Combining Donor Characteristics with Immunohistological Data Improves the Prediction of Islet Isolation Success

Figure 2

Receiver operating characteristic (ROC) curve for predicting islet isolation success. (a) Area under the curve (AUC) = 0.796 for immunohistochemistry data; 95% confidence interval, 0.689 to 0.879; . Using an optimal cutoff point for an insulin-positive area in the pancreas of >1.02% resulted in 89% sensitivity and 76% specificity. (b) Area under the curve (AUC) = 0.653 for donor score; 95% confidence interval, 0.537 to 0.758; . The optimal cutoff point using a donor score of >68 points resulted in 60% sensitivity and 54% specificity. Prediction of islet isolation outcome was therefore superior using the immunohistochemical data ().

(a)

(b)