The widespread use of liver transplantation has been accompanied by the recognition of CMV disease as a serious complication in about 30% of patients [4]. Although some researchers have suggested that cytomegalovirus immune globulin or unselected immune globulin plus acyclovir may help prevent CMV in liver transplantation [5, 6], no single strategy has emerged as the standard of care. Furthermore, studies of CMV prevention in liver transplantation have been limited by the small number of patients studied and the lack of any randomized clinical studies [5-7]. Therefore, we performed a randomized, multicenter, placebo-controlled, double-blind trial of CMV disease prevention in liver transplant recipients.

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample and Protocol

Eligible patients included children and adults who had liver transplants at the participating centers of the Boston Center for Liver Transplantation: Children's Hospital, Massachusetts General Hospital, New England Deaconess, and New England Medical Center. The study protocol was approved by the human-investigation review committee of each institution.

Liver-transplant recipients received either prophylactic intravenous CMV immune globulin (CMVIG) or placebo. All candidates for transplantation and all liver donors were screened for CMV antibody. The CMV antibody status of donors and recipients did not influence eligibility for the trial.

Based on the estimate that CMV disease would develop in 50% to 60% of liver transplant recipients [8], we tested the hypothesis that the rate of CMV disease could be reduced by 50% in globulin recipients. We anticipated that we would need to study 65 patients in each group to have an 80% chance of identifying a significant effect (P = 0.05) of CMV immune globulin.

Cytomegalovirus Immune Globulin

The CMVIG was produced by the Massachusetts Public Health Biological Laboratories, used in previous clinical trials [2, 3] and licensed for use in the United States [9]. Details regarding CMV antibody titers and production have been previously described [2, 3, 10].

Patients who were randomly assigned to receive intravenous CMVIG were given 150 mg/kg body weight within 72 hours of transplantation, repeated doses at 2, 4, 6, and 8 weeks after transplantation, and 100 mg/kg 12 and 16 weeks after transplantation. This regimen was chosen after a pilot study showed that this dose would maintain antibody titers of 8 or higher as measured by indirect hemagglutination for up to 5 months after transplantation [11]. The placebo was 1% serum albumin, packaged in a manner identical to that of CMVIG.

Prenumbered vials for each patient's series of infusions were used. Randomization was performed in blocks of four, with each center having a separate randomization code and its own series of vials.

Clinical Management and Immunosuppression

After enrollment, patients were followed weekly for the first 8 weeks after transplantation, monthly for the next 6 months, and at 1 year after transplantation. Clinical care was individualized by the transplantation team at each institution. Baseline laboratory studies were done at enrollment, weekly for 2 months, when special problems arose, and at scheduled follow-up visits.

Standard immunosuppressive drugs included prophylactic cyclosporine, azathioprine, and steroid, given intravenously soon after transplantation, with a transition to oral administration when gastrointestinal function allowed. Azathioprine was withheld during periods of leukopenia. During the study, 35% of patients received prophylactic murine monoclonal antibody to T3 antigen (OKT3, Orthoclone, Ortho Biotech; Raritan, New Jersey) as part of a concurrent randomized trial comparing prophylactic OKT3 with standard triple immunosuppression. These patients received lower doses of cyclosporine. Acute rejection was treated primarily with intravenous bolus infusion of methylprednisolone; refractory or recurrent rejection was treated with OKT3.

Virologic Studies

Urine and throat-wash specimens (or throat swabs in young children) for viral isolation and serum for analysis of CMV antibody were collected before the first infusion, weekly for 2 months, and then monthly for 6 months. Peripheral-blood leukocytes were obtained for viral isolation every other week for the first 2 months and then every month for 6 months as well as at the time of clinical illness compatible with CMV disease and at the end of 1 year. In addition, tissue specimens obtained by biopsy and at autopsy were cultured for virus [12]. Postinfusion serum samples were also obtained from all patients.

