Graft-versus-host disease following living donor liver transplantation



Graft-versus-host disease (GVHD) is the most common and well-known cause of morbidity and mortality following allogeneic bone marrow transplantation. Sporadic cases have been reported after cadaveric donor liver transplantation with usually fatal outcomes, however, the actual incidence and the characteristics of GVHD after living donor liver transplantation (LDLT) remain unknown. We herein report a person who developed fatal GVHD following LDLT and discuss the applicability of an HLA-homozygous donor to an HLA-haploidentical recipient. A 48-year-old male underwent LDLT for unresectable hepatocellular carcinoma with alcoholic liver cirrhosis. The donor was his 20-year-old son whose pretransplant HLA typing was homozygous at all loci. GVHD occurred 35 days after LDLT and was characterized by fever, diarrhea, maculopapular rash, and leukopenia, which led to the development of fatal pneumonia. We identified 4 cases of GVHD after LDLT in Japan and 1 in the United States, all associated with the use of an HLA-homozygous donor. The use of an HLA homozygous donor which results in a complete 1-way donor-recipient HLA match carries an extremely high risk of developing GVHD after LDLT. Therefore, it is possible that LDLT should be ruled out for such donors. A pretransplant work-up of the HLA type in both the donors and recipients is therefore imperative before determining the indications for LDLT. (Liver Transpl 2004;10:460–464.)

Living donor liver transplantation (LDLT) has now become an accepted and established alternative to cadaveric donor liver transplantation not only for small children, but also for adults.1, 2 In Japan, more than 2,000 cases of LDLT including adult-to-adult combinations have been performed since 1989.3

Graft-versus-host disease (GVHD) is a multiorgan disease occurring when immunologically competent cells from an allogeneic donor are transferred to and react against an immunocompromised patient. GVHD is well known to be a frequent complication after bone marrow transplantation.4–6 However, it is very uncommon after solid organ transplantation and approximately 30 cases have been reported to occur in cadaveric donor liver transplantation until now.7–9 In orthotopic liver transplantation, about 109 donor-derived leukocytes including monocytes, T-cells, and natural killer cells are transplanted with the liver allograft.10 After transplantation, these cells migrate to other tissues in the body and a microchimeric state thereafter develops. However, they rarely cause GVHD, owing to donor-derived immunocompetent cells reacting against recipient tissues. In general, the outcome of GVHD following liver transplantation is almost always fatal, however, some exceptional cases who survive GVHD have been reported.11–13

The use of an HLA-homozygous donor to a haploidentical recipient is a well-documented cause of transfusion-associated GVHD14. However, HLA-typing has been reported to have no impact on the outcome of liver transplantation.15 Such a combination between a donor and a recipient HLA is realistically negligible in the cadaveric donor liver transplantation setting. However, a complete 1-way HLA match between the donor and recipient could be a realistic possibility in LDLT because most donors in LDLT are genetically related to the recipients.

We herein report a case with fatal GVHD after LDLT who received an allograft from his HLA-homozygous son and discuss the validity of using such a donor in the LDLT setting.


GVHD, graft-versus-host disease; LDLT, living donor liver transplantation; ATG, antithymocyte globulin; CF, colonoscopy; CyA, cyclosporine A; G-CSF, granulocyte colony-stimulating factor; GIF, gastroduodenal endoscopy; Bx, biopsy; Dx, diagnosis; POD, postoperative day; BA, biliary atresia; PBC, primary biliary cirrhosis; HCC, hepatocellular carcinoma; AZA, azathioprine; PSL, predonisolone; FK, tacrolimus; MMF, mycophenolate mofetil.

