Figure 1 outlines the case selection method, and Table 1 summarizes the characteristics of the patient. Of the 6106 EOC cases identified in the SEER database, 1700 (27.8%) women retained lymph nodes, while 4406 (72.2%) underwent lymphadenectomy. About 59.3 percent received adjuvant chemotherapy. Throughout the cohort, all patients were female and underwent radical surgery. About 83.3 percent of them were white and 16.7 percent were non-white. The median age of patients was 50 years, of which 66.1% of patients were over 50 years of age, and 2899 (65.8%) of patients received lymphadenectomy. The median tumor size is 10.1 cm (range: 0.1-45 cm). In this cohort, radiotherapy was rarely used (2.7%). Chemotherapy (P=0.141) has no significant association with lymphadenectomy.
Use the IT-AIC method to estimate the effects of lymphadenectomy in a multivariate environment and identify other prognostic factors that may be helpful in selecting patients with lymphadenectomy (Table 2). The top model includes 10 factors, with a 29% chance of being the best approximation model considered. To improve the expected prediction accuracy while maintaining a low overfit, we considered the "confidence sets" of 4 models that, in total, are 97% likely to contain the best model. These models suggest that the following factors are meaningful for predicting survival: (i) grading, (ii) lymphadenectomy, (iii) staging, (iv) marital status, (v) age, (vi) chemotherapy, (vii) ethnicity, (viii) surgery, (ix) tumor size. Lymphadenectomy is certainly a factor in the best model. In any case, once other factors are taken into account, it seems that radiation therapy does not provide any meaningful predictive accuracy (Table 2).
Data on age, marital status, TNM stage, grade, tumor size, surgery, chemotherapy and lymphadenectomy were included in the univariate Cox regression analysis (Table 5). The results showed that age (HR1.52, p<0.001), marital status (HR1.04, 1.51, p<0.001), TNM staging (HR7.26, p<0.001), grade (HR3.96, p<0.001), surgery (HR0.33, p<0.001), chemotherapy (HR1.67, p<0.001), and lymphadenectomy (HR2.21, p<0.001) had a significant correlation with CSS in EOC (Table 5).
In multivariate analysis, lymphadenectomy as a classification and continuous variable remained an independent prognostic factor, and patients with lymphadenectomy had a better prognosis. In addition, age, stage, grade, marriage, chemotherapy, and surgical modality at the time of diagnosis were also found to be important prognostic factors, with a better prognosis < age of 50, chemotherapy, early staging and grading, no debulking surgery, and marriage in EOC women (Figure 8).
The data from the Cox regression model is used to build a column line plot. Each variable contained in a line plot is assigned a value that is related to how much it affects the outcome variables in the model. Each predictor is scored against a set scale. The total sum score (in points) on the line chart is then converted to the probability of survival of 1 year, 3 years, and 5 years. The line plot shows that AJCC staging is the most important factor influencing prognosis, followed by age, lymphadenectomy, marriage, surgical modality, and tumor size. (Figure 5). For a line plot, the AUC of the queue (0.787 for 5 years) indicates that the model has better discriminant power than staging (Figure 6). The calibration plot of the line plot shows that the predicted 5-year survival probability of the cohort is almost identical to the actual observations (Figure 7).
In the overall EOC survival analysis, patients who underwent lymphadenectomy had better survival regardless of whether they were chemotherapy or not, whether they were married or not, regardless of age or tumor size (Table 2; Figure 2b–h).
3.1 Stage I EOC
We analyzed data from stage I EOC cases of lymphadenectomy from 2004 to 2015 in the SEER database. A total of 2738 women (76.3%) received lymphadenectomy (9.1% in patients with 1-3 lymph nodes removed, 67.2% in patients with more than 4 lymph nodes removed, 23.7% in patients who chose conservative treatment of lymph nodes, p < 0.001). There was no difference in lymphadenectomy rates based on year of diagnosis (p = 0.06), tumor grade (p = 0.06), tumor size (p = 0.06), radiation therapy (p = 0.06), and surgical modality (p = 0.06). However, based on available information, women receiving lymphadenectomy were younger (median age, 53 years with more than four lymph nodes removed, 56 years without or removed 1-3 lymph nodes, p < 0.001), and more Caucasians who chose to have more than 4 lymph nodes removed (68.27% Caucasian, 46.51% black, p < 0.001), married women (married 69.31%, single 68.46%, p = 0.001), and patients undergoing chemotherapy (68.99% chemotherapy, 65.68% non-chemotherapy, p = 0.016) more (Table 3). In the survival analysis of patients with stage I EOC, patients who underwent more than 4 lymphadenectomy had better survival regardless of chemotherapy, marriage, age and tumor size, regardless of age and tumor size (all P < 0.001; Figure 3a-f).