For routine CMV cultures, human foreskin fibroblasts and MRC-5 cells were used at the different centers and examined regularly for viral cytopathic effects for up to 6 weeks [13]. Cytomegalovirus culture with early antigen detection using the shell-vial assay was used at several centers to rapidly diagnose CMV infection on various specimen types when clinically indicated [14].

At the time of transplantation, CMV antibody was measured in all patients' and liver donors' serum samples by complement fixation, latex agglutination, and enzyme-linked immunosorbent assay [13]. The enrollment and termination samples and selected serum samples collected during the time of illness were tested by complement fixation, latex agglutination, indirect hemagglutination, and enzyme-linked immunosorbent assay. To confirm seroconversion in the presence of exogenously administered globulin, we tested selected serum samples for CMV-specific immunoglobulin M antibody by enzyme immunoassay (M.A. Bioproducts; Walkersville, Maryland).

The serologic criteria were based on complement fixation as the primary test method for our analysis. We defined seronegativity as a titer of less than 1:8. We defined seroconversion as an increase in complement fixation titer from less than 1:8 to 1:16 or greater. We relied on complement fixation for determinations of seroconversion to avoid the confounding effect of exogenously administered antibody observed in other assays, whether from multiple transfusions administered during or after surgery or globulin administration. We used enzyme-linked immunosorbent assay, indirect hemagglutination, and latex agglutination to supplement the complement fixation assay when needed to clarify a lack of agreement with a donor procurement result or when anticomplementary activity was present. All decisions on serologic classification were made "blinded" as to randomization and before breaking the code.

Antiviral Use

"High-dose" acyclovir (>2 g/d for 30 days or more) was used in a few patients after the report by Balfour and colleagues was published [15]. Until May 1989 ganciclovir use was restricted to established CMV disease on a compassionate protocol. Throughout the study, use was determined at each institution on the basis of clinical judgment and availability from the manufacturer.

For patients in whom CMV pneumonia or clinically severe CMV disease developed, we allowed combination therapy with ganciclovir and CMVIG according to a modification of combination regimens for treating CMV pneumonia after bone marrow transplantation [16-18]. For these patients a standard dose of ganciclovir was given along with CMVIG in a dose of 100 mg/kg every other day for 14 days. Decisions to use combination therapy were left to the discretion of each investigator.

Analytical Framework and Case Definitions

Before beginning the study, we defined as primary end points the effect of CMVIG on rates of CMV disease, CMV pneumonia, severe CMV-associated disease (as defined below), viremia, mortality, and graft survival. All case and severity of illness classifications were made by an outside group of three consultants blinded to the randomization code. Final classification required unanimity of opinion. The analytic framework and all outcome measures were agreed on before breaking the code. We considered as dropouts all patients who died fewer than 15 days after transplantation and all patients who received fewer than two doses of CMVIG or placebo.

We defined CMV disease as the presence of clinical evidence of organ dysfunction along with biopsy proof of CMV in the affected organ, as documented either by virus isolation or histologic evidence of CMV in that organ. We only considered CMV disease to be present if disease was virologically confirmed, that is, at the time of the disease, either CMV was isolated from a clinical specimen, antibody seroconversion occurred within 1 month of the event, or histopathologic evidence of CMV infection was found.

In addition, we defined a clinical CMV syndrome as illness having two or more of the features known to be related to CMV (unexplained fever for at least 3 days in association with one of the following: pneumonitis without other cause, leukopenia [<4.0 x 10⁹/L leukocytes/mm³] or thrombocytopenia (platelets <100 x 10⁹/L) on 3 or more consecutive days after withdrawal of azathioprine or ganciclovir, or atypical lymphocytosis [>5% of peripheral leukocytes]). Because liver transplant recipients frequently have abnormal liver biochemistry test results, chemical evidence of hepatitis was not included in the syndrome definition.