Case Report

A 48-year-old Japanese male (blood type: type A, Rh positive) underwent a LDLT due to alcoholic liver cirrhosis with unresectable hepatocellular carcinoma. He had no history, physical findings, or laboratory findings suggestive of any immunodeficiency. Serologic tests for hepatitis A, B, C, and human immunodeficiency virus were negative. The donor was the 20-year-old recipient's son whose blood type was identical to that of recipient: type A, Rh positive). The HLA class I and class II phenotype of the donor was homozygous at all loci: A33, B44, DR6 while the recipient's was A33, A24; B44, B62; DR6, DR4. The allograft was a left lobe graft weighing 480g, which accounted for 38.6% of the recipient's standard liver volume. During transplantation, the patient received 6 units of group A, Rh-positive packed red blood cells. All blood products from random donors were irradiated and filtered before transfusions. The induction and maintenance immunosuppressive drugs consisted of cyclosporine, mycophenolate mofetil and steroids. The initial function of the graft was excellent and the total bilirubin level normalized by postoperative day 12. The posttransplant course was uneventful and the patient did well until postoperative day 35, when he developed a low grade fever with watery diarrhea of 1 to 2 liters per day (Fig. 1). A stool work-up was negative. An endoscopic examination of the colon and the upper gastrointestinal tract revealed cytomegalovirus gastroenteritis, which was confirmed by a pathological examination of a biopsy specimen. All immunosuppressive drugs were temporarily stopped and ganciclovir (5 mg/kg IV q 12H) was started. On the same day that endoscopy was performed, a skin rash involving the trunk, face, palms, and soles was recognized. Initially, a drug reaction was suspected and all of the suspected drugs including antibiotics were stopped. However, all symptoms continued to worsen over the next few days despite treatment. A skin biopsy specimen on postoperative day 42 revealed epidermal dyskeratosis, basilar vacuolization, and a predominantly mononuclear and lymphocytic infiltrate, which was consistent with GVHD. A repeated total colonoscopy revealed a worsening of the patient's condition, which was characterized by diffuse friability and hemorrhage of the colonic mucosa extending up to the ileo-cecal junction. The presence of donor-derived chimerism in peripheral blood (postoperative day 48) as well as bone marrow (postoperative day 52) was demonstrated (Fig. 2) by polymerase chain reaction-based amplification of a variable number of tandem repeat marker (MCT118). A semiquantitative analysis of microsatellite marker D5S818 showed about a half (47%) of all T cells to be donor in origin (data not shown). A diagnosis of acute GVHD was made based on the typical clinical symptoms and the presence of chimerism in the peripheral blood. The patient was treated with tacrolimus and high dose methylprednisolone (1000 mg, 3 days) which was thereafter tapered. The patient temporally responded to the treatment with an improvement in the diarrhea and skin rash for a few days. However, severe leukopenia (white blood cell count less than 1000/mm3) developed on postoperative day 52 and the general condition of the patient deteriorated with an exacerbation of diarrhea and skin rash. Intravenous horse antithymocyte globulin 900 mg for 5 days as well as granulocyte colony-stimulating factor and namofstat mesilate were started. Despite these intensive treatments, the patient developed pneumonia and died on postoperative day 61.

Figure 1.

The Immunosuppression and clinical course of the recipient. ATG, antithymocyte globulin; CF, colonoscopy; CyA, cyclosporine A; G-CSF, granulocyte colony-stimulating factor; GIF, gastroduodenal endoscopy; Bx, biopsy.

Figure 2.

Analysis of chimerism in the peripheral blood (A) at postoperative day 45 and bone marrow (B) at postoperative day 52 by polymerase chain reaction-based amplification of a variable number of tandem repeat markers (MCT118). A. lane 1, size marker; lane 2, recipient oral mucosa, lane 3, donor peripheral mononuclear cells; lane 4, recipient peripheral mononuclear cells; lane 5, recipient peripheral granulocyte. B. lane 1, size marker; lane 2, recipient oral mucosa, lane 3, donor peripheral mononuclear cells; lane 4, recipient bone marrow mononuclear cells (E-rosette, positive); lane 5, recipient bone marrow mononuclear cells (E-rosette, negative); lane 6, recipient bone marrow granulocyte.


The mechanism of GVHD after LT is unclear. However, it is believed to be basically the same as that of transfusion-associated GVHD, although GVHD after liver translantation differs in some manifestations which characteristically lack liver involvement. The HLA type of the donor and the recipient are such that the recipient does not recognize the donor immune cells as foreign. This may occur when the donor is homozygous for HLA determinants and thus shares a haplotype with a recipient. In this situation, the recipient does not recognize and eliminate donor cells. On the other hand, the donor cells may recognize the unshared haplotype as foreign and react against the recipient. This histocompatibility combination is estimated to occur in 1 of 500 (0.2%) donor-recipient pairs.16 The incidence of homozygous HLA was reported to be 1.6% among the Caucasian population16, while a rate of approximately 3.2% (81/2,635) among blood donors in Japan (unpublished data from Japanese Red Cross Society). Furthermore, in Japan, it was reported that the incidence of 1-way HLA match between non-relatives was 1/800, while 1/100, 1/190 and 1/180 were expected between combinations of parent-child, sibling-sibling, and grand parent-grand child, respectively.17 Accordingly, the risk of selecting 1-way HLA match could be extremely high in the LDLT setting as compared to those in cadaveric situations. In fact, we retrospectively reviewed HLA phenotypes of donors and recipients in our series and found 4/98 (4.1%) of the donors and 1/94 (1.1%) of the recipients to be HLA-homozygous. Furthermore, we retrospectively identified 2 additional cases who had combinations of a 1-way HLA match. These 2 patients are currently doing well and have not developed GVHD with a follow-up period of 39 and 17 months after LDLTs. We tried to prove chimerism in these recipients, however, donor cells could not be detected in the peripheral blood of these 2 recipients (data not shown). Therefore, in our series, the incidence of a complete 1-way HLA match was 3/98 (3.1%) and 1 of 3 recipients with this HLA combination developed GVHD. A Chicago group was the first to report a pediatric case with chronic GVHD who had an allograft from HLA-homozygous parental donor.18 Kiuchi et al.19 analyzed their large series of LDLT and found that 8/280 donors (2.9%) and 11/278 recipients (2.9%) were completely HLA-homozygous. They reported that 1 of 4 cases with complete 1-way HLA match developed GVHD. We surveyed the incidence of GVHD after LDLT in Japan and identified a total of 4 cases including the present cases, which were all associated with the use of a homozygous donor (Table 1). Therefore, as in transfusion-associated GVHD, the use of an allograft from a homozygous donor to a haploidentical recipient should be recognized as a single, definite risk factor for GVHD after LDLT.

