Comparison of sentinel lymph node distribution and lymphatic drainage pathway between high- and low-risk endometrial cancers

This study aimed to compare the distribution and drainage pathway of sentinel lymph nodes between high- and low-risk endometrial cancers. In total, 429 patients with endometrial cancer who underwent sentinel lymph node biopsy in Peking University People’s Hospital from July 2015 to April 2022 were retrospectively enrolled. There were 148 patients in the high-risk group and 281 patients in the low-risk group. The unilateral and bilateral detection rates of sentinel lymph nodes were 86.5% and 55.9%, respectively. The highest detection rate was achieved in the subgroup with a combined use of indocyanine green (ICG) and carbon nanoparticles (CNP) (94.4% for unilateral detection and 66.7% for bilateral detection). The upper paracervical pathway (UPP) was detected in 93.3% of cases in the high-risk group and 96.0% of cases in the low-risk group (p = 0.261). The lower paracervical pathway (LPP) was detected in 10.0% of cases in the high-risk group and 17.9% of cases in the low-risk group (p = 0.048). Remarkably increased detection rates of SLN in the common iliac (7.5%) and para-aortic or precaval areas (2.9%) were observed in the high-risk group. In contrast, a markedly decreased detection rate of SLN in the internal iliac area (1.9%) was observed in the high-risk group. The highest detection rate of SLN was observed in the subgroup with a combined use of ICG and CNP. The detection of UPP is important for both high-risk and low-risk cases, while LPP detection plays a more important role in the low-risk group. Lymphadenectomy in the common iliac and para-aortic or precaval areas is essential for patients with high-risk EC. Removal of internal iliac lymph nodes is essential for patients with low-risk EC, in case of ineffective SLN mapping.


Introduction
Endometrial cancer (EC) is one of the most common malignant tumors among women [1,2]. The incidence and mortality rates of EC are increasing, and approximately 65,950 new cases and 12,500 deaths have been estimated in 2022 [1]. Lymph node metastasis is the main pattern of metastasis in EC, which causes a worse survival and poor prognosis [3]. Outer myometrial invasion, high-grade endometrioid adenocarcinoma, and non-endometrioid adenocarcinomas, including serous carcinoma, clear cell carcinoma, undifferentiated/dedifferentiated carcinoma, and carcinosarcoma, are risk factors for lymphatic metastasis in EC [4][5][6]. Based on risk stratification, lymph node metastasis occurs in 3-9% of patients with EC in the low-risk group and nearly 20% of patients with EC in the high-risk group [7]. An accurate lymphatic assessment is essential for surgical staging and postoperative follow-up and treatment in EC. Indiscriminate systematic lymphadenectomy became controversial because the high incidence of serious complications, such as lymphedema and lymphocyst formation, undermined the quality of life of patients [8].
During the last decade, sentinel lymph node (SLN) biopsy has become a standard method of nodal assessment in EC [9]. Cervical injection of tracers has been recognized as a credible approach for successful SLN mapping, which is widely used for nodal assessment in low-risk EC [10]. Recent studies increasingly focused on the application of SLN biopsy in patients with high-risk EC [4,11,12]. Despite its safety and efficacy in primary studies, more studies are needed. Diagnosis accuracy of SLN biopsy relies on not only a sufficient detection rate but also on a low false negative rate. SLNs are "color nodes" suspicious of metastasis [13]. Anatomy-based lymphadenectomy can play a complementary role for SLN mapping. It has been reported that SLN biopsy can increase the detection rate of metastatic lymph nodes in uncommon anatomical positions [14]. There are two lymphatic drainage pathways in EC [7]. 1) The upper paracervical pathway (UPP) along the uterine artery, drains the external iliac and/or obturator lymph nodes. 2) The lower paracervical pathway (LPP) along the uterine vein, which drains the internal iliac and/or presacral lymph nodes. Besides, the infundibulopelvic pathway (IPP) is located along the infundibulopelvic ligament. The IPP finally converges to the para-aortic lymph nodes, which are isolated from the pelvic lymph nodes [15].
