activity compared with patients with disparity at both (65% vs. 0%, P = 0.009). Logistic regression found that the disparity for both -DR and -DQ was the only factor statistically significantly associated with post-LT UC activity. We conclude that disparity in both HLA-DR and -DQ between donor and recipient is associated with stable UC activity after LT. Liver Transpl 13:552–557, 2007. © 2007 AASLD.
Primary sclerosing cholangitis (PSC) is a chronic cholestatic liver disease associated with inflammatory bowel disease, predominantly ulcerative colitis (UC),1 which occurs in 60–70% of patients with PSC.1–3 The UC is usually less severe clinically but more extensive endoscopically compared with patients with UC alone.4
The cause of PSC is not known, and no specific treatment has been identified to alter the natural course of liver disease. Thus, liver transplantation (LT) remains the only effective therapeutic option for worsening disease.5 However, PSC can recur in some patients after LT, with an incidence of 8.6–37%, and the course of UC can worsen, remain unchanged, or improve after LT.5–8 After LT, the use of cyclosporin, corticosteroids, and azathioprine (AZA), which are therapeutic options for inflammatory bowel disease, should improve the course of UC, but paradoxically, some studies report worsening UC after LT.5, 6
To date, there is no clear understanding why some patients worsen and others remain in stable condition or experience an improvement in their bowel disease after LT. Some data suggest that the early withdrawal (within the first 3 months) of corticosteroids and the use of low doses of AZA after LT is associated with exacerbation of UC.5 Others implicate bile salts affecting the colon's mucosa; a resetting of the immune system after LT6; or, according to a recent study,9 postoperative cytomegalovirus (CMV) infection in the exacerbation or de novo development of inflammatory bowel disease after LT.
Although there is no established explanation, it seems that the immune system may play a major role in the clinical expression of UC after LT. Recently, disparity in both human leukocyte antigen (HLA)-DR and -DQ between mother and fetus has been shown to be markedly associated with less severe inflammatory bowel disease during pregnancy.10 Therefore, we evaluated the course of UC in patients transplanted for PSC and assessed whether postoperative CMV viremia and donor-recipient HLA disparity are associated with the course of post-LT UC in PSC patients.
MATERIALS AND METHODS
Between May 1989 and December 2004, 69 patients (42 men with a median age of 43 [range, 17–66] years) with end-stage liver disease due to PSC underwent LT in our center. The diagnosis of PSC was confirmed by pre-LT cholangiography and histological examination of the explant in all cases. Intraductal cholangiocarcinoma was present in 6 patients; it was detected in 3 before LT, and found incidentally at explant in the other 3 (Table 1).
|No. of patients||69|
|Age (yr), median (range)||43 (17–66)|
|Cholangiocarcinoma, n (%)||6 (8)|
|Before LT (n)||3|
|After LT (n)||3|
|Patients with UC, n (%)||48 (70)|
|Duration of UC before LT (months)||120 (2–480)|
|Pre-LT course, n (%)|
The diagnosis of UC was made on the basis of combined evaluation of clinical, endoscopic, and histological findings, and exclusion of other causes of colitis. In this cohort, none had Crohn disease. All patients with UC were evaluated in detail5 and the course recorded prospectively: type, extent, and duration of UC, maintenance treatment, number and severity of exacerbations, number of hospital admissions before or after LT, courses and total duration of corticosteroid therapy, use and duration of AZA therapy, and colectomy. Irrespective of bowel symptoms and previous UC diagnosis, all patients with PSC underwent total colonoscopy, and biopsy specimens of the colon mucosa were taken in the previous 6 months before LT during the workup evaluation. Yearly surveillance was undertaken after LT.
