The impact of lymph node dissection on stage I ovarian endometrioid carcinoma: A US-China Comparative Analysis.

M.S., F.L., J.S., X.-G.W., L.-S.F., J.Y., and W.W. conceived and crafted the project. M.S., F.L., and J.S. were involved in designing the study, collecting and analyzing data, drafting the manuscript, and critically revising it for substantial intellectual input. Y.L. and J.C. assisted in data collection. Q.G., S.Y., J.F., Y.C., D.W., and Y.H. were responsible for data quality control and management. W.W., J.Y., and F.L. supervised the project and played key roles in manuscript composition and revision. All authors have reviewed and agreed upon the final revision of the manuscript.

A total of 2,717 patients from the SEER database who met the inclusion criteria were analyzed ( Figure 1 ), with LND performed in 2,105 cases (77.5%) and not performed in 612 cases (22.5%). Patients who underwent LND exhibited a younger median age (54.96 vs. 58.66 years; p  < 0.001) compared to those who did not. Additionally, a higher proportion of tumors in the LND group exceeded the median diameter of 105 mm (45.7% vs. 36.4%; p  = 0.001), while there was a lower incidence of FIGO stage IA (56.2% vs. 62.6%), and a higher incidence of stage IB (4.3% vs. 3.1%) and stage IC (39.5% vs. 34.3%) (p for all subgroups <0.05). Furthermore, a higher proportion of patients in the LND group received chemotherapy (51.1% vs. 44.1%, p  < 0.05), and there was a variation in the proportions of different surgical procedures between the two groups ( p  < 0.001) ( Table 1 ). Figure 1 Flowchart of patient selection Table 1 Patient characteristics of SEER database Characteristic Total (N = 2717) LND (N = 2105) Non-LND (N = 612) P Age, years Mean ± SD 55.8 ± 13.17 54.96 ± 12.47 58.66 ± 15.01 <0.001 Race White 2268(83.5%) 1772(84.2%) 496(81.0%) 0.213 Black 131(4.8%) 84(4.0%) 47(7.7%) Others 299(11.0%) 235(11.2%) 64(10.5%) Unknown 19(0.7%) 14(0.7%) 5(0.8%) Marital status Married 1518(55.9%) 1190(56.5%) 328(53.6%) 0.198 Un-Married 1199(44.1%) 915(43.5%) 284(46.4%) Tumor size <105mm 1172(43.1%) 887(42.1%) 285(46.6%) 0.001 ≥105mm 1184(43.6%) 961(45.7%) 223(36.4%) Unknown 361(13.3%) 257(12.2%) 104(17.0%) Grade 1 1044(38.4%) 796(37.8%) 248(40.5%) 0.061 2 967(35.6%) 778(37.0%) 189(30.9%) 3 415(15.3%) 319(15.2%) 96(15.7%) Unknown 291(10.7%) 212(10.1%) 79(12.9%) Lateral Unilateral 2517(92.6%) 1944(92.4%) 573(93.6%) 0.287 Bilateral 200(7.4%) 161(7.6%) 39(6.4%) CA125 Normal 463(17.0%) 358(17.0%) 105(17.2%) 0.114 Elevated 1578(58.1%) 1273(60.5%) 305(49.8%) Unknown 676(24.9%) 474(22.5%) 202(33.0%) Stage IA 1565(57.6%) 1182(56.2%) 383(62.6%) 0.014 IB 110(4.0%) 91(4.3%) 19(3.1%) IC 1042(38.4%) 832(39.5%) 210(34.3%) Chemotherapy Yes 1346(49.5%) 1076(51.1%) 270(44.1%) 0.002 No/Unknown 1371(50.5%) 1029(48.9%) 342(55.9%) Surgery USO/BSO-Hys 147(5.4%) 81(3.8%) 66(10.8%) <0.001 USO/BSO+Hys 574(21.1%) 382(18.1%) 192(31.4%) USO/BSO+ome-Hys 216(7.9%) 178(8.5%) 38(6.2%) USO/BSO+ome+Hys 1179(43.4%) 1026(48.7%) 153(25.0%) Debulking 320(11.8%) 260(12.4%) 60(9.8%) Others 281(10.3%) 178(8.5%) 103(16.8%) All the participants are female. “LND” is lymph node dissection, and “Others” in group of races includes American Indian/Alaska Native, Asian/Pacific Islander. “USO”, “BSO”, “Hys” and “Ome” in the group of surgery respectively denote unilateral oophoro-salpingectomy, bilateral oophoro-salpingectomy, hysterectomy and omentectomy. Flowchart of patient selection Patient characteristics of SEER database All the participants are female. “LND” is lymph node dissection, and “Others” in group of races includes American Indian/Alaska Native, Asian/Pacific Islander. “USO”, “BSO”, “Hys” and “Ome” in the group of surgery respectively denote unilateral oophoro-salpingectomy, bilateral oophoro-salpingectomy, hysterectomy and omentectomy. In the NUWA platform, 116 eligible patients were screened ( Figure 