• © 2005 by American Society of Clinical Oncology

Lymphadenectomy in Apparent Early-Stage Endometrial Carcinoma: Do Numbers Count?

Endometrial carcinoma is the most common gynecologic malignancy in Western countries, with an overall annual incidence in the United States of 25 per 100,000, increasing to more than 85 per 100,000 in women 60 years or older. Because the majority of patients present with early-stage disease, the prognosis of endometrial carcinoma patients is generally good, with 5-year overall and cancer-specific survival rates of 80% to 85% and 90% to 95%.^1-4 Endometrial carcinoma is primarily a cancer of postmenopausal women, with concurrent morbidities such as obesity, hypertension, and diabetes. Given the favorable outcome for the majority of cases, the challenge is to effectively select those patients at increased risk of relapse who might benefit from more extensive surgical procedures and adjuvant therapies, and to avoid overtreatment of low-risk cases that would be exposed unnecessarily to the risk of excess morbidity.

Since the Gynecologic Oncology Group (GOG)-33 staging study,^5,6 which reported an overall 9% risk of pelvic lymph node metastases in clinical stage I endometrial cancer (25% for cases with outer-third myometrial invasion and 18% for grade 3 disease), gynecologic oncologists have advocated standard lymphadenectomy or sampling procedures for all intermediate- and low-risk endometrial cancer patients. It should be noted that GOG-33 included 12% patients, who at laparotomy were found to have macroscopic extrauterine disease; of these, 51% had pelvic metastases, and 23% had aortic node metastases. Among the patients without gross intraperitoneal spread, only 7% and 4% had positive pelvic or aortic nodes, respectively. Women at high risk (> 10%) for pelvic node metastasis were those with grade 3 and outer-third invasion, while all other cases had low to moderate (< 5% to 10%) risk.⁶ This is in accordance with the COSA-NZ-UK trial,⁷ which reported an overall 7% rate of pelvic node metastasis in corpus-confined disease with any of the following features: grade 3; adenosquamous, serous or clear-cell histology, or greater than one-third myometrial invasion. In the GOG-99 trial,¹ patients were eligible only if they had no nodal disease. In the so-called high-intermediate–risk group (which was characterized by age ≥ 70 years with any one of the following risk factors: tumor grade 2 or 3, outer-third myometrial invasion, or presence of lymph-vascular space invasion; age at least 50 years with any two of the other risk factors; or any age with all three other factors), the rates of relapse were 27% without radiation therapy (RT) and 13% with RT, even after lymphadenectomy and confirmed node-negative disease.

An important factor to consider is the optimal extent of lymph node dissection. Should studies set a standard as to the number of nodes to be removed? GOG-99 did not require a minimum number of nodes; however, the GOG panel did evaluate the operative reports for completeness of lymphadenectomy and excluded patients with only unilateral lymph node sampling. Previous studies reporting improved outcome after lymphadenectomy were retrospective single-center studies with selection bias and stage migration.^8-10 Kilgore et al¹⁰ reported a mean number of 11 nodes in patients with multiple site sampling and four nodes in cases with limited node sampling. In the large population-based National Cancer Institute series,¹¹ the median number of lymph nodes sampled was seven. No survival difference was found between 6,363 women with no lymph node sampling and 2,821 with node sampling. However, a survival advantage was associated with node sampling in patients with stage I grade 3 disease.

With lymph node dissection, concerns include the prolongation of operating time, with potential short-term and long-term side effects, and the increased morbidity if more extensive surgery is combined with external-beam radiation therapy¹² or chemotherapy. In GOG-99, the crude 4-year incidence of grade 3 gastrointestinal (GI) complications was 8% in the radiotherapy arm,¹ as compared with an actuarial 5-year rate of 3% in the radiotherapy arm of the PORTEC study without lymphadenectomy.² Chronic morbidity directly related to lymphadenectomy, such as leg edema, occurs in at least 5% of patients.⁸ Para-aortic lymphadenectomy further increases the risk of major complications, but might also identify patients requiring extended-field radiotherapy. A major controversy remains as to whether lymphadenectomy is purely diagnostic or if it is therapeutic, and if radiation therapy could be omitted for all node-negative patients with high risk factors, as the only randomized study still showed a 27% relapse rate without radiotherapy.¹

In this issue of the Journal of Clinical Oncology, Cragun et al¹³ present the results of a large retrospective analysis of the effect of the extent of selective lymphadenectomy on survival and morbidity in women with apparent early-stage endometrial carcinoma. They analyzed 565 patients who were treated with hysterectomy and selective lymphadenectomy (419 pelvic and aortic, 146 pelvic), of a total of 1,656 patients treated between 1973 and 2002. The analysis focuses on the comparison of patients with 11 or fewer to more than 11 pelvic nodes removed, with 11 being the median number of pelvic nodes removed. Patients with at least one node identified by final pathology were included in the analyses. Patients with visible metastatic disease (including gross nodal involvement) at laparotomy were excluded. The rates of pelvic and aortic nodal metastases were 5% and 3%, lower than those in other series^6,7 due to the exclusion of patients with grossly involved nodes. The survival analysis showed a significant survival benefit for patients with grade 3 cancers who had more than 11 pelvic nodes removed, as compared with those with grade 3 disease who had 11 or fewer nodes removed (hazard ratio [HR], 0.25; P < .0001). Patients with grade 1 to 2 disease did not have a survival difference based on the number of nodes removed. Aortic node removal did not significantly influence survival (HR, 1.29). Removal of patients with pelvic or aortic nodal metastases from the analysis still resulted in a survival benefit among patients with grade 3 cancers who had more than 11 nodes removed (HR, 0.20; P < .0001). A highly significant survival difference was found for high-risk patients (grade 3, > 50% invasion, or serous or clear-cell histology), with 82% 5-year survival for those with more than 11 nodes, as compared with 64% for those with 11 or fewer nodes (P = .001). Among low-risk patients, the survival difference (94% v 89%) was not statistically significant. Patients with 11 or more nodes removed had a lower rate of pelvic recurrence (1% v 5%; P = .02) and similar rate of vaginal recurrence (2% v 3%). Although there was no significant difference in complication rates between patients undergoing pelvic and aortic lymphadenectomy and those undergoing pelvic lymphadenectomy alone, the addition of aortic lymphadenectomy resulted in longer anesthesia time, longer hospital stay, more blood loss, and higher transfusion rates. This should be weighed against the potential therapeutic benefit of aortic lymphadenectomy and extended-field radiation for the small subset of patients (3%) with aortic metastases, resulting in a 31% 5-year overall survival rate, which is in concordance with other studies.^14-16

Adverse events were reported in 18% of the patients, with small bowel obstructions requiring medical and surgical treatment in 2.6% and 1.8%, respectively; 2.6% deep venous thrombosis; one death within 30 days of surgery (unknown cause), another death 2 months after surgery due to subarachnoid hemorrhage after anticoagulation for deep venous thrombosis, and a third death after 8 months due to gastrointestinal bleeding following whole abdominal radiation therapy. Data on long-term morbidity are lacking, especially on the incidence of leg edema.

The results of this analysis are important and relevant for clinical practice. The role of lymphadenectomy remains a topic of continuing debate, and the selection of patients for lymphadenectomy, the extent of lymphadenectomy (pelvic and/or aortic), and whether a minimum number of nodes should be required to consider lymphadenectomy adequate are unresolved issues, as data from randomized studies are lacking. The analysis by Cragun et al suggests that lymphadenectomy should be considered for patients with grade 3 cancers, but no benefit is obtained for grade 1 to 2 disease. Lymphadenectomy should yield more than 11 pelvic nodes from multiple sites.

Strengths of the study are the large number of patients, the thoroughness of the analysis with specification of patient and tumor characteristics for all patients treated during the period studied, and the reporting of adverse effects. Due to the retrospective nature of the analysis and the fact that a minority of patients treated during the study period was included in the analysis, selection bias cannot be ruled out. The 5-year overall survival of 83% for the study group, including 56% patients with stage IA or IB disease, was not different from other series, with or without lymphadenectomy. There was a clear association between treatment period and median number of pelvic nodes removed (nine between 1973 and 1987 v 14 between 1988 and 2002). The selection of patients for lymphadenectomy was based on surgeon preference, tumor factors, and patient comorbidities. All cases with at least one node identified were included, but a single-node biopsy can hardly be regarded as a lymphadenectomy. The cutoff of 11 nodes is relatively arbitrary, though supported by the statistical analysis. Total node count is not only a reflection of completeness of lymphadenectomy, but is also dependent on the pathologist's evaluation of the surgical specimen. Total node count and thorough evaluation of all lymph nodes in the specimen depend on the pathologist's macroscopic evaluation, and are therefore subject to bias and are not amenable to independent pathology review.

An intriguing finding is the persistence of improved survival for removal of more than 11 nodes among patients with grade 3 cancers, when those with proven nodal metastasis were excluded from the analysis. The authors hypothesize that the removal of occult (micro)metastatic disease that remains undiagnosed by the pathologist accounts for this therapeutic benefit. This raises the question as to whether the sentinel node concept would be the ideal solution to the problem of weighing the potential benefit of extensive lymphadenectomy limited to some patients, against the added toxicities to which all patients are exposed. In breast cancer, it has been shown that removal and meticulous analysis of sentinel nodes with serial sectioning and immunohistochemistry more adequately identifies occult nodal involvement, while sparing the majority of patients without nodal metastases the risks of more extensive procedures. First results from sentinel node studies in endometrial carcinoma^17-19 have shown that the combined use of radiocolloid labeling and patent blue results in an sentinel node detection rate of 82% to 94%, with identification of two to three sentinel nodes per patient, which are located in the obturator, external iliac, and/or para-aortic regions. In the study by Barranger et al,¹⁸ macrometastases were identified in three sentinel nodes from two patients, while immunohistochemical analysis identified six micrometastatic sentinel nodes and one sentinel node containing isolated tumor cells in three other patients. No false-negative sentinel nodes were observed. Additional larger studies are needed to further explore this concept and develop reliable sentinel node detection procedures, which might greatly benefit the management of patients with endometrial cancer by adequately detecting microscopic spread, while sparing most patients an unnecessary and potentially harmful lymphadenectomy. Patients with proven nodal involvement can then be enrolled on studies of radiation therapy with concomitant chemotherapy to further improve their outcome.

The first multicenter randomized trial to determine the value of standard lymphadenectomy for stage 1 and 2 endometrial carcinoma, the UK Medical Research Council-ASTEC study, is nearing accrual completion and will hopefully provide definitive, unbiased information as to the role of lymphadenectomy in intermediate- to high-risk endometrial cancer. In the light of the current findings, its analysis should include clear data on the completeness of lymphadenectomy and the number of nodes removed.