Lymphadenectomy or sentinel node biopsy for the management of endometrial cancer

Endometrial cancer, which affects the lining of the uterus, is the most common form of uterine cancer (96%), and the sixth most common cancer in females worldwide, accounting for 4.5% of all cancers in females. In 2022, there were 420,242 cases of uterine cancer and 97,704 deaths from the disease worldwide. Most women have early-stage endometrial cancer at diagnosis. Traditionally, surgical staging included removal of all lymph nodes (lymphadenectomy) in the pelvis (pelvic lymphadenectomy) with or without para-aortic areas (pelvic/para-aortic lymphadenectomy), to determine the need for further treatment. However, rates of lymph node involvement are relatively low and may be predicted by uterine histopathology and molecular markers. Lymphadenectomy carries a significant risk of long-term morbidity from lymphoedema and previous studies comparing pelvic lymphadenectomy with no lymphadenectomy found no survival benefit. Detecting the first draining lymph node(s) from each side of the uterus, called sentinel lymph node biopsy, can replace lymphadenectomy in terms of accuracy of detecting nodes, but no studies have shown whether sentinel lymph node biopsy is beneficial to women, despite its wide use.

To evaluate the benefits and harms of lymphadenectomy and sentinel lymph node biopsy for the management of endometrial cancer comparing different head-to-head comparisons in a network meta-analysis allowing ranking of treatment strategies.

We searched CENTRAL, MEDLINE, Embase, ClinicalTrials.gov and the WHO ICTRP for studies up to 22 March 2024.

We included randomised controlled trials (RCTs) of women with early-stage endometrial cancer, comparing combinations of no lymphadenectomy, pelvic lymphadenectomy, pelvic/para-aortic lymphadenectomy and sentinel lymph node biopsy. We excluded non-randomised studies and studies assessing diagnostic test accuracy of lymph node sampling.

Overall survival; progression-free survival; morbidity and mortality related to surgery; early and late adverse events, including lymphoedema and lymphocyst formation; and quality of life.

We used RoB 2 to assess risk of bias.

We conducted meta-analyses using random-effects models to calculate hazard ratios (HR) for time-to-event data and risk ratios (RR) and mean difference (MD) for other outcomes, with 95% confidence intervals (CI). We used GRADE to summarise the certainty of evidence. We intended to compare treatments in a network meta-analysis.

We included five RCTs (one remains ongoing) with 2074 women. Studies were conducted in the UK, South Africa, Poland, New Zealand, Chile, Italy, Egypt and Brazil, and published between 2008 and 2023. Another 10 studies are ongoing. Three studies (1955 participants) compared no lymphadenectomy with pelvic lymphadenectomy, one study (50 participants) compared no lymphadenectomy with pelvic/para-aortic lymphadenectomy, and one study (69 participants - ongoing) compared sentinel lymph node biopsy with pelvic/para-aortic lymphadenectomy.

No lymphadenectomy versus pelvic lymphadenectomy No lymphadenectomy probably results in little to no difference in overall survival (HR 0.85, 95% CI 0.66 to 1.10; 2 studies, 1922 participants; moderate-certainty evidence) and improves progression-free survival (HR 0.78, 95% CI 0.63 to 0.96; 2 studies, 1922 participants; high-certainty evidence) compared to pelvic lymphadenectomy. No lymphadenectomy may reduce early adverse effects from direct surgical morbidity slightly (RR 0.68, 95% CI 0.27 to 1.71; 3 studies, 1955 participants; low-certainty evidence) and probably reduces early adverse effects due to surgically related systemic morbidity (RR 0.28, 95% CI 0.09 to 0.93; 3 studies, 1955 participants; moderate-certainty evidence). No lymphadenectomy probably results in a large reduction in lymphoedema (RR 0.12, 95% CI 0.05 to 0.26; 3 studies, 1955 participants; moderate-certainty evidence) and likely reduces lymphocyst formation (RR 0.20, 95% CI 0.04 to 0.91; 1 study, 1403 participants; moderate-certainty evidence). There were no quality of life data. Sentinel lymph node biopsy versus pelvic/para-aortic lymphadenectomy One study shared unpublished data and the evidence is very uncertain about the effect of sentinel lymph node biopsy on overall survival, progression-free survival, early adverse events, lymphocyst formation and quality of life at 12 months. Sentinel lymph node biopsy compared with pelvic/para-aortic lymphadenectomy probably reduces the development of lymphoedema (RR 0.30, 95% CI 0.09 to 0.97; 1 study, 69 participants; moderate-certainty evidence). No lymphadenectomy versus pelvic/para-aortic lymphadenectomy One study closed after the recruitment of 50 participants due to slow uptake, and we were unable to extract data for use in the meta-analysis. Because of this, we were unable to form a linked network for meta-analysis. Other comparisons Studies of other comparisons are ongoing or results are yet to be published.

Data suggest 'less is probably more' in terms of surgical staging for women with presumed endometrial cancer, as no lymphadenectomy is favoured over pelvic lymphadenectomy in terms of important outcomes, with overall moderate certainty. Preliminary results for sentinel lymph node biopsy versus pelvic/para-aortic lymphadenectomy have a similar direction of effect, but the evidence is very uncertain. Data from several studies are ongoing. However, given the weight of evidence that supports no lymphadenectomy over lymphadenectomy, our ability to make adjuvant treatment decisions based on uterine factors, and the advent of molecular profiling, it is disappointing that only one study compared no lymphadenectomy with sentinel lymph node biopsy, potentially putting many women at continued risk of short- and significant long-term consequences of extensive lymphadenectomy.

This Cochrane review had no dedicated funding.

This review is based on an updated protocol including network meta-analysis methods and new RoB 2 assessment of a previously published review. Updated protocol 2023 available via https://doi.org/10.1002/14651858.CD015786.