We appreciate Drs. Singh and Gayowski's interest in our work. We believe that considerable evidence is accumulating regarding the effect of donor and recipient CMV serologic status on mortality rates in orthotopic liver transplant recipients. By using rigorous analytical methods (including multivariate techniques to control for possible confounders), we showed an independent, clinically and statistically significant association between donor and recipient CMV serologic status and 1-year mortality rates among transplant recipients. One of the problems in interpreting results of other studies is that donor and recipient CMV serologic status may be commonly misclassified in liver transplantation because commercial assays do not have good specificity for detecting true CMV infection. Our primary method was complement fixation, which is more specific for CMV infection.

Preliminary findings in support of our data were reported by two groups in Europe. In a study of 688 orthotopic liver transplant recipients, French investigators [1] reported higher survival rates in seronegative recipients who received seronegative transplants compared with patients from other CMV serologic strata (5-year survival rates, 81% compared with 70%; P < 0.01) [1]. Spanish investigators [2] also reported higher survival rates in transplant recipients without CMV disease than in patients with CMV disease (3-year survival rates, 75.4% compared with 48.5%; P = 0.0048). Although the data from Stratta and colleagues are not statistically significant, there is a trend for decreased survival in patients who developed CMV disease compared with those who did not [3]. A trend for a difference in mortality rates among donor and recipient CMV serologic groups is also seen in the data reported by Singh and Gayowski, especially when the group of CMV-seronegative donors and recipients is compared with the other groups. Finally, similar findings on the effect of CMV on mortality rates have been reported in recipients of other solid organs (kidney, heart, and lung) [4, 5].

We recognize that current practices, including the use of tacrolimus (FK-506) as the primary immunosuppressive agent and preventive and therapeutic measures directed against CMV, may influence the magnitude of the effect of CMV on mortality rates in transplant recipients. We agree with Singh and Gayowski that other transplant centers should analyze their data to study the impact of CMV on mortality rates in transplant recipients.

Given the scarcity of liver donors (which leads to prolonged waiting time) and the high probability of death before liver transplantation, it may not be appropriate to match donor and recipient for CMV serologic status. However, our data strongly suggest that donor and recipient CMV serologic status has an independent effect on mortality rates in recipients of orthotopic liver transplants and thus that intense CMV prophylaxis is needed.