Several studies described and validated these pathways during SLN mapping in EC [16,17]. However, the differences in the location or drainage pathways of SLN between high-risk and low-risk ECs remain to be known.
In this study, we aimed to describe the location and drainage pathways of SLN in high-risk and low-risk ECs after cervical injection of tracers. This study provides information about anatomical heterogeneity between highrisk and low-risk group of EC.

Materials and methods
Patients who underwent SLN biopsy with or without systematic lymphadenectomy in Peking University People's Hospital from July 2015 to April 2022 were retrospectively enrolled in our study. The inclusion criteria were as follow: (a) Diagnosed with EC by pathological assessments. (b) Staging surgery, including total hysterectomy, bilateral adnexectomy, and SLN biopsy with or without systematic lymphadenectomy, was performed. The exclusion criteria were as follow: clinical records were missing.
Clinical records include age, height, weight, FIGO stage, histopathological type (grade), depth of myometrial infiltration, tracer use (type and injection site), number of resected SLNs, number of resected pelvic lymph nodes, site, and number of positive lymph nodes. Patients with grade 1-2 endometrial adenocarcinoma and inner half myometrial invasion were assigned to the low-risk group. Patients with high-grade cancer or outer half myometrial invasion were assigned to the high-risk group.
Indocyanine green (ICG) alone, carbon nanoparticles (CNP) alone, and the combination of ICG and carbon nanoparticles were used as tracers. Carbon nanosuspension (50 mg/mL) or indocyanine green diluent (25 mg/10 mL) were prepared before surgery. The injection time was precisely recorded after successful anesthesia and before the operation. All patients received cervical injection. The injection sites include two-point (3 and 9 o 'clock positions on the cervix) or four-point (3, 6, 9, and 12 o 'clock positions or 2, 4, 8, and 10 o 'clock positions on the cervix). A 1 mL syringe was used and 0.2 mL was injected into each point. Deep (1 cm) and superficial (2-3 mm) injections were employed for better mapping. The median time from injection to removal of SLN was 32 (18-51) minutes. H&E staining was performed on all resected lymph nodes.
The clinicopathologic characteristics of patients were described. Categorical variables were compared by Chisquare and Fisher's exact tests. Continuous variables were compared by student's t test or Mann-Whitney U test. p value < 0.05 was considered statistically significant. All analyses were performed using SPSS version 26.0 (IBM SPSS Inc., Chicago, IL).

Results
A total of 429 patients were finally enrolled in our study, including 148 patients in the high-risk group and 281 patients in the low-risk group. The median age of patients was 56 years (range 51-62), and the median body mass index was 26.0 kg/m2 (range 23.5-28.3). There were 384 (89.5%) patients diagnosed with endometrial adenocarcinoma. In total, 140 patients (32.6%) underwent SLN dissection only, while 289 patients (67.4%) underwent pelvic lymph node dissection with or without para-aortic lymph node dissection. ICG was used in 31.0% (133/429) patients, CNP was used in 48.0% (206/429) patients, and the combination of ICG and CNP was used in 21.0% (90/429) patients. Tracers showed no significant differences between the highrisk and low-risk groups (p = 0.866). The clinicopathologic characteristics are presented in Table 1.
The unilateral and bilateral detection rates of SLN were 86.5% (371/429) and 55.9% (240/429), respectively. The detection rate was remarkably lower in the high-risk group than in the low-risk group (unilateral detection rate: highrisk vs. low-risk: 81.1% vs. 89.3%, p = 0.018; bilateral detection rate: high-risk vs. low-risk: 43.9% vs. 62.3%, p < 0.001). There was no significant difference in unilateral and bilateral detection rates between the groups when the combination of ICG and CNP was used (unilateral detection rate: high-risk vs. low-risk: 96.6% vs. 93.4%, p = 0.547; bilateral detection rate: high-risk vs. low-risk: 55.2% vs. 72.1%, p = 0.111). The detection rates of SLN in EC are presented in Table 2. The overall sensitivity and negative predictive value were 70.6% and 98.6%, respectively. The sensitivity and negative predictive value in the high-risk group were 64.3% and 95.5%, respectively. The sensitivity and negative predictive value in the low-risk group were 100.0% and 100.0%, respectively (Table S1).
