A detailed analysis of lymph node recurrence in endometrial carcinoma.

Background: The lymph node (LN) is one of the main sites of recurrence in patients with endometrial carcinoma (EC). Literature specifically analyzing LN recurrence (LNR) in EC remains limited in number.
Methods: Patients with EC undergoing surgery between 2006 and 2021 in Peking University People's Hospital was included, clinicopathological data of whom were collected and analyzed retrospectively by R 4.0.3.
Results: A total of 792 patients were included, with 73 patients having recurrence, among whom 21 patients had LNR. Median recurrence-free survival (RFS) in patients with LNR was 16 [4-39] months. LNR was extensive, with pelvic LNs most commonly involved (9/21). There are various patterns of LNR, with 33.3% (7/21) LN-only recurrence. Multivariable analysis suggested advanced stage, larger tumor diameter and poor histology were independent risk factors for LNR. Patients with LN metastasis (LNM) diagnosed at initial treatment accounted for 47.6% (10/21) of cases with LNR, 60.0% (6/10) of whom had recurrent LNs beyond the region of LNM, 90.0% (9/10) of whom had recurrence nodes overlapping with the range of lymphadenectomy. Uni- and multi-variable analysis suggested lymphadenectomy was not a protective factor for LNR, with both the range and number of LNs harvested considered. Conclusions: LNR is common in patients with EC, with an extensive range and various patterns of recurrence. The International Federation of Gynecology and Obstetrics (FIGO) stage, tumor diameter and histology were independent risk factors for LNR, but lymphadenectomy seemed not a protective factor for LNR. 2022 Translational Cancer Research. All rights reserved.

Introduction

Uterine cancer, the second most common gynecological cancer in China, accounted for 2.18×104 death in 2015 (1). Endometrial carcinoma (EC), constituting the majority of uterine cancer, has a fairly low recurrence rate (2), and the lymph node (LN) is one of the main sites of recurrence (3-5). LN recurrence (LNR) was seldomly discussed as a separate topic, and was usually incorporated into the comparison between ‘distant recurrence’ and ‘locoregional recurrence’ (3). Studies discussing the risk or protective factors of LNR remain limited in number (5). Our study is designed to discuss the incidence, distribution and risk factors of LNR, including the potential protective effect of lymphadenectomy. We present the following article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-21-2588/rc).

Methods

Study population and treatment

We retrospectively analyzed patients with EC in Peking University People’s Hospital, who underwent surgery between 2006 to 2021. Clinicopathological data such as the International Federation of Gynecology and Obstetrics (FIGO) 2009 stage, tumor diameter, histology, lymphovascular space invasion (LVSI), peritoneal cytology, lymphadenectomy and adjuvant therapy, were collected from the medical record system. Histology was classified by invasiveness into ‘endometrioid G1/2’, ‘endometrioid G3’ and ‘non-endometrioid’. Generally, patients with EC (I) <2 cm; (II) myometrial invasion (MI) <1/2; (III) endometrioid G1/2 would have a surgery without systemic lymphadenectomy, otherwise a staging surgery ± sentinel LN (SLN) biopsy would be adopted. Pelvic lymphadenectomy (PLND) involved external, internal and common iliac LNs, obturator LNs and deep inguinal LNs in our institution. Whether to perform para-aortic lymphadenectomy (PALND) depended on the range and pathological characteristics of the tumor, including intraoperative pathological findings of SLNs. Risk grouping for recurrence was in accordance with the guideline by the European Society of Gynaecological Oncology, the European Society for Radiotherapy and Oncology and the European Society of Pathology (ESGO/ESTRO/ESP) published in 2015 (6). Adjuvant therapy for EC included radiotherapy and chemotherapy, which were adopted generally according to patients’ risk stratification, with age, health status and economic condition included in the practice as well. Patients with no less than intermediate risk generally received radiotherapy. The dose and range of radiotherapy depended on clinicopathological parameters, such as cervical involvement, LVSI, LN metastasis (LNM), residual foci, etc. Patients with no less than high-intermediate risk generally received chemotherapy ± radiotherapy, with platinum plus taxane being the most common regimen. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by institutional ethics board of Peking University People’s Hospital (No. 2020PHB013-01) and individual consent for this retrospective analysis was waived.

