Background
The accuracy of sentinel lymph node mapping has been shown in endometrial cancer, but studies to date have primarily focused on cohorts at low risk for nodal involvement. In our practice, we acknowledge the lack of benefit of lymphadenectomy in the low-risk subgroup and omit lymph node removal in these patients. Thus, our aim was to evaluate the feasibility and accuracy of sentinel node mapping in women at sufficient risk for nodal metastasis warranting lymphadenectomy and in whom the potential benefit of avoiding nodal procurement could be realized.
Objective
To evaluate the detection rate and accuracy of fluorescence-guided sentinel lymph node mapping in endometrial cancer patients undergoing robotic-assisted staging.
Study Design
One hundred twenty-three endometrial cancer patients undergoing sentinel lymph node sentinel node mapping using indocyanine green were prospectively evaluated. Two mL (1.0 mg/mL) of dye were injected into the cervical stroma divided between the 2−3 and 9−10 o’clock positions at the time of uterine manipulator placement. Before hysterectomy, the retroperitoneal spaces were developed and fluorescence imaging was used for sentinel node detection. Identified sentinel nodes were removed and submitted for touch prep intraoperatively, followed by permanent assessment with routine hematoxylin and eosin levels. Patients then underwent hysterectomy, bilateral salpingo-oophorectomy, and completion bilateral pelvic and periaortic lymphadenectomy based on intrauterine risk factors determined intraoperatively (tumor size >2 cm, >50% myometrial invasion, and grade 3 histology).
Results
Of 123 patients enrolled, at least 1 sentinel node was detected in 119 (96.7%). Ninety-nine patients (80%) had bilateral pelvic or periaortic sentinel nodes detected. A total of 85 patients met criteria warranting completion lymphadenectomy. In 14 patients (16%) periaortic lymphadenectomy was not feasible, and the mean number of pelvic nodes procured was 13 (6−22). Of the 71 patients undergoing pelvic and periaortic lymphadenectomy, the mean nodal count was 23.2 (8−51). Of patients undergoing lymphadenectomy, 10.6% had lymph node metastasis on final hematoxylin and eosin evaluation. Notably, the sentinel node was the only positive node in 44% of cases. There were no cases in which final pathology of the sentinel node was negative and metastatic disease was detected upon completion lymphadenectomy in the non-sentinel nodes (no false negatives), yielding a sensitivity of 100%. Of the 14 sentinel nodes ultimately found to harbor metastases, 3 were negative on touch prep, yielding a sensitivity of 78.6% for intraoperative detection of sentinel node involvement. In all 3 of the false-negative touch preps, final pathology detected a single micrometastasis (0.24 mm, 1.4 mm, 1.5 mm). As expected, there were no false-positive results, yielding a specificity of 100%. No complications related to sentinel node mapping or allergic reactions to the dye were encountered.
Conclusion
Intraoperative sentinel node mapping using fluorescence imaging with indocyanine green in endometrial cancer patients is feasible and yields high detection rates. In our pilot study, sentinel node mapping identified all women with Stage IIIC disease. Low false-negative rates are encouraging, and if confirmed in multi-institutional trials, this approach would be anticipated to reduce the morbidity, operative times, and costs associated with complete pelvic and periaortic lymphadenectomy.
Endometrial cancer (EC) is the most common reproductive tract malignancy in North America. In the majority of cases, women will be diagnosed at an early stage with uterine-limited disease, amenable to surgical extirpation alone. In 1988, The International Federation of Gynecology and Obstetrics (FIGO) recognized the shortcomings of clinical staging and formally advocated surgical staging to delineate disease distribution and guide adjuvant therapy. More than 2 decades later, a consensus regarding the criteria for assessment of lymphatic involvement remains elusive, and patient care generally reflects institutional or physician biases.
Several aspects of EC have contributed to this debate, including an overall low risk of nodal metastasis, a high proportion of grade 1 or 2 tumors with a favorable prognosis and low risk of extrauterine disease, and an apparent lack of a survival advantage with lymphadenectomy (LAN) documented in 2 large randomized trials that failed to show a beneficial effect of pelvic LAN on overall or recurrence-free survival.
A recent single-institution trial has redirected focus from a debate between universal staging vs abandoning surgical assessment of regional nodes in EC to the identification of a low-risk group of patients who do not benefit from either LAN or adjuvant radiation therapy. Mariani et al. reported that in more than 300 patients with FIGO grade 1 or 2 endometrioid adenocarcinoma with less than 50% myometrial penetration, no patients with tumor size less than 2 cm had documented lymph node metastases, developed lymph node recurrences, or died of their disease. Subsequently, their findings were validated by other groups as well as in a prospective series at the Mayo Clinic.
