Previous studies of predictors of death in liver transplant recipients have focused on the urgency of liver transplantation, concomitant illness before surgery, surgical risk factors, bacterial and fungal infections after transplantation, rejection of the transplanted organ, and the need for retransplantation [3-9]. Cytomegalovirus (CMV) is known to cause illness in liver transplant recipients [1, 10], and a trend for an association between CMV disease and decreased survival has been reported [10]. Evidence among heart and lung transplant recipients indicates that CMV infection is associated with decreased graft and patient survival [11, 12]. However, no data obtained using multivariate analysis are available for examination of the association between death and the status of CMV infection in liver transplant recipients.

We explore the possible effect of the CMV serologic status of donor and recipient on 1-year mortality rates in a cohort of liver transplant recipients after controlling for pretransplantation characteristics; post-transplantation interventions, such as the immunosuppressive regimen; and other outcomes, such as rejection of the transplanted organ and bacterial and fungal infections. All of the factors have been known to affect survival.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

We did an intention-to-treat analysis of prospectively gathered data from a cohort of 146 liver transplant recipients enrolled in a multicenter, randomized, placebo-controlled, double-blind trial of CMV immune globulin. We analyzed data from patients who were followed for 1 year in that study, which was done from December 1987 though June 1990 in four transplant centers in Boston, Massachusetts (the participating centers of the Boston Center for Liver Transplantation). Details of enrollment; follow-up; clinical management; all microbiological studies, including cultures and studies of serologic status for CMV; and immunosuppressive therapy and the definitions of CMV infection, CMV disease, bacteremia, and invasive fungal disease have been reported elsewhere [13, 14].

Statistical Analysis

Pretransplantation, intratransplantation, and post-transplantation variables were analyzed for their association with death. In the univariate analysis, we used a log-rank test from Kaplan-Meier survival analysis for discrete pretransplantation variables, a Wald chi-square from Cox proportional-hazards survival analysis for continuous pretransplantation and intratransplantation variables, and a Wald chi-square from time-dependent Cox proportional-hazards survival analysis for post-transplantation variables [15]. Post-transplantation variables were analyzed as time-dependent covariates, except for the total amount of solumedrol and murine monoclonal antibody (Orthoclone OKT3, Ortho Biotech, Inc., Raritan, New Jersey) and the development of CMV infection or CMV viremia, which were analyzed as time-independent covariates. Only variables that showed a statistical association (P < 0.05) with the 1-year mortality rate on univariate analysis were included as candidate variables for stepwise selection in the Cox proportional-hazards models. To explore the possible role of donor and recipient CMV serologic status on mortality rates, we constructed a model for multivariate analysis that did not include variables that were highly correlated with donor and recipient CMV serologic status. Multivariate analysis of pretransplantation and intratransplantation variables alone was also done.

Results

Top Methods Results Discussion Author & Article Info References

Univariate Analysis

Thirty-eight of 146 (26%) patients died within the first year after liver transplantation. The 1-year mortality rate was 25.4% (31 of 122 patients) in adults and 29.2% (7 of 24 patients) in children. In this intention-to-treat analysis, we included 5 patients who had had retransplantation after the first 30 days after enrollment and who had been censored from the original report after loss of the first graft. We also included 5 patients who had been excluded from analysis because they had received fewer than two doses of the study medication [13].

Univariate analysis showed a highly statistically significant increase in the relative risk for death associated with CMV disease, invasive fungal disease, bacteremia, retransplantation, number of units of blood products administered during transplantation, abdominal surgery (excluding retransplantation) after liver transplantation, CMV serologic status of donor and recipient before transplantation, and fungal isolation (colonization or invasive fungal disease) (Table 1). When analyzed individually, the number of units of each type of blood product (red blood cells, fresh frozen plasma, and platelets) administered during transplantation was also associated with higher mortality rates (Table 1).

We saw highly statistically significant differences in 1-year mortality rates among the four combinations of donor and recipient CMV serologic status before transplantation. One-year mortality rates by CMV serologic status were as follows: 11% if donor and recipient were both seronegative, 22% if donor was seronegative and recipient was seropositive, 30% if donor and recipient were both seropositive, and 44% if donor was seropositive and recipient was seronegative (P = 0.0091) (Table 1). This difference was seen for patients who received either CMV immune globulin or placebo (data not shown). Stratified analysis by donor CMV serologic status showed a 2.7-fold increase in relative risk for death up to 1 year for recipients of livers from CMV-seropositive donors compared with CMV-seronegative donors (95% CI, 1.4 to 5.3; P = 0.003).

Multivariate Analysis

In a multivariate analysis of pretransplantation variables only, the CMV serologic status of donor and recipient was the only variable that was significantly associated with survival (P = 0.0091) (Table 2). A Cox proportional-hazards model was used to analyze possible associations of pretransplantation and intratransplantation characteristics with death. This analysis showed that the number of units of blood products administered during transplantation (relative risk, 1.007 per unit; P < 0.001) and the combination of donor and recipient CMV serologic status (relative risks compared with seronegative donors and recipients were as follows: 1.94 for seronegative donor and seropositive recipient [P = 0.24], 3.17 for seropositive donor and recipient [P = 0.051], and 4.61 for seropositive donor and seronegative recipient [P = 0.0034]) were the only variables independently associated with a higher 1-year mortality rate.