Follow-up and recurrence classification

Follow-up data were collected using medical record system, phone call and text message, including physical examination, ultrasonography, tumor marker, etc. CT and MRI scan were adopted if necessary. Follow-up was conducted every 3 months for the first 2 years, every 6 months for the next 3 years, and yearly since the fifth year. LNR was defined as recurrence in LN bearing area, e.g., pelvic sidewall, para-aortic area, consistent with the previous study (5), with/without recurrence of other patterns. Besides LNR, pelvic recurrence and peritoneal recurrence were adopted to refer to recurrence occurring in pelvic cavity and peritoneal cavity, e.g., recurrence in the vaginal cuff and on the surface of the liver. Distant recurrence referred to recurrence in distant organs or deep in the parenchyma of pelvic and abdominal organs, e.g., recurrence in the liver parenchyma. Recurrence-free survival (RFS) was defined as the time between surgery and the first recurrence, or death from any cause, whichever occurred first, or the last visit for patients alive without recurrence. Thus, for a cohort only composed of patients with LNR and patients without recurrence (‘test-LNR’) to explore risk factors of LNR, RFS is equivalent to LNR-free survival.

Statistical analysis

To explore possible risk/protective factors, univariable analysis and multivariable analysis were conducted by Kaplan-Meier analysis and Cox regression model. Mann-Whitney test was adopted to evaluate the distribution of continuous variables between groups while Chi-square test was adopted for categorized variables. All variables had an effective rate above 95% except for molecular subtype. Therefore, missing data were omitted from the analysis, while the molecular subtype was reported separately in Table S1. P<0.05 was considered statistically significant. Data were collected and analyzed by R 4.0.3.

Results

Overview of LNR in EC patients

A total of 792 patients with EC were included in this study, with a median follow-up of 47 [1-110] months. Clinicopathological data were presented in . Seventy-three (9.2%) patients experienced recurrence, with a 3-year RFS (3yRFS) of 98.2% and a 5yRFS of 97.2%. Excluding 14 patients with sites of recurrence unknown, 21 patients (35.6%) had LNR, and 38 (64.4%) had recurrence without LNR. Patients with LNR had a median RFS of 16 [4-39] months, with 76.2% (16/21) experiencing recurrence within 2 years (Table S1).

Table 1

Clinicopathological characteristics of the EC patients, excluding recurrence with sites unknown

Clinicopathological characteristics	LNR† (n=21)	Recurrence without LNR (n=38)	No recurrence (n=733)
Age (years), median [range]	57.0 [46–76]	57.0 [23–77]	56.0 [28–83]
Tumor diameter (cm), median [range]	3.8 [0.8–11.0]	3.5 [0.1–10.0]	2.5 [0.1–16.0]
CA125 (U/mL), median [range]	38.0 [5.0–3,060.0]	23.4 [4.2–530.8]	20.5 [4.5–1,296.0]
Stage, n (%)
I	6 (1.0)	16 (2.6)	602 (96.5)
II	1 (2.3)	2 (4.7)	40 (93.0)
III	8 (9.2)	9 (10.3)	70 (80.5)
IV	6 (25.0)	11 (45.8)	7 (29.2)
Histology, n (%)
Endometrioid G1/2	6 (1.0)	17 (2.8)	574 (96.1)
Endometrioid G3	7 (8.2)	5 (5.9)	73 (85.9)
Non-endometrioid	8 (8.3)	16 (16.7)	72 (75.0)
LVSI, n (%)
Negative	14 (2.1)	25 (3.8)	619 (94.1)
Positive	7 (5.8)	13 (10.8)	100 (83.3)
Peritoneal cytology, n (%)
Negative	13 (1.8)	23 (3.2)	676 (94.9)
Positive	5 (9.4)	14 (26.4)	34 (64.2)
Chemotherapy, n (%)
No	4 (0.9)	8 (1.8)	438 (97.3)
Yes	17 (5.2)	30 (9.2)	280 (85.6)
Radiotherapy, n (%)
No	13 (2.6)	28 (5.7)	454 (91.7)
Yes	8 (5.7)	10 (7.1)	122 (87.1)
Lymphadenectomy, n (%)
None	3 (2.5)	7 (5.7)	112 (91.8)
Biopsy only	0	3 (3.5)	82 (96.5)
PLND	5 (4.3)	8 (7.0)	102 (88.7)
PPALND	13 (2.9)	20 (4.4)	423 (92.8)
LNM, n (%)
No	11 (1.5)	27 (3.8)	673 (94.7)
Yes	10 (14.9)	11 (16.4)	46 (68.7)

†, with/without recurrence in sites beyond LNs. EC, endometrial carcinoma; CA125, carbohydrate antigen 125; LVSI, lymphovascular space invasion; PLND, pelvic lymphadenectomy; PPALND, pelvic and para-aortic lymphadenectomy; LNM, lymph node metastasis; LNR, lymph node recurrence; LNs, lymph nodes.