1 ), including 88 (75.9%) undergoing LND and 28 (24.1%) not. Similar to the SEER database, patients who underwent LND in the NUWA database were of a younger age (median age 43.37 vs. 45.89; p  < 0.001). Among patients who received LND compared to those who did not, the proportion of tumors smaller than the median diameter of 110 mm was slightly lower (38.6% vs. 46.4%; p  = 0.556), and the percentage of patients receiving chemotherapy was higher, though not statistically significant (83.0% vs. 71.4%, p  = 0.183). Contrary to the SEER data, the proportion of FIGO stage IA patients with LND was higher than those without LND in the NUWA platform (31.8% vs. 25.0%), while the proportion of stage IC patients was lower (64.8% vs. 75.0%), with no significant difference observed ( p  = 0.623 for all subgroups). The distribution of various surgical procedures showed no statistical differences ( p  = 0.506 for all subgroups) ( Table 2 ). Table 2 Patient characteristics of NUWA database Characteristic Total (N = 116) LND (N = 88) Non-LND (N = 28) P Age, years Mean ± SD 44 ± 10.05 43.37 ± 9.99 45.89 ± 10.16 <0.001 Marital status Married 104(89.7%) 77(87.5%) 27(96.4%) 0.177 Un-Married 12(10.3%) 11(12.5%) 1(3.6%) Tumor size <110mm 47(40.5%) 34(38.6%) 13(46.4%) 0.556 ≥110mm 49(42.2%) 38(43.2%) 11(39.3%) Unknown 20(17.2%) 16(18.2%) 4(14.3%) Grade 1 34(29.3%) 25(28.4%) 9(32.1%) 0.605 2 40(34.5%) 32(36.4%) 8(28.6%) 3 19(16.4%) 13(14.8%) 6(21.4%) Unknown 23(19.8%) 18(20.5%) 5(17.9%) Lateral Unilateral 92(79.3%) 72(81.8%) 20(71.4%) 0.772 Bilateral 16(13.8%) 12(13.6%) 4(14.3%) Unknown 8(6.9%) 4(4.5%) 4(14.3%) CA125 Normal 56(48.3%) 42(47.7%) 14(50.0%) 0.206 Elevated 12(10.3%) 11(12.5%) 1(3.6%) Unknown 48(41.4%) 35(39.8%) 13(46.4%) Stage IA 35(30.2%) 28(31.8%) 7(25.0%) 0.623 IB 1(0.9%) 1(1.1%) 0(0%) IC 78(67.2%) 57(64.8%) 21(75.0%) Unknown 2(1.7%) 2(2.3%) 0(0%) Chemotherapy Yes 93(80.2%) 73(83.0%) 20(71.4%) 0.183 No/Unknown 23(19.8%) 15(17.0%) 8(28.6%) Surgery USO/BSO-Hys – – – 0.506 USO/BSO+Hys 4(3.4%) 3(3.4%) 1(3.6%) USO/BSO+ome-Hys 13(11.2%) 12(13.6%) 1(3.6%) USO/BSO+ome+Hys 68(58.6%) 51(58.0%) 17(60.7%) Debulking – – – Others 31(26.7%) 22(25.0%) 9(32.1%) All the participants are female and of Asian ethnicity. “LND” is lymph node dissection. And “USO”, “BSO”, “Hys”, and “Ome” in the group of surgery, respectively, denote unilateral oophoro-salpingectomy, bilateral oophoro-salpingectomy, hysterectomy, and omentectomy. Patient characteristics of NUWA database All the participants are female and of Asian ethnicity. “LND” is lymph node dissection. And “USO”, “BSO”, “Hys”, and “Ome” in the group of surgery, respectively, denote unilateral oophoro-salpingectomy, bilateral oophoro-salpingectomy, hysterectomy, and omentectomy. The 5-year survival rate for the entire cohort of 2,717 individuals selected from the SEER database was 91.2%. Specifically, the 5-year survival rate for patients who underwent LND was 93.4%, compared to 84.2% for those who did not. The Kaplan-Meier analysis exhibited a higher survival probability for patients who underwent LND ( Figure 2 A). However, due to the limited sample size, no statistically significant difference was observed in the survival curves between the LND and non-LND groups in the NUWA database ( Figure 2 B). Figure 2 Survival analysis of stage I OEC for lymph nodes dissection (LND) or no lymph nodes dissection (non-LND) (A) SEER database. (B) NUWA platform. Survival analysis of stage I OEC for lymph nodes dissection (LND) or no lymph nodes dissection (non-LND) (A) SEER database. (B) NUWA platform. Tables 3 and 4 show the impact of LND on the OS in stage I OEC patients. The protective effects of LND remained robust among multiple strategies for covariate adjustment in the Cox regression analysis. In the null model, the HR of LND on the OS of stage I OEC patients in SEER data were 0.461 (95% confidence