Article Text
Abstract
Background Sentinel lymph node (SLN) biopsy is increasingly used in endometrial cancer staging; however, success of the technique is variable, and the learning curve needs to be better understood. Success is defined as identification of a SLN specimen containing nodal tissue in bilateral hemi-pelvises.
Objective To assess the learning curve of surgeons at an academic institution in performing successful SLN mapping and biopsy during robotic staging for endometrial cancer.
Methods After institutional review board approval, patients who underwent staging with robotic SLN mapping using indocyanine green at a single academic program between July 2012 and December 2017 were identified. Demographic, pathologic, and surgical data were retrospectively collected from the medical records. Descriptive and comparative statistics were performed. Surgeon rates of successful bilateral SLN mapping and removal of lymphoid-containing SLN specimens were compared. A logistic model was used to analyze the probability of successful SLN mapping and removal of lymph node-containing tissue with increasing number of procedures performed.
Results Three hundred and seventeen patients met the eligibility criteria. Most had early-stage, low-grade endometrial cancer. A total of 194 (61%) patients had successful bilateral mapping. Among seven surgeons, a plateau in rates of successful bilateral mapping was achieved after 40 cases. No linear correlation was seen between the number of surgeries performed and the rate of removal of lymph node-containing tissue among surgeons. Each additional 10 procedures performed was associated with a 5% and an 11% increase in the odds of successful SLN mapping and removal of lymph node-containing tissue, respectively.
Discussion The successful removal of lymph node-containing specimens appears to be a surgeon-specific phenomenon. The plateau of the learning curve for successful bilateral mapping seems to be reached at around 40 cases. These first 40 cases offer a time for auditing of individual rates of SLN mapping and removal to identify surgeons who may benefit from procedure-specific remediation.
- sentinel lymph node
- surgical procedures, operative
- uterine cancer
Statistics from Altmetric.com
HIGHLIGHTS
The learning curve for successful sentinel lymph node (SLN) mapping among attending surgeons stabilizes after 40 cases.
Cumulative rates of removal of lymph node-containing tissue among low-volume surgeons appears to be surgeon dependent.
Each additional 10 procedures performed was associated with an increase in the odds of bilateral SLN mapping and removal of lymph-node containing tissue.
INTRODUCTION
Surgical staging of endometrial cancer has historically included both pelvic and para-aortic lymphadenectomy, as occult lymph node metastases can be identified microscopically in almost 20% of patients.1 Although systematic pelvic lymphadenectomy directs adjuvant therapy,2 multiple phase III trials, a Cochrane review, and a meta-analysis have shown that it provides no significant therapeutic benefit.3–5 Sentinel lymph node (SLN) mapping and biopsy is non-inferior to lymphadenectomy in endometrial cancer staging and has a high negative predictive value and a sensitivity of >97% in identifying nodal metastases.6
Successful SLN mapping is measured by the ability of a surgeon to identify SLN specimens that contains nodal tissue in both (bilateral) hemi-pelvises. The reported SLN detection rate in endometrial cancer staging is 23–100%, with a pooled averaged of 81%; success of bilateral mapping is 6–88%, with a pooled average of 50%.7 Although the literature is limited, increasing surgical volume appears to increase the rate of successful SLN mapping in endometrial cancer among experienced surgeons, with an increase in detection rate after 20–30 cases.8 9
A troubling phenomenon with SLN biopsy is the removal of empty SLN specimens, in which tissue, thought to be a SLN, contains no lymph node material on final pathologic evaluation. The rate of consistent removal of lymph node-containing tissue appears to increase with surgeon experience, reportedly stabilizing after approximately 30 procedures.10 The use of indocyanine green tracer and access to near-infrared imaging has improved rates of SLN mapping.11 It is also known that patient factors and potential tumor factors influence successful SLN mapping, with obesity and advanced cancer states associated with decreased mapping.12 13
We sought to assess the learning curve of surgeons at an academic institution, with learner (fellow) participation, in performing successful SLN mapping. SLN mapping was defined by both bilateral SLN identification and removal of lymph node-containing specimens, which are both critical elements of complete staging for endometrial cancer using the SLN technique. We also sought to evaluate the influence of tumor and patient-related factors in the success of bilateral SLN mapping.
