Endometrial cancer remains the most common malignancy of the female genital tract. Lymph node metastasis is one of the most important prognostic factors, and stratification into pelvic lymph node invasion (stage IIIC1) and para-aortic lymph node invasion (stage IIIC2) improved the predictive value of the 2009 International Federation of Gynecology and Obstetrics (FIGO) classification.
Radiological examination methods such as magnetic resonance imaging and positron emission tomography–computed tomography do not have good-enough sensitivity to avoid lymphadenectomy for the assessment of lymph node invasion. Prediction scores are becoming increasingly valuable to exclude lymph node metastasis in low-risk groups, and biomarkers could help to identify patients with high-risk lymph node metastatic probability.
The therapeutic role of lymph node dissection remains a matter of debate. Several end points can be considered to evaluate the opportunity of lymphadenectomy in endometrial cancer.
First, we compare survival according to the realization, the extent, and the numbers of nodes removed during lymphadenectomy. Second, we assess the opportunity of lymphadenectomy in order to tailor adjuvant treatment modalities. Third, we analyze the surgical complication rate after pelvic lymphadenectomy.
Introduction
In Europe, endometrial cancer (EC) remains the third most common cancer diagnosed in 2012 . Most cases occur in postmenopausal women. Based on histopathology, ECs are divided into two categories. Type I are typically low-grade adenocarcinomas that are usually estrogen related, are diagnosed early, and have a favorable prognosis. Type II ECs are not hormone dependent and are usually high-grade endometrioid adenocarcinomas, and papillary serous and clear cell carcinomas and carcinosarcomas .
Distant organ metastasis is the most important factor in determining patient survival in cancer. Metastasis is thought to occur through the blood vascular and lymphatic systems. Cancer cells are able to express growth factors and consequently have the capacity to create conduits for tumor metastasis. In EC, pelvic lymph nodes (LNs) are the most frequent location of metastasis. The major lymphatic trunks are the utero-ovarian (infundibulopelvic), parametrial, and presacral, which drain into the hypogastric, external iliac, common iliac, presacral, and para-aortic nodes . Although a direct route of lymphatic spread from the corpus uteri to the para-aortic through the infundibulopelvic ligament has been suggested from anatomical and sentinel LN studies , direct metastases to the para-aortic LN are uncommon. This reflects the complexity of lymphatic drainage patterns of the uterus, particularly when compared to those observed in mapping studies of cutaneous or breast malignancies. At a larger scale, the underlying controversy around the clinical benefit of lymphadenectomy has not been solved. Here, we aim to summarize the main findings related to this commonly performed surgical staging procedure.
Prognostic value of LN metastasis
LN metastasis constitutes one of the most important prognostic factors of EC. In 1988, this information was included in the International Federation of Gynecology and Obstetrics (FIGO) classification. In 2009, the FIGO revised the classification of EC. Apart from merging FIGO 1988 stages IA and IB, LN invasion has been stratified into pelvic LN invasion (stage IIIC1) and para-aortic LN invasion (stage IIIC2). The prognostic performances of the 1988 and 2009 FIGO staging systems have been compared using the concordance indexes. The FIGO staging systems were not significantly different . However, other studies have suggested that the 2009 FIGO staging system for EC is highly prognostic , particularly because stages IIIC1 and IIIC2 have different prognoses (5-year overall survivals are 57% and 49%, respectively )
An independent prognostic factor?
Primary pathological characteristics (myometrial invasion, cervical stromal invasion, histological type, lymphovascular space invasion (LVSI), and grade) are associated with the occurrence of LN metastasis. Consequently, we cannot exclude the negative impact of uterine risk factors associated with LN metastasis on survival independently of LN involvement . To answer this question, Barrena Medel et al. compared the prognostic significance of uterine risk factor and nodal metastases and determined the independent effects on outcome . Their results suggest that LN metastasis is an independent and the major prognostic factor, as survival in patients with LN metastasis and without uterine risk factors was worse when compared with patients without LN metastasis and with uterine risk factors.