The UPP was identified mostly (353/371, 95.1%) in EC patients with successful SLN mapping, and the bilateral UPP was detected in 55.8% (207/371) of these patients. In the high-risk group, the unilateral and bilateral detection rates of UPP were 93.3% (112/120) and 51.7% (62/120), respectively (Fig. 1). In the low-risk group, the unilateral and bilateral detection rates of UPP were 96.0% (241/251) and 57.8% (145/251), respectively. The detection rate of UPP was similar in high-and low-risk groups (p = 0.261). The LPP was found in only 57 patients (57/371, 15.4%). The detection rate of LPP was significantly lower in the high-risk group than in the low-risk group (10.0% vs. 17.9%, p = 0.048). Isolated LPP was rarely detected in EC patients (5/371, 1.3%), and there was no difference in this regard between the two  Table 3. About 1801 SLNs were dissected from 371 EC patients, including 520 SLNs in the high-risk group and 1281 SLNs in the low-risk group. External iliac SLNs (46.6% in total; high-risk group vs. low-risk group: 44.0% vs. 47.6%, p = 0.167) and obturator SLNs (37.8% in total; high-risk group vs. low-risk group: 40.0% vs. 36.9%, p = 0.223) were the most common SLNs. Common iliac SLNs were significantly more common in the high-risk group than in the lowrisk group (7.5% vs. 5.1%, p = 0.045), while internal iliac SLNs were significantly less common in the high-risk group than in the low-risk group (1.9% vs. 6.9%, p < 0.001). Interestingly, SLNs in the para-aortic or precaval area were more common in the high-risk group, compared with the low-risk group (2.9% vs. 0.9%, p = 0.001). The distribution of SLNs in the two groups of EC patients is shown in Table S2 and Fig. 1.
We found that lymphatic metastasis mostly affected external iliac (18/103, 17.5%) and obturator (34/103, 33.0%) lymph nodes. Internal iliac metastatic lymph nodes were significantly less common in the high-risk group than in the low-risk group (1.1% vs. 27.3%, p = 0.001), while para-aortic or precaval metastatic lymph nodes were more common in the high-risk group (32.6% vs. 9.1%, p = 0.208). The distribution of metastatic lymph nodes in the high-and low-risk patients is shown in Table 4.

Discussion
Previous studies have reported different detection rates of SLN. The total detection rate ranged from 60 to 100%, and the bilateral detection rate ranged from 40 to 80% in previous studies [9]. A sufficiently high bilateral detection rate is needed for optimal application of SLN. The type of tracer can influence SLN detection rate. ICG is widely used for SLN mapping in EC. Previous studies reported that ICG has relatively higher unilateral and bilateral detection rates than  Fig. 1 The lymphatic drainage pathway of EC and the distribution of sentinel lymph nodes between high-risk and low-risk groups Several studies have also compared ICG alone with the combined use of blue dye and 99mTc. The results showed that the overall detection rate of ICG, and particularly its bilateral detection rate is superior to that of blue dye and 99mTc [14,[18][19][20][21]. CNP is a potential tracer for SLN mapping with a higher tendency for the lymphatic system and prolonged presence in lymph nodes. It is also suitable for laparotomy or laparoscopic surgery. In our study, ICG, CNP, and the combination of ICG and CNP were employed. The highest unilateral (94.4%) and bilateral (66.7%) detection rates were observed after using the combination of ICG and CNP. The enrollment of patients with FIGO III-IV, lymph node involvement, and enlarged lymph nodes may be predictive factors for failed mapping [22]. In our study, 42 patients (42/429, 9.8%) were diagnosed with FIGO III-IV EC, which might cause a failure of the mapping and lead to the low bilateral detection rate of SLN. It is important to select more suitable cases for SLN biopsy.
