Endometrial cancer is the sixth most common cancer in females, worldwide. There were an estimated 382,000 new cases and 90,000 deaths in 2018. In Australia, endometrial cancer is the fifth most common cancer with 3040 new cases and 582 deaths in 2018 [1]. The majority of endometrial cancers are diagnosed as early stage, low grade tumours and have an excellent prognosis with 5-year survival rates exceeding 90% [2]. 5-year survival for node positive advanced stage disease in contrast is 50–58% [3,4], with lymph node status being prognostic for recurrence and survival.

Pre-operative diagnosis is an important step to obtain a cell type and histologic grade and make a decision of treatment plan. Currently, the diagnostic methods of endometrial cancer include dilatation and curettage (D&C), suction biopsy (Pipelle), vacuum aspiration biopsy random essay (Novak curette and Vabra aspirator), hysteroscopic endometrial biopsy, etc. Outpatient endometrial biopsy (Pipelle, Novak curette and Vabra aspirator) is a non-invasive technique comparing with D&C. Although this method has a high overall accuracy in diagnosing endometrial cancer [5], this is a blinded technique. An inadequate specimen causes a decrease in the diagnostic accuracy. Preoperative hysteroscopic biopsy can provide an adequate tissue. Grasp endometrial biopsy is one of a technique providing more accurate diagnosis of histology type and tumour grade when in presence of endometrioid type tumour compared to blind endometrial biopsy obtained using the Novak curette [6].

Standard treatment for early-stage endometrial cancer includes extrafascial hysterectomy with bilateral salpingo-oophorectomy [7]. For patients with disease apparently confined to the uterus and cervix, knowledge of pelvic and para-aortic lymph node status is important for accurate surgical staging, and for guiding adjuvant radiation therapy. Two randomized controlled trials have however failed to demonstrate a therapeutic advantage for systematic lymphadenectomy, although neither trial had a consistent policy with respect to the management of the paraaortic lymph nodes and did not utilize nodal status to direct adjuvant therapy [8,9].

Low-risk patients, defined as grade 1 or 2 endometrioid histology, $\le$50% myometrial invasion and tumour diameter $\le$2 cm, have 1% risk of lymph node metastasis, compared to intermediate- or high-risk patients who have a 20–25% risk. The risk of lymph node metastasis is up to 40% in those with non-endometrioid histology [10]. With such a low incidence of lymph node metastasis in the low-risk patients, the risks of exposing such patients to systematic lymphadenectomy are not justified.

Studies of systematic pelvic and para-aortic lymphadenectomy have consistently shown that patients with positive pelvic nodes have a 50% incidence of positive para-aortic nodes [11,12]. In those with confirmed stage 3 disease, Mariani and colleagues concluded that both adequacy of pelvic and para-aortic dissection and delivery of adjuvant radiation to nodal beds independently reduced pelvic and paraaortic failures [13]. Hence knowledge of pelvic lymph node status is integral for the appropriate prescription of adjuvant treatment, especially in the node-positive patients, where the addition of systemic chemotherapy to radiation has also demonstrated an improvement in survival [2,14].

Sentinel lymph node (SLN) mapping is an alternative technique that enables assessment of lymph node status and avoids the surgical complications associated with complete pelvic and para-aortic lymphadenectomy such as blood vessel or nerve injury, lymphocyst formation and lymphedema [15-17]. The techniques for lymphatic mapping are colorimetric mapping with coloured dyes, radionuclear mapping with technetium 99 (Tc99) and near-infrared fluorescence (NIRF) imaging with indocyanine green dye (ICG). A systematic review and meta-analysis of SLN mapping in endometrial cancer showed a pooled SLN detection rate of 81%, sensitivity of 96% and negative predictive value (NPV) of 99.7% [18]. The use of ICG with NIRF imaging, intracervical tracer injection, a SLN mapping algorithm and an ultra-staging protocol are all associated with improved detection rates, increased sensitivity to detect node positive patients and improved NPV [18-20].

At Mater Hospital Brisbane, we have been using ICG with NIRF imaging for SLN mapping as a part of surgical staging in all women with clinically uterus-confined endometrial cancer undergoing laparoscopic or robotic surgery since 2017. We conducted this pilot study to determine our detection rate and localization of mapped SLNs, to assess the safety of the technique within our institution as well as to assess the utilization of SLN biopsy results in informing decisions on adjuvant therapy compared to our existing staging and adjuvant therapy protocols.

This retrospective study was approved by Mater Misericordiae Ltd Human Research Ethics Committee (HREC). From January 2017 to May 2019, all patients with histologically confirmed endometrial cancer were consented for laparoscopic or robotic hysterectomy, bilateral salpingo-oophorectomy and SLN mapping with ICG at Mater Hospital Brisbane, Queensland, Australia. The inclusion criteria were all patients with histologically confirmed endometrial cancer and radiologically confirmed FIGO stage I disease. Patients with extra-uterine disease on imaging were excluded. We further excluded any case where ICG injection was omitted or abandoned for any reason and any cases with incomplete medical records.