The results of the prospective Gynecologic Oncology Group (GOG) Lap II trial confirmed that there is a low-risk subgroup in which LAN is associated only with potential morbidity and no survival benefit. In it, 7% of women with endometrioid histology participating in Lap II had nodal metastasis. Of these, 48% had less than 50% myometrial invasion, 60% had grade 1 or 2 tumors, and 23% had a tumor diameter less than 2 cm. Notably, it was only when all 3 of these criteria were simultaneously present that the risk of lymph node metastases decreased to an acceptably low level at 0.8%, further validating the Mayo Clinic criteria.
Sentinel lymph node (SLN) mapping has become the standard of care to assess lymphatic dissemination in breast cancer, melanoma, and vulvar cancer. The clinical value of SLN mapping is based on a high sensitivity and reliable negative predictive value to detect nodal involvement. Ideally, SLN mapping would be highly feasible and provide the accuracy of systematic LAN in identifying patients with lymph node metastasis while decreasing surgical morbidity and operative times. The accuracy of SLN mapping has been shown in patients with early-stage EC at low risk for nodal involvement. Studies in low-risk cohorts report rates of high detection (>80%) with false-negative rates ranging from 2 to 16%. In our practice, we acknowledge the lack of benefit of LAN in the low-risk subgroup and omit systematic LAN in these patients. Thus, the aim of our study was to prospectively evaluate the feasibility of routine SLN mapping and the accuracy of SLN histology in women at sufficient risk for nodal metastasis that warranted systematic LAN in whom the potential substantial benefit of avoiding full LAN could be realized.
Materials and Methods
Patients and procedure
From March 2012 through May 2015, all patients with biopsy-proven EC of any histologic subtype presenting to 1 of 4 gynecologic oncologists (P.J.P., D.S.V., C.A.S., J.Z.P.) with planned robotic-assisted surgical management at 1 of 2 Swedish Medical Center campuses were offered enrollment. The study was approved by the Swedish Medical Center institutional review board, and all participants provided written informed consent. All data were collected prospectively and entered into a secure database. Patients were followed daily while in the hospital, and outcomes at 1−2 weeks and 10−12 weeks postoperatively were collected.
Two milliliters (1.0 mg/mL) of indocyanine green (ICG; Alcorn Pharmaceuticals, Lake Forest, IL) was injected into the cervical stroma divided between the 2−3 and 9−10 o’clock positions at the time of uterine manipulator placement. Before hysterectomy, the retroperitoneal spaces were developed and robotic camera fluorescence imaging used for SLN mapping. The fluorescence properties of ICG were visualized by the use of an imaging system integrates the SPY scope near-infrared (NIR) imaging technology (Novadaq Technologies, Bonita Springs, FL) with the daVinci Surgical System (Intuitive Surgical, Sunnyvale, CA). Identified SLNs were removed and submitted intraoperatively for touch prep evaluation, followed by routine hematoxylin and eosin (H&E) levels.
Patients subsequently underwent robotic hysterectomy, bilateral salpingo-oophorectomy, followed by completion bilateral pelvic and periaortic LAN (PPALAN) based on intrauterine risk factors as determined by intraoperative frozen section. PPALAN was performed in patients with at least 1 of the following findings: serous, clear-cell, or carcinosarcoma histology on preoperative biopsy, tumor size >2 cm, >50% myometrial wall invasion, grade 3 histology, grossly suspicious nonsentinel pelvic or periaortic lymph nodes, or SLN positive for metastatic disease on touch prep. In cases of mapping failures, completion PPALAN was performed on the basis of the same criteria followed in successful mapping cases. The number and location of SLN was documented prospectively.
Histopathology
All identified SLNs were submitted intraoperatively for imprint cytology. Each SLN was cut perpendicular to the long axis with 1 or more touch preps per SLN being reviewed. Each SLN was submitted entirely with multiple sections at an interval of 100−200 microns between each level evaluated by routine H&E staining. The nonsentinel nodes were processed similarly, being entirely submitted for routine H&E levels.