interval [CI]: 0.369–0.578, p  < 0.001). For the model with full adjustment for covariates, the corresponding HR was 0.561 (95% CI: 0.440–0.715, p  < 0.001). Similarly, a protective effect of LND was also evident in the NUWA datasets, although the estimated differences were not statistically significant due to the limited sample size. Table 3 Cox proportional hazards model of LND in SEER database Hazard ratio 95%CI P Model 1 0.461 0.369–0.578 <0.001 Model 2 0.602 0.478–0.758 <0.001 Model 3 0.572 0.452–0.723 <0.001 Model 4 0.561 0.440–0.715 <0.001 Model 1, LND. Model 2, LND+Age+Marital status. Model 3, Model 2+Tumor size+Grade+Lateral+CA125+Stage. Model 4, Model 3+Chemotherapy+Surgery. Table 4 Cox proportional hazards model of LND in NUWA database Hazard Ratio 95%CI P Model 1 0.637 0.264–1.540 0.317 Model 2 0.631 0.253–1.574 0.323 Model 3 0.587 0.180–1.920 0.379 Model 4 0.697 0.189–2.570 0.588 Model 1, LND. Model 2, LND+Age+Marital status. Model 3, Model 2+Tumor size+Grade+Lateral+CA125+Stage. Model 4, Model 3+Chemotherapy+Surgery. Cox proportional hazards model of LND in SEER database Model 1, LND. Model 2, LND+Age+Marital status. Model 3, Model 2+Tumor size+Grade+Lateral+CA125+Stage. Model 4, Model 3+Chemotherapy+Surgery. Cox proportional hazards model of LND in NUWA database Model 1, LND. Model 2, LND+Age+Marital status. Model 3, Model 2+Tumor size+Grade+Lateral+CA125+Stage. Model 4, Model 3+Chemotherapy+Surgery. To further demonstrate the potential effect modifiers influencing the impact of LND on OS in stage I OEC patients, stratification analyses were conducted based on various patient characteristics. Regardless of age, marital status, tumor size, histological grade, unilateral or bilateral involvement, presence of elevated carbohydrate antigen 125 (CA125), FIGO stage, chemotherapy status, or surgical procedure, the HRs for LND consistently remained below 1 and were statistically significant ( p  < 0.05) in the SEER database ( Table 5 ). Differences in HRs across different subgroups of each stratification did not reach statistical significance, suggesting that personal characteristics would not confound the protective effects of LND ( Table 5 ). Table 5 The hazard ratio of OS in subgroups in SEER database Characteristics LND ( N  = 2105) Non-LND ( N  = 612) Hazard Ratio 95%CI P P for difference Age, years <55 1038(80.5%) 250(19.5%) 0.546 0.351–0.850 0.007 ≥55 1067(74.6%) 362(25.4%) 0.478 0.367–0.622 <0.001 0.613 Marital status Married 1190(78.3%) 328(21.7%) 0.495 0.355–0.690 <0.001 Un-married 915(76.3%) 284(23.7%) 0.443 0.326–0.601 <0.001 0.63 Tumor size <105mm 887(75.6%) 285(24.4%) 0.463 0.321–0.668 <0.001 ≥105mm 961(81.1%) 223(18.9%) 0.483 0.342–0.683 <0.001 0.869 Grade 1 796(76.2%) 248(23.8%) 0.414 0.266–0.644 <0.001 2 778(80.4%) 189(19.6%) 0.423 0.292–0.613 <0.001 0.942 3 319(76.8%) 96(23.2%) 0.53 0.349–0.805 0.003 0.781 Lateral Unilateral 1944(77.2%) 573(22.8%) 0.457 0.362–0.579 <0.001 Bilateral 161(80.5%) 39(19.5%) 0.433 0.198–0.945 0.036 0.897 CA125 Normal 358(77.3%) 105(22.7%) 0.556 0.309–0.998 0.049 Elevated 1273(80.7%) 305(19.3%) 0.452 0.333–0.613 <0.001 0.539 Stage IA 1182(75.5%) 383(24.5%) 0.481 0.358–0.646 <0.001 IB 91(82.7%) 19(17.3%) 0.683 0.222–2.103 0.506 0.554 IC 832(79.8%) 210(20.2%) 0.387 0.268–0.558 <0.001 Chemotherapy Yes 1076(79.9%) 270(20.1%) 0.543 0.385–0.764 <0.001 No/Unknown 1029(75.1%) 342(24.9%) 0.403 0.298–0.545 <0.001 0.201 Surgery USO a /BSO b -Hys c 81(55.1%) 66(44.9%) 0.553 0.261–1.169 0.121 USO/BSO+Hys 382(66.6%) 192(33.4%) 0.5 0.318–0.786 0.003 0.822 USO/BSO+ome d -Hys 178(82.4%) 38(17.6%) 0.461 0.204–1.043 0.063 0.748 USO/BSO+ome+Hys 1026(87.0%) 153(13.0%) 0.377 0.255–0.557 <0.001 0.374 Debulking 260(81.3%) 60(18.7%) 0.657 0.306–1.411 0.281 0.752 Others 178(63.3%) 103(36.7%) 0.421 0.207–0.853 0.016 0.604 a USO, unilateral oophoro-salpingectomy. b BSO, bilateral oophoro-salpingectomy. c