3.2 Stage III-IV EOC
We analyzed data from stage III-IV EOC cases of lymphadenectomy from 2004 to 2015 in the SEER database. A total of 786 (59.0%) women underwent lymphadenectomy (11.6% of patients with 1-3 lymph nodes removed, 47.4% of patients with more than 4 lymph nodes removed, 41.0% of those who chose conservative lymph node therapy, p < 0.001). There was no difference in lymphadenectomy rates based on diagnostic year (p = 0.208), tumor grade (p = 0.175), tumor size (p = 0.3), chemotherapy (p = 0.063), and surgical modality (p = 0.317). According to available information, women who received lymphadenectomy were younger (median age, 56 years with more than four lymph nodes removed, 61 years with 1-3 lymph nodes removed, 59 years without resection of lymph nodes, p < 0.001), and whites who chose to have more than 4 lymph nodes removed (47.85% white, 30.95% black, p < 0.001), married women (married 53.73%, single 46.60%, p = 0.001), and patients undergoing chemotherapy (chemotherapy 48.22%, non-chemotherapy 43.94%, p = 0.016) more (Table 4). In the survival analysis of patients with stage III-IV EOC, patients who underwent more than 4 lymphadenectomy had better survival regardless of chemotherapy, marriage, age and tumor size (all P < 0.001; Figure 4a-f).
Table 1 Characteristics of patients diagnosed.
Characteristics
|
n
|
Lymph(%)
|
Lymph conservation(%)
|
p
|
Age(years)
|
|
|
|
|
≤50
|
2069
|
1507 (72.84%)
|
562 (27.16%)
|
|
50-60
|
1923
|
1457 (75.77%)
|
446 (23.19%)
|
<0.001
|
60-70
|
1227
|
892 (72.70%)
|
335 (27.30%)
|
|
>70
|
887
|
550 (62.01%)
|
337 (37.99%)
|
|
Race
|
|
|
|
White
|
5090
|
3711 (72.91%)
|
1379 (27.09%)
|
<0.001
|
Black
|
310
|
179 (57.74%)
|
131 (42.26%)
|
|
Others
|
706
|
516 (73.09%)
|
190 (26.91%)
|
|
Marital status
|
|
|
|
Single
|
1382
|
981 (70.98%)
|
401 (29.02%)
|
<0.001
|
Married
|
3261
|
2449 (75.10%)
|
812 (24.90%)
|
|
Others
|
1463
|
976 (66.71%)
|
487 (33.29%)
|
|
Year of diagnosis
|
|
|
|
2004-2007
|
2058
|
1423 (69.14%)
|
635 (30.86%)
|
0.001
|
2008-2011
|
2009
|
1473 (73.32%)
|
536 (26.68%)
|
|
2012-2015
|
2039
|
1510 (74.06%)
|
529 (25.94%)
|
|
TNM stage
|
|
|
|
|
I
|
3659
|
2810(76.80%)
|
849 (23.20%)
|
II
|
977
|
741 (75.84%)
|
236 (24.16%)
|
<0.001
|
III
|
1049
|
676 (64.44%)
|
373 (35.56%)
|
|
IV
|
318
|
145 (45.60%)
|
173 (54.40%)
|
|
Unknown
|
103
|
34 (33.01%)
|
69 (66.99%)
|
|
Tumor stage
|
|
|
|
|
T1
|
3820
|
2923 (76.52%)
|
897 (23.48%)
|
|
T2
|
1127
|
851 (75.51%)
|
276 (24.49%)
|
<0.001
|
T3
