Sentinel lymph node intraoperative analysis in endometrial cancer

Abstract

Purpose

Surgical staging in endometrial cancer has evolved and sentinel lymph node (SLN) mapping has replaced a full pelvic and paraaortic lymphadenectomy in several cases. An intraoperative evaluation of SLN might identify patients who could benefit the most from a full lymphadenectomy. The aim of this study is to evaluate the clinical relevance of frozen section of SLN.

Methods

A retrospective analysis in patients with endometrial cancer who underwent SLN mapping with intraoperative evaluation at frozen section between February 2016 and September 2019 was performed. In case of metastatic involvement, a full lymphadenectomy was performed.

Results

Fifty-eight patients met the inclusion criteria. Clinical–pathologic characteristics of the patients and surgical data were analyzed. Overall, bilateral and unilateral detection rates were 100% (58/58), 89.7% (52/58), and 10.3% (6/58), respectively. Eight patients had a stage IIIC disease at permanent section. Frozen section detected SLN metastases in four of eight patients. Of these, two were micrometastases and two were macrometastases. At frozen section of the SLNs, no macrometastases were misdiagnosed. Sensitivity, specificity, accuracy, positive predictive value and negative predictive value of frozen section in detecting metastases was 50%, 100%, 93%, 100% and 92.6%, respectively.

Conclusion

The intraoperative evaluation of SLN in endometrial cancer accurately identifies patients with macrometastases. This is the cohort that might benefit the most of a full lymphadenectomy for a higher risk of additional lymph node metastases.

Introduction

Lymph nodal status is one of the most important prognostic factors in endometrial cancer (EC). Whether removal of metastatic lymph nodes is therapeutic is still debated. In the last decade, sentinel lymph node (SLN) mapping has emerged as a reliable approach for surgical management, reducing the morbidity of a complete lymphadenectomy, while preserving prognostic information that guide adjuvant therapy (Hollaway 2017; Tanner 2017). The role of para-aortic lymphadenectomy in the SLN surgical algorithm for EC is currently undefined. The overall frequency of paraaortic lymph node dissemination has been described between 0 and 17% (Alhilli 2013). Kumar et al. demonstrated that in the presence of pelvic metastases, 51% of patients had paraaortic lymph node dissemination, whereas in absence of pelvic metastases, 3% had isolated paraaortic lymph node dissemination (Kumar 2014). Similarly, other studies have shown that the risk of isolated paraaortic node metastases ranges from 1 to 5% (Turan 2011).

A key component of the SLN algorithm proposed in the National Comprehensive Cancer Network (NCCN) guidelines for EC is its enhanced pathological examination, also known as ultra-staging (Multinu 2019). This technique consists of an evaluation for the presence of micrometastases (tumor clusters > 0.2–2.0 mm) and isolated tumor cells (ITCs) (single tumor cells or tumor clusters ≤ 0.2 mm) by immunohistochemistry (IHC) in lymph nodes that are negative at initial examination performed using hematoxylin–eosin (H&E) staining.

In a series from MSKCC, positive nodes were detected at the initial pathological examination in 6.9% of the patients, whereas through an ultra-staging protocol and IHC an additional 4.5% of patients with low-volume metastases that would otherwise have been missed were diagnosed (Kim 2013).

Knowing the pathological status of SLNs at the time of surgery may allow to tailor intraoperatively the extension of the retroperitoneal lymphadenectomy, providing more accurate pathological information, selecting patients who could benefit from a full lymphadenectomy at the same operative time and defining the most appropriate adjuvant therapy.

The aim of this study is to compare the evaluation of the intraoperative SLNs frozen section (FS) with the permanent section (PS) in patients with a preoperative diagnosis of EC who underwent laparoscopic ICG (indocyanine green) SLN mapping.

Materials and methods

A retrospective analysis on patients with grade 1, 2 and 3 EC on preoperative endometrial biopsy undergoing laparoscopic ICG-SLN mapping at the Department of Obstetrics and Gynecology, Ospedale Civico di Lugano, in Lugano, Switzerland between February 2016 and September 2019 was performed. The study was approved by ethics committee (2019-01376-CE 3506).

The SNL mapping was performed via cervical injection of 8 ml ICG in the four quadrants of the cervix (2 ml per quadrant). One vial of 5 mg ICG powder was diluted with 20 ml of sterile water. The SLN was defined as the first juxtauterine fluorescent node with a clear afferent lymphatic vessel in each separate lymphatic compartment. The NIR signal was identified and followed to the SLN which was excised. All the SLNs were examined at FS and underwent an ultra-staging at PS. Non-SLNs (NSLNs) were analyzed at final pathological analysis according to standard protocols.

A systematic pelvic and paraaortic lymphadenectomy (PPALND) was performed for grade 3 EC and/or for those patients in whom macrometastatic disease to the SLNs or suspicious NSLNs was detected intraoperatively. A side-specific lymphadenectomy (LND) was performed when the detection of SLNs failed on one side. Macrometastases were defined as groups of malignant cells > 2.0 mm in size, micrometastases were defined as metastatic lesions between 0.2 and 2.0 mm and isolated tumor cells (ITCs) as small clusters of cells not greater than 0.2 mm, present as either single tumor cells or clusters of < 200 cells.

Clinical–pathologic characteristics of the patients were evaluated using the basic descriptive statistics. Characteristics of the SLN mapping were defined by sensitivity and specificity; clinical relevance was defined by positive and negative predictive values. Sensitivity was defined as the percentage of true positives; specificity was defined as the percentage of true negatives. As true positives (TP) were considered those patients in whom a LND was indicated based on FS of SLN and who presented metastatic disease to the SLNs at PS. As false negatives (FN) were considered those patients in whom a LND was not indicated based on FS of SLN and who presented metastatic disease to the SLNs at PS. As true negatives (TN) were defined those patients in whom a LND was not performed based on FS of SLN and who has negative SLNs at PS. The false positive rate was defined as zero. Sensitivity, specificity, positive predictive value and negative predictive value of FS were calculated on the number of SLNs and on the number of patients. Statistical analyses were performed using the StataMP software (version 14).

