The effects of cytomegalovirus serology on graft and recipient survival in cadaveric renal transplantation: implications for organ allocation.

The potential benefits from allocating donated cadaveric kidneys based on donor and recipient cytomegalovirus (CMV) serology remain controversial. We estimated graft survival and recipient survival using bivariate Kaplan-Meier models and multivariate Cox proportional hazards models for 24,543 first cadaveric renal transplantations performed in the United States between 1989, coinciding with the introduction of ganciclovir, and 1994. The effects of donor and recipient CMV serology were estimated, and the implications of these estimates for CMV-based allocation of cadaveric kidneys were considered. From Kaplan-Meier estimates, the 3-year impact of CMV-seropositive donor kidneys was a 3.6% reduction in graft survival and a 2.4% reduction in recipient survival for CMV-seronegative recipients, and a 3.9% reduction in graft survival and a 3.0% reduction in recipient survival for CMV-seropositive recipients. Multivariate Cox analysis demonstrated an adverse impact of donor CMV seropositivity regardless of recipient CMV status. D-/R- CMV serologic pairs had the best 3-year outcomes, with 73.4% graft survival and 87.7% recipient survival. D+/R+ CMV serologic pairs were found to have the worst 3-year outcomes, with 68.4% graft survival and 83.1% recipient survival, and were significantly worse than D+/R- pairs in terms of recipient survival. The maximum estimated impact of a program allocating donor kidneys to maximize the number of D-/R- CMV serologic pairs, assuming no impact on HLA mismatches, was a 0.1% reduction in aggregate 3-year graft survival and a 0.2% reduction in aggregate recipient survival. An alternative program allocating donor kidneys to minimize the number of D+/R+ pairs had no estimated effect on either graft or recipient survival. We conclude that during the ganciclovir era, CMV continues to have an important impact on first cadaveric renal transplantation. However, even under ideal conditions, CMV-based kidney allocation to either maximize the number of D-/R- pairs or minimize the number of D+/R+ pairs is likely to provide little benefit to the population of cadaveric renal transplant recipients.