|
1140
|
626 (54.91%)
|
514 (45.09%)
|
|
Unknown
|
19
|
6 (31.58%)
|
13 (68.42%)
|
|
Nodal stage
|
|
|
|
N0
|
5521
|
4001 (72.47%)
|
1520 (27.53%)
|
<0.001
|
N1
|
446
|
399 (89.46%)
|
47 (10.54%)
|
|
Unknown
|
139
|
6 (4.32%)
|
133 (95.68%)
|
|
Grade
|
|
|
|
|
Well differentiated
|
1774
|
1337(75.37%)
|
437 (24.63%)
|
Moderately differentiated
|
2088
|
1542 (73.85%)
|
546 (26.15%)
|
<0.001
|
Poorly differentiated
|
1339
|
903 (67.44%)
|
436 (32.56%)
|
|
Undifferentiated or anapla
|
222
|
144 (64.86%)
|
78 (35.14%)
|
|
Unknown
|
683
|
480 (70.28%)
|
203 (29.72%)
|
|
Tumor Size (cm)
|
|
|
|
|
≤10
|
2516
|
1797 (71.42%)
|
719 (28.58%)
|
|
10.1-20
|
2119
|
1591 (75.08%)
|
528 (24.92%)
|
|
20.1-30
|
373
|
273 (73.19%)
|
100 (26.81%)
|
<0.001
|
30.1-40
|
15
|
14 (93.33%)
|
1 (6.67%)
|
|
>40
|
8
|
5 (62.50%)
|
3 (37.50%)
|
|
Unknown
|
1075
|
726 (67.53%)
|
349 (32.47%)
|
|
Radiotherapy
|
|
|
|
No
|
5919
|
4250 (71.80%)
|
1669 (28.20%)
|
0.002
|
Unknown
|
25
|
20 (80.00%)
|
5 (20.00%)
|
|
Yes
|
162
|
136 (83.95%)
|
26 (16.05%)
|
|
Chemotherapy
|
|
|
|
|
No
|
2486
|
1768 (71.12%)
|
718 (28.88%)
|
0.141
|
Yes
|
3620
|
2638 (72.87%)
|
982 (27.13%)
|
|
Sugery
Oophorectomy
Debulking
|
4755
1351
|
3479 (73.17%)
927 (68.62%)
|
1276 (26.83%)
424 (31.38%)
|
0.001
|
Table 2. Set of models created with forward-stepwise selection, ranked by corrected AIC.
Shaded boxes signify the factors included within the model. Models with darker shading represent the confidence set, which is > 95% likely to contain the factors of the best-approximating model (based on ∑Wt). df is the number of parameters. logLik is the log-likelihood. ∆AICc is the difference in corrected AIC compared to the top-ranked model (values < 2 indicate informational equivalence). AICcWt represents the proportional AICc weight of the model in the total set of models (values approximate the likelihood that a given model is the best of those in the set). ∑Wt is the cumulative sum of AICc weights. Grd: Cancer Grade(I/I/III/V/Unknown); Ly: Lymphadenectomy(Lymphadenectomy/ Lymph nodes conservation);Tnm: Cancer Stage(I-II/III-IV/Unknown);Mrg: Marital Status(Married/Single/Others); Age: Age by class (≤50/>50); Chm: Chemotherapy (Yes/No); t: tumor T-Stage(I/I/III/V/Unknown);n: Tumor N-Stage(N0/N1/Unknown);Rac: Race (White/Black/Others; Sur:Surgery(oophorectomy/ Debulking);Tum: Tumor Size(≤10/10.1-20/20.1-30/30.1-40/>40/Unknown);Rad: Radiotherapy(Yes/No/Unknown);Yea: Year of diagnosis(2004-2007/2008-2011/2012-2015).
