Predicting Graft Loss by 1-Year in Pediatric Heart Transplant Candidates: An Analysis of the PHTS Database
Background—Pediatric data regarding the impact of pre-heart transplant (HTx) risk factors on early post-HTx outcomes remain inconclusive. Thus, among patients with previous congenital heart disease (CHD) or cardiomyopathy (CMP), disease-specific risk models for graft loss were developed using pre-HTx recipient and donor characteristics.
Methods and Results—Patients enrolled in the Pediatric Heart Transplant Study (PHTS) from 1996-2006 were stratified by pre-HTx diagnosis into CMP and CHD cohorts. Logistic regression identified independent, pre-HTx risk factors. Risk models were constructed for 1-year, post-HTx graft loss. Donor factors were added for model refinement. The models were validated using patients transplanted from 2007-2009. Risk factors for graft loss were identified in CMP (n=896) and CHD (n=965) pts. For CMP, independent risk factors were earlier year of transplant, non-white race, female gender, diagnosis other than dilated cardiomyopathy, higher BUN, and panel reactive antibody (PRA) >10%. The recipient-characteristic risk model had good accuracy in the validation cohort with predicted vs. actual survival of 97.5% vs. 95.3% (C-statistic 0.73). For CHD patients, independent risk factors were non-white race, history of Fontan, ventilator dependence, higher BUN, PRA >10%, and lower body surface area. The risk model was less accurate with 86.6% predicted vs. 92.4% actual survival in the validation cohort (C-statistic 0.63). Donor characteristics did not enhance model precision.
Conclusions—Risk factors for 1-year post-HTx graft loss differ based on pre-HTx cardiac diagnosis. Modeling effectively stratifies the risk of graft loss in CMP patients and may be an adjunctive tool in allocation policies and center performance metrics.
- Received January 28, 2014.
- Revision received December 23, 2014.
- Accepted January 6, 2015.