We classified CMV pneumonia as definite if there was histologic, biopsy, or autopsy evidence of typical intranuclear inclusions or if a bronchoalveolar lavage specimen had cytologic evidence of intranuclear inclusions in a patient with organ dysfunction (hypoxia, interstitial infiltration). We defined CMV pneumonia as possible if the bronchoalveolar lavage specimens were positive for culture or for shell-vial rapid culture assay in a patient who had an interstitial pulmonary infiltrate with organ dysfunction. In addition, if a patient had an interstitial pulmonary infiltrate, lung dysfunction, evidence of CMV disease or infection elsewhere, and no other explanation for the infiltrate and a clinical response to ganciclovir, we considered that patient as having a possible case of CMV pneumonia. Cytomegalovirus hepatitis was defined as the presence of CMV inclusions on histologic testing or isolation of CMV from a liver biopsy specimen along with evidence of organ dysfunction.

We defined severe CMV-associated disease as biopsy-proven CMV disease in two or more organs or the presence of CMV pneumonia (definite or possible) or invasive fungal disease in the presence of CMV infection. Although inclusion of invasive fungal disease associated with CMV infection may be considered unusual, an association of CMV with opportunistic fungal disease has been reported [2, 19-21]. Further, CMVIG has been shown to reduce opportunistic fungal disease in kidney transplant recipients [2].

Fungal and opportunistic disease were also classified before the code was broken. We defined fungemia as the presence of a fungal species in the blood in one or more cultures. Invasive fungal disease was the presence of any opportunistic fungal pathogen in an organ on biopsy or at autopsy or the presence of fungemia. We considered the presence of Candida species in mucosal sites, including multiple mucosal sites, such as throat, urine, bronchoalveolar lavage, and T-tube or biliary drainage as colonization and classified those patients as not having invasive disease unless evidence of dissemination was found on biopsy, at autopsy, or from a blood culture. Isolation of Candida species from ascites was also considered noninvasive infection because many of these patients have this complication from a biliary anastomotic leak, that is, mucosal colonization.

Statistical Analysis

Before the study began, we decided to determine, in stratified analyses, whether effects of CMVIG differed according to donor-recipient CMV serologic status and OKT3 use. Because the data were censored, overall and stratified risk were compared using survival analyses and tested with the generalized Peto-Prentice statistic [22]. Estimates of relative risk and confidence intervals were calculated from the Cox proportional-hazard model [23]. Multivariate analyses were done also using a Cox proportional-hazards model [24]. The chi-square test and Student t-test were also used where appropriate. All P values are two-tailed.

Results

Top Methods Results Discussion Author & Article Info References

Enrollment and Randomization

From December 1987 until 5 June 1990, 146 patients were enrolled in the study from the four different centers. Seventy-three were randomized to each treatment group. Five patients dropped out, four in the CMVIG group and one in the placebo group. Reasons for dropping out included death within 14 days of transplantation (four patients) and refusal of further infusions after the first infusion (one patient in the CMVIG group). Therefore, 141 patients were available for analysis of efficacy. A mean of 30 samples for CMV culture were obtained from each patient for viral isolation.

Table 1 lists the demographic characteristics of the two study groups. No statistical differences were found with respect to age, sex distribution, type of underlying liver disease, cytomegalovirus serologic status for donor and recipient pairs, transplant urgency status, type of initial immunosuppressive regimen, number of rejection episodes, or type of treatment for rejection in the two groups. "High-dose" acyclovir was used sparingly. A slight disproportion of primary biliary cirrhosis and alcoholic liver disease existed between the two groups (Table 1).

