JSUR Journal

Introduction

Performing Sentinel Lymph Node (SLN) mapping in women with presumed early-stage Endometrial Cancer (EC) increased enormously, as it is applied by 50.3% of physicians or consultants in gynaecologic oncology and recommended by multiple international groups. [1-5] The SLNs are the first lymph nodes  that receive lymphatic fluid through the regional lymphatic drainage pathway from the uterus, and these lymph nodes are detected and removed in both hemipelvis. [6] By examining a SLN on metastasis, instead of assuming a chance of metastasis, over- and undertreatment of patients can be prevented: only patients with actual lymph node metastases require adjuvant treatment. SLN mapping in EC patients performed by experienced gynaecologists detects or excludes pelvic lymph node metastases, with a sensitivity over 90% and a false-negative rate below 5%. [7-9] In the Netherlands, a hysterectomy with bilateral salpingooophorectomy (BSO) is still standard for presumed low- and intermediate-risk EC, with adjuvant therapy based on a risk assessment including both tumour and patient characteristics, more commonly including molecular markers as well. [10] The actual lymph node status is no part of the risk assessment, even though lymph node metastases can be found in up to 15% of presumed low- and intermediate-risk EC patients. [10-12] If lymph node metastases are present, survival can be improved by treating patients adequately. The PORTEC-3 study showed 10% overall survival benefit for radiotherapy with chemotherapy as compared to radiotherapy alone in women with stage IIIC EC. [13] As SLN mapping in EC is not common practice in the Netherlands, it is important to acknowledge and establish a learning curve prior to the implementation of SLN mapping. The aim of this study was to evaluate the learning curve of gynaecologists with different levels of surgical oncological experience.

Methods

Since SLN mapping is no part of the standard treatment of EC in the Netherlands, all patients included in this learning curve analysis were patients who participated in a multicenter, intervention, prospective, cohort study (i.e. SLIM study; Sentinel node biopsy in Low- and Intermediate-risk EC Management, NL52051.091.15). [14] They were included in the SLIM study between March 2016 and December 2021 and underwent SLN mapping for histological confirmed, presumed early-stage, low- and intermediate-risk EC. All patients provided written informed consent. After reaching the required number of inclusions for the SLIM study, no more patients could be added to our learning curve analysis, as SLN mapping in grade 1 and 2 endometrioid EC is only performed in study context in the Netherlands. All patients underwent minimally invasive surgery: SLN mapping followed by a hysterectomy and BSO. A subsequent (side-specific) lymphadenectomy in case of mapping failure was not performed. SLN mapping was performed using indocyanine green (ICG, 1.25 mg/ml), injected superficial (1-2 mm) and deep (20-30 mm) at 3 and 9 o’clock in the cervix after induction of anesthesia and 15-30 minutes before surgery for a total of 4 ml. The dye was injected slowly, at a rate of 5-10 seconds per quadrant. ICG was chosen as marker because of its preferable status based on technical ease, high success rate and reliability. [2,8] The mapping procedure was performed with a near-infrared camera (laparoscopy: Olympus; robot-assisted: Da Vinci X or Xi). SLN mapping was performed conform the operation guideline as described by Moloney et al. [15] The mapped SLNs on both hemipelvis were dissected separately. Tracer re-injection was an option if the SLNs were not visualised upfront. If multiple draining pathways were seen then both draining lymph nodes were dissected. Removed ICG positive tissue was not palpated routinely to determine if nodal tissue was felt. Histological assessment was performed by experienced gynaecological pathologists and consisted of ultrastaging according to a standardised protocol, by 5-step paraffin sections, Haematoxylin And Eosin (H&E) staining and immunohistochemistry on all levels.

In the Netherlands, EC patients with presumed earlystage, low- and intermediate-risk disease are not centralised in the eight tertiary referral hospitals, unlike other gynaecological malignancies, but treated by general gynaecologists in all other hospitals, of whom some only treat less than ten  patients per year. Implementation of SLN mapping to standard treatment requires new surgical skills for general gynaecologists and a highvolume of procedures. And although gynaecological oncologists are common with lymph node dissections, SLN mapping may be a relatively new skill for them as well. Therefore, we aimed to evaluate the learning curve of gynaecologists with different levels of surgical oncological experience. All surgeries were performed by general gynaecologists or gynaecological oncologists with experience in minimally invasive surgery, but without previous experience in SLN mapping. All surgeons had at least three years of experience in laparoscopic or robot-assisted hysterectomies, with a minimum of 50 minimally invasive hysterectomies performed; the gynaecological oncologists were also experienced in minimally invasive pelvic lymph node dissection. The first five cases of sentinel lymph node mapping were performed under direct supervision of an experienced proctor. The learning curve started from the first SLN mapping on. Participants in this study were two gynaecological oncologists (gynaecologist 1 and 2, working in different tertiary referral hospitals, both over ten years of experience in laparoscopic or robot-assisted hysterectomies) and two general gynaecologists (gynaecologist 3 and 4, working in a general hospital, respectively over three and over ten years of experience in laparoscopic or robot-assisted hysterectomies). The general gynaecologists did not have any experience with lymphadenectomies. The learning curve for bilateral SLN mapping was determined for each gynaecologist. Successful SLN mapping was defined as bilateral SLN mapping with actual lymphatic tissue in the SLN specimen (pathologically confirmed). By the use of this definition, unilateral and bilateral empty packets (i.e. SLN specimens without actual lymph nodes) were not considered as successful SLN mapping. The false negative rate of SLN mapping could not be determined, as a subsequent lymphadenectomy is not performed in the study population.

