Since the publication of the Gynecologic Oncology Group (GOG) study [1], the staging of endometrial cancer (EC) is based on surgical evaluation. By 1988 the International Federation of Gynecology and Obstetrics (FIGO) and GOG recommended systematic surgical procedure that included total abdominal hysterectomy with bilateral salpingo-oophorectomy (TAH&BSO), pelvic washing, and pelvic and para-aortic lymph node dissection (LND) in patients that presented risk factors including unfavorable histology and myometrial invasion [2]. The protocol stood until 2005, the year in which the American College of Obstetrics and Gynecology (ACOG) and FIGO recommended a complete surgical staging in every patient regardless of the risk factors [3]. This opinion was based on 2 retrospective studies which evidenced survival improvement in patients with EC stage I, grade 3 or more advanced tumors when the complete surgical staging was held [45]. However, no randomized studies supported these data. For this reason, the systematic LND has been lately a reason of debate. These doubts are more than justified considering the serious consequences and possible complications of the LND just to stratify the patient, if in fact there is no impact on the overall survival (OS). In consequence, more selective and individualized treatment regimens have been proposed in order to avoid an overtreatment. Several international guidelines already agree that patients classified as European Society of Medical Oncology (ESMO) low risk for recurrence tumors [6] could benefit from the conservative staging procedures without LND [78]. However, the controversy remains for tumors identified as ESMO intermediate or high-risk for recurrence. Recently a match pair retrospective study has observed the lack of prognostic value of LND in high risk EC [9], but studies analyzing intermediate-risk have not been performed.

The aim of our study was to evaluate the impact of LND on disease-free survival (DFS) in EC with intermediate-risk as primary objective; OS, morbidity, and cost as secondary end-points.

After approval of Institutional Review Board in each participating center, we conducted a multicenter retrospective study between the years 1995 and 2013 at 6 tertiary Spanish medical centers. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement has been followed for this study. A sample of 720 women diagnosed of intermediate-risk EC (IREC) according to the guidelines of the Spanish Society of Gynecology and Obstetrics (SEGO) and ESMO (endometrioid histology with myometrial invasion <50% and histological grade 3; or myometrial invasion ≥50% and grades 1–2; or cervical involvement and grades 1–2) [67] were identified and their medical charts reviewed.

Fig. 1 shows a flow chart of subjects and reasons for withdrawal. We excluded 6 (0.8%) out of 720 patients who did not undergo surgery as primary treatment because of severe morbidity prior to surgery. These patients were treated with progestin or radiotherapy. We excluded 66 (9.1%) cases with follow-up less than 12 months.

Patients with enlarged lymph nodes (LNs), as noted on computed tomography (CT) scan or magnetic resonance imaging (MRI), were included in the study. The remaining 648 (90%) cases were analyzed using a case-matched method to obtain equal cohorts of patients respect to age, body mass index (BMI), pre-surgical co-morbidities, American Society of Anesthesiologists (ASA) score, myometrial invasion and surgical approach to assess short-term surgical and oncological outcomes regarding the performance or not of any kind of LND (pelvic or pelvic and para-aortic LND). Finally, a total of 356 (49.4%) patients (178 matched pairs) were included in the final comparison.

All patients underwent TAH&BSO. Pelvic or pelvic and para-aortic LND were performed at institution discretion based on surgical team experience. The same oncological team in each center performed all surgical interventions. All patients were managed following the guidelines approved by the SEGO, which considered recommended but not mandatory the performance of LND for IREC cases.

Preoperatively, mechanical bowel preparation could be carried out, and prophylactic antibiotics and low molecular weight heparin always administered. Intravenous fluids were maintained until patients tolerated oral fluids, usually within first 24 hours after the surgery. Foley catheter was usually removed the day after the surgery. Serum hemoglobin levels were routinely obtained within 24 hours after the procedure. Patients were discharged home if they demonstrated ability to ambulate independently, tolerated a regular diet, had stable vital signs and pain was under control.

Route of surgery depended on the surgical team and institution experience. Due to its influence in morbidity outcomes, match paired controls were also selected for this variable. External beam radiotherapy (EBRT), brachytherapy and chemotherapy were selectively used postoperatively depending on the Institutional Tumor Board decision, as well as patients' follow-up. These data were also included in the match-pair analysis. The presence of recurrence, type of therapies, and patient status were collected during the patient's follow-up.