Population With UC Before LT
UC was present in 48 patients (70%) with PSC before LT; 44 had total and 4 distal UC. The median duration of UC before LT was 120 (range, 2–480) months (Table 1). The definitions used for the clinical course of UC during the period up to 5 years before LT have been described elsewhere.5 In summary, the course was as follows: quiescent: ≤1 mild to moderate exacerbation of colitis, corticosteroid treatment for <2 months, and no hospital admission or need for AZA; active: ≥2 mild to moderate or ≥1 severe exacerbations of colitis, ≥1 hospital admission, corticosteroid therapy for ≥2 months, or treatment with AZA. Truelove and Witts's criteria were used for classification of severity of exacerbation of UC.11 Corticosteroid-dependent patients were those who continuously needed corticosteroids to maintain remission of UC. Colectomy had been performed in 6 (12.5%) of 48 before LT (intractable disease in 4, high-grade dysplasia or colon cancer in 2). In those (n = 42) without colectomy in the period up to 5 years before LT, UC was quiescent in 26 (62%) and active in 16 (38%). Forty-three patients were receiving mesalazine or sulfasalazine, and 5 patients received no maintenance UC treatment.
Population After LT
The 56 (81%) of 69 patients who survived more than 12 months were evaluated for activity after LT. The patient cohort comprised 32 men with a median age of 42 (range, 17–66) years and a median follow-up of 34 (range, 12–96) months. Survival after 12 months was not different with respect to the presence of pre-LT UC or not (38 [79%] of 48 vs. 18 [85%] of 21, P = 0.65). Survival after 12 months occurred in 21 (81%) of 26 patients with quiescent, 12 (75%) of 16 patients with active, and 5 (83%) of 6 patients with pre-LT colectomy (P = 0.88). The cause of death in the first 12 months after LT was multiple organ failure in 7, graft dysfunction in 2, severe rejection in 1, rupture of hepatic hematoma in 1, and recurrence of cholangiocarcinoma in 2 patients.
Post-LT maintenance immunosuppression was monotherapy with cyclosporine (CYC) 10 mg/kg/d (n = 10) or tacrolimus (FK) 0.1 mg/kg/d (n = 12), both in 2 divided doses or combination therapies. Combination therapies were as follows: AZA 1 mg/kg/d plus CYC (n = 17); AZA plus FK (n = 10); or mycophenolate mofetil (MMF) 1 g twice a day plus FK (n = 7). Corticosteroids were withdrawn within 6 months after LT in all but 6 patients, who had received prednisolone for 7–13 months to treat previous episodes of repeated rejection (4 of these patients had pre-LT UC). All patients with known UC were treated with long-term mesalazine therapy (1.5–2 g/d) within the first 3 months after LT.
All patients attended a specialized inflammatory bowel disease clinic and underwent surveillance colonoscopy with 1 to 2 yearly colon biopsies, irrespective of bowel symptoms and pre-LT UC status. Post-LT UC activity during follow-up after LT was classified by the same criteria for pre-LT UC activity. CMV viremia (assessed by in-house CMV polymerase chain reaction and CMV polymerase chain reaction Taq Man–based assay) was evaluated 3 times a week during the first 3 months after LT.
The course of post-LT UC over the entire follow-up period was compared with that in the period up to 5 years before LT in the 36 of 51 patients with UC before or after (de novo) LT who had >12 months' post-LT survival and without colectomy before LT. The course of UC was considered worse when it was becoming more active after LT than before LT—for example, moving from quiescent to active, requiring corticosteroid dependency, or requiring colectomy.
The laboratory center provided serological typing for HLA class I (-A and -B) and class II (-DR and -DQ) antigens. Typing for all HLA antigens was performed by the microcytotoxicity technique, as previously described.12 Assessment of HLA disparity was performed by 2 investigators (E.C. and P.Z.) blinded to clinical data, and data were checked by an in-house computer program (G.S.). For each HLA, disparity was considered to be present if there was complete dissimilarity between loci of donor and recipient.
All data were analyzed with SPSS statistical software, version 13.0 (SPSS, Chicago, IL). Quantitative variables were expressed as mean values ± 1 SD, and as median values (range) if their distribution was skewed. The χ2 test was used for categorical variables, and the Mann-Whitney U test was used for comparison of quantitative variables. Significance testing was 2-sided and set to less than 0.05. To identify the factors that were independently associated with the course of UC after LT, variables close to or under the statistical significance in the univariate analysis (P < 0.10) or clinically relevant to UC activity were included in the multivariate analysis by using logistic regression models.