Distinct from LNM mainly involving pelvic and para-aortic LNs, LNR seemed more extensive, with a detailed description in Table S1. Common sites of LNR were as follows: pelvic LNs (9/21), retroperitoneal LNs (not specific) (5/21), inguinal LNs (4/21), para-aortic LNs (3/21), mediastinal LNs (2/21), supraclavicular LNs (2/21), the others.

Only 33.3% (7/21) of patients with LNR experienced LN-only recurrence, while the others experienced combined recurrence, with LN + distant ± others being the most common combination (7/21) (Figure S1).

Risk factors of LNR in EC patients

A cohort consisting of patients with LNR and patients with no recurrence was defined as ‘test-LNR’ (n=740), in order to discuss risk factors of LNR, excluding recurrent cases without LNR or with unknown sites of recurrence.

Advanced stage, poor histology, LVSI, tumor diameter ≥2 cm, LNM and adjuvant chemotherapy were significantly related to LNR by univariable analysis, among which only advanced stage [hazard ratio (HR): 6.8, 95% confidence interval (CI): 2.3–20.1], larger tumor diameter (HR: 1.2, 95% CI: 1.0–1.4) and poor histology (G3 vs. G1/2: HR: 6.9, 95% CI: 1.9–25.0; non-endometrioid vs. G1/2: HR: 10.6, 95% CI: 3.0–37.4) were independent risk factors of LNR by multivariable analysis ().

Table 2

Survival analysis for risk factors of LNR in test-LNR

Clinicopathological features	Univariable analysis			Multivariable analysis
	N	P value		HR (95% CI)	P value
FIGO stage		<0.001*
I/II	649			Reference
III/IV	91			6.8 (2.3–20.1)	0.001*
LVSI		0.004*
Negative	633			–	–
Positive	107
Tumor diameter (cm)†		0.004*
<2 cm	289			1.2 (1.0–1.4)	0.032*
≥2 cm	447
Histological subtype		<0.001*
Endometrioid G1/2	580			Reference
Endometrioid G3	80			6.9 (1.9–25.0)	0.003*
Non-endometrioid	80			10.6 (3.0–37.4)	<0.001*
Peritoneal cytology		<0.001*
Negative	689			–	–
Positive	39
LNM		<0.001*
No	684			–	–
Yes	56
Adjuvant radiotherapy		0.080
No	467			–	–
Yes	130
Adjuvant chemotherapy		<0.001*
No	442			–	–
Yes	297
Lymphadenectomy		0.421
Yes	626			–	–
No	114

†, adopted into multivariable analysis as a continuous variable; *, P<0.05. LNR, lymph node recurrence; FIGO, the International Federation of Gynecology and Obstetrics; LVSI, lymphovascular space invasion; LNM, lymph node metastasis; HR, hazard ratio; CI, confidence interval.

LNM and LNR

LNM was significantly associated with LNR in univariable analysis, and among patients with LNR, 47.6% (10/21) exhibited LNM at initial treatment, defined as ‘LNMR’. In LNMR patients, 80.0% (8/10) had recurrent LNs related to metastatic LNs, e.g., left deep inguinal LNs metastasis → left inguinal LNs recurrence. However, 60.0% (6/10) of such LNMR patients had recurrent LNs beyond the regions of initial LNM, e.g., pelvic LNs metastasis → pelvic, para-aortic and inguinal LNs recurrence (Table S1).

Intriguingly, all of LNMR patients underwent systemic lymphadenectomy, with 80.0% (8/10) undergoing pelvic and para-aortic lymphadenectomy (PPALND) and 20.0% (2/10) undergoing PLND. However, 90.0% (9/10) of LNMR patients had recurrent LNs overlapping with the range of lymphadenectomy, suggesting possible failure of lymphadenectomy to eradicate tumor cells in the lymphatic system, even within the scope of this operation.

Effect of lymphadenectomy on LNR

Univariable analysis suggested lymphadenectomy was not a protective factor for LNR in EC patients (). Further multivariable analysis including lymphadenectomy and independent risk factors reached in didn’t support the protective effect of lymphadenectomy for LNR (P=0.351), either. A Chi-square test stratified by LNM exhibited no correlation between lymphadenectomy and LNR for both LNM cases and non-LNM cases (P>0.999 and P=0.701, respectively, Fisher’s exact test).

