Demographic characteristics
In our study, we enrolled 644 and 277 cases into the training cohort and the validation cohort. There was no difference in various indicators between the two cohorts (P > 0.05, Table 1). Most of the patients were white (76.9%), the histological type was serous carcinoma (44.4%), the most histological grade was G3 (43.8%), 66.6% of the patients were with positive serum CA125 and 77.6% of the patients were with positive LNM.
Table 1
Characteristics of patients in the training and validation cohorts
Characteristics
|
Training cohort
(N=277)
|
Internal Validation cohort
(N=644)
|
P
|
Age (years)
|
|
|
0.985
|
≤40
|
22(7.9%)
|
51(7.9%)
|
|
40-70
|
199(71.8%)
|
458(71.1%)
|
|
≥70
|
56(20.2%)
|
135(21.0%)
|
|
Insurance
|
|
|
0.317
|
Insured
|
193(69.7%)
|
427(66.3%)
|
|
Uninsured and others
|
84(30.3%)
|
217(33.7%)
|
|
Race
|
|
|
0.473
|
White
|
220(79.4%)
|
488(75.8%)
|
|
Black
|
20(7.2%)
|
52(8.1%)
|
|
Other
|
37(13.4%)
|
104(16.1%)
|
|
Marital status
|
|
|
0.552
|
Married
|
212(76.5%)
|
481(74.7%)
|
|
Unmarried and others
|
65(23.5%)
|
163(25.3%)
|
|
Histology
|
|
|
0.751
|
Serous adenocarcinoma
|
123(44.4%)
|
286(44.4%)
|
|
Mucinous carcinoma
|
26(9.4%)
|
56(8.7%)
|
|
Endometrioid carcinoma
|
74(26.7%)
|
158(24.5%)
|
|
Others
|
54(19.5%)
|
144(22.4%)
|
|
Differentiation
|
|
|
0.358
|
Grade I
|
66(23.8%)
|
143(22.2%)
|
|
Grade II
|
49(17.7%)
|
89(13.8%)
|
|
Grade III
|
112(40.4%)
|
291(45.2%)
|
|
Grade IV
|
50(18.1%)
|
121(18.8%)
|
|
Laterality
|
|
|
0.970
|
Left
|
129(46.6%)
|
305(47.4%)
|
|
Right
|
125(45.1%)
|
285(44.3%)
|
|
Unspecial
|
23(8.3%)
|
54(8.4%)
|
|
Tumor size (mm)
|
110.27±67.28
|
112.55±69.65
|
0.645
|
Preoperative serum CA125 level
|
|
|
0.175
|
Positive
|
195(70.4%)
|
418(64.9%)
|
|
Negative
|
38(13.7%)
|
119(18.5%)
|
|
Not documented
|
44(15.9%)
|
107(16.6%)
|
|
lymph node metastasis
|
|
|
0.869
|
Negative
|
61(22.0%)
|
145(22.5%)
|
|
Positive
|
216(78.0%)
|
499(77.5%)
|
|
Nomogram Construction
We used the univariate logistic regression to analyze the association between age, insurance, race, marital status, histology type, histology grade, laterality, tumor size, preoperative serum CA125 level and LNM. We found that histology type (Mucinous carcinoma, OR=0.346, p<0.001; Endometrioid carcinoma, OR=11.555, p=0.018; Others, OR=2.807, p=0.007), histology grade (Grade II, OR=6.059, p<0.001; Grade III, OR=6.658, p<0.001; Grade IV, OR=1.894, p=0.005) and preoperative serum CA125 level (positive, OR=2.749, p<0.001) were all significant predictors of LNM (Table2).