All surgeons were gynaecologic oncologists with significant experience in endoscopic surgery including lymphadenectomy and SLN mapping. KARL STORZ Image1 S OPAL1 for NIR/ICG was used in laparoscopic surgery and Firefly$\mathrm{\circledR }$ NIRF on da Vinci Xi was used in robotic surgery. ICG, a soluble tricarbocyanine dye, was diluted in 20 mL of sterile water (final concentration 1.25 mg/mL). After anaesthetic induction, a total of 4 mL (5 mg) of ICG was injected into the ectocervix: 1 mL superficially (3–4 mm into cervix) and 1 mL deep (1 cm into cervix) at the 3 and 9 o’clock positions. After exploration of abdominal cavity and peritoneal washing, the pelvic retroperitoneum was opened bilaterally. Mapped SLNs were visualized by a camera detecting NIRF light. The first mapped lymph node (first lymph node detected with fluorescence nearest to the uterine body) was defined as a SLN and was resected. Any suspicious or bulky lymph nodes were removed regardless of mapping and labelled non-SLN.

A comprehensive pelvic lymph node dissection was performed selectively in cases where SLN mapping failed to detect a SLN in either hemi-pelvic, only in those patients with intermediate and high-intermediate risk uterine features (grade 3 endometrioid, deep myoinvasion or bulky tumour), as per current institutional practice (Supplementary Fig. 1). Low risk patients (grade 1–2 endometrioid with superficial myoinvasion) were not staged further if SLN mapping failed because of a very low risk of nodal metastasis. Patients with deeply invasive high grade or invasive non-endometrioid tumours i.e., serous, clear cell, carcinosarcoma, were also not subjected to side specific systematic lymphadenectomy if SLN mapping failed, as under or institutional protocol these patients received adjuvant therapy with 4 cycles of chemotherapy and external beam pelvic radiation irrespective of nodal dissection. After SLN dissection, routine hysterectomy, bilateral salpingo-oophorectomy (BSO) was performed.

All specimens were analysed by pathologists specialized in gynaecologic cancers. All SLNs were sectioned and stained with haematoxylin and eosin (H&E). Ultra-staging was performed if the SLN was negative on initial H&E staining. Serial sectioning was performed with 250 $\mu$m intervals. One H&E and 2 unstained slides were taken from each 5 levels. When additional H&E stained slides were negative, immunohistochemistry (IHC) staining for cytokeratin was used on unstained secctions to detect microscopic metastasis (Supplementary Fig. 2). Non-SLNs were evaluated by H&E staining only. According to the American Joint Committee on Cancer definitions, macro-metastasis is defined as a tumour focus larger than 2 mm [21]. Tumour foci between 0.2 and 2 mm. are considered as micro-metastasis. Isolated tumour cells (ITCs) are defined as tumour deposits less than 0.2 mm. Mater Hospital Multi-Disciplinary Tumour Board Meeting made recommendations for adjuvant therapy for patients based on the final histopathology including SLN status utilising the adjuvant therapy protocol of Queensland Centre for Gynaecological Cancer (QCGC) (Supplementary Fig. 3). All patients attended for follow up at 6 and 12 months after surgery.

All statistical analysis was performed using IBM SPSS Statistics for Windows, Version 22.0. (IBM Corp Armonk, NY, USA). Descriptive data were presented as mean with standard deviation (SD) for continuous variables and percentage for categorical variables. A Chi-square test was used to analyse the differences between two variables. Overall detection rate was defined as the percentage of patients undergoing SLN mapping in which at least one SLN was pathologically identified in either hemipelvis. Bilateral detection rate was defined as the percentage of patients with at least one SLN pathologically identified in each hemipelvis. A p-value of 0.05 was considered statistically significant.

From January 2017 to May 2019, 140 patients with histologically confirmed endometrial cancer were consented for a minimally invasive hysterectomy, BSO and SLN mapping. ICG injection was cancelled due to patient factors in 9 cases (Fig. 1). Three cases had an extra-uterine disease, 3 cases had some surgical technique and anatomical problems, 1 case underwent full pelvic lymphadenectomy, 1 case had a suspicious synchronous vulvar cancer and 1 case did not record any reason. A further 9 cases were omitted due to incomplete medical records. SLN mapping was abandoned intra-operatively (despite prior ICG injection) in a further 11 cases due to morbid obesity (BMI $\ge$40 kg/m${}^2$), technical difficulty and advanced disease. We included these 11 patients in the failed bilateral SLN detection group. Decision regarding the abandonment of SLN mapping in these cases was made by the operating surgeons where the risk of potential complications associated with the technique exceeded the foreseeable benefit of the procedure.

