Acute graft-versus-host disease after liver transplant: Novel use of etanercept and the role of tumor necrosis factor α inhibitors



Acute graft-versus-host disease following orthotopic liver transplantation is a rare but feared complication arising in 1% to 2% of cases with a dismal prognosis. It most often presents as fever, rash, and diarrhea with or without pancytopenia. Patients die from complications of marrow failure such as sepsis or bleeding. Because of its low incidence, there is no clear treatment protocol for this complication. Both increasing and withdrawing immunosuppression have been attempted with variable success. Although anti–tumor necrosis factor α therapy has been widely used for the treatment of steroid-resistant acute graft-versus-host disease in the hematopoietic stem cell transplant setting, there previously have been no reported cases of its use in liver transplantation. The aim of this report is to review a case of acute graft-versus-host disease and the use of etanercept to manage this complication. Etanercept has never previously been used in liver transplantation complicated by acute graft-versus-host disease. In the hematology literature, the success of its use is offset by significant rates of serious infectious (especially fungal) complications. However, preliminary results are encouraging and offer insight into its use as a potentially viable therapeutic option. We report the first successful use of etanercept in liver transplantation–associated graft-versus-host disease, albeit complicated by invasive aspergillosis, and recommend concurrent antifungal prophylaxis when the drug is used in this setting. Liver Transpl 15:421–426, 2009. © 2009 AASLD.

Acute graft-versus-host disease (GVHD) is a rare complication after liver transplantation with a high mortality of 85% to 90%.1 It occurs when donor passenger lymphocytes mount an alloreactive response against the host's histocompatibility antigens, and it presents as fever, rash, diarrhea, and/or pancytopenia. Almost any organ except the liver can theoretically be involved.2, 3 Diagnosis is confirmed by demonstration of chimerism.4 Only 78 cases have been reported to date worldwide,5 and this makes it difficult to establish effective therapeutic options.

We report a case of GVHD post–liver transplant at our institution and describe the first successful use of etanercept in treating this complication.


G-CSF, granulocyte stimulating factor; GVHD, graft-versus-host disease; HLA, human leukocyte antigen; HSCT, hematopoietic stem cell transplant; POD, postoperative day; TNF-α, tumor necrosis factor α; VNTR, variable number tandem repeat.


A 65-year-old Chinese man with chronic hepatitis B virus infection underwent orthotopic liver transplantation for a 6.3-cm hepatocellular carcinoma. His hepatitis B virus DNA viral load was undetectable on entecavir. He was pretreated with 2 courses of transarterial chemoembolization with drug eluting beads before undergoing liver transplantation. The donor was a 74-year-old female whose human leukocyte antigen (HLA) type was A*02,32-B*40,44-DRB1*07,15 and whose blood group was O+. The recipient's HLA type was A*02,11-B*40,46-DRB1*14,15, and his blood group was O+. Both donor and recipient were cytomegalovirus-positive. The patient had an uneventful perioperative recovery and was discharged after 10 days with prednisolone, tacrolimus, and azathioprine.

On postoperative day (POD) 20, the patient presented with rigors and pancytopenia, which were discovered on routine follow-up only 3 days earlier when the culprit drugs were ceased. His hemoglobin was 71 g/L, his white cell count was 0.4 × 109, his neutrophil count was 0.03 × 109, and his platelet count was 111 × 109. He developed a rapidly progressive rash over his trunk and arms (Fig. 1). Over the next 24 hours, his temperature reached 37.8°, the rash deteriorated, and he developed florid, watery diarrhea with mucus. Blood cultures, viral serology, and polymerase chain reaction were negative for infection as a cause of his illness. Repeated stool specimens were negative for Clostridium difficile. He underwent bone marrow biopsy, which revealed very hypocellular marrow, and skin biopsy, which was histologically suggestive of GVHD. In the mean time, the patient was treated with broad-spectrum antibiotics, including meropenem and vancomycin, as well as amphoteracin intravenously for his ongoing febrile neutropenia. He was also managed supportively by granulocyte stimulating factor (G-CSF) and blood transfusions.

Figure 1.

Diffuse desquamative rash over the limbs and trunk.