Post-LT UC Activity
Post-LT UC activity compared with that during the 5 years before LT was the same in 16 (48.5%), and worse in 17 (51.5%) of 33 patients. Three other patients had de novo UC after LT. UC activity worsened after LT in 9 (43%) of 21 with quiescent disease (active in 2 and corticosteroid dependent in 5; 2 required colectomy), and in 8 (67%) of 12 with active pre-LT UC (corticosteroid dependent in 3; 5 required colectomy) (P = 0.18). In total, 7 (19%) of 36 patients required total colectomy after LT (3 for severe acute exacerbations and intractable disease and 4 for high-grade dysplasia or colon cancer). Post-LT UC course was not associated with graft survival. Patients with quiescent or active UC course after LT had a similar incidence of episodes of rejection diagnosed by analysis of protocol biopsy samples and presence of chronic rejection (80% vs. 85%, P = 0.74), as well as similar mean number of courses of bolus corticosteroids to treat rejection (1 and 1.2 in the 2 groups, respectively, P = 0.83). Repeated LT was not significantly different (6% vs. 27%, P = 0.29).
De Novo UC
De novo UC, confirmed by clinical and histopathological data, developed in 3 (16%) of 18 patients without pre-LT UC who survived more than 12 months after LT. Age at LT, duration of PSC, and type of immunosuppression and CMV viremia did not differ between patients with de novo UC and those without UC. UC activity was quiescent (only one mild or moderate exacerbation at presentation, which responded immediately to a short course of corticosteroids) in 2 patients diagnosed at 9 and 45 months after LT and followed for 6 and 48 months after onset of UC. UC had an active course in the third patient, ending in corticosteroid dependence 8 months after the first presentation.
Factors Associated With Post-LT UC Activity
Post-LT UC activity (compared with pre-LT UC activity) and de novo post-LT UC were not found to be related to patients' gender, age at LT, UC duration, extent of UC before LT, treatment with AZA or corticosteroids before LT, type of immunosuppression (CYC or FK alone or in combination with AZA or MMF), course and duration of corticosteroids after LT, or post-LT CMV viremia (Table 2). In addition, there were no differences when preexisting or de novo UC were analyzed separately. Finally, accurate smoking status before and after LT was available in 29 of 36 patients: 25 patients did not change their smoking status (24 were nonsmokers before and after LT, and 1 continued smoking after LT), but 4 patients changed their smoking habits after LT (3 stopped smoking and 1 became a smoker after LT). In the remaining 7 patients, 3 had the same activity of UC, and in 4, it was worse. Post-LT UC activity was not significantly associated with the changes in smoking status (Table 2).
|Characteristic||Post-LT UC activity||P value|
|Same (n = 16)||Worse or de novo (n = 20)|
|Age (yr)||40 ± 12||36 ± 13||0.28|
|Duration of UC (months) median (range)||108 (3–480)||84 (0–312)||0.45|
|Pre-LT course of UC (quiescent/active)||12/4||9/8||0.19|
|Extent of UC before LT (total/distal)||13/3||19/0||0.11|
|Corticosteroid dependence UC before LT||13/2||15/3||0.39|
|AZA before LT||14/1||15/2||0.27|
|Colon histology before LT (normal/mild/moderate/severe)||6/3/1/1||5/7/4/2||0.37|
|Maintenance immunosuppression (CYC/CYC+AZA/FK/FK+AZA/FK+MMF)||3/5/4/1/3||3/8/4/5/–||0.20|
|AZA after LT in general||6||13||0.12|
|Rejection episodes after LT (no/mild/moderate/severe)||3/4/5/4||4/3/5/8||0.76|
|Duration of corticosteroid therapy (months)||4.4||2.6||0.23|
|Duration of corticosteroid therapy ≥6 months (yes/no)||3/13||1/19||0.19|
|Unchanged smoking status (yes)*||11||14||0.49|
|Cytomegalovirus viremia after LT||6||7||0.65|
Disparity at a Single HLA Locus
Disparity with respect to HLA-A, -B, -DR, and -DQ was found in 21 (58%), 10 (27%), 16 (44%), and 14 (39%) donor-recipient pairs, respectively, of the 36 patients with UC before or after LT with >12 months' post-LT survival and without pre-LT colectomy. There was no significant difference in the course of post-LT UC based on individual HLA disparity. In particular, worse or de novo UC after LT was present in patients with similar donor-recipient pairs vs. those who showed disparity at one locus, as follows: for HLA-A: 11 (55%) of 20 vs. 9 (45%) of 20, P = 0.06; for HLA-B: 13 (65%) of 20 vs. 7 (35%) of 20, P = 0.28; for HLA-DR: 12 (60%) of 20 vs. 8 (40%) of 20, P = 0.55; for HLA-DQ: 14 (70%) of 20 vs. 6 (30%) of 20, P = 0.22 (Table 3).