Statistics

Statistical analyses were performed using R (version 4.2.1; R Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics were used to summarize the data. To determine when proficiency was reached [16], the cumulative summation test for learning curve (LC-CUSUM) was applied. The LC-CUSUM test is derived from the CUSUM test, which is a method used in statistical process control to monitor a process over time and to detect inadequate runs of performance. [17] Unlike other methods which can be used to determine a learning curve, the LC-CUSUM indicates when a process has reached a predefined level of performance, and allows to determine when proficiency is achieved. [18] The LC-CUSUM test presumes that the gynaecologist is not proficient at the start of monitoring and signals when the trainee has reached the acceptable predefined level of performance. Therefore, the Null Hypothesis (H₀) is that the process is out of control (gynaecologist’s performance deviates from adequate performance, p₀, by at least delta), and an Alternative Hypothesis (H₁) is that process is in control (gynaecologist’s performance is equal to the adequate performance). The gynaecologist is considered not proficient as long as the LC-CUSUM score remains below the limit h. When the LC-CUSUM score crosses this limit, the gynaecologist is considered to be proficient.

The learning curve was established for bilateral SLN mapping with actual lymph node tissue in the specimen during the pathological assessment. Based on available literature [2], we considered p₀ = 34% as the criterion for an acceptable failure rate (i.e. the bilateral SLN mapping rate should be at least 66%), p₁ = 45% as an unacceptable failure rate, and δ = 5% as the acceptable deviation from adequate performance for pathological bilateral SLN mapping. Based on these settings and using computer simulation (10,000 simulations of series of 100 procedures) a limit of h = 1.1 was chosen for LC-CUSUM test to provide a true discovery rate (TDR; i.e. the probability of an alarm being raised over 100 procedures if the true performance of the gynaecologist is 34% failure) of 79% and an false discovery rate (FDR; i.e. the probability of an alarm being raised over 100 procedures if the true performance of the gynaecologist is 45% failure) of 11%. A standard CUSUM test was applied once gynaecologists demonstrated competency to ensure retention of proficiency. A limit h = 3.75 was chosen for the CUSUM test to provide TDR of 81% and an FDR of 7%.

Ethical Approval

Ethical approval was obtained from the Radboudumc Committee for Ethics in Research (CMO) in the region Arnhem and Nijmegen (file number 2015-1783).

Results

A total of 122 SLN mappings were performed by the four participating gynaecologists. After 122 procedures, the inclusion of the SLIM study was completed and no more patients could be added in this the learning curve analysis. The patient and operative characteristics are presented in (Table 1). None of patients had prior lymphadenectomy or prior radiation therapy. The majority of patients (82%) underwent robot-assisted surgery. SLN mapping was deemed successful per operatively in both hemipelvis in 81% of patients (n=99). However, in 11 patients with apparent bilateral SLN mapping, the dissected SLN specimens on one or both hemipelvis did not contain actual lymph node tissue (empty packet dissection; Table 2). After pathological assessment, the actual bilateral detection rate of SLNs was 72% (n=88). The overall (bilateral and unilateral) SLN detection rate per operatively was 94% (n=115). Final pathological assessment showed a true overall detection rate of 91% (n=111). In seven patients (6%), no SLN at all was found in both hemipelvis. There were few complications: two patients developed a complication during the operative procedure (blood loss > 500 mL, vaginal vault lesion). Five patients developed postoperative complications: extensive skin haematoma (n=1), haematoma of vaginal vault (n=1), and wound infection (n=3). In none of the cases this could be directly linked to SLN mapping.

*Operation time was the time from start anaesthesia until end of anaesthesia

** Patients without SLN detection during surgery are included

Table 1: Patient characteristics and operative characteristics. Data presented as median (range in brackets) or count (percentage in brackets).

Table 2: Overview of surgical versus pathological SLN detection. Data presented as counts.

Learning Curve of SLN Mapping

The LC-CUSUM and CUSUM plots of bilateral SLN mapping are shown in Figure 1. The h-limit was crossed by gynaecologists 1 and 3 after 22 and 45 cases, respectively, meaning they reached proficiency in SLN mapping (pathologically confirmed). During their short maintenance period, there were no signs of progressing from an in-control to an out-of-control state, and they maintained the performance at an acceptable success rate there onwards, based on the CUSUM curve. Gynaecologists 2 and 4 did not reach proficiency for bilateral SLN mapping after the 22 cases that they performed (after which no more patients could be included). But a tendency to reach the h-limit is seen, especially in gynaecologist 2.

Figure 1: LC-CUSUM and CUSUM curve for bilateral SLN mapping.

LC-CUSUM is applied until acceptable performance has been reached and CUSUM is used thereafter to ensure that adequate level is maintained. For the LC-CUSUM, as long as the score remains below the h-limit (dotted line), the gynaecologist is not considered as proficient; when the LC-CUSUM score crosses this limit, the gynaecologist is considered to have learned the procedure. For the CUSUM, as long as the score remains under the limit, the gynaecologist is considered to maintain an acceptable performance.