Statistical analysis
The detection rate was calculated as the number of patients with at least 1 identified SLN divided by the total number of enrolled patients. The accuracy of SLN mapping was determined by calculating the sensitivity, specificity, and positive and negative predictive values along with 95% confidence intervals (95% CIs) only in those cases in which at least 1 SLN was identified and completion LAN was performed. A value of 100% represents perfect performance of the marker and 50% indicates a level of performance that is expected by chance alone. All statistical analysis was performed with STATA (14.0 for Mac OS X, College Station, TX).
Results
One hundred twenty-three patients with EC underwent attempted SLN mapping with key demographic and clinical features outlined in Table 1 . The mean age of all participants was 62.5 years (range, 42−86 years) with a mean body mass index of 32 kg/m 2 (range, 19−63 kg/m 2 ). Characteristic of the EC population, 51% of participants were obese and 20% morbidly obese. All patients who provided consent experienced successful completion of robotic procedures with no conversions to laparotomy and no malfunctions of the SPY scope NIR imaging technology. The average operating room time, defined as skin incision to skin closure, was 161 minutes (range, 60−360 minutes), with a mean estimated blood loss of 48.7 mL (range, 15−200 mL) and an average length of hospital stay of 24.6 hours (range, 15−92 hours). An obstetrics and gynecology resident or gynecologic oncology fellow performed a key component of the case at the surgeon console in 66% of the cases.
| Characteristic | |
|---|---|
| Age, y | 62.5 (42−86) |
| Body mass index, kg/m 2 , n (%) | 32 (19–63) |
| >30 | −63 (51) |
| >40 | −25 (20) |
| Operating room time, min | 161 (60−360) |
| Estimated blood loss, mL | 48.7 (15−200) |
| Length of hospital stay | 24.6 (15−92) |
| Resident and/or fellow on surgeon console, n (%) | 81 (66) |
In 123 patients in whom SLN mapping was attempted, 342 total SLNs were identified and procured intraoperatively ( Table 2 ). The average time from intracervical ICG injection to SLN identification was 19.8 minutes, with a range of 5−55 minutes. At least 1 SLN was identified in 119 patients for a detection rate of 96.7%. In 99 patients (80%) bilateral pelvic and/or periaortic SLNs were mapped successfully. On average 1.6 right SLN (range 1−6) and 1.5 left SLN range 1−5) were detected per patient. SLNs were most commonly identified in the external iliac basins (49% of cases), followed by the obturator fossa (21%), common iliac (12%), periaortic region (8%), presacral (1.5%), and parametrial web (<1%). In 10 cases (3%), the location was not specified and could not be ascertained from review of the operative report or pathology report in the patient’s electronic medical record.
| Variable | |
|---|---|
| Time from ICG injection to SLN identification, min, mean (range) | 19.8 (5−55) |
| Number of right SLN per patient, mean (range) | 1.6 (1−6) |
| Number of left SLN per patient, mean (range) | 1.5 (1−5) |
| SLN detection rate, n (%) | |
| Overall | 119 (96.7%) |
| Bilateral | 99 (80%) |
| Failed | 4 (3.3) |
| SLN location, n (%) | |
| External iliac | 167 (49%) |
| Obturator | 72 (21%) |
| Common iliac | 42 (12%) |
| Periaortic | 29 (8%) |
| Internal iliac | 16 (5%) |
| Not recorded | 10 (3%) |
| Presacral | 5 (1.5%) |
| Parametrial | 1 (<1%) |
Eighty-five patients met criteria warranting PPALAN. In 14 patients (16%), PALAN was not feasible and the mean number of pelvic nodes procured was 13 (range, 6−22). Of the 71 women undergoing PPALAN, the mean nodal count was 23.2 (range, 8−51). Table 3 illustrates the clinicopathologic characteristics of the 85 patients undergoing completion LAN. As anticipated, most women had endometrioid histology (82%), and on final pathology 38% had FIGO grade 1, 28% grade 2, and 34% grade 3 tumors.
| Variable | n (%) |
|---|---|
| Stage (final pathology) | |
| Ia | 49 (57%) |
| Ib | 22 (26%) |
| II | 1 (1%) |
| IIIa | 4 (5%) |
| IIIc1 | 4 (5%) |
| IIIc2 | 5 (6%) |
| Grade (final pathology endometrioid only) | |
| 1 | 32 (46%) |
| 2 | 24 (34%) |
| 3 | 14 (20%) |
| Histology (final pathology) | |
| Endometrioid | 70 (82%) |
| Serous | 10 (12%) |
| Clear cell | 3 (3.5%) |
| Carcinosarcoma | 2 (2.5%) |
| Intraoperative tumor size, cm | |
| ≤2×2 | 8 (9%) |
| >2×2 | 77 (91%) |
a Overall, 29 of 85 (34%) patients in whom completion lymphadenectomy was performed had high-grade histologies of any histologic subtype.