|Characteristic||Post-LT UC activity||P value|
|Same (n = 16)||Worse or de novo* (n = 20)|
|Disparity/similarity at single locus|
|HLA disparity/similarity at 2 loci|
|HLA class I (A+B)||2/14||4/16||0.54|
|HLA class II (DR+DQ)||5/11||0/20||0.009|
Disparity at 2 or More Loci
Disparity with respect to HLA class I (both HLA-A and -B) and HLA class II (both HLA-DR and -DQ) was found in 6 (16%) for both pairs. Worse or de novo UC after LT was present in 16 (53%) of 30 patients with similar HLA class I (disparity in none or only one HLA-A or -B) donor-recipient pairs compared with 4 (66%) of 6 patients with disparity in both HLA-A and -B (P = 0.54). However, patients with similar HLA class II (disparity in none or only one HLA-DR or -DQ) had significantly worse UC activity or de novo UC after LT compared with those patients with disparity in both HLA-DR and -DQ loci (20 [65%] of 31 patients vs. 0 of 5, P = 0.009) (Table 3). In contrast, disparity with respect to 3 loci (in any combination) or all loci was not significantly associated with the course of UC after LT (P = 0.85 and P = 0.72, respectively). Disparity with respect to HLA was not significantly associated with recurrence of PSC after LT: for example, 14% of patients with disparity in both -DR and -DQ loci experienced recurrence of PSC after LT, compared with 16% of patients with disparity in none or only one -DR or -DQ locus (P = 0.88).
In the past, it was anticipated that inflammatory bowel disease would remain quiescent and not worsen after LT because of the continued immunosuppression. However, we and others have shown that inflammatory bowel disease does recur after LT or develops de novo, and the course of disease may be more aggressive.5, 6 In the present study, we evaluated a larger cohort of patients from our center, and again, we found that post-LT UC activity improved in none, worsened in 51.5% (9 of 21 patients with quiescent and 8 of 12 patients with active pre-LT UC), and developed de novo in 3 patients.