A multivariate, time-dependent Cox proportional-hazards model with stepwise selection using the eight pretransplantation, intratransplantation, and post-transplantation variables that had a significant (P < 0.05) association on univariate analysis showed that retransplantation (relative risk, 4.6 [CI, 1.9 to 10.7]; P < 0.001), total number of units of blood products administered during transplantation (relative risk, 1.006 per unit [CI, 1.003 to 1.010]; P < 0.001), presence of CMV disease (relative risk, 3.9 [CI, 1.8 to 8.5]; P < 0.001), presence of invasive fungal disease (relative risk, 3.3 [CI, 1.5 to 7.1]; P = 0.0020), and presence of bacteremia (relative risk, 2.5 [CI, 1.2 to 5.2]; P = 0.0136) were independently associated with higher mortality rates (Table 2). Variables that were independently associated with survival in the main model remained unchanged in a subset analysis of the 124 adult liver transplant recipients (data not shown).

To fully explore the effect of donor and recipient CMV serologic status before transplantation on 1-year mortality rates, we studied the correlation between the four CMV serologic strata and the post-transplantation variables that were found to be independently associated with higher 1-year mortality rates. Donor and recipient CMV serologic status before transplantation was associated with presence of CMV disease (P < 0.001), invasive fungal disease (P < 0.001), and bacteremia (P = 0.0153). In another multivariate analysis that included CMV serologic status as a candidate variable but did not include the three post-transplantation characteristics that were highly associated with CMV serologic status, the combination of donor and recipient CMV serologic status was found to be independently associated with higher 1-year mortality rates (relative risks compared with seronegative donor and recipient were 2.23 [CI, 0.71 to 6.99] for seronegative donor and seropositive recipient, 5.15 [CI, 1.55 to 17.0] for seropositive donor and recipient, and 5.0 [CI, 1.8 to 13.9] for seropositive donor and seronegative recipient; global P value for differences among groups, 0.002). In that model, the total number of units of blood products administered during transplantation (relative risk, 1.01 per unit; P = 0.001) and retransplantation (relative risk, 3.50; P = 0.0012) remained highly associated with increased mortality rates.

Discussion

Top Methods Results Discussion Author & Article Info References

We show that the development of CMV disease in a multicenter, prospectively followed cohort of 146 recipients of orthotopic liver transplants is associated with almost a fourfold increase in the relative risk for death within 1 year after transplantation. Furthermore, we show a difference in survival among the four CMV serologic strata of donor and recipient; the highest risk occurred among CMV-seronegative patients who received a transplant from a CMV-seropositive donor (the group known to have the highest risk for CMV disease and CMV-associated complications). These results generally agree with those seen in heart and lung transplant recipients [11, 12]. Other risk factors for death, namely, the total number of units of blood products administered during transplantation, the need for retransplantation, and the development of fungal infections or bacteremia, were similar to those seen in previous studies of liver transplant recipients [5-8].

The strengths of the study from which we obtained our cohort are its multicenter nature, intention-to-treat analysis, prospective follow-up, blinded case definitions of CMV and invasive fungal disease, and blinded serologic analysis [13]. However, the study had some limitations. First, only six patients received FK506, an immunosuppressive agent that is associated with a slight increase in survival in liver transplant recipients [2]. Second, data on the nutritional status of transplant recipients were not collected [9]. Third, donor and recipient HLA testing had not been done in most of the patients examined in our study; HLA testing was not the standard of practice at the time and the relation of HLA matching to survival remains controversial [16]. Finally, these patients were treated before more effective means of preventing or treating CMV disease in liver transplant recipients were instituted [10, 17, 18].

Although it is difficult to determine whether deaths are directly attributable to CMV in patients who have multifactorial causes of death, its pathogenic and immunomodulating effects make CMV a unique pathogen. The possibility that CMV is primarily a marker of immunosuppression rather than a direct or indirect cause of death cannot be excluded. Although CMV is independently associated with decreased survival, we should emphasize that other variables may be more directly associated with death. Such variables may include fungal or bacterial infections, which are known to be associated with CMV and may be facilitated by the immunosuppressive properties of the virus [14, 19].

Cytomegalovirus serologic status of the organ donor and recipient is not currently considered to be a risk factor for death by the United Network for Organ Sharing; it is also not considered to be a factor in organ allocation strategies. When the probabilities for death according to donor and recipient CMV serologic status are examined, patients who receive an organ from a CMV-seropositive donor are clearly targets for more intensive prophylaxis [17, 18].

In conclusion, CMV disease and receipt of an organ from a CMV-seropositive donor were found to be independently associated with an increased 1-year mortality rate in liver transplant recipients. These data suggest that, despite the scarcity of liver donors, either a policy of matching donor and recipient for CMV serologic status or intensified CMV prophylaxis in liver transplant recipients who receive an organ from a CMV-seropositive donor is warranted, particularly in transplant recipients who are at risk for primary CMV infection.

Presented in part at the 33rd Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), October 1993, New Orleans, Louisiana.

Dr. Werner: Massachusetts State Laboratory Institute, 305 South Street, Boston, MA 02130.

Appendix

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