Introduction

Lymph node (LN) evaluation in endometrial cancer is controversial [1]. The data regarding its use are limited and practice varies between surgeons [2]. However, the status of the LNs is potentially crucial information as part of surgical staging. It provides both prognostic information and guides adjuvant therapy [3].

Atypical hyperplasia (AH), or previously referred to as complex hyperplasia with atypia, is a pre-malignant condition, and 30–40% of women have a concurrent diagnosis of endometrial cancer [3]. Given the high frequency of endometrial cancer in this population, there is rationale to consider LN assessment in these women, as well. However, there is no consensus on the role of LN assessment in this population [1, 4, 5].

A full pelvic lymph node dissection (PLND) has the potential for significant morbidity and has not been shown to have a survival benefit in early stage disease [6, 7]. Hence, the role of a sentinel lymph node biopsy (SLNB) bridges the gap between “all” or “nothing” approach [2, 3, 8]. SLNB aims to reduce the number of nodes removed for staging, by targeting those most likely to contain metastasis while still maintaining the ability to find microscopic nodal disease [9]. This allows for assessment of LN status to guide staging and adjuvant treatment. SLNB offers an efficient alternative to selecting patients to be surgically staged based on risk factor-based models [10, 11]. SLNB minimises the risks of a full PLND—including reduction in operative time, peri-operative complications, length of hospital stay, costs, and impact on QOL such as lymphoedema [12,13,14,15]. Furthermore, the accuracy of an SLNB is supported by three prospective trials, demonstrating high sensitivity and negative predictive value, and low false-negative rate [8, 16, 17].

In addition, many surgeons have embraced robotic-assisted surgery for endometrial cancer [2]. It has advantages to laparoscopy including superior 3D vision, improved range of motion, and comfortable ergonomics. It has also been shown to increase the number of LNs dissected, and reduce the length of stay and estimated blood loss [18].

Therefore, we aimed to examine the characteristics and peri-operative outcomes of women with AH or endometrial cancer undergoing robotic-assisted hysterectomy. We also aimed to examine the utilisation, feasibility, and role of SLNB, and compare the peri-operative outcomes of women undergoing SLNB versus no LN assessment versus PLND.

Methods

We performed a retrospective cohort study at a single tertiary centre in Sydney. Ethics approval was obtained (2021/ETH01079 and 2021/STE02091). We included all women who underwent a robotic-assisted hysterectomy (RAH) ± adnexal procedure for either AH or endometrial cancer. Two consultant surgeons experienced in robotic-assisted surgery using the DaVinci XI (Intuitive Surgery, Sunnyvale, CA, USA) performed all the robotic surgeries, using the multi-port system, with Gynaeoncology fellows and/or registrars assisting. Each case was reviewed in collaboration at the Gynaeoncology unit multidisciplinary team meeting. In the AH cohort, the decision to perform a SLNB was made prior to surgery based on the consensus at the Gynaeoncology unit multidisciplinary team meeting after the pre-operative histopathology was examined by a dedicated Gynaeoncology pathologist. SLNB was recommended in women with AH who were post-menopausal and therefore considered to be at higher risk for endometrial cancer.

To perform a SLNB, indocyanine green (ICG) fluorescent tracer was injected into the cervix at the beginning of the case following general anaesthesia. The standard technique of using an ICG dose of 0.5 mg/mL was created by diluting the stock solution into sterile water. This was injected into the cervix superficial (1–3 mm) and deep (1–2 cm) at 3 o’clock and 9 o’clock of the ectocervix [3, 13]. After entry to the peritoneal cavity, fluorescence imaging was used to visualise the ICG tracer in the lymphatics. Mapped LNs were then excised, as well as any other suspicious bulky nodes. If there was no mapping to a hemi-pelvis in women with a pre-operative diagnosis of endometrial cancer, then a side-specific LND was performed [3]. If there was no mapping in women with a pre-operative diagnosis of AH, an LND was not performed unless visual assessment of the specimen suggested gross invasion of greater than 50% of the myometrium. Intra-operative frozen sections were not used.

The study period was from December 2018 to February 2021. Patients were excluded if they had diagnoses other than AH or endometrial cancer. Women were also excluded if they were planned for single-site robotic surgery, laparoscopic surgery, or laparotomy approach. LN assessment was classed as either no LN sampling, SLNB, or PLND with intention to treat analysis.

Patients were identified from a prospectively maintained gynaecology oncology database. Data were collected from the Oncology and general medical electronic records, including age, BMI, indication for surgery (pre-operative diagnosis), final pathological diagnosis, type of surgery performed, length of hospital stay, operation duration, estimated surgical blood loss, and adverse events (including mortality within 30 days, conversion to laparotomy, return to theatre, urological injury, and readmissions or representations to the emergency department). Surgical duration was recorded from the time of skin incision to the completion of closure of wounds. For the SLNB cohort, detection rate was defined as the proportion of cases with any successful SLN mapping. Bilaterality was defined as the proportion of cases in which SLNs were identified intra-operatively. Empty packages were defined as samples that did not yield an LN on pathological analysis.

Comparisons were analysed using t test for normally distributed continuous data and Chi-square test for categorical data. Yates correction and Fisher’s exact tests were used as required. P value of < 0.05 was considered significant.

Results

115 women had RAH with or without LN assessment. Mean age was 61.9 ± 10.3 years and mean BMI 34.3 ± 10 0.5 kg/m2. 45% were born in Australia. 59% had SLNB, 29% had no LN assessment, and 12% had PLND (Table 1). 41.9% of women with a pre-operative indication of AH had a final diagnosis of cancer. 35.7% of the PLND cohort had positive LNs.