Hys, hysterectomy. d Ome, omentectomy. The hazard ratio of OS in subgroups in SEER database USO, unilateral oophoro-salpingectomy. BSO, bilateral oophoro-salpingectomy. Hys, hysterectomy. Ome, omentectomy. We similarly calculated the HRs of LND in different stratifications within the NUWA database. Due to a limited sample size, HR values could not be obtained for the unmarried, grade 3, bilateral, CA125 elevated, FIGO stage IB, and the surgical procedures unilateral/bilateral oophoro-salpingectomy+hysterectomy (USO/BSO+Hys), and unilateral/bilateral oophoro-salpingectomy+hysterectomy+omentectomy without hysterectomy (USO/BSO+ome-Hys) stratifications. And the subgroup comparisons were not available for the abovementioned groups. In other stratifications where HR values were calculated, despite the lack of statistical differences, the calculated HR values were consistently less than 1 ( Table 6 ), which was consistent with the results from the SEER database. Table 6 The hazard ratio of OS in subgroups in NUWA database Characteristics LND ( N  = 88) Non-LND ( N  = 28) Hazard Ratio 95%CI P Age, years <44 39(79.6%) 10(20.4%) 0.527 0.170–1.633 0.267 ≥44 49(73.1%) 18(26.9%) 0.914 0.196–4.261 0.909 Marital status Married 77(74.0%) 27(26.0%) 0.634 0.257–1.559 0.321 Un-married 11(91.7%) 1(8.33%) NA Tumor size <110mm 34(72.3%) 13(27.7%) 0.566 0.160–2.003 0.377 ≥110mm 38(77.6%) 11(22.4%) 0.66 0.081–5.374 0.698 Grade 1 25(73.5%) 9(26.5%) 0.803 0.315–2.046 0.646 2 32(80.0%) 8(20.0%) 0.489 0.114–2.105 0.337 3 16(84.2%) 3(15.8%) NA Lateral Unilateral 72(78.3%) 20(21.7%) 0.698 0.282–1.726 0.436 Bilateral 14(87.5%) 2(12.5%) NA CA125 Normal 42(75.0%) 14(25.0%) 0.288 0.090–0.922 0.036 Elevated 11(91.7%) 1(8.33%) NA Stage IA 28(80.0%) 7(20.0%) 0.993 0.191–5.157 0.994 IB 1(100%) 0(0%) NA IC 57(73.1%) 21(26.9%) 0.475 0.162–1.399 0.177 Chemotherapy Yes 73(78.5%) 20(21.5%) 0.885 0.291–2.689 0.83 No/Unknown 15(65.2%) 8(34.8%) 0.311 0.052–1.865 0.201 Surgery USO a /BSO b -Hys c – – – – – USO/BSO+Hys 3(75.0%) 1(25.0%) NA USO/BSO+ome d -Hys 12(92.3%) 1(7.7%) NA USO/BSO+ome+Hys 51(75.0%) 17(25.0%) 0.53 0.178–1.578 0.254 Debulking – – – – – Others 22(71.0%) 9(29.0%) 0.869 0.159–4.759 0.872 NA, not available. a USO, unilateral oophoro-salpingectomy. b BSO, bilateral oophoro-salpingectomy. c Hys, hysterectomy. d Ome, omentectomy. The hazard ratio of OS in subgroups in NUWA database NA, not available. USO, unilateral oophoro-salpingectomy. BSO, bilateral oophoro-salpingectomy. Hys, hysterectomy. Ome, omentectomy. Figure 3 presents the dose-response association between the number of LNs and survival rate of patients with stage I OEC in both the SEER and NUWA databases. The HR values exhibited a consistent decline as the number of LNs increased in both databases, reaching a plateau after 20 LNs, and increased after 40 LNs. Based on the detailed HR values associated with number of LNs, the most substantial protective effect of LND was observed between 21 and 34 LNs for the SEER data, and between 29 and 36 lymph nodes for the NUWA data ( Table S1 , Figure 3 ). The combined results of the SEER and NUWA cohorts indicated that the optimal range of the dissected LNs number was between 29 and 34. Figure 3 Frequency distribution and optimal number of lymph node dissection (A) A histogram of frequency distribution and RCS curve of HR values for varying dissected lymph nodes in the SEER cohort. (B) A histogram of frequency distribution and the RCS curve depicting HR values for varying numbers of dissected lymph nodes in the NUWA cohort. Frequency distribution and optimal number of lymph node dissection (A) A histogram of frequency distribution and RCS curve of HR values for varying dissected lymph nodes in the SEER cohort. (B) A histogram of frequency distribution and the RCS curve depicting HR values for varying numbers of dissected lymph nodes in the NUWA cohort.