Chimerism studies were performed on the marrow and skin specimens via fluorescent in situ hybridization, whereas molecular techniques using variable number tandem repeats were requested on peripheral blood, bone marrow, and buccal mucosal washings. In this method, DNA was extracted from bone marrow and peripheral blood samples with Qiagen 200-μL blood kits on the EZ1 DNA extraction robot (Qiagen, Valencia, CA). DNA from the recipient and donor before transplantation was obtained from the Immunology Department, Path-West, Royal Perth Hospital. Samples were amplified by polymerase chain reaction for the variable number tandem repeat locus D1S80 as previously described.6 This was identified as fully informative, showing unique-sized polymerase chain reaction product bands for both the donor and recipient. However, the method was only qualitative, and chimerism could not be quantitated.

On POD 22, the diagnosis of GVHD was suspected, and a trial of withdrawal of immunosuppression was commenced on the basis of previous successful case reports.7, 8

On POD 28, studies of the bone marrow and skin specimens identified the presence of 18% XX cells (the donor was female). Through the use of molecular techniques, the banding pattern of the patient's samples post–liver transplant revealed the presence of both donor-specific and recipient-specific alleles, confirming chimerism (Table 1).

Table 1. Results of Chimerism Studies Using VNTRs
DateClinical DataSpecimenVNTR LocusResultsConclusion
  1. Abbreviation: VNTR, variable number tandem repeat.

 73-year-old female donorPeripheral bloodDS1802 bandsDonor-specific, fully informative
 65-year-old male recipientPeripheral bloodDS1801 bandRecipient-specific, fully informative
12/17/073 weeks post–liver transplantBone marrowD1S803 bandsMixed chimerism
12/17/073 weeks post–liver transplantPeripheral bloodD1S803 bands, very faintMixed chimerism
12/31/075 weeks post–liver transplantPeripheral bloodD1S803 bands, very faint; visual limit of detectionMixed chimerism
02/01/089 weeks post–liver transplantPeripheral bloodD1S801 band, all for recipientNo chimerism detected

Neither clinical improvement nor marrow response was seen after immunosuppression was ceased for 5 days. Immunosuppression was re-instituted with high-dose methylprednisolone (1 g daily) for 3 days followed by 500- and 250-mg doses over the next 2 days. Thereafter, oral prednisolone (100 mg daily) was commenced, and the patient was weaned off it gradually over several weeks. Baseline immunosuppression was recommenced with lower doses of cyclosporine (250 mg twice daily) and mycophenolate mofetil (250 mg twice daily). In addition, etanercept (25 mg subcutaneously twice a week for 8 weeks) was started after discussions with the hematologists. Antifungal prophylaxis with fluconazole was not commenced at this point in time. Within 48 hours, a marrow response was noted with improved blood counts, negating the need for platelet transfusion (Fig. 2), and within 3 days, he was in clinical remission. His neutrophil count at this point was 3.0 × 109/L in the context of ongoing G-CSF administration. His hemoglobin was 114 g/L, his white cell count was 3.7 × 109/L, and his platelet count was 14 × 109/L. His fever subsided, and his rash and diarrhea rapidly resolved. Only 2 days later, his hemoglobin was 104 g/L, his white cell count was 17.3 × 109/L, his neutrophil count was 15.74 × 109/L, and his platelet count was 42 × 109/L. G-CSF was therefore ceased, and his white cell count remained in the normal range. His platelet count and hemoglobin eventually normalized within 4 weeks. His liver function tests remained normal throughout this period, except for a peak of bilirubin to 31 (<20 μmol/L), alanine aminotransferase to 55 (<40 U/L), alkaline phosphatase to 142 (35-135 U/L), and gamma glutamyl transpeptidase to 90 (<60 U/L) 4 days after immunosuppression was recommenced. Hepatic Doppler ultrasound was performed and demonstrated normal vascular flow. The hepatitis B DNA viral load was checked for viral reactivation, and this was undetectable. His liver function tests returned to normal over 4 weeks. He continued receiving hepatitis B prophylaxis with hepatitis B immunoglobulin and entecavir. The patient was discharged on POD 42, and 3 weeks later, chimerism was no longer detectable in his peripheral blood (Table 1). However, he developed an invasive Aspergillus infection of the face that required surgical debridement and a prolonged course of antifungal treatment. He has since recovered and is still alive 8 months post–liver transplantation.