The cost analysis was performed using economic data provided by the economical department of each center. The value for each item analyzed was the median cost in euros of the participating centers. All costs with regards to surgery (including Robotic Da Vinci Surgical System and laparoscopy towers with an amortization over 7 years), human resources, the use of surgical instruments, the operating room per minute, and cost of conversion to other surgical routes. The hospitalization cost included the hospital stay and expenses resulting from complications, blood transfusions, intensive care unit stay, and re-operations. We did not include adjuvant treatment and societal costs.

For the statistical analysis, after obtaining the matched pairs cohorts, differences between the study groups (178 cases with LND vs. 178 cases with no-LND) were analyzed. Continuous variables were expressed as median and range, and were compared using the t-test in normal distributions or the Mann-Whitney test in non-parametrical distributions. Discrete variables were represented with absolute frequencies and percentages, and they were compared by a chi-squared test or the Fisher's exact test in case of small cell comparisons. Receiver operating characteristic (ROC) curve was calculated to analyze the relation between the number of nodes removed and nodes status to identify the cut-off predicting the number of nodes removed to find an affected node. For survival analysis, Cox's method was used to assess which factors were directly associated with survival. Multivariate modeling using Cox's proportional hazard models, including the significant variables in univariate analysis, was performed to obtain a subset of independent predictors of DFS and OS. Hazard ratio (HR) with 95% confidence interval (CI) was calculated. The Kaplan-Meier method was used to estimate the survival distribution in the study groups. The log-rank test was used to calculate the statistical signification between the groups in relation to disease recurrence and death. All statistical tests were 2-sided and statistical significance was defined as p-value less than 0.05. All computations were performed using STATA Statistics/Data Analysis version 11.2 (StataCorp LP, College Station, TX, USA).

A total of 356 patients (178 matched pairs) with IREC were included in the study. The lymphadenectomy group (LND group) included 178 patients who underwent TAH&BSO and LND. In the no-lymphadenectomy group (no-LND group), 178 matched patients underwent only TAH&BSO.

Patient's characteristics, preoperative findings, pathology and therapeutic details are shown in Table 1. Both surgical groups were similar in all variables, except in FIGO stage.

In the LND group, the median number of nodes removed was 15 (range, 1–65): 16 nodes for pelvic LND (range, 1–34) and 6 nodes for para-aortic LND (range, 1–42). In 67 cases (37.6%) the total nodal count was ≥18, and in 10 patients (5.6%) positive nodes were observed. All cases with pelvic positive nodes had a para-aortic LND, but para-aortic metastases were detected only in 1 case.

Adjuvant treatment was not added in 78 patients (21.9%). The complications derived from surgery, advanced age, co-morbidities, and refusal of consent were the reasons for not receiving it. In the 10 patients with positive nodes, EBRT was used as adjuvant treatment in 4, chemotherapy alone in 1 patient, and combined chemo-radiotherapy in 5 patients.

The benefit in DFS and OS of lymphadenectomy in IREC is still very controversial and there is no consensus among international scientific societies. This study did not found any benefit for survival in these types of tumors. The ACOG recommends the comprehensive surgical staging as an initial management of EC [11]; whereas others institutions suggest an individually adjusted LND approach [12]. The recently published thorough European consensus of the 3 societies of European Society for Radiotherapy and Oncology (ESTRO), European Society of Gynecologic Oncology (ESGO), and ESMO again do not clarify the surgical staging in patients with IREC [8].

There are just 2 prospective randomized studies analyzing the role of LND in EC [1314]. Both, Benedetti Panici group and ASTEC study concluded that there was apparently no survival benefit of LND neither in low risk nor intermediate/high risk EC. However, there are no prospective randomized studies that assess expressly the role of LND in IREC. The main drawback of the A Study in the Treatment of Endometrial Cancer (ASTEC) was the absence of systematic para-aortic LND performance in the LND group, the low mean number of nodes retrieved, and the mathematical model used in the analysis [1516]. These results were verified by a Cochrane group meta-analysis [17], and suggested the need of studies for intermediate-risk patients.