Another approach to determining the prognostic impact of LN metastasis consists in building a score to predict survival in EC and in analyzing the impact of lymphadenectomy on the accuracy of this score. Such a score has been developed by Abu Rustum et al. and been validated externally . This score was designed to predict the 3-year survival probability, and it comprised the LN status. More precisely, the number of negative LNs was taken into account in this score as well as the 1988 FIGO stage, age at diagnosis, final FIGO grade, and histologic subtype. In this score, the LN status was considered for LN metastatic risk assessment as the FIGO classification was one of the nomogram components, and consequently LN-positive cases were assigned stage IIIC. Based on their preliminary findings that patients who had more regional LNs examined were likely to be more accurately assigned to the correct FIGO stage and because adding positive LNs in the multivariate model did not increase the score’s predictive accuracy, the authors decided to include the number of negative LNs into the multivariate model. This characteristic remained significantly associated with survival.
LN ratio
LN ratio is defined as the number of metastatic LNs to the total number of removed LNs. This parameter incorporates not only the burden of nodal disease and cancer spread but also the extent and quality of surgical staging. Polterauer et al. reported that patients with LN ratios ≤10%, >10–50%, and >50% had 5-year overall survival rates of 79%, 61%, and 36%, respectively ( P < 0.001). In multivariable analysis, only the LN ratio was associated with both progression-free survival and overall survival.
Prognostic value of LN metastasis
LN metastasis constitutes one of the most important prognostic factors of EC. In 1988, this information was included in the International Federation of Gynecology and Obstetrics (FIGO) classification. In 2009, the FIGO revised the classification of EC. Apart from merging FIGO 1988 stages IA and IB, LN invasion has been stratified into pelvic LN invasion (stage IIIC1) and para-aortic LN invasion (stage IIIC2). The prognostic performances of the 1988 and 2009 FIGO staging systems have been compared using the concordance indexes. The FIGO staging systems were not significantly different . However, other studies have suggested that the 2009 FIGO staging system for EC is highly prognostic , particularly because stages IIIC1 and IIIC2 have different prognoses (5-year overall survivals are 57% and 49%, respectively )
An independent prognostic factor?
Primary pathological characteristics (myometrial invasion, cervical stromal invasion, histological type, lymphovascular space invasion (LVSI), and grade) are associated with the occurrence of LN metastasis. Consequently, we cannot exclude the negative impact of uterine risk factors associated with LN metastasis on survival independently of LN involvement . To answer this question, Barrena Medel et al. compared the prognostic significance of uterine risk factor and nodal metastases and determined the independent effects on outcome . Their results suggest that LN metastasis is an independent and the major prognostic factor, as survival in patients with LN metastasis and without uterine risk factors was worse when compared with patients without LN metastasis and with uterine risk factors.
Another approach to determining the prognostic impact of LN metastasis consists in building a score to predict survival in EC and in analyzing the impact of lymphadenectomy on the accuracy of this score. Such a score has been developed by Abu Rustum et al. and been validated externally . This score was designed to predict the 3-year survival probability, and it comprised the LN status. More precisely, the number of negative LNs was taken into account in this score as well as the 1988 FIGO stage, age at diagnosis, final FIGO grade, and histologic subtype. In this score, the LN status was considered for LN metastatic risk assessment as the FIGO classification was one of the nomogram components, and consequently LN-positive cases were assigned stage IIIC. Based on their preliminary findings that patients who had more regional LNs examined were likely to be more accurately assigned to the correct FIGO stage and because adding positive LNs in the multivariate model did not increase the score’s predictive accuracy, the authors decided to include the number of negative LNs into the multivariate model. This characteristic remained significantly associated with survival.
LN ratio
LN ratio is defined as the number of metastatic LNs to the total number of removed LNs. This parameter incorporates not only the burden of nodal disease and cancer spread but also the extent and quality of surgical staging. Polterauer et al. reported that patients with LN ratios ≤10%, >10–50%, and >50% had 5-year overall survival rates of 79%, 61%, and 36%, respectively ( P < 0.001). In multivariable analysis, only the LN ratio was associated with both progression-free survival and overall survival.
Locations of LN metastasis
The incidence of metastases to the pelvic LNs in patients with corpus-confined EC who undergo lymphadenectomy varies between 5% and 18% . Para-aortic LN metastases are observed in 3–11% of patients with early stageEC depending on the primary tumoral characteristics, and 29–67% of patients with pelvic LN metastasis have para-aortic involvement . On the contrary, only a small percentage of patients with negative pelvic LN have para-aortic invasion . Although a direct route of lymphatic spread from the corpus uteri to the para-aortic through the infundibulopelvic ligament has been suggested from anatomical studies , direct metastases to the para-aortic LN is uncommon, ranging from 0% to 6% .