Table 3 Characteristics of patients diagnosed with stage I endometrioid carcinoma of ovary in 2004–2015
|
n
|
1 to 3 (%)
|
4 or more (%)
|
None(%)
|
p
|
Age (years)
|
|
|
|
|
<0.001
|
0-50
|
1309
|
104(7.94%)
|
881(67.30%)
|
324(24.75%)
|
50-60
|
1168
|
109(9.33%)
|
831(71.15%)
|
228(19.52%)
|
60-70
|
700
|
66(9.43%)
|
471(67.29%)
|
163(23.29%)
|
>70
|
410
|
47(11.46%)
|
229(55.85%)
|
134(32.68%)
|
Race
|
|
|
|
|
<0.001
|
White
Black
|
3015
172
|
268(8.89%)
24(13.95%)
|
2058(68.27%)
80(46.51%)
|
689(22.85%)
68(39.53%)
|
|
Others
|
400
|
34(8.50%)
|
274(68.50%)
|
92(23.00%)
|
|
Marital status
|
|
|
|
0.001
|
Married
|
1929
|
161(8.35%)
|
1337(69.31%)
|
431(22.34%)
|
Others
|
840
|
99(11.79%)
|
515(61.31%)
|
226(26.90%)
|
Single
|
818
|
66(8.07%)
|
560(68.46%)
|
192(23.47%)
|
Year of diagnosis
|
|
|
|
0.06
|
2004-2007
|
1124
|
114(10.14%)
|
718(63.88%)
|
292(25.98%)
|
2008-2011
|
1193
|
98(8.21%)
|
820(68.73%)
|
275(23.05%)
|
2012-2015
|
1270
|
114(8.98%)
|
874(68.82%)
|
282(22.20%)
|
Grade
|
|
|
|
|
0.568
|
Well differentiated
|
1399
|
116(8.29%)
|
936(66.90%)
|
347(24.80%)
|
Moderately differentiated
|
1289
|
125(9.70%)
|
883(68.50%)
|
281(21.80%)
|
Poorly differentiated
|
424
|
44(10.38%)
|
273(64.39%)
|
107(25.24%)
|
Undifferentiated
|
71
|
6(8.45%)
|
46(64.79%)
|
19(26.76%)
|
|
Unknown
|
404
|
35(8.66%)
|
274(67.82%)
|
95(23.51%)
|
|
Tumor Size (cm)
|
|
|
|
0.288
|
0-10
|
1565
|
136(8.69%)
|
1040(66.45%)
|
389(24.86%)
|
10.1-20
|
1207
|
115(9.53%)
|
833(69.01%)
|
259(21.46%)
|
20.1-30
|
232
|
29(12.50%)
|
151(65.09%)
|
52(22.41%)
|
|
30.1-40
|
10
|
1(10.00%)
|
8(80.00%)
|
1(10.00%)
|
|
40+
|
5
|
0(0.00%)
|
3(60.00%)
|
2(40.00%)
|
|
Unknown
|
568
|
45(7.92%)
|
377(66.37%)
|
146(25.70%)
|
Radiotherapy
|
|
|
|
0.197
|
No
|
3512
|
319(9.08%)
|
2353(67.00%)
|
840(23.92%)
|
Unknown
|
14
|
1(7.14%)
|
11(78.57%)
|
2(14.29%)
|
|
Yes
|
61
|
6(9.84%)
|
48(78.69%)
|
7(11.48%)
|
|
Chemotherapy
Yes
No
|
1696
1891
|
161(9.49%)
165(8.73%)
|
1170(68.99%)
1242(65.68%)
|
365(21.52%)
484(25.59%)
|
0.016
|
Treatment
Oophorectomy
Debulking
|
3211
376
|
292(9.09%)
34(9.04%)
|
2146(66.83%)
266(70.74%)
|
773(24.07%)
76(20.21%)
|
0.236
|
Table 4 Characteristics of patients diagnosed with stage III-IV endometrioid carcinoma of ovary in 2004–2015
|
n
|
1 to 3 (%)
|
4 or more (%)
|
None(%)
|
p
|
Age (years)
|
|
|
|
0.001
|
0-50
|
386
|
44(11.40%)
|
197(51.04%)
|
145(37.56%)
|
50-60
|
372
|
30(8.06%)
|
196(52.69%)
|
146(39.25%)
|
60-70
|
294
|
39(13.27%)
|
136(46.26%)
|
119(40.48%)
|
>70
|
280
|
42(15.00%)
|
102(36.43%)
|
136(48.57%)
|
Race
|
|
|
|
|
0.015
|
Black
|
84
|
15(17.86%)
|
26(30.95%)
|
43(51.19%)
|
|
Others
|
153
|
13(8.50%)
|
81(52.94%)
|
59(38.56%)
|
|
White
|
1095
|
127(11.60%)
|
524(47.85%)
|
