49 Retrospective analyses on toxicity in esophageal cancer patients Table 4 NTCP models on toxicity endpoints using a logistic regression analysis and cox regression models on overall survival Endpoint (logistic regression) Explanatory variable Intercept Odds ratio CI (95%) Significance performance Adjusted AUC Radiation pneumonitis Mean dose lung -2,56 1,18 [1,07-1,30] 0,00 (Nagelkerke R2=0.08) AUC 0,67[0,58 -0,75] 0,63 Pericardial effusion Mean dose pericard -3,26 1,11 [1,06-1,16] 0,00 (Nagelkerke R2=0.17) AUC 0,73[0,660,80] 0,70 Pericardial effusion Mean dose heart -3,11 1,09 [1,05-1,12] 0,00 (Nagelkerke R2=0,16) AUC 0,72[0,650,79] 0,69 Pericardial effusion Mean dose RV -3,24 1,08 [1,05-1,11] 0,00 (Nagelkerke R2=0,20) AUC 0,73 [0,670,80] 0,71 Endpoint (cox regression) Explanatory variable Hazard Ratio CI (95%) Significance performance Adjusted AUC Overall Survival (Cox regression) V 45 Lung 1.23 [1.09-1.39] 0.00 Stage I and II vs Stage III and IV 2.34 [1.38-3.96] 0.00 Diabetes 0.33 [0.13-0.83] 0.02 AUC 0.73 [0.670.80] AUC 0.70 After the bootstrapping procedure, the same variables were selected and included in the preferred model. The adjusted AUC was 0.69 (suppl. data figure 5). 4
RkJQdWJsaXNoZXIy MTk4NDMw