Clinical Outcome

Thirty-five patients with CMV disease were in the study sample: 13 (19%) in the globulin group and 22 (31%) in the placebo group (relative risk, 0.56; 95% CI, 0.28 to 1.11; P = 0.11) (Table 2). The types of CMV disease seen included 17 cases of pneumonia, 28 cases of hepatitis, 1 case of colitis, and 4 CMV-associated syndromes. Histologically proven CMV pneumonia was documented in 6 patients: 1 globulin recipient and 5 placebo recipients (P = 0.10). Four patients in the placebo group were bronchoalveolar-lavage positive compared with two in the globulin group. Five patients had interstitial pneumonia with either CMV disease elsewhere, viremia, or both; all five responded to ganciclovir therapy. Three of the 12 patients with histologically proven or bronchoalveolar-lavage-positive CMV pneumonia were in the globulin group and 9 were in the placebo group (P = 0.08). Eleven of the 17 patients with pneumonia died (3 in the globulin group and 8 in the placebo group). Among patients with documented CMV infection or disease, ganciclovir therapy was used in 10 globulin and 16 placebo recipients (P > 0.2).

Virologic Outcome

Cytomegalovirus infection developed in 57% of globulin recipients and 61% of placebo recipients. The rates of viremia did not differ between the two groups (41% in globulin recipients compared with 43% in placebo group). The mean onset of viral excretion (any site) in globulin recipients occurred 62.7 (± 94.0) days after transplantation compared with a mean onset in controls of 40.8 (± 41.8) days. The mean onset of viremia was 45.3 (± 71.3) days in globulin recipients compared with 41.9 (± 38.0) days in placebo recipients.

Severe Cytomegalovirus-associated Disease

Twenty-seven patients had severe CMV-associated disease (CMV disease in two or more organs, CMV pneumonia, or disseminated or invasive fungal disease associated with CMV infection). The use of CMVIG prophylaxis resulted in a reduction of severe CMV-associated disease by more than 50% (relative risk, 0.39; CI, 0.17 to 0.89; P = 0.02) (Table 2).

Sixteen (59%) patients with severe CMV-associated disease died, compared with 11.4% of participants in the rest of the study sample (P < 0.01). Among patients who developed severe CMV-associated disease, four of eight globulin recipients died compared with 12 of 19 placebo recipients.

Opportunistic Disease

Of the 35 patients with CMV disease, complicating fungemia or disseminated fungal disease occurred in 13 (5 globulin recipients and 8 placebo recipients). The overall mortality rate in this group was 85% (four globulin and seven placebo recipients). Four more CMV viremic placebo recipients had candidemia, two of whom died. Two other candidemic patients had no evidence of CMV infection (one in each group) and one patient in the placebo group had both viruria and candidemia. Invasive fungal disease associated with CMV infection or disease occurred less frequently in the CMVIG group (relative risk, 0.35; CI, 0.13 to 1.02; P = 0.04) (Table 2).

Stratified Analyses

We analyzed the effect of CMVIG in subgroups (CMV donor and recipient serologic status, OKT3 use, and study center) using a stratified analysis (Table 3). For all analyses discussed, controlling for center ef-fects did not statistically affect the results (data not shown).

The relative risk for CMV disease in globulin recipients was 0.55 (CI, 0.27 to 1.08; P = 0.08) when controlled for CMV serologic status (Table 3). We found differences in the effect of globulin on the different CMV serologic strata. The most statistically significant effect of globulin was found in the group of patients with severe CMV-associated disease. Globulin recipients had more than a 50% reduction in such events (P = 0.02). When we controlled for serologic status, we found a relative risk of 0.34 (CI, 0.15 to 0.78; P < 0.01). The effect of globulin in the donor-positive, recipient- negative group, however, was much less than that seen in the other three serologic categories. When we controlled for OKT3 use, globulin was still protective (relative risk, 0.39; CI, 0.17 to 0.90; P = 0.02).

Morbidity and Mortality

At 1 year 11 of 69 (17%) of globulin recipients had died compared with 18 of 72 (25%) placebo recipients (P = 0.21). Three patients in the placebo group received second transplants as did five patients in the globulin group. The mean bilirubin level in 1-year survivors was 18.8 µmol/L (± 18.8 µmol/L) and 22.2 µmol/L (± 37.6 µmol/L) in globulin and placebo groups, respectively.