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Wei Wang ( [email protected] ). This study is based on retrospective data analysis and did not generate new unique reagents. Consequently, no materials from this study are available for distribution. • Data: This study did not generate any new data. Data reported in this paper were downloaded from two specialized repositories: the Surveillance, Epidemiology, and End Results (SEER) Database and the NUWA platform. The SEER Database is dedicated to cancer research, encompassing incidence, mortality, and survival statistics. Meanwhile, the NUWA platform focuses on real-world patient management and research specifically in the field of ovarian cancer. The datasets from SEER are publicly available and can be accessed at the SEER website ( https://seer.cancer.gov/ ). Noteworthily, the specific datasets or studies may not have a DOI but are accessible through the SEER∗Stat software and the SEER Program website. For the NUWA datasets, we obtained them from the electronic medical records of multiple medical institutions in China and underwent deidentification, and used them under the agreement to not engage in the unauthorized distribution of the raw data to a third party and to use the data for scientific research only. • Code: The code will be shared by the lead contact upon available request. • Additional information: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Data: This study did not generate any new data. Data reported in this paper were downloaded from two specialized repositories: the Surveillance, Epidemiology, and End Results (SEER) Database and the NUWA platform. The SEER Database is dedicated to cancer research, encompassing incidence, mortality, and survival statistics. Meanwhile, the NUWA platform focuses on real-world patient management and research specifically in the field of ovarian cancer. The datasets from SEER are publicly available and can be accessed at the SEER website ( https://seer.cancer.gov/ ). Noteworthily, the specific datasets or studies may not have a DOI but are accessible through the SEER∗Stat software and the SEER Program website. For the NUWA datasets, we obtained them from the electronic medical records of multiple medical institutions in China and underwent deidentification, and used them under the agreement to not engage in the unauthorized distribution of the raw data to a third party and to use the data for scientific research only. Code: The code will be shared by the lead contact upon available request. Additional information: Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

LND plays a crucial role in the staging and treatment planning for various malignancies. Previous studies have suggested a positive impact of LND on patient survival, 26 , 27 , 28 , 29 prompting surgeons to strive for the removal of a significant number of LNs to ensure thorough elimination of potential cancer cells in the respective regions. However, recent research has challenged the notion that removing more LNs leads to better outcomes. No survival benefits were found in the LION trial, as higher complication and mortality rates were observed. While retroperitoneal staging doesn’t enhance survival rates, the management of isolated nodal recurrences may contribute to improved survival outcomes. 30 For instance, in early-stage cervical cancer, patients receiving comprehensive LND did not exhibit higher DFS and OS rates compared to those who underwent sentinel LN biopsy alone. 31 In recurrent NSCLC, the removal of more than 16 LNs was found to provide no additional benefits, and even adversely affected the efficacy of immunotherapy. 23 , 32 A study showed that in non-metastatic adrenocortical carcinoma, the degree of surgical intervention and LND did not correlate with survival rates. 33 Additionally, lymphedema in the lower limbs caused by LND in gynecologic malignancies has been associated with a lower quality of life for patients. 34 OEC generally exhibits a relatively favorable prognosis, with a 5-year OS exceeding 80%. 35 This led us to contemplate the crucial role of LND in improving survival outcome for patients with early-stage OEC. However, given the relatively low incidence of OEC, there remains ongoing debate regarding prognostic factors and surgical treatments for stage I OEC patients. 