Purpose: In Germany, approximately one-third of the harvested donor corneas are not suitable for transplantation, mostly due to insufficient endothelial cell density (ECD). The ECD can only be reliably determined after harvesting and processing of the cornea. Our group has previously developed a predictive model for corneal ECD: \( {Predicted\, ECD} = 2919-6^{\ast}\;{age}\; [{years}]-189\; [{if\, male}]\\ -7^{\ast}\;{death-to-explantation\, interval\,} [{hours}]\\ - 378\; [{if\, pseudophakic}] \;{cells/mm}^2 \).
Methods: A total of 2.999 consecutive donor corneas harvested between 2017 and 2021 from the Eye Bank of Rhineland-Palatinate in Mainz, Germany, were included. An actual ECD of >2000 cells/mm2 was defined as the cutoff value. To evaluate the clinical utility of the prognostic model as a screening instrument for transplant eligibility in an independent cohort, we performed a decision curve analysis.
Results: The median predicted ECD was 2061 cells/mm2 (interquartile range [IQR] = 1834 to 2221), whereas the median actual ECD was 2377 cells/mm2 (IQR = 1907 to 2624). There was a positive correlation between predicted and actual ECD (correlation coefficient = 0.411; P < 0.01). Our predictive model for ECD is a strong predictor for an actual ECD greater than 2000 (odds ratio = 1.374, 95% confidence interval [CI]) per 100 cells; P < 0.001, area under the curve [AUC] of 0.73). Decision curve analysis showed that the predictive model yielded a positive net benefit in clinical settings.
Conclusions: Decision curve analysis demonstrated a positive net benefit of the ECD predictive model in clinical settings with limited eye bank resources.
Translational relevance: In possible scenarios where a choice between corneal grafts is required, or in countries with limited eye bank infrastructure and staff, the initial estimate of ECD from the formula may be beneficial.