Table 2.
Performance of RMI, CA125, ROMA, and new models GOT-1 (RMI + HE4), GOT-2 (CA125 + HE4) and GOT-3 (ROMA + TVU), comparing benign disease with EOC.
| Group (n) | Model | p-Value | ROC | SN % (75% SP) | SP % (Target SN) | |
|---|---|---|---|---|---|---|
| AUC | 95% CI | |||||
| Benign (445) vs. EOC (135) | RMI3 (cut-off > 200) | <0.001 | 0.95 | 0.93–0.97 | 97 | 84 |
| GOT 1 (RMI + HE4) | 0.95 | 0.93–0.98 | 96 | 86 | ||
| Benign (445) vs. EOC (135) | CA125 (cut-off > 35 U/mL) | <0.001 | 0.92 | 0.89–0.94 | 88 | 68 |
| GOT 2 (CA125 + HE4) | 0.94 | 0.92–0.97 | 93 | 79 | ||
| Benign (230) vs. EOC (23) Pre-M | ROMA (cut-off ≥11.4%) | <0.001 | 0.93 | 0.86–1.00 | 87 | 81 |
| GOT 3 (ROMA + TVU) | 0.94 | 0.87–1.00 | 91 | 88 | ||
| Benign (215) vs. EOC (112) Post-M | ROMA (cut-off ≥ 29.9%) | <0.001 | 0.94 | 0.91–0.96 | 93 | 77 |
| GOT 3 (ROMA + TVU) | 0.94 | 0.91–0.96 | 94 | 80 | ||
p-values calculated with the likelihood ratio test, comparing the regression models to their baseline models. p-value < 0.05 was considered statistically significant. Specificity achieved calculated using target sensitivity (Target SN). Target SN GOT-1 = 92%, target SN GOT-2 = 93%, and target SN GOT-3 = 87%/91% (Pre-M/Post-M). AUC = area under the curve; GOT 1 = Gothenburg index 1; GOT 2 = Gothenburg index 2; GOT 3 = Gothenburg index 3; Pre-M = premenopausal; Post-M = postmenopausal; RMI = Risk of malignancy index; ROMA = Risk of malignancy algorithm; ROC = receiver operating characteristics; SN = sensitivity; SP = specificity; TVU = Transvaginal ultrasound.