The factors that are involved in the post-LT UC activity remain unknown. In the nontransplantation setting, the development of UC may be the result of an insufficiently suppressed immune response to an unknown luminal antigen.13 UC is also considered an autoimmune disease because autoantibodies have been demonstrated, such as antibodies against neutrophils (pANCA) and against tropomyosin.13 However, these immunological disturbances have to be considered with the genetic background of the individual patient.14 Thus, there is evidence that this background may not only contribute to disease susceptibility (e.g., loci on chromosomes 16, 12q, 6p, and 14q),15 but more importantly, define or modify disease phenotype. An association between the HLA-DRB1*0103 and more extensive and severe colitis requiring colectomy has been found.16, 17
In the LT setting, donor and recipient HLA can have important implications for the patient's clinical course after LT. For example, in LT for autoimmune disease, the presence of donor-recipient HLA mismatching improves outcome in mismatched transplants.18 Moreover, Francavilla et al.19 reported that pediatric grafts with 2 HLA-A or 2 HLA-B mismatches had lower rates of acute rejection and better 5-year survival rates, respectively. In our study, we tried to elucidate another possible implication of HLA mismatching—that is, UC activity after LT—on the basis of a report of the course of UC in pregnancy and HLA matching between mother and fetus, in which disparity at both HLA-DRB1 and -DQ between mother and fetus was associated with milder activity score of inflammatory bowel disease during pregnancy.10 Interestingly, we found that donor-recipient pairs with similar HLA class II (disparity in none or only one -DR or -DQ) had significantly worse UC post-LT activity and/or de novo post-LT UC compared with pairs with disparity in both -DR and -DQ loci. Our findings are in accordance with those of a recent study that formed the hypothesis for our study.10
Other studies have also found that a maternal-fetal disparity in HLA class II antigens is correlated with the amelioration of other autoimmune diseases during pregnancy.20 In the case of patients with rheumatoid arthritis, it is thought that peptides derived from HLA class II of the fetus might compete with or displace maternal self-peptides, based on the hypothesis that a defect in the presentation of self-peptides is involved in the pathogenesis of the disease. Similar pathogenetic mechanisms have also been proposed in other autoimmune diseases, such as myasthenia gravis, multiple sclerosis, and collagen-induced arthritis.21 These studies suggest that fetal cells or HLA peptides derived from fetal cells may have a beneficial effect in these disorders. Thus, microchimerism may play a beneficial or an adverse role in autoimmune diseases activity during pregnancy, depending on the particular HLA relationship of host and nonhost cells.22–24
In organ transplantation, it has been hypothesized that donor cell microchimerism facilitates graft acceptance.25 It has also been suggested that donor cells with HLA more similar to host cells are more likely to result in microchimerism and potential risk for autoimmune diseases.23 Like pregnancy, organ transplantation has been associated with better course or remission in cases of autoimmune diseases. For example, islet transplant patients with previous HLA-mismatch renal transplantation have lower recurrent autoimmunity and/or islet graft rejection,26 and renal transplant patients for immunoglobulin A nephropathy have higher disease recurrence when there is good donor-recipient HLA matching.27 In LT patients, there are sporadic case reports for remission of autoimmune diseases, such as multiple sclerosis and rheumatoid arthritis after LT.28, 29 In these cases, although the donor and recipient HLAs are not given, the remission was attributed to LT itself and not only to immunosuppressant therapy, through a possible systemic microchimerism that modified the disease's course.28
In our study, we found that neither post-LT UC activity nor de novo post-LT UC was associated with CMV viremia, which was prospectively monitored 3 times a week for 3 months. Thus, we could not confirm a recent study, in which CMV infection was implicated with exacerbation and de novo post-LT UC.9 In our study, 1 of the 3 patients with de novo post-LT UC had no detectable CMV DNA at any time. The discrepant results may be because in the study by Verdonk et al.,9 both patients with PSC and autoimmune hepatitis were included and the presence of CMV-pp65 antigenemia (and not viremia) was evaluated. In addition, in our cohort, HLA mismatch was the only factor associated with post-LT UC course (Table 3). Our data are robust because the course of UC was prospectively collected and these data were evaluated with a specific hypothesis in mind. A potential source of bias was removed by having 2 investigators (E.C. and P.Z.) blinded to the clinical course of UC until after the matching had been evaluated and checked by another author (A.K.B.). However, because molecular testing was not performed, further studies are needed to ascertain the relative contribution of each locus. Moreover, other groups of loci may be worth evaluating in future studies.
In conclusion, combined HLA-DR and -DQ disparity between donor and recipient is associated with stable UC activity after LT. Although the exact explanation for our findings is not known, the findings are similar to effect in pregnancy with UC10 and to other autoimmune diseases in pregnancy.21 Whether the recognition of non-host-derived HLA antigens or some other factor that downregulates the immune response is involved with the reduced activity of UC remains to be investigated.
- 10HLA disparity determines disease activity through pregnancy in women with inflammatory bowel disease. Am J Gastroenterol 2004; 99: 1523–1526., , , .Direct Link: