Introduction

Endometrial cancer (EC) is the most common gynecological cancer and International federation of gynecology and obstetrics (FIGO) recommended surgical staging in 1988 due to limitation of clinical staging and updated the staging system in 2009 [1, 2]. In most clinics, pelvic lymph node (PLN) and para-aortic lymph node (PALN) dissection addition to total hysterectomy and bilateral salphingo-oophorectomy is a routine operative procedure in patients with poor prognostic factors such as cervical involvement, non-endometrioid histologic subtypes, or >1/2 myometrial invasion [3, 4]. By significant improvement in training and instrumentation and innovative adaptation of instruments and techniques, minimally invasive approaches for the treatment of women with all gynecologic malignancies had become popular [5].

Lymphatic spread is one of the ways of tumor dissemination and the presence of metastatic PLN and PALN is one of the most important factor related to stage and prognosis [6, 7]. Recent studies about sentinel LN in EC improved our knowledge about lymphatic spread in EC and obturator, internal iliac and the common iliac LNs were found to be the most common sites of metastasis [20–24]. However, little is known about the way of endometrial tumor’s lymphatic spread. Does the lymphatic spread follows the same route in all patients or does it change due to presence of different prognostoc factors? The answers are not clear yet and this issue still remains controversial [8–12].

The aim of this study was to evaluate whether the presence of cervical invasion has altered the site of LN metastasis in stage IIIC EC patients.

Materials and methods

Between 1993 and 2009, patients who had open systematic pelvic and para-aortic lymphadenectomy surgery extending to renal vessels for EC at gynecologic oncology department were studied retrospectively. All stage IIIC patients who had at least 15 LNs removed from their pelvic region and 10 LNs from para-aortic region were included in the study. Further, patients were divided into two groups according to cervical involvement. Patients with cervical invasion was defined as Group A and patients without cervical invasion was defined as Group B. Patients who had other malignancies diagnosed within 2 years before or after their diagnosis were excluded. All surgical procedures were performed by the same gynecologic oncology team and frozen sections (FS) by the same pathologists. FS was studied routinely for all cases and provided information on the grade of neoplasia, histology, depth of myometrial invasion, tumor diameter, sites of extrauterine spread and lymphovascular invasion.

Bilateral PLN and PALN dissections were performed when one of the following criteria was determined by frozen section examination:

  1. 1.

    Grade 2–3

  2. 2.

    ≥1/2 myometrial invasion

  3. 3.

    Tumor size >2 cm

  4. 4.

    Adnexial metastasis

  5. 5.

    Cervical involvement

  6. 6.

    High risk histopathology (clear cell, adenosquamous or serous papillary).

The staging was defined in accordance with the 1988 FIGO surgical staging system. Bilateral pelvic lymphadenectomy was performed to complete skeletonization, with all lymphatic tissue of the common, external and internal iliac vessels and the obturator fossa removed after visualization of the obturator nerve. The superior surgical dissection margin for the pelvic nodes was the aortic bifurcation, and the anterior distal surgical dissection margin was the circumflex iliac vein. The presacral lymphatic tissue was harvested separately. The upper limit of para-aortic lymphadenectomy was renal vessels. All lymphatic tissue was then harvested from all patients from the lateral, anterior, and medial aspects of the vena cava and aorta to the renal veins. The lymph-node-bearing tissue from the PLNs (bilateral common iliac arteries, external iliac arteries, internal iliac arteries, obturator fossa and presacral) and the PALNs was submitted for analysis. All surgeries and pathologic findings were performed and interpreted at a single institution. All relevant patients’ data, including history, surgical data and histological reports, were abstracted from the patients’ records.

Statistical analysis was performed using SPSS version 17.0. Chi-square and Annova table test were used to examine the effect of cervical invasion on LN metastasis. The cut-off value for statistical significance was set at p < 0.05.

Results

Between study period, 1,500 patients were operated for endometrial cancer at our hospital and 56.9% of them had lymphadenectomy. Total of 204 consecutive patients who had open systematic pelvic and para-aortic lymphadenectomy surgery extending to renal vessels and whose total number of harvested LN was adequate were evaluated. Among 204 patients, 46 fulfilled inclusion criteria and were included in the study. The mean age and tumor size of patients were 59 years (range 38–81) and 47 mm (range 10–80), respectively. Twenty-three patients had cervical involvement (Group A) and 23 had no cervical metastasis (Group B). Of these 23 patients with cervical involvement 4 had glandular, 19 had glandular and stromal invasion. Clinicopathologic features of all patients are shown in Table 1.

Table 1 Surgical and pathologic characteristics of patients
Full size table

Tumor was disseminated to PALN in 26 patients and to PLN in 41 patients. Obturator LN was the most and presacral LN was the least common site of PLN involvement (Table 1).

The mean number of removed LNs was 21.4 (range 9–41) for PALN, 41.6 (range 23–78) for PLN region and number of metastatic LNs was 4.7 (range 1–30) for PALN, 4.5 (range 1–35) for PLN. Detailed distribution of removed and metastatic LNs is shown in Table 2.

Table 2 Number of removed and metastatic lymph nodes
Full size table

Furthermore, groups were compared with regard to cell type, grade, depth of myometrial invasion, tumor size, adnexal involvement, peritoneal metastasis and lymphovascular space invasion. The difference was not statistically significant, however, patients in Group A were younger than patients in Group B (55.8 vs. 62, p = 0.017) (Table 3). The statistical analysis was evaluated to determine whether the age had an effect on lymph node spread. When the mean age was set as cut-off value, statistical analysis showed that age had no effect on lymph node spread pattern (for para-aortic region p = 0.678, pelvic region p = 0.262, common iliac LN p = 0.809, external iliac LN p = 0.446, internal iliac LN p = 0.708 and obturator LN p = 0.264).

Table 3 Distribution of surgical and pathologic factors (%)
Full size table

Mean number of removed LNs was similar between groups. The relation between cervical invasion and site of LN metastasis was evaluated. Cervical invasion was found to have increased the rate of external iliac LN metastasis from 21.7 to 60.9% (p = 0.07). Cervical invasion has increased the risk of PLN and obturator LN involvement from 82.6 to 95.7% and 39.1 to 52.2%, respectively (p = 0.155 and p = 0.375). Moreover, internal iliac and PALN metastasis were found to have decreased in patients with cervical invasion, but the difference was not significant (Table 4).

Table 4 Distribution of removed and metastatic LN number according to groups
Full size table

We analysed whether presence of glandular or stromal involvement had changed lymph node spread pattern. Our results showed that glandular or stromal involvement did not change the pattern of lymphatic spread. P values for each region was: para-aortic region p = 0.231, pelvic region p = 0.639, common iliac LN p = 0.651, external iliac LN p = 0.106, internal iliac LN p = 0.957, obturator LN p = 0.315 and presacral LN p = 0.497.

Discussion

EC is the most common gynecologic cancer and in approximately 75% of the cases, the tumor is confined to the uterus at the initial diagnosis [13]. Patients with cervical invasion may present with stage II, III, or IV disease (FIGO 2009) [2]. Of note, histologically documented cervical involvement is relatively uncommon and accounts for only 11% of all endometrial cancers [14]. Five-year survival of patients with EC involving the cervix is approximately 64% [15].

Lymphatic spread is the main way of dissemination of gynecologic cancers but the route of spread may differ between malignancies. For primary cervical cancer, lateral pathway flowing to obturator, internal iliac, external iliac and common iliac LNs is known to be the most important way. Also anterior pathway flowing to external iliac LNs and posterior pathway flowing to common iliac, sacral and PALNs are the other ways for lymphatic spread in primary cancer of cervix [16, 17]. Is the route of spread is similar in patients with cervix invaded EC? Studies about sentinel LN in EC were performed to find the answer [18–20]. First, it was thought that EC with cervical involvement may have similar lymphatic tumor dissemination to primary cervical cancer, however, further studies revealed that the pattern of tumor dissemination was quite different. Obturator, internal iliac and common iliac LNs were found to be the most common involved sites in EC [21, 22]. In the present study, similarly obturator LN was the most common site of metastasis, however, external iliac LN was found to be the most common site when cervix was invaded.