36 , 37 , 38 , 39 Consequently, we conducted an in-depth analysis to evaluate the impact of LND on the OS of stage I OEC patients. In this study, we analyzed the clinical characteristics of stage I OEC patients sourced from both the SEER database and the NUWA platform. Our comparative analysis of the SEER and NUWA cohorts showed notable differences in treatment patterns and demographic characteristics among patients. Within the NUWA cohort, 58.6% of patients underwent surgical procedures, including USO/BSO+ome+Hys, while 75.9% of the cohort underwent LND, a proportion closely aligning with the SEER cohort (77.5%). However, significant variations were observed in patient profiles: the average age was 56 years in the SEER cohort compared to 44 years in the NUWA cohort, and the proportion of married patients was markedly higher in the NUWA cohort (89.7%) than in the SEER cohort (55.9%). These discrepancies may reflect underlying differences in diagnostic criteria, treatment protocols, healthcare infrastructure, and cultural influences between the United States and China, which collectively contribute to above variations in clinical outcomes. Kaplan-Meier analysis in the SEER cohort revealed that patients who underwent LND exhibited superior OS compared to those without LND. Cox regression analyses, ranging from the null model to the fully adjusted model in the SEER cohort, identified LND as an independent protective factor for stage I OEC patients. This observation held true across patient with different characteristics including age, marital status, tumor size, histological grade, laterality, CA125 levels, FIGO stage, receipt of chemotherapy, and surgical procedure. However, although the analysis results of the NUWA data did not achieve statistical significance, they exhibited similar data trends. These results strongly suggest that LND can provide significant survival benefits for patients with stage I OEC. Variability in LND practices for early-stage OEC patient, ranging from isolated sampling to comprehensive pelvic LND, is shaped by differences in surgical expertise and resource availability, which in turn impacts the quality of systematic LND. The inconsistent findings from retrospective studies regarding LN number and survival rates contribute to the absence of a globally accepted standard for OEC. This study analyzed the SEER and NUWA cohorts to investigate the correlation between the quantity of LND and the HR for mortality risk in stage I OEC patients. Following smoothing, both the LND and HR curves for the SEER and NUWA cohorts exhibited a “U”-shaped pattern, indicating similar effect of LND on the survival rate among stage I OEC patients although differences in diagnostic criteria, patient demographics, treatment protocols, and healthcare systems between the United States and China were evident. And the intersection of the lowest value intervals from the SEER and NUWA cohorts yielded the optimal protective range of 29–34 for LND in patients with stage I OEC. These findings offer valuable guidance for clinical decision-making regarding LND in this patient population and may serve as a basis for further surgical interventions in cases where initial LN removal was insufficient. It is essential to acknowledge the inherent limitations of our retrospective analysis, arising from its nature and SEER database constraints, including the unavailability of specific CA125 levels, genetic test results, endometriosis presence, and detailed surgical data. Moreover, the relatively limited dataset from real-world Chinese patients may have introduced certain biases into our research results. In conclusion, this study, based on large, multicenter datasets, provides an in-depth analysis of the demographic characteristics, survival outcomes, and optimal range of LND in patients with stage I OEC. Our findings, supported by Cox regression analyses, demonstrate a notable enhancement in OS for stage I OEC patients who undergo LND. These results present clear surgical recommendations and comprehensive preoperative guidance for clinicians. Furthermore, we have determined that the ideal LND range for stage I OEC patients’ lies within 29–34 LNs. This not only introduces a novel metric for evaluating survival outcomes but also furnishes scientific evidence for assessing the necessity of additional surgical interventions. There are several limitations in the present study that should be considered. Firstly, the retrospective nature of the SEER database may lead to potential bias in information collection. Secondly, further analysis was limited in the SEER database due to lacking of several important variables, such as specific CA125 levels, genetic testing outcomes, information on endometriosis presence, and comprehensive surgical data. Furthermore, the relatively small sample size derived from real-world Chinese patients, embodied by the NUWA cohort, may compromise the generalizability of the findings.