METHODS
After approval by the University of North Carolina institutional review board (UNC IRB protocol No 19–1138), a single institution retrospective case collection was performed on patients who underwent endometrial cancer staging with SLN biopsy at the University of North Carolina, Chapel Hill between January 2012 and December 2017. This time period included the first 3 years of introduction of SLN mapping and biopsy techniques performed in the research protocol for the Fluorescence Imaging for Robotic Endometrial Sentinel lymph node biopsy (FIRES) trial (clinicaltrials.gov, number NCT01673022). Patients were included if they underwent robot-assisted surgical staging with planned SLN biopsy. Patients were excluded if they underwent open staging or if SLN biopsy was not planned or attempted. Patients were also excluded if the outcome of the SLN mapping procedure could not be ascertained from the operative report or pathology report (n=3). Completion unilateral or bilateral pelvic with or without para-aortic lymphadenectomy was performed if the patient was enrolled in the FIRES trial, if bilateral mapping dependent on intra-operative pathology results failed, or at the discretion of the surgeon.
Indocyanine green tracer was injected at the outset of surgery, before uterine manipulator placement, according to the FIRES protocol.6 This technique was adopted and generalized by the entire complement of surgeons after completion of the trial in July 2016. During surgery, retro-peritoneal spaces were opened in accordance with the FIRES protocol and the first in chain lymph node was removed as the SLN.
Demographic, clinical, pathologic, and surgical data for each surgeon were retrospectively collected from the electronic medical record. Bilateral, unilateral, or unsuccessful SLN mapping was recorded for each patient based on review of the operative note and final pathology report. The pathology report was used to determine whether patients had lymph node-containing tissue removed from all SLN specimens or whether any specimens lacked lymphoid tissue (empty node packets).
Descriptive statistics were used for demographic and clinical characteristics of the cohort. Chi-square and t tests were used to compare those patients who had successful bilateral SLN mapping with those who did not.
Among all surgeons, the learning curves for successful bilateral SLN mapping and removal of lymph node-containing tissue were analyzed by year over the 7-year study period. When creating individual surgeon learning curves, the first three surgeries were excluded owing to variability of the estimated cumulative mapping and lymph-node containing tissue removal rates when computed from one to three surgeries. Seven surgeons performed endometrial cancer staging surgeries with SLN mapping and biopsy during the study period and were included in our analysis. A logistic model was used to analyze the probability of successful SLN mapping and removal of lymph node-containing tissue with increasing number of procedures performed. The generalized estimating equation approach was used to estimate the co-variance structure in the data to account for possible correlation of outcomes within each surgeon.
RESULTS
A total of 317 patients met the eligibility criteria and were included in our analysis. Baseline patient demographic and clinicopathologic characteristics are shown in Table 1. During the 7-year study period, 5 patients underwent surgery in 2012, 26 in 2013, 33 in 2014, 29 in 2015, 47 in 2016, and 177 in 2017. The median age at surgery was 62 years (range 27–87) and the majority of patients were Caucasian (74%) and non-Hispanic (94%) with a mean body mass index of 35 kg/m2 (range 18–72). Most patients had early-stage, low-grade endometrial cancer on final pathology (Table 1). At least one SLN was found in 288 patients (91%). Bilateral mapping occurred successfully in 194 patients (61%) while unilateral mapping (to one hemi-pelvis) was successful in an additional 94 patients (30%). Equal rates of successful bilateral mapping were found among the patient subgroups with surgical stage I low-intermediate risk versus surgical stage I high-intermediate risk as defined by Gynecologic Oncology Group 99 risk factors (Table 1).2
Among patient- and tumor-related factors, advanced age (median 64 years vs 61 years, P=0.02) was associated with failed bilateral mapping. The presence of lymphovascular space invasion was associated with successful bilateral mapping (P=0.03). A trend was seen towards a higher mean body mass index in those with unsuccessful bilateral mapping compared with those with successful bilaterally mapping (36 vs 34), but this did not reach statistical significance (P=0.06). No other patient or pathology risk factors were associated with failed bilateral mapping (Table 1).