Fig. 1.

Study profile.

The characteristics of the remaining 122 patients who were included in the analysis are summarized in Table 1. Eighty-five patients (70%) had a laparoscopic procedure and 37 patients (30%) had robotic surgery. Endometroid adenocarcinoma was the most common cell type diagnosed in pre- and post-operative histologic results.

3.4 Alteration of adjuvant treatment based on SLN status

After surgery, the adjuvant therapy was prescribed based on the adjuvant therapy protocol of Queensland Centre for Gynaecological Cancer (QCGC) (Supplementary Fig. 3). Without knowledge of lymph node status (based on uterine features alone) 37 of 122 patients (30.3%) would receive adjuvant treatment. Using the adjuvant therapy protocol, it was predicted that radiation therapy (RT) would be prescribed in 37 cases (30.3%), with the addition of chemotherapy (CMT/RT) in 27 cases (22.1%), based on high-risk features.

Based on SLN status, predicted adjuvant therapy as per uterine features was altered in 16 out of the 103 patients (15.5%) that underwent successful SLN mapping. This represents 13% of the entire cohort of patients (Fig. 2). Case NO.4 would have been prescribed CMT/RT based on the local protocol, but as only isolated tumor cells were detected in the SLN (considered a negative node at our institution) systemic chemotherapy was withheld. Case NO. 8 had low-risk disease but had micro-metastasis in SLN on final pathology and received CMT/RT. For the remaining 14 patients with negative SLN after ultra-staging, adjuvant treatment was de-escalated from CMT/RT to RT in 5 cases, CMT/RT to observation in 7 cases and RT to observation in 2 cases.

Fig. 2.

Adjuvant treatment alteration based on sentinel lymph node status.

Lymphadenectomy in endometrial cancer is required for accurate staging and the knowledge of lymph node status determines adjuvant treatment [3]. SLN mapping has been proposed to provide information on lymph node status, without the morbidity of full lymph node dissection. According to National Comprehensive Cancer Network (NCCN) guideline, SLN mapping can be considered for the surgical staging of uterine-confined tumour without metastasis or extra-uterine disease demonstrated by imaging or exploration [22]. The recommended tracers were Tc99, coloured dye (isosulfan blue 1%, methylene blue 1% and patent blue 2.5% sodium) and ICG.

ICG is a small particle tracer detected by NIR. This tracer has been used for lymph node mapping in many cancers include endometrial cancer. ICG is accepted for SLN mapping by many authorities including NCCN [22] because of the high detection rate, sensitivity and NPV [18]. The other benefit of ICG was reported in endometriosis surgery. ICG with NIR is useful in the evaluation of vascularization in intestinal anastomoses and identifying ureters or ureteric injury after ureterolysis for parametrial deep infiltrating endometriosis [23].

Our study used ICG as the tracer for SLN mapping in minimally invasive surgery. The cervical submucosa and stroma were chosen as the site of injection as it has been reported to improve detection rates. This protocol provided a high detection rate (86.1% of overall detection rate and 73% of bilateral detection rate) consistent with other studies using this technique. Overall and bilateral SLN detection rates for ICG are reported between 86–100% and 52–88%, respectively [24-31]. This indicates that there is a short learning curve for the technique as SLNs were reliably detected in this early phase of introducing the ICG and NIFR SLN mapping protocol in our institution.

The percentage of SLN detected outside the routine lymphadenectomy zone is between 3–22% [25-28,30]. In our series, 30 patients (24.6%) had SLN detected outside the routine lymphadenectomy zone suggesting that these disease bearing nodes may be missed on standard pelvic lymphadenectomy. 2 of 8 (25%) of SLN metastasis were detected in these aberrant locations.

Our study showed that BMI 30 kg/m${}^2$ was a significant predictor for successful overall and bilateral detection. This is supported by a previous retrospective study which reported that median BMI was significantly different between the patients with and without at least one mapped SLN (30.1 vs. 41.2 kg/m${}^2$,p-value = 0.001), and in patients that mapped unilateral vs. bilaterally (34 vs. 29.6 kg/m${}^2$,p-value = 0.02) [24]. This information may be helpful for counselling patients about the success rate of SLN mapping.

Recent literature reports SLN mapping as having a good detection of positive lymph nodes. Fifteen percent (range 10–30%) of all mapped SLNs were positive and 50% of positive SLNs were diagnosed after ultra-staging [19]. In our study, 6.6% of all patients had positive SLN which is consistent with 80% of our participants having a low to moderately differentiated endometroid adenocarcinoma and 70% having disease confined to the inner half of the myometrium, putting this group at low risk of lymph node involvement.