Figure 2.

Progress of the neutrophil count with treatment.


Acute GVHD following orthotopic liver transplantation is a rare but feared complication arising in 1% to 2% of cases with a high mortality rate of 85%.1 At our institution, we have performed approximately 200 transplants, 2 of which have resulted in GVHD. The complication arises 1 to 6 weeks after orthotopic liver transplantation and presents as fever, rash, and diarrhea with or without pancytopenia1 but has been reported to present as late as 4 months after liver transplant.9 Patients succumb to complications of marrow failure, such as overwhelming sepsis or bleeding and metabolic derangements of desquamated skin and gut. Diagnosis is often delayed, and it is mistaken for viral infections and drug toxicities. Prognosis seems to be poorer in pancytopenia, although this does not reach statistical significance,10 and if diagnosis is delayed, this leads to deferred treatment.5

Essential to the diagnosis and confirmation of GVHD is the demonstration of chimerism, notably macrochimerism, in which donor cells compose >1% of circulating nucleated cells in peripheral blood.4 However, macrochimerism appears transiently in the majority of patients in the early postoperative period after orthotopic liver transplantation, especially in the first week,11–13 and rapidly declines by the third to fourth week post-transplant.13 In the context of fever, rash, diarrhea, and pancytopenia, demonstration of macrochimerism is highly suggestive of GVHD.4 There are various ways in which chimerism can be established. In essence, these methods examine the presence of donor cells in the recipient's circulation or tissues. These include the retesting of the recipient's HLA type by serology or DNA-based methods such as polymerase chain reaction using sequence-specific primers4, 13, 14 for the presence of the donor's HLA type, demonstration of sex-mismatched chromosomes,2, 15, 16 and microsatellite markers such as variable number tandem repeats or short tandem repeats.3, 13 We employed the last method in our case, but our assay was not able to quantify donor lymphocytes. Nevertheless, the disappearance of chimerism (Table 1) paralleled the clinical improvement of the patient.

Risk factors for the development of acute GVHD in liver transplantation include a close degree of HLA match,1, 17, 18 recipient age (typically >65),1 recipient immunocompromise,1 and difference in donor and recipient age of more than 40 years.1 Our patient had standard levels of immunosuppression and had only a moderate degree of HLA match with his donor. He was 65 years old and received a liver from a 74-year-old woman.

There is no clear protocol for the treatment of acute GVHD post–liver transplant. However, a closer look at the immunobiology of GVHD offers a variety of targeted treatments that have been reported with variable success in the literature. Taylor et al.10 reported a conceptual model of the pathogenesis of GVHD. The essential elements include intestinal epithelial destruction and release of proinflammatory cytokines including tumor necrosis factor α (TNF-α) and interleukin-1 by recipient macrophages and epithelial cells along with lipopolysaccharides, which augment the activity of antigen presenting cells. This in turn promotes the proliferation of alloreactive donor T cells, which target the damaged intestinal epithelium, propelling the vicious cycle of inflammation. In hematopoietic stem cell transplant (HSCT), the preparative regimen is the inciting agent of intestinal damage; however, in liver transplantation, the surgical process is thought to be the inflammatory stressor that up-regulates T cell activation, initiating this inflammatory cascade.

Various treatment modalities attempt to target different aspects of this model, usually in the form of increased immunosuppression using the broad lympholytic effects of steroids and/or an increase in baseline immunosuppressants.1, 4, 19–26 Other modalities of treatment include anti–T cell therapies,18, 27–29 anti–interleukin 2 therapy,30, 31 and activated host lymphocyte infusions.32 Success has also been reported with withdrawing immunosuppression after the onset of symptoms7, 8 in an attempt to restore immunological competence. Our case has demonstrated the application of some of these therapeutic concepts.