Table 1 Number of patients undergoing each type of lymph node assessment
Full size table

76.4% of women in the SLNB cohort had a pre-operative diagnosis of endometrioid adenocarcinoma (EAC), mostly Grade 1 (54.4%) (Table 2). The final diagnosis was mostly early stage low-grade disease (Stage 1A—50%, Grade 1 EAC—56%) (Table 2). The detection rate for SLNB was 90% (Table 3). The rate of bilaterality was 83.8% and empty packages were found in 7.8% (Table 3). There was a statistically significant trend towards performing SLNB over time (47% December 2018–December 2019 versus 75% January 2020–February 2021, P value 0.004).

Table 2 Pre-operative and post-operative diagnosis of patients who underwent an SLNB
Full size table
Table 3 Effectiveness of the SLNB technique
Full size table

On comparison of the SLNB and no LN sampling cohort, there was no difference in age and peri-operative complications (Table 4). However, there was a statistically lower BMI, shorter length of stay, less estimated blood loss, and shorter surgical duration in the SLNB cohort, compared to the no LN assessment cohort (P values 0.02, 0.01 and 0.03 respectively) (Table 4).

Table 4 Demographics and peri-operative outcomes SLNB versus no LN cohorts
Full size table

On comparison of the SLNB and PLND cohorts (Table 5), there was no difference in age, BMI, length of stay, or complications. However, there was statistically significant less estimated blood loss and surgical duration in the SLNB compared to the PLND cohort (P values 0.03 and 0.001, respectively).

Table 5 Demographics and peri-operative outcomes SLNB versus PLND cohorts
Full size table

Discussion

In this study, SLNBs were utilised and feasible using robotic-assisted technology for women undergoing surgery for AH or endometrial cancer. There was a total of 115 patients in our cohort, 59% of which had an SLNB, and there was a statistically significant trend towards an increased utilisation of SLNB over time. This trend towards performing an SLNB is consistent with the changing paradigm in surgical staging for these patients [17].

There was a high detection rate of 90%, comparative to the previous studies [2, 13]. Our bilaterality rate was also moderate at 84%; there is a wide range in bilaterality rates in the previous studies, likely a reflection of the learning curve of different surgeons and centres [13, 16, 17]. Since most of our patients had a pre-operative indication of grade 1 EAC and final diagnosis of early stage low-grade endometrial cancer, it is a suitable approach to surgical staging in this cohort, comparative to previous studies [2, 3, 17].

The no LN sampling cohort had a longer length of stay, higher EBL, and longer surgery duration (Table 4). This is likely a reflection of selection bias; that is, women who did not have LN assessment were more likely to have a higher BMI (P value 0.0002), and hence a higher surgical risk. This may explain why the no LN assessment group had a longer length of stay [19], higher EBL [20], and longer surgical duration [19, 21]. Although not examined in this study, it is likely that these women also had more comorbidities or illnesses, which meant that they were a higher surgical risk and deemed not suitable for longer, more complicated surgery, and these risk factors may have also influenced their peri-operative outcomes.

In the cohort of AH, 42% had a final diagnosis of endometrial cancer, which supports the role of performing an SLNB in these women. This rate of findings cancer in the final pathology in this cohort is consistent with the previous trials [4, 5]. Most of the patients in this cohort that were diagnosed with cancer were diagnosed with Stage 1A Grade 1 disease and no women had positive nodes; hence, there is an argument that SLNB is potentially overtreating these women and an argument for selecting only high-risk women for an SLNB [4, 5]. For example, the previous studies have identified that those patients at greater risk of endometrial cancer on final pathology had a pre-operative diagnosis of AH where cancer could not be excluded, and hence that SLN mapping could be a valuable staging procedure in these patients [4]. Furthermore, they identified that an elevated pre-operative CA 125 was associated with the risk of LN metastasis on final pathology. No patients in our study had positive nodes from the AH cohort, which may be due to a small sample size. In the study by Touhami et al. [4], they included 120 patients with atypical hyperplasia, of which 3.3% had positive LN metastasis. This supports a role for performing an SLNB in high-risk women with a pre-operative diagnosis of AH.

We also demonstrated that SLNB using robotic-assisted technology was safe, consistent with previous findings [16, 17]. We had a low complication rate, which was comparable to the cohort which had no LN sampling performed and previous studies [17]. It also had advantages to a full PLND included statistically significantly less EBL and surgical time [13, 14, 16].

Some of the strengths of this study include that it was conducted at a large tertiary centre and the pathology was reviewed by a dedicated Gynaeoncology pathologist. Some limitations of this study include that is in a single-centre study, it is retrospective, subject to selection bias and confounding, we did not assess long-term outcomes (such as cancer recurrence, survival, and lymphoedema) and we were unable to calculate the sensitivity, specificity, and false-negative rate based on the available data. Future directions could include larger prospective studies, studies that examine the learning curve of surgeons and trainees, long-term outcomes (such as lymphoedema and cancer recurrence), a comparison of robotic and laparoscopic SLNB, and an assessment of patient and surgeon attitudes.

Conclusion

In conclusion, SLNB at RAH was utilised and feasible with a statistically significant trend towards increased utilisation over time and high detection rate. SLNB at RAH was safe with a low complication rate comparable to No LN sampling cohort and had advantages compared to PLND cohort. SLNB should be considered in suitable selected women undergoing surgery for endometrial cancer or AH.