Skip to main content
. 2021 Aug 4;11:682925. doi: 10.3389/fonc.2021.682925

Table 4.

Discrimination of different models in their ability to distinguish patients with a low risk of LNM.

Model Criteria for low risk of LNM Proportion of low-risk group Number of LNM in low-risk group Sensitivity Specificity PPV NPV
Model A (3) Probability of LNM calculated by the nomogram <0.2 73.4% (384/523) 18 (65 in total) 72.3% * 79.9% 33.8% * 95.3%
Model B (28) Pathological grade 1; Myometrial invasion <1/2;
Serum CA125 <35 IU/ml
51.7% (89/172) * 1 (18 in total) 94.4% * 57.1% * 20.5% * 98.9%
Model C (15) Ratio of [(P53 + Ki67)/(ER + PR)] < 0.71 78.1% (375/480) 28 (57 in total) 50.9% * 82.0% 27.6% * 92.5% *
Model D (11) Serum CA125 < 30.0 IU/mL, PR > 50% and Ki67 < 40%. 61.9% (229/370) * 6 (39 in total) 84.6% * 67.4% * 23.4% * 97.3%
Model E (29) Endometrioid histology;
For FIGO stage IA grade 1 or 2: 1) ER ≥30%; 2) ER < 30% and PR ≥15%. For FIGO stage IA grade 3, or FIGO stage IB grade 1 or 2: 1) no LVSI; 2) LVSI and PR ≥15%;
72.7% (346/476) 15 (58 in total) 74.1% * 79.2% 33.1% * 95.7%
Model proposed in this study Probability of LNM calculated by the nomogram <0.18 72.2% (393/544) in training cohort 15 (87 in total) 82.8% 82.7% 47.7% 96.2%
67.2% (156/232) in validation cohort 2 (42 in total) 95.2% 80.2% 52.6% 98.7%

NPV, negative predictive value; LVSI, lymphovascular space invasion; ER, estrogen receptor; PR, progesterone receptor. *P < 0.05 compared with the model proposed in this study.