Table 4 summarizes the performance of SLN mapping in 85 patients undergoing completion LAN. Of these patients, 10.6% had lymph node metastasis on final H&E evaluation. Notably, the SLN was the only positive node in 44% of cases. Of the 14 sentinel nodes ultimately found to harbor metastases in 9 patients, 3 were negative on intraoperative touch prep, yielding a sensitivity of 78.6% (95% CI, 49.2−95.5%) for detection of SLN involvement with this technique. In all 3 of the false-negative touch preps, H&E detected a single micrometastasis (0.24, 1.4, and 1.5 mm). There were no cases in which H&E of the sentinel node was negative and metastatic disease was detected on PPALAN in the nonsentinel nodes (no false negatives), yielding a sensitivity of 100% (95% CI, 66.4−100%). There were no false-positive results, yielding a specificity of 100%. Of the women with stage IIIC disease, 33.3% had only 1 criteria for proceeding with PPALAN, 67% of which was tumor size >2×2 cm in women with grade 1 endometrioid tumors with <50% myometrial invasion.
| Variable | n (%) |
|---|---|
| Patients with any nodal metastasis after completion lymphadenectomy on final H&E | 9 (10.6%) |
| Sensitivity of SLN mapping | |
| Touch prep | 11/14 (78.6%) 95% CI (49.2−95.5%) |
| Final H&E | 9/9 (100%) 95% CI (66.4−100%) |
| Patients with isolated metastasis in SLN after completion lymphadenectomy on final H&E | 4/9 (44%) |
No complications directly attributable to SLN mapping or the ICG dye were encountered. Minor complications occurred in 8 of 123 patients (6.5%), including transient numbness and weakness in the distribution of the left ulnar nerve (n = 1), uncomplicated urinary tract infection (n = 3), lymphocele (n = 2), and puckering of a skin closure site (n = 1). Major complications occurred in 4 of 123 patients (3.3%), including vaginal cuff dehiscence (n = 1), pulmonary embolism (n = 1), postoperative anemia resulting in atrial fibrillation requiring transfusion (n = 1), and pelvic abscess requiring antibiotics and drainage (n = 1).
Results
One hundred twenty-three patients with EC underwent attempted SLN mapping with key demographic and clinical features outlined in Table 1 . The mean age of all participants was 62.5 years (range, 42−86 years) with a mean body mass index of 32 kg/m 2 (range, 19−63 kg/m 2 ). Characteristic of the EC population, 51% of participants were obese and 20% morbidly obese. All patients who provided consent experienced successful completion of robotic procedures with no conversions to laparotomy and no malfunctions of the SPY scope NIR imaging technology. The average operating room time, defined as skin incision to skin closure, was 161 minutes (range, 60−360 minutes), with a mean estimated blood loss of 48.7 mL (range, 15−200 mL) and an average length of hospital stay of 24.6 hours (range, 15−92 hours). An obstetrics and gynecology resident or gynecologic oncology fellow performed a key component of the case at the surgeon console in 66% of the cases.
| Characteristic | |
|---|---|
| Age, y | 62.5 (42−86) |
| Body mass index, kg/m 2 , n (%) | 32 (19–63) |
| >30 | −63 (51) |
| >40 | −25 (20) |
| Operating room time, min | 161 (60−360) |
| Estimated blood loss, mL | 48.7 (15−200) |
| Length of hospital stay | 24.6 (15−92) |
| Resident and/or fellow on surgeon console, n (%) | 81 (66) |
In 123 patients in whom SLN mapping was attempted, 342 total SLNs were identified and procured intraoperatively ( Table 2 ). The average time from intracervical ICG injection to SLN identification was 19.8 minutes, with a range of 5−55 minutes. At least 1 SLN was identified in 119 patients for a detection rate of 96.7%. In 99 patients (80%) bilateral pelvic and/or periaortic SLNs were mapped successfully. On average 1.6 right SLN (range 1−6) and 1.5 left SLN range 1−5) were detected per patient. SLNs were most commonly identified in the external iliac basins (49% of cases), followed by the obturator fossa (21%), common iliac (12%), periaortic region (8%), presacral (1.5%), and parametrial web (<1%). In 10 cases (3%), the location was not specified and could not be ascertained from review of the operative report or pathology report in the patient’s electronic medical record.