As a measure of morbidity, we analyzed the mean length of stay in all viremic patients. The globulin group had a mean length of stay of 57.3 (± 26.8) days for the first year after transplantation, whereas the placebo group had a mean length of stay of 73.7 (± 56.4) days (P = 0.07).

Multivariate Analyses

Because of the complexity of interactions among immunosuppression, globulin, CMV donor and recipient serologic status, blood product use, age, underlying diseases, and center, we constructed logistic regression models for CMV disease and severe CMV-associated disease using a Cox proportional-hazards model (Table 4).

For CMV disease, the multivariate model showed that donor and recipient serologic status (P = 0.001) and OKT3 use (P = 0.05) were significantly and independently associated with CMV disease. Globulin had a borderline protective effect (P = 0.11) (Table 4). When we controlled for OKT3 use, United Network Organ Sharing class 4, and serologic status, CMVIG had a protective effect (relative risk, 0.22; CI, 0.06 to 0.81; P = 0.01) for all serologic groups except the donor-positive, recipient-negative group.

For severe CMV-associated disease, the multivariate analysis showed that donor and recipient serologic status (P < 0.01), administration of more than 30 units of packed red blood cells (P < 0.01), and younger age (P < 0.01) were all independently associated with an increased risk for disease. Receipt of globulin was independently protective (P = 0.04) (Table 5).

Analysis of the Donor-Positive, Recipient-Negative Subgroup

To understand the lack of effect of globulin on the donor-positive, recipient-negative subgroup, we analyzed this group separately for potential confounders. The donor-positive, recipient-negative group received more OKT3 (71.4 ± 68.5 mg compared with 47.8 ± 44.2 mg, P = 0.12) and were more likely to be treated with OKT3 for rejection (P = 0.05) than the rest of the patients.

The subgroup of donor-positive, recipient-negative placebo recipients received slightly less OKT3 and more fresh frozen plasma than did globulin recipients. None of these differences, however, was statistically significant (data not shown).

Safety of Globulin

Seventeen of 73 (23%) patients in the group randomized to receive globulin developed a possible reaction compared with 8 of 73 (11%) placebo recipients (P = 0.08). Twenty-nine possible reactions occurred during 436 (6.7%) globulin infusions compared with 16 possible reactions during 419 (3.8%) placebo infusions (P = 0.07) (Table 5). Only one patient had an infusion discontinued because of side effects. This patient developed hemolysis, presumed to have been caused by anti-a in the globulin.

Discussion

Top Methods Results Discussion Author & Article Info References

The prevention of cytomegalovirus disease is of paramount importance to successful transplantation [1, 2, 24]. Not only is cytomegalovirus disease clinically important but cytomegalovirus infection itself may lead to further immunosuppression and the occurrence of opportunistic infection [19, 20, 24, 25]. We have shown that the prophylactic use of CMVIG reduces the risk for severe CMV-associated disease in patients having orthotopic liver transplantation. When the analysis controlled for the use of OKT3, globulin was still shown to be protective for severe CMV-associated disease.

Although the rates of CMV disease in globulin recipients were reduced by 40%, this finding was not quite statistically significant. The CMV disease rate in placebo recipients was lower than expected; therefore the sample size might have impeded our ability to detect a reduction in CMV disease. However, the biological consistency with the findings was remarkable: Both CMV disease and CMV pneumonia rates were reduced by approximately 50%. In this study and in our previous renal transplant study [2], we have shown that prophylactic CMVIG reduces the severity of CMV-associated disease without changing the rate of infection. Although we could not show a direct effect of globulin on rates of viremia, the mean length of hospitalization among viremic patients was reduced in globulin recipients by 21 days.

The interaction of globulin administration with other potential confounders such as donor-recipient CMV serologic status, the use of OKT3, blood transfusions, and plasma transfusions is clearly complex. In our multivariate analysis we found that the use of globulin was an independent factor in preventing severe CMV-associated disease. The most prominent risk factor, however, was the transplantation of a CMV-seropositive organ into a CMV-seronegative recipient.

