106 Chapter 4 Supplementary Table 2: LOOCV univariable logistic regression analysis on the diagnostic performance of the nine individual markers and LOOCV multivariable logistic regression analysis on the diagnostic performance of the optimal marker combinations for endometrial cancer detection in the different sample types. Individual methylation markers with the highest AUC value per sample type are depicted in bold. Urine ADCYAP1 BHLHE22 CDH13 CDO1 GALR1 GHSR HAND2 SST ZIC1 CDH13+GHSR+SST AUC (LOOCV) 0.81 0.84 0.89 0.90 0.78 0.93 0.69 0.58 0.76 0.93 Sens (%) 62 66 79 82 62 80 44 29 54 87 Spec (%) 93 98 92 87 86 93 92 96 94 90 Self-sample ADCYAP1 BHLHE22 CDH13 CDO1 GALR1 GHSR HAND2 SST ZIC1 CDO1+GHSR+ZIC1 AUC (LOOCV) 0.65 0.76 0.68 0.91 0.69 0.84 0.63 0.60 0.58 0.92 Sens (%) 47 60 44 80 44 72 46 28 31 84 Spec (%) 92 88 93 91 93 89 85 93 86 90 Scrape ADCYAP1 BHLHE22 CDH13 CDO1 GALR1 GHSR HAND2 SST ZIC1 CDH13+CDO1+ZIC1 AUC (LOOCV) 0.79 0.84 0.66 0.96 0.61 0.90 0.60 0.73 0.77 0.97 Sens (%) 57 72 48 86 53 75 43 64 67 92 Spec (%) 91 85 87 93 74 95 86 74 78 89 The sensitivity and specificity corresponding to the LOOCV AUCs are computed using the Youden’s Index (J) threshold. AUC = area under the receiver operating characteristic curve; LOOCV = leave-one-out cross-validated.
RkJQdWJsaXNoZXIy MjY0ODMw