In 2001, Mariani et al. [23] compared patients with EC limited to the uterine corpus and invading the cervix; and found that EC with cervical metastasis spread more readily to the common iliac LNs. Furthermore, they concluded that para-aortic LN metastases spread via a route shared by the common iliac LNs when tumor involves the cervix but spread predominantly via a route common to the obturator LNs (and/or external iliac LNs) when the primary tumor site is only the corpus.

In addition, in 2008 Mariani et al. [24] reported that ovarian vessels are one of the way of tumor spread to para-aortic region and further concluded that tumor localized to fundus and corpus uteri may spread more frequently to PALN compared to tumors localized to isthmus uteri. Similarly, we found 17% less PALN metastasis in patients with stage IIIC cervix invaded EC.

Cervical invasion is considered to increase the risk of nodal metastasis and parametrial invasion [25]. However, in a current study it was reported that parametrial spread cannot be predicted by cervical involvement alone but may be predicted by various lymphovascular space invasion-related histopathologic factors and parametrial spread may not be an independent prognostic factor in individuals with uterine endometrial cancer [25].

Preoperative cervical evaluation for predicting cervical involvement in EC was studied and found to predict cervical metastasis in most cases [26]. Furthermore, authors have suggested pap smear, colposcopy, cervical palpation and rectal parametrial examination prior to surgery; however, this procedure is not routinely performed in clinical practice at present [26]. Of note, extension of cervical invasion was measured and impact on recurrence and survival was studied but no relation was found [27].

Retrospectivity and small number of patient are the limitations of our study, but considering the number of removed LNs as inclusion criteria and performing PALN dissection up to the renal vessels in all patients is the favorable sides of our study. Since it would not be reliable to comment about LN metastasis in a patient who had only one LN removed, patients who had <15 removed LNs from the pelvic region and <10 LNs from the para-aortic region were not included in the study. Mean 62.9 LNs removed from each patient was considered to be enough to comment about LN status. Also surgical and pathologic features including total number of removed LNs were similar between groups. This was also a desirable object for exact statistical analysis.

Of note, in our clinic, accordance of LVSI determined by FS and final pathology was evaluated and results showed that LVSI may be determined by FS with a 50% of sensitivity, 100% of specificity, 94.4% of NPV and 100% of PPV.

As we know, LN dissection increases the morbidity compared with TAH+BSO alone and therefore patient selection who will undergo LN dissection is important. From this point of view, much effort has been expended to define the factors affecting the route of lymphatic spread in EC up to now because once the predictive factors and the route are clarified the question of “In which patients and to what extent should we perform LN dissection” can be resolved. The “ASTEC trial”, a randomised prospective large trial, was set out to determine whether pelvic lymphadenectomy could improve the survival of women with EC. The lymphadenectomy study concluded that there was “no evidence of benefit in terms of overall or recurrence-free survival for pelvic lymphadenectomy in women with early endometrial cancer,” and that “pelvic lymphadenectomy cannot be recommended as routine procedure for therapeutic purposes outside of a clinical trial” [28]. However, some authors claimed that this conclusion cannot be made because a systematic lymphadenectomy was not performed in most patients in the lymphadenectomy [29]. Also, other authors criticised that the risk factors were not equalized in the two arms of the study [30]. As can be seen, there is still no consensus about indication and extent of lympadenectomy in EC. This dilemma should be more clarified until the risk factors for LN metastasis and the route of lymphatic spread were determined.

In summary, our study revealed a slight change in LN involvement whether cervix was invaded. External iliac LNs were more common and PALNs were less common sites of LN metastasis in patients with cervix invaded stage IIIC EC. However, obturatuar LNs were the most common involved LNs in patients without cervical invasion. Large prospective studies are needed to clarify the alteration of LN metastasis in cervix invaded EC patients.