In contrast to what had been previously observed in kidney transplantation, globulin did not provide significant protection for CMV-associated disease or severe CMV disease in the group at highest risk; namely, the CMV donor-positive, recipient-negative subgroup. There are several potential explanations for this finding. The levels of immunosuppression were greater in our liver transplantation patients compared with those who had renal transplants; rates of rejection therapy and use of OKT3 were more frequent. In our liver transplant group, the donor-positive, recipient-negative group received more OKT3 and blood products than did the study group as a whole. These therapies may have increased the risk for CMV disease. The apparent ineffectiveness of globulin may have been caused by small numbers of donor-positive, recipient-negative patients, representing an insufficient sample size. A final possibility, although theoretical, is that because the transplanted liver is larger than the kidney, the virus burden may be greater in liver than in kidney transplants.

The reduction in invasive fungal disease associated with CMV disease or infection in liver transplant recipients given prophylactic immune globulin is consistent with results seen in renal transplant studies using CMVIG [2] and in studies of bone marrow and liver transplant recipients using large volumes of unselected intravenous immune globulin [6, 26]. The mechanism by which CMVIG or any immune globulin might reduce invasive fungal disease is not known but could be caused by an immunomodulatory effect.

The apparent decreased effectiveness of globulin in donor-positive, recipient-negative liver transplant recipients accentuates the need for improved measures to control CMV disease. The donor-positive, recipient- negative population composed about 27% of the total group studied yet contributed 57% of the cases of CMV disease and 59% of the cases of severe CMV-associated disease.

Two recent studies using prophylactic ganciclovir have shown effective CMV disease prevention among CMV-seropositive heart and liver transplant recipients [27, 28]. However, neither study showed a protective effect for ganciclovir prophylaxis among the seronegative group receiving an organ from a CMV-seropositive donor [27, 28] nor did either study examine any effect on fungal disease. Studies in experimental animal models of CMV infection suggest a synergistic antiviral effect of combination ganciclovir and immune serum [29]. Whether a prophylactic strategy of combined CMVIG plus ganciclovir will prevent CMV disease among donor-positive, recipient-negative liver transplant recipients remains to be studied.

We have shown that prophylactic CMV immune globulin can reduce severe CMV-associated disease in liver transplant recipients without changing the overall rate of infection. We estimate the cost of such prophylaxis to be approximately $13 000. Formal cost–benefit studies are needed in these patients.

Appendix. The following institutions and individuals comprised the Boston Center for Liver Transplantation CMVIG Study Group: New England Medical Center: David R. Snydman, MD; Richard Rohrer, MD; Richard Freeman, MD; Karim Fawaz, MD; Mark A. Hoffman, MD; Marshall Kaplan, MD; Maura Gill, RN; John Griffith, MS. Massachusetts General Hospital: Robert H. Rubin, MD; Jules L. Dienstag, MD; Maureen Doran, RN, MPH. Children's Hospital: Edward O'Rourke, MD; Joseph Vacanti, MD. New England Deaconess Hospital: Roger Jenkins, MD; W. David Lewis, MD; Scott Hammer, MD; Maureen Martin, MD; Ralph Fairchild, MD. Massachusetts State Laboratory Institute: Barbara G. Werner, PhD; George F. Grady, MD; Jeanne Leszczynski, DrPH; Nancy Dougherty, RN, MPH; Andrea Katz, BSN, RN; Gary Fausett, MA. Clinical Data Analysis Committee: Richard Platt, MD; Sarah H. Cheeseman, MD; Mark Pasternack, MD; Sherwood L. Gorbach, MD.

Presented in part at the Thirty-first Interscience Conference on Antimicrobial Agents and Chemotherapy, Chicago, Illinois, October 1991, and at the Eleventh Annual Meeting of the American Society for Transplant Physicians, Chicago, Illinois, May 1992.