According to the National Comprehensive Cancer Network (NCCN) guidelines, comprehensive staging surgery is advocated as the primary strategy for early-stage ovarian cancer, and lymph node dissection (LND) is mandatory for pathological assessment. 1 However, there is a controversy regarding the necessity of LND in early-stage ovarian cancer. Researchers argue that lymphadenectomy may improve the accuracy of ovarian cancer staging, thereby aiding in treatment decisions and prognostic predictions. 2 It is linked to improved progression-free survival, disease-free survival, and overall survival (OS) in patients with specific stage or early-stage ovarian cancer. 3 , 4 Conversely, some studies suggest that lymphadenectomy does not notably prolong the survival of early-stage ovarian cancer patients, 5 , 6 , 7 particularly, given the generally low rate of LN metastasis, which is below 20%. This suggests a potential risk of overtreatment for over 80% of these patients. 8 , 9 Ovarian endometrioid carcinoma (OEC) comprises 10–15% of epithelial ovarian cancers (EOCs). 10 Unlike other EOC types, OEC often presents with distinctive clinical manifestations in its early phase, including lower abdominal mass, pain, distension, and postmenopausal bleeding. 11 Consequently, approximately 75% of OEC patients are diagnosed at stages I–II according to International Federation of Gynecology and Obstetrics (FIGO) criteria. 12 OEC exhibits a more favorable prognosis (5-year survival rate >80%) than other EOC types. 13 Prognostic determinants for OEC may involve stage, grade, surgical intervention, residual tumor status, and adjuvant therapy. 14 , 15 , 16 Some studies have indicated a correlation between the incidence and progression of OEC with variables such as the number of pregnancies, age at menopause, and tubal ligation. 17 Nonetheless, the scarcity of stage I OEC cases across research facilities has impeded the development of a unified consensus on the clinical and pathological attributes, independent prognostic factors, and optimal surgical strategies. Currently, the surgical principles for OEC typically align with those of serous ovarian cancer, often including LND. 18 Standards for quality LND surgery have been established for various malignancies including non-small cell lung cancer (NSCLC), 2 , 19 and assessing the adequacy of LND surgery relies on the quantity of LNs removed, 20 , 21 but research focusing on early-stage OEC remains limited. Among patients diagnosed with stage I OEC preoperatively, the rate of LN positivity following dissection is only 5.5%. 4 LND may lead to complications such as lower limb lymphedema, lymphatic cysts, and pelvic infections. 22 Research in recurrent NSCLC suggests that an elevated number of LND (cutoff: 16) may reduce the efficacy of immunotherapy. 23 Consequently, further inquiry is warranted to elucidate its implications for prognosis and determine the optimal number of nodes to be removed. The rarity of OEC poses significant barriers for prospective, multicenter, and large-scale studies. Furthermore, notable regional differences in disease characteristics hinder the standardization of research methodologies and treatment protocols. For instance, among IC-stage OEC patients in the United States, well-differentiated tumors account for a higher proportion (42.1%) compared to moderately (40.8%) and poorly differentiated tumors (17.1%). 24 In contrast, most Chinese OEC patients are diagnosed with moderately differentiated tumors, accounting for 61.5%. 25 Such variations may influence clinical outcomes and treatment approaches, potentially resulting in divergent assessments regarding the necessity for comprehensive LN dissection. Therefore, it is particularly important to conduct retrospective studies using multiple databases to advance our understanding of this rare cancer. This study aimed to comprehensively assess the impact of LND on the OS rate of patients with stage I OEC. Data from 2,717 individuals sourced from the Surveillance, Epidemiology, and End Results (SEER) database in the United States, and 116 individuals sourced from the Chinese National Union of Real-world Gynecological Oncology Research and Patient Management Platform (NUWA) were utilized to determine the ideal number of lymph nodes to be dissected during the surgical procedure.

The authors declare no competing interests.