Mariani et al. have evaluated LN metastatic risk with distinction between the pelvic and para-aortic areas . In their prospective study, pelvic and para-aortic lymphadenectomies were systematically performed in 310 patients with EC (the exclusion criteria were as follows: grade 1 or 2 endometrioid type with myometrial invasion ≤50% and primary tumor diameter ≤2 cm). Among those patients, 57 patients (22%) had LN invasion, and this concerned the pelvic area in 19 cases (33%), the para-aortic area alone in nine cases (16%), and both the pelvic and the para-aortic areas in 29 cases (51%). Because of the low number of patients with metastatic LN in this study (and in patients with EC more generally), the results of this study should be interpreted with caution.
Odagiri et al. recently focused on the precise mapping of LN metastasis sites in EC . Among the 42 patients with positive LN metastasis, the pelvic area alone was involved in 16 cases (38%), the para-aortic area alone in seven cases (16%), and both the pelvic and para-aortic areas in 19 cases (45%). One-third of positive para-aortic LNs were above the inferior mesenteric artery (IMA) without positive LN below IMA. Metastasis to the deep inguinal nodes was found to be extremely rare (0.4%).
Consequently, for an optimal surgical staging, resection of the deep inguinal nodes is not recommended, whereas para-aortic lymphadenectomy should be extended up to the level of the renal veins. Moreover, because metastasis to circumflex iliac nodes distal to external iliac nodes (CINDEIN) is extremely rare in patients with positive LN , those nodes should not be systematically removed.
Preoperative identification of LN metastasis
Pelvic and para-aortic LN dissection is the recommended method for LN staging in EC. However, the probability of LN involvement is low and, considering the invasive nature of lymphadenectomy, avoidance of unnecessary lymphadenectomy is mandatory.
Magnetic resonance imaging
Few studies report the accuracy of magnetic resonance imaging (MRI) for predicting LN metastasis. The major limitation is using size criteria alone as thresholds for identifying metastatic LNs based on enlargement (usually 10 mm). This size criterion was used in three out of the four studies that showed similar diagnostic accuracy for MRI and sentinel-node biopsy in women with EC in the meta-analysis published by Selman et al. This meta-analysis included 18 studies and 693 women. A limitation with this meta-analysis is that the included studies did not use diffusion-weighted imaging. LN-specific magnetic resonance contrast agents have been developed to improve LN staging of pelvic cancers. Sensitivity has been reported to be between 17% and 66%. Specificity appears to be better (73–99%), and as LN metastasis is a considerably rare event in EC, the negative predictive value (NPV) is excellent (95–98%).
Computed tomography scan
Few studies have evaluated the accuracy of computed tomography (CT) scan for predicting LN metastasis in EC. The accuracy for LN metastasis ranges from 67% to 88%, with a correct NPV (81–94%) but poor sensibility and positive predictive value (PPV) (57% and 31–50%, respectively) .
Positron emission tomography/CT
Positron emission tomography (PET)/CT with fluorine-18 fluorodeoxyglucose (18F-FDG) has been used for staging with different results. The results of a recent meta-analysis including 16 studies with pooled diagnostic indexes for LN staging were calculated. LN status based on pathology was compared to the results of the PET/CT. Unfortunately, the type of metastasis (micro- or macrometastases) was not reported. We presume that only macrometastases were considered in the studies included in this meta-analysis, as the detection and diagnosis of micrometastasis are not performed in routine practice. The results from this meta-analysis are shown in Table 1 . Only patient-basis data are presented.
| Sensitivity,% | Specificity,% | LR+ | LR− | DOR | |
|---|---|---|---|---|---|
| LN overall | 72 (64–80) | 93 (91–95) | 8.4 (5.9–11.8) | 0.4 (0.2–0.5) | 27.7 (15.8–48.6) |
| Pelvic LN | 61 (48–73) | 97 (95–99) | 15.4 (8.6–27.5) | 0.4 (0.2–0.9) | 42.0 (18.8–93.6) |
| PA LN | 87 (66–97) | 99 (97–100) | 46.6 (18.4–117.9) | 0.1 (0.05–0.4) | 309 (70.6–1352) |
The pooled sensitivity of 18F-FDG for detection of LN involvement was low, which reduces its interest in LN staging of EC. However, the specificity was considerably better. In other words, the result of positive 18F-FDG PET in the LN is reliable. Considering pelvic and para-aortic LNs separately, the accuracy of 18F-FDG PET was higher for para-aortic LN staging when compared with the pelvic region.
