Assessment of prognostic factors and projection of outcomes in renal transplantation.

Analysis of data on renal transplantation collected in two large multicenter observational studies resulted in the concordant identification of five factors that correlated highly and at a substantial level of statistical significance with the outcome of unrelated cadaveric donor transplantation (i.e., they were associated with differences in one-year graft survivals of 0.07-0.21 and P values less than 0.05). These factors were: blood transfusions prior to the transplant, race of the recipient (white or black), prior failure in transplantation, level of sensitization to lymphocyte alloantigens, and diabetes as the cause of end-stage renal failure. Multivariate analysis with a mathematical survival model confirmed the importance and independence of these prognostic factors. Matching of HLA antigens appeared to be beneficial in both studies, but failed to attain high statistical significance in one. Systematic differences in the use of pretransplant splenectomy and, probably, in the nature of the antilymphocyte serum or globulin led to discordance in assessment of the importance of these factors in the two studies. Although advanced age (greater than 45 years) of the recipient was associated with reduced graft survival in both studies, analysis by means of the model failed to detect a significant correlation between the recipient's age and the outcome in one of the studies because the relation was not monotonic. In an illustration of their utility in the detailed assessment of performance, the prognostic factors were found to substantially account for the markedly superior results at one center and partly for lower graft survivals at another. These prognostic factors may be used to predict probable outcomes for populations and for individual patients subjected to particular arrays of conditioning strategies.