Table 2
Univariate and multivariate logistic regression model for predicting lymph node metastasis in the model of training cohort
Variables
|
Univariate analysis
|
|
Multivariate analysis
|
OR (95% CI)
|
P value
|
OR (95% CI)
|
P value
|
Age (years)
|
|
|
|
|
|
≤40
|
1
|
|
|
1
|
|
40-70
|
1.344(0.566,3.190)
|
0.503
|
|
0.530(0.195,1.439)
|
0.213
|
≥70
|
0.791(0.493,1.269)
|
0.331
|
|
0.544(0.323,0.917)
|
0.022
|
Insurance
|
|
|
|
|
|
Insured
|
1
|
|
|
1
|
|
Uninsured and others
|
0.787(0.527,1.176)
|
0.243
|
|
0.805(0.515,1.256)
|
0.339
|
Race
|
|
|
|
|
|
White
|
1
|
|
|
1
|
|
Black
|
1.445(0.894,2.337)
|
0.133
|
|
1.213(0.708,2.079)
|
0.483
|
Other
|
1.160(0.542,2.481)
|
0.702
|
|
0.964(0.406,2.286)
|
0.933
|
Marital status
|
|
|
|
|
|
Married
|
1
|
|
|
1
|
|
Unmarried and others
|
0.922(0.600,1.418)
|
0.712
|
|
0.844(0.523,1.362)
|
0.487
|
Histology
|
|
|
|
|
|
Serous adenocarcinoma
|
1
|
|
|
1
|
|
Mucinous carcinoma
|
0.346(0.211,0.567)
|
0.000
|
|
0.390(0.227,0.669)
|
0.001
|
Endometrioid carcinoma
|
11.555(1.526,87.466)
|
0.018
|
|
7.946(0.977,64.622)
|
0.053
|
Others
|
2.807(1.327,5.939)
|
0.007
|
|
2.400(1.1041,5.534)
|
0.040
|
Differentiation
|
|
|
|
|
|
Grade I
|
1
|
|
|
1
|
|
Grade II
|
6.059(3.129,11.733)
|
0.000
|
|
2.423(1.098,5.345)
|
0.028
|
Grade III
|
6.658(2.954,15.004)
|
0.000
|
|
1.982(0.777,5.056)
|
0.152
|
Grade IV
|
1.894(1.209,2.967)
|
0.005
|
|
1.594(0.976,2.605)
|
0.063
|
Laterality
|
|
|
|
|
|
Left
|
1
|
|
|
1
|
|
Right
|
1.081(0.550,2.139)
|
0.815
|
|
1.067(0.498,2.284)
|
0.868
|
Unspecial
|
1.117(0.564,2.214)
|
0.751
|
|
1.148(0.535,2.463)
|
0.724
|
Tumor size (cm)
|
1.001(0.999,1.004)
|
0.292
|
|
0.999(0.996,1.003)
|
0.636
|
Preoperative serum CA125 level
|
|
|
|
|
|
Negative or Not documented
|
1
|
|
|
1
|
|
Positive
|
2.749(1.753,4.309)
|
0.000
|
|
2.236(1.373,3.641)
|
0.001
|
Using the multivariate logistic regression, we analyzed the association between age, insurance, race, marital status, histology type, histology grade, laterality, tumor size, preoperative serum CA125 level and LNM (Table 3). We found that age(≥70, OR=0.544, p=0.022), histology type (Mucinous carcinoma, OR=0.390, p=0.001; Endometrioid carcinoma, OR=7.946, p=0.053; Others, OR=2.400, p=0.040), histology grade (Grade II, OR=2.423, p=0.028; Grade III, OR=1.982, p=0.152; Grade IV, OR=1.594, p=0.063) and preoperative serum CA125 level (positive, OR=2.236, p=0.001) were all significant predictors of LNM (Table2). In addition, we found the tolerance was >0.1 and VIF was <10 for the predictors, suggesting no collinearity among these independent variables (Supplement Table 1). Based on the above risk factors, we established the nomogram for predicting LNM.
Table 3
Performance of the nomogram in predicting lymph node metastasis in the training cohorts
Performance
parameter
|
AUC
|
Accuracy
|
Specificity
|
Sensitivity
|
PLR
|
NLR
|
DOR
|
Nomogram
|
0.77
|
0.72
|
0.73
|
0.72
|
2.69
|
0.38
|
7.02
|
AUC: Area Under the Curve |
PLR: positive likelihood ratio |
NLR: negative likelihood ratio |
DOR: Diagnostic Odds Ratio |
Nomogram Validation
Internal Validation
The AUC of the model training cohort and validation cohort were 0.78 (figure3A) and 0.79 (figure3B) respectively, which indicated favorable discrimination. The calibration curves showed that the predicted outcome fitted well to the observed outcome in the training cohort (p=0.825, figure3D) and validation cohort (p=0.503, figure3E). The decision curves showed the nomogram had more benefits than the All or None scheme if the threshold probability is >50% and <100% in training cohort and validation cohort (figure3G, H). The AUC, accuracy, specificity, sensitivity, PLR, NLR, DOR were 0.77, 0.72, 0.73, 0.72, 2.69, 0.38, 7.02 respectively (Table3).
External Validation
A total of 104 OC patients in the Department of Gynecology, General Hospital of Northern Theatre Command were collected. In the age, the proportion of ≤40, 40-70 and ≥70 was 11.5%, 49.0% and 39.4% respectively. Most of the patients were white (72.1%), the insured was 62.5%, the married was 70.2%, the positive serum CA125 was 58.7%, the positive LNM was 78.8%. In the differentiation, the proportion of grade I, grade II, grade III and grade IV was 33.7%, 18.3%, 37.5%, and 10.6% respectively. In the histology type, the proportion of serous adenocarcinoma, mucinous carcinoma, endometrioid carcinoma and others was 26.0%, 17.3%, 28.8%, and 27.9% respectively. The mean tumor size was 120.67±74.15 mm. Characteristics of patients are shown in supplementary Table 2. The AUC of the external validation cohort were 0.76 (figure3C), which indicated favorable discrimination. The calibration curves showed that the predicted outcome fitted well to the observed outcome in the external validation cohort (p=0.108, figure3F). The decision curves showed the nomogram had more benefits than the All or None scheme if the threshold probability is >30% and <90% in the external validation cohort (figure3I).