PET/CT has been reported to be more useful than PET alone, its CT component yielding helpful anatomical and morphological information. However, subgroup analyses regarding PET versus PET/CT instruments showed higher sensitivity and lower specificity for PET compared with PET/CT studies. These differences were not statistically different between PET and PET/CT studies ( P = 0.3 for sensitivity and P = 0.8 for specificity) .
Altogether, the sensitivity of 18F-FDG PET imaging is not sufficient to justify its routine use for preoperative LN staging in EC. On the other hand, the specificity of 18F-FDG PET imaging is adequate. Further studies with large sample size as well as consistently applied pelvic and para-aortic lymphadenectomy are definitely needed to be able to draw further specific conclusions.
Recently, fused PET/MRI images of EC have been compared with 18F-FDG contrast-enhanced PET/CT and MRI for the prediction of LN metastasis . Despite patient-based sensitivity, specificity, and accuracy for detecting pelvic nodal metastasis of 100%, 96%, and 97% for both fused PET/MRI, the low number of patients included (30 cases) makes it a non-recommendable systematic radiological examination method for the management of patients with EC.
Predicitng LN metastasis
Primary tumoral characteristics, demographical characteristics, and CA125 level are associated with LN metastasis in EC. The strongest association is observed with primary tumoral characteristics that include local extension (myometrial and stroma cervical invasion), subtype, grade, diameter, and LVSI. Most of these characteristics are assessable preoperatively (local extension, subtype, grade, and diameter) excluding LVSI status. In particular, many studies reported on the accuracy of MRI for the evaluation of the locoregional extension of EC. A systematic review that included 11 studies and 548 women reported on the detection of deep myometrial invasion . The analysis concluded that contrast-enhanced MRI has a good diagnostic performance in assessing the myometrial invasion of EC and is superior to T2-weighted imaging alone. For assessing deep myometrial involvement, its NPV appears relatively high, and negative findings strongly suggest the absence of deep myometrial involvement. However, both contrast-enhanced and T2-weighted have suboptimal PPV. New imaging sequences, such as diffusion-weighted imaging, were only recently introduced and have not been evaluated enough yet. Ultrasound has similar performances for predicting myometrial invasion with accuracy between 67% and 84% . Concerning the prediction of cervical stromal invasion, the results of MRI are more heterogeneous and less satisfying: accuracy between 46% and 92%, poor sensitivity, and poor PPV. Few studies have evaluated the accuracy of CT scan for the evaluation of the locoregional extension of EC. The accuracy of CT scan for assessment of the outer half of the myometrium ranges between 62% and 87%, and very uneven sensibility and specificity are reported (10–83% and 42–100%, respectively) . Concerning cervical stromal invasion, the performance of CT scan is insufficient, with low sensibility and PPV (20–25% and 9–14%, respectively) .
Scores to predict LN metastasis
Several authors have proposed different scores in order to predict LN metastasis ( Table 2 ). Most of these use primary tumoral extension (myometrial invasion particularly), and a cut point was chosen to maximize the NPV and minimize the risk of mislabeling patients with nodal involvement. For all scores, this strategy permits to identify a low-risk group of patients accounting for approximately half of patients in whom the LN metastatic probability is extremely low (≤3%). Unfortunately, most of these scores have not been validated externally, which reduces their generalizability. Similarly, some of those scores are based on definitive pathological results and are only applicable once hysterectomy has been performed.