Among all surgeons collectively, the rates of successful bilateral mapping improved with surgeon experience (Figure 1). The cumulative rate of successful bilateral mapping increased to at least 50% after 30 surgeries and to at least 65% after 70 surgeries. After performing 40 surgeries, the cumulative rates of bilateral mapping appeared to stabilize among these higher volume surgeons. Among lower volume surgeons (<40 surgeries), no clear trend was seen. Although a plateau was achieved after 40 cases, the direction of trend towards the plateau was variable (Figure 2). There was no association between the cumulative rate of successful SLN mapping and the frequency with which surgeries were performed by an individual surgeon.
Next, the rates of removal of lymph node-containing tissue in all SLN specimens were compared. The rate of lymph node-containing tissue in all SLNs removed ranged from 67% to 92% among surgeons. Among all surgeons collectively, no linear correlation was found between the number of surgeries performed and the rate of lymph node-containing tissue in all specimens removed (P=0.56) (Figure 3). In higher volume surgeons (>40 surgeries), cumulative rates of lymph node-containing tissue removed in all specimens were ≥80%. Among lower volume surgeons, cumulative rates of lymph node-containing tissue removed in all specimens were divergent; they were either >90% or <70%, suggesting that, at least among low-volume surgeons, this was a surgeon-dependent phenomenon. Among individual surgeons, the learning curve for removal of lymph node-containing tissue was variable (Figure 4). No association was found between the cumulative rate of removal of lymph-node containing tissue and the frequency with which surgeries were performed by an individual surgeon.
Finally, the probability of successful SLN mapping and successful removal of lymph-node containing tissue increased with increasing number of procedures performed. Each additional 10 procedures performed was associated with a 5% (95% CI −5% to 16%) increase in the odds of successful SLN mapping. Each additional 10 procedures performed was also associated with an 11% (95% CI −5% to 13%) increase in the odds of removing lymph node-containing tissue.
DISCUSSION
This study aimed at measuring the impact of time and surgeon experience, as well as patient and tumor factors, on the success of SLN mapping with respect to bilateral detection and lymph node-containing specimens—two critical elements in complete staging using the SLN technique. With each 10 additional procedures performed, there was an increase in the odds of successful bilateral mapping and in the removal of lymph-node containing tissue. The results of this study suggest that surgeons can expect their rates of bilateral mapping to stabilize after 40 cases. Until that time, there is an opportunity for quality review and potentially, remediation of surgeons who are retrieving empty nodal specimens more frequently.
The limited number of existing studies exploring the learning curve for SLN mapping in endometrial cancer examine the question of mapping rates. It is also important, however, to consider successful lymph-node tissue identification. Our study deals with both questions and thus is a more complete and comprehensive representation of overcoming the learning curve for successful SLN mapping. The assessment of success in both components of the SLN technique can help to dictate when a surgeon can omit completion lymphadenectomies as part of their transition to the SLN technique.
Among surgeons, the learning curve for bilateral mapping seems to stabilize after 40 cases. Success rates of removal of lymph node-containing SLNs among surgeons appear to be a surgeon-specific phenomenon. This may highlight an objective measure of surgical quality, which can be audited and re-mediated if identified early in a surgeon’s experience. The clinical importance of extracting lymph node-containing specimens is related to the potential for unrecognized failed mapping and inadequate staging if a solitary empty nodal specimen is retrieved from a side. If the surgeon is unaware of failed mapping intra-operatively, they are unable to perform a side-specific completion lymphadenectomy as recommended by the previously established SLN algorithm.14 One way to audit individual surgeons early in their SLN learning curve would be to send SLNs for intra-operative pathology review to confirm the presence of nodal tissue.