TNF-α has emerged as a key cytokine in the inflammatory cascade of acute GVHD.10 Although TNF-α inhibitors have been used with success for the treatment of steroid-resistant acute GVHD in the HSCT setting, there were no reported cases of its use in GVHD post–solid organ transplant at the time of this report. In a multicenter retrospective study33 of 32 patients with steroid-refractory grade II-IV acute GVHD from allogeneic HSCT, infliximab was added as a second-line agent for 14 patients and as a third-line agent for 18 patients with acute GVHD. Patients were more likely to respond if they were younger or had bowel-only involvement or if the time between HSCT and infliximab administration (thus lower TNFα levels) was longer. All 13 unresponsive patients died of GVHD at a median time of 43 days, whereas 13 of 19 responsive patients (68%) were still alive at a median follow-up of 449 days after infliximab. Three of the 6 remaining responders died from overwhelming sepsis, and 2 of the 3 had disseminated fungal infection. In all, 72% had at least 1 infectious complication. Couriel et al.34 reported 21 patients who had infliximab added for steroid-resistant GVHD after HSCT and found a complete response rate of 62% at the expense of increased infectious complications. The significant rates of serious infections have been replicated in other series, in which 38% to 54% treated with infliximab were complicated by fungal infections, especially invasive aspergillosis.34–36

The experience with etanercept in GHVD post–allogeneic HSCT is also encouraging but not conclusive. Uberti et al.37 assessed combination therapy with etanercept (25 mg twice weekly for 8 weeks), methylprednisolone, and tacrolimus as first-line therapy versus methylprednisolone alone in a pilot trial of 20 patients with acute GVHD. Complete response was achieved in 75% within 4 weeks of treatment. Once again, benefits were best noted in patients with extensive gut involvement, a factor that may have contributed to the success of our case, who had florid diarrhea. Another retrospective analysis38 of 16 patients with refractory GVHD treated with a combination of anti-thymocyte globulin, tacrolimus, and etanercept found an overall response rate of 81%, which was significantly better than that of historical controls using anti-thymocyte globulin alone. Busca et al.39 described a similar overall response rate of 62% in a small study of etanercept use in steroid-refractory/dependent acute and chronic GVHD; however, lower rates of infectious complications were noted (14% rate of bacterial infections and 19% rate of fungal infections) in comparison with the higher rates observed in studies using infliximab. These differences may be due to the cytolytic effect of infliximab, which theoretically should render etanercept a safer alternative. It is unclear if these high rates of infections in hematology patients apply to liver transplant patients who suffer GVHD, although in our report the patient developed an invasive Aspergillus infection. In this setting, we would consider the use of antifungal prophylaxis upon institution of etanercept treatment in the future.

A common limitation of the studies using TNF-α inhibitors in HSCT is the small number of heterogeneous patients with differences in preparatory regimens, immunosuppressant agents used prior to the addition of the anti–TNF-α agent, and severity of GVHD at onset. There is also the frequent lack of an interventional control group. Levine et al.40 attempted to address this issue in a phase II study by comparing 61 patients treated with etanercept and methylprednisolone to 99 controls treated with methylprednisolone alone as first-line management for acute GVHD post-HSCT. Statistically significant improvements in early complete response were seen in those treated with etanercept and methylprednisolone at 4 weeks (69% versus 33%, P < 0.0001), and this persisted to 12 weeks (77% versus 50%, P < 0.0009). Survival was statistically significant at 100 days post-transplant but did not reach statistical significance at 180 days overall, except in the subgroup of unrelated donors. Of note, there were no differences in rates of serious infections at 100 days.

In conclusion, we have reviewed a case of acute GVHD from our institution for which previously described treatment modalities were employed. This included both the withdrawal of immunosuppression and an increase in immunosuppression with the novel addition of etanercept, which is currently used in hematology patients with acute GVHD. In light of the compelling preliminary evidence regarding the use of etanercept in the hematology literature and the first successful case of its use in acute GVHD post–liver transplant, we propose that etanercept is another viable therapeutic option that should be explored. We also recommend the concurrent use of antifungal prophylaxis to prevent serious fungal infections.


The authors thank Dr. Campbell Witte for his time and scientific contribution.