Significantly lower bilateral detection rates were found in the ICG (p = 0.01) and CNP (p = 0.037) subgroups of the high-risk group and the whole high-risk group (p < 0.001), compared with the low-risk group. Although there was an increasing bilateral detection rate of the combination groups for high-risk patients, the difference was not statistically significant (p > 0.05). The role of SLN biopsy in high-risk group remains unclear. A systematic review and meta-analysis were performed to draw a conclusion that SLN biopsy with ICG cervical injection could be adopted for nodal assessment in both high-risk and low-risk group [23]. More prospective studies are urgently needed to provide a higher level of evidence for the safety of SLN in high-risk group.
The UPP drains external iliac and obturator lymph nodes to the common iliac lymph nodes. It has been reported as the most common drainage pathway in EC [24]. It was detected in most cases during SLN mapping in EC, regardless of the injection site (cervical or fundal injection of tracers) [25]. In our study, cervical injection was done in all cases. More than 95% of cases (353/371, 95.1%) with successful SLN mapping had UPP involvement. In addition, 81.1% (301/371) of cases had isolated UPP involvement. No significant difference was found between the high-risk and low-risk groups (93.3% vs. 96.0%, p = 0.261) regarding UPP involvement. Consistently, previous studies indicated that the detection of at least one SLN in the UPP is necessary regardless of the risk group. The LPP drains the internal iliac and/or presacral lymph nodes. Its involvement is less common in SLN mapping of EC. LPP involvement seems to be limited to high-risk histology phenotypes, including grade 3 endometrioid adenocarcinoma, serous adenocarcinoma, clear cell adenocarcinoma, and carcinosarcoma [24]. In our study, 15.4% (57/371) of cases had LPP involvement. LPP involvement was significantly lower in the high-risk group (10.0% vs. 17.9%) (p = 0.048). This shows that LPP involvement is more common in low-grade EC and superficial myometrial invasion.
It has been proven that the external iliac and obturator fossa are the most common sites for detecting SLN [18]. In our study, 46.6% (839/1801) and 37.8% (681/1801) of SLNs were detected in external iliac and obturator lymph nodes, respectively. There was no significant difference in SLN detection rate in these areas between the high-risk and low-risk groups (p > 0.05). However, a significant difference was found in SLNs detection rate in common iliac, internal iliac, para-aortic, and precaval areas between the groups. A remarkably increased SLN detection rate was observed in the common iliac and para-aortic or precaval areas of the high-risk group, while a remarkably decreased SLN detection rate was observed in the internal iliac area of the high-risk group. Regarding the location of metastatic lymph nodes, the external iliac area and obturator fossa were common sites of involvement in both groups. Para-aortic or precaval nodal metastasis was more common in the high-risk group, while internal iliac nodal metastasis was more common in the low-risk group.
We recognize the limitations of the current study. Since this is a retrospective study in nature, patients undergoing ICG, CNP, and both tracers with either laparotomy or laparoscopy were all enrolled. The tracers used (ICG, CNP, or both) and surgical approaches (laparoscopy or laparotomy) might cause selection biases in this study. In the initial time, we performed the combination of ICG and CNP by laparoscopy to investigate the additional value of CNP. As the safety and value of CNP were proven, we performed a randomized controlled study to compare the use of ICG and CNP. The bias caused by tracers' use could be minimized through randomization. Thus, further prospective studies focusing on high-risk patients must be conducted.

Conclusion
The highest detection rate of SLN was observed in the subgroup with a combined use of ICG and CNP. The UPP is important for both high-risk and low-risk cases, while the LPP is more important in the low-risk group. Common iliac and para-aortic or precaval lymphadenectomy is essential for patients with high-risk EC, while internal iliac lymphadenectomy is essential for patients with low-risk EC, in case of ineffective SLN mapping.