A cause-specific hazard rate analysis of prognostic factors among 877 adults who received primary orthotopic liver transplantation

BACKGROUND. In orthotopic liver transplantation (OLT) distinct causes of graft failure (GF) and death with a functioning graft (DFG) exist. Prognostic factors for one failure type may be distinctly different from those predictive of other types, and an accurate portrayal of these relationships may more clearly explain each factor's importance. METHODS. A multivariable cause-specific hazard (CSH) rate analysis using Cox stepwise regression was performed among 877 adults who received primary OLT during 1996-2004 with tacrolimus+steroids as immunosuppression. RESULTS. Older donor age (P=0.004) implied greater primary dysfunction GF, while primary sclerosing cholangitis (PSC; P=0.0002) implied greater vascular thrombosis GF. Recurrent nonmalignant liver disease GF was higher among hepatitis C virus patients (P<0.00001), and younger recipient age (P=0.005) implied greater death from recurrent (metastatic) hepatocellular carcinoma. African-American race (P<0.00001), PSC (P=0.003), and younger recipient age (P=0.005) were independently associated with greater GF due to chronic rejection. Older donor age (P=0.003) implied greater infection DFG, while older recipient age (P=0.003) and pretransplant diabetes (P=0.03) were independently associated with greater cardiovascular/cerebrovascular DFG. Finally, most of these cause-specific predictors were not significant in an overall Cox model for graft survival. CONCLUSIONS. The CSH approach should be more widely used in investigations of prognostic factors. The result of older donor age implying greater primary dysfunction GF and infection DFG but having no association with other failure types demonstrates that its impact is specific to the graft's early posttransplant functional status. In addition, while recipient age was an important prognosticator, its direction of association reverses depending upon the outcome being analyzed.