Ultra-staging was required for the diagnosis of metastatic disease (in the form of micro-metastasis or ITCs) in 6 of the 8 patients with positive SLNs (75%). This is relevant as despite a comprehensive pelvic lymphadenectomy, these 6 patients would not have had their LN metastasis detected by standard H&E staining methods, confirming the value of an ultra-staging protocol is confirmed in our series.

In our series 5.7% of low risk patients, who would not have undergone lymph node assessment according to our pre-existing staging algorithm, were found to have positive SLNs and were upstaged. This allowed these patients to have adjuvant therapy to reduce their risk of recurrence.

Our study reported high detection rates (90.9% for overall detection and 81.8% for bilateral detection) for patients with high-risk histological subtypes. Several studies in patients with high-risk histology have confirmed excellent performance of SLN mapping using cervical injection with bilateral detection in the order of 60% and NPV exceeding 98% [32-34]. In this cohort, 21–30% of mapped SLNs are positive, with a low false negative rate (2–5%) [32,33]. In our study 15% of patients with high-risk histology had positive SLNs. This percentage may be an underestimation as our study population only included 20 patients with high-risk histology and we did not perform full lymphadenectomy following SLN mapping in our high-risk patients.

Large retrospective data review from the Surveillance, Epidemiology, and End Results (SEER) Program showed a survival benefit from more extensive lymphadenectomy in the high-risk group [3]. It is likely that the same survival benefits can be achieved from SLN mapping, with SLN status used to prescribe chemotherapy and pelvic +/- para-aortic radiation to patients with positive pelvic nodes.

The limitations of this study are that this is a retrospective study susceptible to biases, including surgical bias to abandon SLN mapping in difficult cases, and bias in prescription of adjuvant therapy based on various patient factors. We did not perform full pelvic lymphadenectomy in patients after SLN mapping and hence cannot provide sensitivity, specificity or false negative rate for our institution. These are the major disadvantage of our study so we cannot conclude that SLN mapping can replace full lymphadenectomy in high-risk group. However, in our unit, prior to SLN mapping, selective pelvic lymphadenectomy was performed in those with intermediate and high intermediate risk features on frozen section and our series does demonstrate that SLN mapping was effectively implemented into the existing staging paradigm as the vast majority of patients in the intermediate and high-risk groups were able to avoid comprehensive staging lymphadenectomy. Further limitations include incomplete medical records which resulted in patient exclusion. An additional limitation is that long-term survival data is not yet available on all participants treated.

The strengths of this study are that it was conducted in a tertiary centre with a high incidence of endometrial cancer. The study has captured the majority of sentinel lymph node mapping cases from the introduction of the technique into our institution’s staging protocol. This study is the first to our knowledge that illustrates the potential impact of SLN status on recommendations for adjuvant therapy in apparent early stage endometrial cancer compared with a protocol that prescribed adjuvant therapy based on uterine features. Long term survival data would be required to validate this approach however our high bilateral SLN detection rates and detection of SLN metastasis are equivocal to published reports indicating reliable detection of node positive patients.

Despite no prior experience with the technique, our detection rates were comparable with those in the literature showing effective implementation of the SLN mapping technique at our institution. We have shown that SLN mapping can be incorporated into the treatment protocol for endometrial cancer without compromising operative safety. Positive nodes located outside the standard lymphadenectomy field may be missed without SLN mapping. Ultra-staging with IHC increased the detection of positive SLNs.

The clear advantage for sentinel node mapping is in low-risk patients, who would not otherwise have a lymphadenectomy. The detection of occult nodal metastasis in 5.7% of these patients in our study allowed them to receive appropriate adjuvant therapy.

Knowledge of the SLN status in our cohort resulted in a reduction in the amount of adjuvant therapy required in a subgroup of patients, hence reducing treatment related morbidity. We await longer follow-up to observe the impact of this on patient recurrence risk and survival. Further studies should be conducted prospectively to evaluate the impact of SLN mapping on prescribed adjuvant therapy, patterns of disease recurrence and disease free and overall survival.

WC collected the data, managed the database and wrote the manuscript. VC and EW collected the data. NC and LP performed the staging surgery and sentinel lymph node mapping. NJ designed the study, performed the staging surgery and sentinel lymph node mapping and edited the manuscript. SB designed the study, and edited the manuscript. All authors read and approved the final manuscript.

This retrospective study was approved by Mater Misericordiae Ltd Human Research Ethics Committee (HREC). Approval number is HREC/MML/46596.

We thank to professor Neville Hacker for his opinion and suggestions during preparation of the manuscript.

This study received no external funding.

The authors declare no conflict of interest.

Supplementary material associated with this article can be found, in the online version, at https://ejgo.imrpress.com/EN/10.31083/j.ejgo4204106.