Table 1. Characteristics of the 4 Reported Cases With GVHD After Living-Donor Liver Transplantation
No.Age/GenderDxImmunosuppressionHLAOnset (POD)TreatmentOutcome (POD)
  • Abbreviations: Dx, diagnosis; POD, postoperative day; BA, biliary atresia; PBC, primary biliary cirrhosis; HCC, hepatocellular carcinoma; CsA, cyclosporine A; AZA, azathioprine; ATG, antithymocyte immunoglobulin; PSL, predonisolone; FK, Tacrolimus; MMF, mycophenolate mofetil.

  • Present case.

19m/FBACsA, AZA, PSLA1,-; B8,-; DR3,-DR3, DR760CsA, ATGAlive (250)
28m/FBAFK, PSLA24,-; B52,-; DRB1∗1502,-; DQB1∗0601,-A2, A24; B52, 61; C10,-; DRB1∗0802, 1502; DQB1∗0302, 060127 Dead (43)
350y/FPBCFK, PSLA3301/03,-; B4402,-; Cw∗14021/03,-; DRB1∗1301-02,-;A2601-07, A3301/03; B4402, B4006; Cw∗14021, Cw∗0801/3; DRB1∗1301-02, DRB1∗09012;114FK, PSLAlive
448y/MHCC AlcoholCsA, MMF, PSLDQB1∗06011,- A33,-; B44,-; DR6,-DQB1∗06011, DQB1∗03032 A33, A24; B44, B62; DR6, DR442ATGDead (61)

The prognosis of GVHD after liver transplantation is generally poor and survival is very rare especially in adult patients. A few cases have been reported to survive such complications by means of a reduction or withdrawal of immunosuppression12 as well as increased immunosuppression.10 However, no established treatment for GVHD exists.

The prevention of GVHD before transplant is very difficult. However, it may be prevented by the immunosuppression of the donor before organ harvest, the physical removal of obvious nodal tissue from the graft, the removal of T-cells by elutriation or perfusion of the graft with anti-T-cell drugs (T-cell depletion), and the addition of more potent immunosuppressive drugs such as methotrexate. However, the efficacy of these measures is unknown and their feasibility in living donors is either questionable or very limited. One possible measure could be to irradiate the graft before or after implantation. Although no study has been done in the setting of liver transplantation, the efficacy of irradiation has been already established in blood transfusion. Furthermore, in the setting of renal transplantation, the efficacy of local graft irradiation has been reported to reverse refractory rejection.20 As a result, a further basic study on the efficacy of graft irradiation is warranted for the prevention of GVHD after liver transplantation.

A history of discontinuing of immunosuppression before developing GVHD could be an important characteristic of GVHD after LDLT. To the best of our knowledge, a total of 4 cases of GVHD after LDLT, including the present case, all discontinued immunosuppression before developing GVHD. The mechanism of this action remains unclear, however, such a discontinuation may have permitted an expansion of the donor-derived T-cell population, thereby leading to the onset of GVHD.

In conclusion, the use of a graft from an HLA-homozygous donor with 1-way donor-recipient HLA-match carries an extremely high risk of developing GVHD in LDLT. Therefore, LDLT under such combinations should be avoided or restricted to highly emergent cases. A pretransplant work-up including complete HLA typing is therefore essential in LDLT to avoid such combinations.