No association was suggested between the scope of lymphadenectomy and LNR by univariable analysis (). Since no LNR occurred to patients undergoing LN biopsy only, they were merged with patients not undergoing lymphadenectomy into one group to conduct another univariable analysis, yet still failing to exhibit an association between systemic lymphadenectomy and a decrease in LNR (P=0.300).

Figure 1

No association exhibited between range of lymphadenectomy and decrease of LNR in test-LNR. PLND, pelvic lymphadenectomy; PPALND, pelvic and para-aortic lymphadenectomy; LNR, lymph node recurrence.

Median number of LNs harvested in LNR cases was 20 [0-56], while that of cases without recurrence was 26 [0-78] (). No significant difference was observed between the groups (P=0.254, Mann-Whitney test).

Figure 2

Distribution of LNs harvested between no-recurrence and LNR group in test-LNR. LNR, lymph node recurrence.

Discussion

As the majority of uterine cancer, EC has a relatively low rate of recurrence (2,7), but the incidence of recurrence still indicates a poor prognosis (8). Recurrence is often detected in vagina, pelvic cavity, peritoneal cavity, lung, liver, bone, LN (pelvic, para-aortic), etc. (4,7,9). LNR, as one of the main patterns of EC recurrence (3,4), is seldom documented specifically (5). LNR indicates failure in the lymphatic system, and there might exist unique characteristics. We discussed several interesting topics of LNR, including its incidence, common sites, risk factors, as well as relationship between LNR and LNM, between LNR and lymphadenectomy.

LNR existed in over 1/3 of recurrent EC cases, with a more extensive pattern involving lymphatic system around the whole body (4,10,11). Still, LNs mostly involved in recurrent EC were those in metastatic cases at initial treatment, i.e., retroperitoneal LNs, suggesting a possible mechanism of LNR from growth of potential metastatic lesions in the lymphatic system, which survived primary surgery and adjuvant therapy. This was consistent with the univariable analysis identifying LNM as a risk factor for LNR. Several studies demonstrated a similar correlation (5,11). However, 88.9% of LNMR patients had recurrent LNs exceeding the range of LNM at initial treatment, suggesting that LNM might have the ability to further disseminate in the lymphatic network, rather than simply be an exhibition of invasiveness of the primary lesion. This research seems the first to describe this phenomenon in EC.

The benefit of lymphadenectomy in EC patients has been a controversial topic for over a decade (12-14). The survival advantage associated with lymphadenectomy exists primarily in patients with a relatively high risk of recurrence (14-18), and improperly expanding the indication of lymphadenectomy might result in more adverse effects instead (19-25). Our research didn’t support the role of lymphadenectomy in preventing LNR in terms of the scope and number of LNs harvest, with agreement with Mariani et al. (5). Lymphadenectomy removes LNs in certain node-bearing regions, instead of the whole lymphatic network. Considering the expanding property of the tumor in LNMR patients stated above, we propose chemotherapy agents with sufficient concentration and efficacy in lymphatic system might be crucial for prevention of LNR, rather than the attempt to eradicate LNs in drainage directions.

This study is the second study focused specifically on LNR in EC. In 2002, Mariani et al. firstly defined a similar concept, lymphatic failure, as a relapse occurring on node-bearing area as the primary site of failure. Different from our discussing risk factor in a cohort consisting of LNR and non-recurrence patients, Mariani et al. explored the difference between LNR and recurrence in other sites based upon a cohort of recurrent EC patients. Three predictors of lymphatic failure were LVSI, positive LNs, and cervical stromal invasion (5), two of which were also identified as risk factors of LNR in our research. Independent risk factors identified in our research, namely FIGO stage, tumor diameter and histology, were all related to malignant behavior of the tumor. Future research on the invasion, survival and proliferation of EC cells in the lymphatic system will help understand the formation of LNR and provide potential therapeutic targets.

Still there exist some limitations in our study. Over the span of 15 years, indications and practice of lymphadenectomy in EC have evolved in our institution, increasing the risk of confounding bias, e.g., random LN biopsy in the past vs. SLN biopsy nowadays. Also, due to relatively few cases of LNR, systematic comparison between LNR and recurrence in other sites as Mariani et al. was not conducted, and there was no separate discussion of cases with LN-only recurrence. Finally, a negative result might originate from a small sample size, rather than the objective lack of difference.

In summary, our study suggested LNR was common in patients with EC, with an extensive range and various patterns of recurrence. FIGO stage, tumor diameter and histology were independent risk factors of LNR, but lymphadenectomy seemed not protective for LNR.

The article’s supplementary files as