REAGENT or RESOURCE SOURCE IDENTIFIER Software and algorithms SEER ∗ Stat Software National Cancer Institute https://seer.cancer.gov/seerstat/ STATA Software StataCorp Version 14.0 R Software R Foundation for Statistical Computing Version 4.1.2 The Surveillance, Epidemiology, and End Results (SEER) database compiles cancer case data from 18 geographically diverse regions across the US. We selected the SEER database due to its comprehensive and detailed repository of cancer incidence, mortality, and survival statistics sourced from population-based registries nationwide. The patient population and selection process within the SEER Database involved the utilization of SEER ∗ Stat 8.4.1 software ( https://seer.cancer.gov/seerstat/ ) to identify patients diagnosed with OEC from January 2004 to November 2018. These patients were monitored until December 2019, following the guidelines outlined in the SEER documentation. The inclusion criteria encompassed patients diagnosed with stage I endometrioid carcinoma (OEC) with ICD- O -3 histologies 8380/3, 8381/3, 8382/3, and 8383/3 who underwent surgical intervention. The exclusion criteria involved patients at stage II–IV, those with unknown lymph node dissection status, those without follow-up data, individuals who did not undergo surgical intervention. As the present study focused on the ovarian cancer, all study participants were female. Ultimately, a total of 2,717 patients from the SEER database were selected for further analysis ( Figure 1 ). To gain a more comprehensive understanding of the characteristics of populations in different countries and regions, we have selected the multi-center NUWA database from China as a comparative benchmark. The NUWA database encompasses medical institutions from 20 diverse regions across the country and focuses on real-world research and patient management in the field of ovarian cancer. By utilizing both the SEER and NUWA databases, we can conduct a more comprehensive analysis regarding the impact of LND on survival outcomes for patients with OEC. The analysis utilized data from patients diagnosed with stage I OEC who underwent surgical intervention between January 2010 and December 2022, sourced from multiple institutions via the NUWA platform in China. The LNs were dissected during the surgical procedure for OEC patients, with the quantity of dissected lymph nodes assessed by pathologists in the postoperative phase. Patients were staged according to the FIGO Staging System for Ovarian, Fallopian Tube, and Primary Peritoneal Cancer (8th edition, 2017). The inclusion and exclusion criteria were similar to those of the SEER database. Finally, 116 patients diagnosed with OEC from the NUWA platform were incorporated into this study ( Figure 1 ). Being a retrospective study, no random allocation was performed in both SEER and NUWA databases. The SEER database is a publicly available resource that does not require individual ethical approval. For NUWA database, ethical approval was obtained from the review boards of the First Affiliated Hospital of Guangzhou Medical University; Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology; the First Affiliated Hospital, Sun Yat-sen University; the First Affiliated Hospital of Guangxi Medical University; the First Affiliated Hospital of Soochow University; Liaoning Cancer Hospital and Hubei Cancer Hospital. Data were anonymized and de-identified in both databases. The study was approved by the Ethics Committee of the First Affiliated Hospital of Guangzhou Medical University and informed consent was waived (2020-147). No cell lines, animals, plants, or microbe strains were used in this study. We incorporated multiple clinical and pathological factors into our analysis, which were obtainable from both the SEER database and the NUWA platform. To address the inconsistent staging versions in the SEER database, we standardized the staging to the 2014 FIGO system. The NUWA platform similarly used the 2014 FIGO system. The main outcomes were the OS of stage I OEC patients. OS referred to the length of time from the start of treatment or diagnosis of OEC until the time of death from any cause. Covariates included baseline age at diagnosis, race (white, black, other, unknown), marital status (married, un-married), tumor size (<the median diameter, ≥the median diameter, unknown), histological grade (1, 2, 3, unknown), laterality (unilateral, bilateral), serum CA125 level (normal, elevated, unknown), FIGO stage (IA, IB, IC), receipt of chemotherapy (yes, no/unknown) and surgical procedure (USO/BSO-Hys, USO/BSO+Hys, USO/BSO+ome-Hys, USO/BSO+ome+Hys, debulking, others). The definition of positive CA125 was that the value of serum CA125 exceeded the upper limit of the reference value. 40 Pearson’s χ 2 test was utilized for comparing categorical data, whereas independent samples t-tests were employed for comparing normally distributed continuous data. Survival curves were constructed using the Kaplan–Meier method and were compared via the log rank test. Cox proportional hazards regression was conducted to evaluate the influence of LND on the prognosis of stage I OEC. In order to identify the optimal number of dissected LNs, Cox proportional hazards regression combined with a restricted cubic spline (RCS) function using 4 degrees of freedom was utilized to explore the relationship between number of dissected LNs and prognosis of stage I OEC. A stratification analysis was subsequently conducted to identify potential risk factors, including age, marital status, tumor size, histological grade, unilateral or bilateral involvement, presence of elevated levels of CA125, FIGO stage, chemotherapy status, and surgical procedure. The differences between subgroups were evaluated using a Z-test: Z = E 1 − E 2 S E ( E 1 ) 2 + S E ( E 2 ) 2 , where E 1 and E 2 represent the logarithm of hazard risks (HRs), and SE (E 1 ) and SE (E 2 ) represent the corresponding standard errors. All statistical analyses were performed using STATA version 14.0 (StataCorp, College Station, TX, USA) and R software version 4.1.2 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided P-value of less than 0.05 was deemed as statistically significant.

We would like to express our gratitude to the SEER and NUWA for providing open access to clinical data, LinkDoc Technology (Beijing) Co. Ltd. for their technical assistance, and AstraZeneca (inv) China for their practical support. This work was supported by the Hospital Technology Transfer Project ( ZH201905 ), the Postdoctoral Fellowship Program of CPSF ( GZC20230611 ), and the University Student Innovation Capacity Enhancement Program ( 02-408-2304-19104XM and 02-408-2304-19096XM ).