The detection rate of bilateral SLNs in two recent meta-analyses was 50–60%,7 15 with a pooled bilateral detection rate of almost 80% with the use of indocyanine green.15 Higher SLN detection rates have been reported to be associated with intra-cervical injection and the use of indocyanine green.7 15 Additionally, patient- and tumor-related factors, including increased body mass index, clinically enlarged lymph nodes, and metastatic lymph nodes, have been associated with decreased bilateral mapping rates.12 13 16 17
Our findings agree with prior studies which show that obesity is associated with failed bilateral SLN mapping. The effect of obesity on failure can be explained by the difficulty of dissecting through fatty bundles to visualize the SLN. Although age and the presence of lymphovascular space invasion were statistically different between patients who mapped bilaterally and those who did not, the absolute effect was small, and this is unlikely to represent any clinically significant difference.
Multiple prospective studies have shown that SLN mapping and detection rates improve with surgeon experience,8 9 with 72% bilateral mapping success seen after 30 cases.9 Khoury-Collado et al8 found that after the first 30 cases, the SLN detection rate (defined as the identification of a SLN) increased from 77% to 94% (P=0.03). Data are limited, but there appears to be a similar learning curve trend for rates of removal of lymph node-containing specimens. A 3.6% reduction in the odds of an empty lymph node packet was seen with each procedure until about 30 procedures were performed, at which time the rate of empty lymph node packets stabilized.10
Surgeries performed within the first 3–4 years of our study period (August 1, 2012 to October 20, 2015) were carried out within a clinical trial (FIRES). Compared with published literature, our learning curves for successful bilateral mapping among surgeons appear more attenuated. We suspect that this could have been caused by the confines and consent process of the clinical trial, which probably limited the volume of patients and regularity with which SLN biopsy was performed. The attenuated learning curve might also have been related to fellow involvement in the procedure and the associated learning curves among trainees over the short duration of their training. Among individual surgeons, it is notable that the learning curve of some high-volume surgeons decreased initially over time, although still achieving a plateau. We speculate that this could also be related to their early surgeries being performed within the confines of a clinical trial and exclusively by the attending surgeon, with limited involvement of learners such as fellows, as compared with later in the study period where the learners are primarily performing the surgery under attending surgeon supervision.
Our study has multiple limitations, including the retrospective nature of our study design, which can limit the completeness of data and prevents us from knowing which portion of the SLN procedure was performed by the surgeon and which by the fellow. Operative time, which is an important measure of competency, was not available for the SLN portion of the procedure. Although the surgical times were available for each patient, they were variable given the involvement of multiple levels of learners in other parts of the staging surgery. Finally, although we could verify the presence of lymphoid tissue within samples removed as SLNs, removal of a lymph node positive for indocyanine green does not guarantee that this was in fact the true SLN, another important element in perfecting the SLN biopsy technique.
In conclusion, both modifiable and unmodifiable factors influence the learning curve for SLN biopsy. Surgeons at a teaching institution appear to achieve stable competency after 40 cases of bilateral SLN mapping and retrieval of lymph node-containing specimens. Within the first 40 cases, it may be beneficial to audit surgeon outcomes (bilateral detection and lymph node-containing specimens) to identify those surgeons requiring remediation of their technique.
References
Footnotes
Editor's note This paper will feature in a special issue on sentinel lymph node mapping in 2020.
Contributors KT and ER contributed to the conceptualization of the study. KT contributed to data collection. Data analysis and interpretation was done by KT, AI and ER. Figure and table creation was done by KT, AI, and ER. All authors were involved in the writing or review of the manuscript and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval Institutional Review Board at the University of North Carolina, study #19-1138, reference ID 245813.
Provenance and peer review Commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.