Validating available Donor-recipient risk assessment tools in a German Heart Transplant cohort

Abstract

Abstract Objective Several Risk assessment tools have been developed to help predict mortality after heart transplantation. While validated within their regional transplant cohorts, it remains unclear whether they can also be applied to other cohorts. We therefore aimed to test and compare the predictive value of an US-Score validated in the UNOS Database by Joyce et al., a French Score by Jasseron et al., and the BO-Score within the Eurotransplant area developed by Schramm et al. Methods We applied the three scores to our cohort using published model coefficients and variables (see Figure 1 for distribution). The initial BO-Score as well as the French Score were hereby modified to produce positive integer values within our cohort (mod. French Score = [initial French Score + 2.24]x10, mod. BO Score = [initial BO initial score+10]x3). C-statistics using ROC-analysis (Figure 2) were used to compare the three scores regarding their ability to discriminate between 30-day, 1 year and 5-year survivors and non-survivors within our heart transplant cohort. Results The French Score performed best predicting the 1 survival with an AUC of 0.69 [95% CI: 0.59-0.79, p<0.0001], followed by the BO-Score with 0.66 (95% CI: 0.56-0.77, p=0.003) and the US-Score with an AUC of 0.67, CI 0.57-0.76, p=0.02). Regarding 30-day-Survival, only the French Score with an AUC of 0.64 [95% CI: 0.52-0.76, p=0.006] as well as the BO-Score with 0.71 (95% CI: 0.60-0.81, p=0.008) could predict mortality sufficiently, while the US-Score could not discriminate statistically significant between survivors and non-survivors (AUC 0.64, CI 0.52-0.76, p=0.06). All scores performed worst predicting 5-year survival, with only the BO-Score being statistically significant (AUC 0.61, 95% CI: 0.56-0.77, p=0.003). Conclusion In our heart transplant cohort, the available donor recipient risk assessment tools performed best predicting 1 year survival, with worse and heterogenous results for 30 day or 5 year survival. Further developments will have to incorporate risk factors not yet taken into account in order to make even more reliable predictions, especially for the Eurotransplant area.Figure 1 and 2