| Score | Components | Evaluated | Externally validated | Low risk |
|---|---|---|---|---|
| Kang et al. (2012) | CA125, MRI locoregional extension (myometrial invasion, LN enlargement, and extension beyond uterine corpus) | Preoperatively | Yes | 43% 1.4% |
| Lee et al. (2010) | CA125, grade, MRI locoregional extension (disease extent and myometrial invasion) | Preoperatively | No | 57% 0% |
| Kamura et al. (1999) | Pathological local extension (tumor diameter and myometrial invasion) | Postoperatively | No | Not reported |
| Todo et al. (2007) | CA125, tumor grade/histology, MRI local extension (volume index) | Preoperatively | Yes | 54% 3% |
| Luomaranta et al. (2013) | Biochemical factors (leukocytosis, thrombocytosis, and CA125 level) and preoperative tumor characteristics | Preoperatively | No | Not reported |
| Bendifallah et al. (2012) | Pathological hysterectomy characteristics (subtype; grade and myometrial and cervical invasion), age, and race | Preoperatively and Postoperatively | Yes | 54% 0% |
| Milam et al. (2012) | Uterine pathology characteristics (myometrial invasion, tumor size, and differentiation) | Postoperatively | No | 40% 0.8% |
Group risk based on LN metastatic probability
In 2000, Mariani et al. proposed an algorithm based on readily ascertainable intraoperative pathologic indicators that would discriminate a subgroup of early EC, which would not require lymphadenectomy or adjuvant radiotherapy . A total of 328 patients with endometrioid EC, grade 1 or 2 tumor, myometrial invasion ≤50%, and no intraoperative evidence of macroscopic extrauterine spread were treated surgically. Pelvic lymphadenectomy was performed in 187 cases (57%), and nodes were positive in nine cases (5%). The primary tumor diameter and lymphatic or vascular invasion significantly affected longevity. No patient with tumor diameter ≤2 cm had positive LNs or died of disease. Convery et al. have validated the Mayo algorithm in a multicenter retrospective chart review . Of 110 patients satisfying the Mayo algorithm with an adequate lymphadenectomy, 2% were diagnosed with LN metastasis and 4% subsequently developed recurrent disease. The Mayo algorithm identified women who would not benefit from a lymphadenectomy with a 98% NPV. There was no significant difference in the recurrence rate between women who underwent lymphadenectomy and those who did not when the Mayo algorithm was satisfied. Although a small number of patients with advanced-stage disease may be missed when applying the Mayo criteria, there is no apparent survival benefit to lymphadenectomy for patients satisfying this algorithm, and these data support its use at other institutions.
In 2013, the European Society for Medical Oncology (ESMO) separated ECs into three risk groups .
- –
Low risk: stage IA, grade 1 or 2
- –
Intermediate risk: stage IA, grade 3 and stage IB, grade 1 or 2
- –
High risk: stage IB, grade 3 or type 2
Because of the low-risk LN metastatic probability, the ESMO does not recommend lymphadenectomy practice in the low-risk group. For the intermediate- and high-risk groups, no recommendation is made by the ESMO.
Group risk based on biomarkers to predict LN metastatic probability
Other approaches have been conducted in an attempt to estimate the LN metastatic risk . For example, double-negative hormone receptor status in EC curettage has recently been suggested to independently predict LN metastasis . In the study by Trovik et al., 26% of patients with combined loss of estrogen and progesterone receptor expression had LN metastasis. Similarly, stathmin immunohistochemical staining appears to be able to identify EC with LN metastases . The value, as a predictive marker for response to PI3Kinase inhibition and as a tool to stratify patients for LN sampling, remains to be determined. Unfortunately, the impact of lymphadenectomy in terms of survival has not been compared in high-risk groups based on such biomarkers. Due to the relatively low frequency of LN metastasis and recurrence in low-risk groups, adequately powered randomized surgical trials have been difficult to conduct. Improved tools (such as biomarkers) identifying patients with high risk for LN metastasis could reduce the required sample size of a randomized clinical trial assessing the survival impact of lymphadenectomy.
A part from those biomarkers, LVSI, which is thought to be the beginnings of lymphatic and hematologic metastases, has been evaluated in several studies. The interobserver variability in the evaluation of lymphatic and blood vessel invasion cannot be neglected because of the difficulties in recognizing lymphatic channels and blood vessels using standard hematoxylin-and-eosin staining alone . This could explain why reported rates of LVSI in EC are particularly variable (from 14% to 20% in stage I tumors and between 4% and 37% for all EC stages) . Even in prospective studies, LVSI determination is not systematically performed by pathologists. In such cases, two possibilities are raised: the pathologist did not mention the presence or absence of LVSI because he did not observe any LVSI or because there was no LVSI. Previous studies evaluating LVSI have shown mixed results concerning the impact of LVSI on the recurrence rate . Despite those contradictory results, distinction by LVSI status has recently been suggested to be more relevant than the distinction between stages IA and IB for predicting survival in stage I EC .
Concerning the impact of LVSI on LN metastasis, the association has been suggested in numerous studies and LVSI should be considered to be an independent risk factor for LN metastasis. Interestingly, in a multicenter study, the risk of LN metastasis is similar when the LVSI is negative or is not detailed in the pathological report .
Unfortunately, LVSI status is only available once hysterectomy is performed. Because the presence of LVSI increases to a large extent, the decision to perform lymphadenectomy due to the presence of LVSI on the hysterectomy specimen is practical from the perspective of removing the LN metastasis.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