In the present study, the results can be limited because systematic LND including pelvic and para-aortic dissection was not carried out in all patients, and the number of para-aortic nodes removed was lower when compared with other studies [18]. Although, no minimum nodal counts have been established so far, some publications that focus on pelvic lymphadenectomy suggest that removal at least 10 pelvic nodes is needed to show a survival improvement [19]. Lutman et al. [20] suggested the excision of 12 LNs to improved survival in patients of intermediate/high-risk EC. A systematic review analyzed data from 2 randomized controlled trials (RCTs) and 7 observational studies, involving 16,995 patients with EC, showed that the efficacy of systematic LND, defined as removal of at least 10 LNs, resulted in improved OS in patients with intermediate/high-risk EC. The same group recently analyzed 476 patients with intermediate/high-risk EC and concluded that removing as many pelvic nodes as possible is required to warrant accurate nodal staging and improve survival [21]. Additionally, in our study the number of positive LN cases was low (10 cases), and probably a larger study could be needed to address those differences.

LND in intermediate/high-risk EC has been also studied in numerous retrospective trials, showing controversial results in relation to its survival benefit [2223]. Our group recently performed a retrospective matched pair study, which included 194 (97 pairs) diagnosed of high-risk EC, finding no benefit in DFS and OS of LND [10]. Similarly, differences between pelvic and para-aortic LND compared to pelvic LND alone have been studied, again with controversial results [1824]. The Survival Effect of Para-aortic Lymphadenectomy in Endometrial Cancer (SEPAL) study concluded that para-aortic LND had survival benefits for patients with intermediate/high-risk EC compare to pelvic LND alone [18]. However, the no-LND group was not reported, which could bias its final conclusions. Although, the role of para-aortic LND is established in the case of high-risk tumors, it remains to be determined in intermediate-risk tumors [25].

An additional value of LND in EC could be the impact on the adjuvant treatment modification, specifically radiation therapy. In these sense, the nodal metastases have been described as independent prognostic factor of survival, that it is why need to be explored surgically, supporting systematic LND [26]. However, other studies suggested that surgical nodal evaluation could be unnecessary [1027]. In our study, no impact on adjuvant treatment was observed.

In our study, we observed a low rate of perioperative complications and morbidity in the LND group. Nevertheless, we noted a significant difference between the 2 groups regarding late postoperative complications as lymphocyst and lower-limb lymphedema. These findings were previously observed in other studies [1028]. The infection rate was significantly higher in the no-LND group, probably due to the same factors that could have the impact on the decision to not perform LND, such as clinical characteristic of the patients and associated diseases. In addition, we considered infection not only the related ones with the surgical wound, but whatever sign of infection, included isolated fever.

Finally, the cost effectiveness study could also argue for the implementation of LND. There were no significant differences between the 2 analyzed groups with respect to the surgical treatment, hospitalization related costs or global costs. These results concur with studies in high-risk EC [10] but not in low-risk EC [28]. Other authors, however, suggested that a selective LND strategy based on a preoperative prediction model is shown to be more cost-effective than routine LND for EC [17]. The differences are probably due to the inclusion of different risk tumors and the differences in National Health Service budgets. More prospective trials are needed, performing specific analyzes for each risk group to establish the efficiency of LND in EC.

This study has some drawbacks that must be considered. Although the matched pair analysis is the best way to study a retrospective cohort, this is not a randomized trial and its results have a limited evidence. On the other hand, the exclusion of 45% of cases could have caused the loss of cases with recurrence or death that could affect the survival, and the low LN count could have induced the omission of some cases with positive LN. However, the presented study has a strength due to a matched-pair design that reduces the bias in the patient selection, creating homogenous and comparable groups for the main features, that could alter the election of the LND and survival.

In conclusion, LND in IREC have no influence in DFS or OS rates. However, the role of the type of LND and its biological impact need to be further explored. LND did not increase early morbidity or costs, but it may present higher late postoperative complications rate. Considering the low rate of positive nodes found in this type of tumors and lack of benefit, LND cannot be recommended systematically, but it could be an option when the adjuvant treatment has to be planned. In absence of prospective randomized trials, the present study contributes to refocus the surgical management for patients with IREC.