444(40.55%)
|
Marital.status
|
|
|
|
<0.001
|
Married
|
683
|
79(11.57%)
|
367(53.73%)
|
237(34.70%)
|
Others
|
340
|
48(14.12%)
|
120(35.29%)
|
172(50.59%)
|
Single
|
309
|
28(9.06%)
|
144(46.60%)
|
137(44.34%)
|
Year of diagnosis
|
|
|
|
0.208
|
2004-2007
|
531
|
61(11.49%)
|
235(44.26%)
|
235(44.26%)
|
2008-2011
|
426
|
56(13.15%)
|
207(48.59%)
|
163(38.26%)
|
2012-2015
|
375
|
38(10.13%)
|
189(50.40%)
|
148(39.47%)
|
Grade
|
|
|
|
|
0.175
|
Well differentiated
|
114
|
11(9.65%)
|
69(60.53%)
|
34(29.82%)
|
|
Moderately differentiated
|
373
|
43(11.53%)
|
180(48.26%)
|
150(40.21%)
|
Poorly differentiated
|
607
|
74(12.19%)
|
280(46.13%)
|
253(41.68%)
|
Undifferentiated
|
109
|
14(12.84%)
|
47(43.12%)
|
48(44.04%)
|
|
Unknown
|
129
|
13(10.08%)
|
55(42.64%)
|
61(47.29%)
|
|
Tumor Size (cm)
|
|
|
|
0.3
|
0-10
|
496
|
63(12.70%)
|
228(45.97%)
|
205(41.33%)
|
10.1-20
|
461
|
54(11.71%)
|
235(50.98%)
|
172(37.31%)
|
20.1-30
|
83
|
11(13.25%)
|
37(44.58%)
|
35(42.17%)
|
|
30.1-40
|
3
|
0(0.00%)
|
3(100.00%)
|
0(0.00%)
|
|
40+
|
3
|
0(0.00%)
|
2(66.67%)
|
1(33.33%)
|
|
Unknown
|
286
|
27(9.44%)
|
126(44.06%)
|
133(46.50%)
|
Radiotherapy
|
|
|
|
0.021
|
No
|
1277
|
146(11.43%)
|
597(46.75%)
|
534(41.82%)
|
Unknown
|
7
|
0(0.00%)
|
4(57.14%)
|
3(42.86%)
|
|
Yes
|
48
|
9(18.75%)
|
30(62.50%)
|
9(18.75%)
|
|
Chemotherapy
|
|
|
|
0.063
|
Yes
|
1068
|
131(12.27%)
|
515(48.22%)
|
422(39.51%)
|
No
|
264
|
24(9.09%)
|
116(43.94%)
|
124(46.97%)
|
treatment
|
|
|
|
0.317
|
oophorectomy
|
660
|
69(10.45%)
|
323(48.94%)
|
268(40.61%)
|
Debulking
|
672
|
86(12.80%)
|
308(45.83%)
|
278(41.37%)
|
Table 5 Univariate analysis of endometrioid carcinoma of ovary cancer-specific survival and overall survival.
Characteristics
|
HR
|
CSS
95% CI
|
p
|
HR
|
OS
95% CI
|
p
|
Age(years)
|
|
|
|
|
|
|
≤50
|
1
|
|
|
1
|
|
|
>50
|
1.52
|
1.33-1.72
|
<0.001
|
1.56
|
1.40-1.73
|
<0.001
|
Marital status
|
|
|
|
|
|
|
Married
|
1
|
|
|
1
|
|
|
Single
|
1.04
|
0.90-1.20
|
0.651
|
1.03
|
0.92-1.17
|
0.655
|
Others
|
1.51
|
1.32-1.72
|
<0.001
|
1.65
|
1.48-1.84
|
<0.001
|
TNM stage
|
|
|
|
|
|
|
I-II
|
1
|
|
|
1
|
|
|
III-IV
|
7.26
|
6.47-8.15
|
<0.001
|
4.64
|
4.22-5.10
|
<0.001
|
Grade
|
|
|
|
|
|
|
I-II
|
1
|
|
|
1
|
|
|
III-IV
|
3.96
|
3.53-4.44
|
<0.001
|
2.73
|
2.49-3.00
|
<0.001
|
Tumor Size (cm)
|
|
|
|
|
|
|
≤10
|
1
|
|
|
1
|
|
|
>10
|
1.05
|
0.94-1.17
|
0.476
|
1.09
|
0.99-1.19
|
0.15
|
Surgery
|
|
|
|
|
|
|
Debulking
|
1
|
|
|
1
|
|
|
Oophorectomy
|
0.33
|
0.29-0.36
|
<0.001
|
0.43
|
0.39-0.47
|
<0.001
|
Chemotherapy
|
|
|
|
|
|
|
No
|
1
|
|
|
1
|
|
|
Yes
|
1.67
|
1.47-1.90
|
<0.001
|
1.19
|
1.08-1.32
|
0.003
|
Lymphadenectomy
|
|
|
|
|
|
|
Yes
|
1
|
|
|
1
|
|
|
No
|
2.21
|
1.97-2.48
|
<0.001
|
2.1
|
1.91-2.31
|
<0.001
|