Results

Fifty-eight patients met the inclusion criteria and were included in the study. Clinical–pathologic characteristics of the patients and surgical data are presented in Tables 1 and 2, respectively.

Table 1.

Patients’ characteristics

Number of patients	58
Median age in years (range)	65 (30–85)
Median BMI in Kg/m2 (range)	27 (16.0–48.0)
FIGO Stage (%)
IA	35 (60.3)
IB	11 (19.0)
II	4 (6.9)
IIIA	0 (0)
IIIB	1 (1.7)
IIIC	7 (12.0)
Tumor size (%)
< 2 cm	29 (50)
≥ 2 cm	29 (50)
Grading (%)
NA	1 (1.7)
G1	22 (37.9)
G2	27 (46.6)
G3	8 (13.8)
LVSI (%)
Negative	45 (77.6)
Positive	13 (22.4)
Hystology (%)
Endometrioid	54 (93.1)
UPSC	4 (6.9)

Table 2.

Surgical data

OR time in minutes (range)	167 (80–330)
EBL in mL (range)	123 (50–600)
Hospital stay in days (range)	6 (2–20)
Postoperative complications (%)
Absent	54 (93.1)
Present	2* (3.5)
Urinary infection	1 (1.7)
Nerve injury	1 (1.7)

*One patient had an UTI and a nerve injury

Overall and bilateral detection rates were 100% and 89.7%, respectively, as reported in Table 3.

Table 3.

SLN mapping performance and test characteristics

Detection rate (%)
Overall	58 (100)
Bilateral	52 (89.7)
Unilateral	6 (10.3)
Median number of SLNs (range)
Pelvic	3 (1–8)
Para-aortal	0 (0–2)
Patients with SLNs metastases (%)
At frozen section	4 (6.9)
At permanent section	8 (13.8)
Sensitivity (%)
SLNs	25
Patients	50
Specificity (%)
SLNs	100
Patients	100
Positive predictive value (%)
SLNs	100
Patients	100
Negative predictive value (%)
SLNs	92.9
Patients	92.6
Test accuracy (%)
SLNs	93
Patients	93

Four (6.9%) patients had positive SLNs at FS. In eight cases (13.8%) SLNs revealed metastases at PS, of which two were macrometastases and three micrometastases (Fig. 1). ITCs were present in four SLNs (6.9%) and one patient had both micrometastases and ITC at PS (Table 4).

Fig. 1.

Representative frozen sections of sentinel lymph node macrometastases (a). Representative permanent section performed using hematoxylin–eosin (H&E) staining of sentinel lymph node macrometastases (b). Representative permanent section performed using hematoxylin–eosin (H&E) staining of sentinel lymph node micrometastases (c). Representative permanent section performed using immunohistochemistry (IHC) for pancytokeratin AE1/AE3 of sentinel lymph node micrometastases (d). Representative permanent section performed using hematoxylin–eosin (H&E) staining of sentinel lymph node with isolated tumor cells (e). Representative permanent section performed using immunohistochemistry (IHC) for pancytokeratin AE1/AE3 of sentinel lymph node with isolated tumor cells (f). Arrows indicate metastatic tumor foci

Table 4.

SLN mapping: frozen versus permanent section

	Frozen section (%)	Permanent section (%)
Number of SLNs	185	202
SLNs metastases	4 (2.2)	16 (7.9)
Type of metastases
Macrometastases	2 (1.1)	2 (1.0)
Micrometastases	2 (1.1)	8 (3.4)
ITC	0 (0)	6 (3.0)
Localization of metastatic SLNs
Right external iliac artery	4 (100)	7 (43.8)
Left external iliac artery	0	1 (6.3)
Right obturator fossa	0	3 (18.8)
Left obturator fossa	0	5 (31.3)
Localization of metastatic NSLNs
Right external iliac artery	NA	1 (20%)
Left external iliac artery		1 (20%)
Right obturator fossa		2 (40%)
Paraaortic		1 (20%)
Empty nodes	12 (6.5)	9 (4.5)

After the identification of metastatic disease to the SLNs at FS, 2 (3.4%) and 1 (1.7%) patients underwent a PLND and a PPALND, respectively. Due to a G3 EC and/or clinically suspicious NSLNs four (6.9%) patients underwent a PPALND. Three (5.2%) patients performed a side-specific PLND for unilateral mapping.

In patients who underwent a completion PPALND, five additional metastatic lymph nodes were identified in two patients and one (1.7%) patient had metastatic paraaortic lymph nodes at PS. Metastatic SLNs were localized in the following compartments: right external iliac artery (43.8%), right obturator fossa (18.8%), left external iliac artery (6.3%) and left obturator fossa (31.3%). In four (6.9%) cases, FS of SNLs failed to identify metastatic disease. The characteristics of the SLN mapping and its clinical relevance are reported in Table 3.

The false negative (FN) rate of the FS in identifying patient with metastatic lymph nodes was 50% with a NPV and a PPV of 92.6% and 100%, respectively. The overall FN rate of the FS in identifying metastatic lymph nodes was 75%. The FN rate was 100% for ITCs, 75% for micrometastases and 0% in macrometastases. NPV and PPV of the FS for the identification of metastatic lymph nodes was 92.9% and 100%, respectively.

Discussion

In our study, the overall and bilateral detection rates were 100% and 89.7%. Papadia et al. (2016) reported a 96% and 88% overall and bilateral detection rate, respectively, in a series of 75 patients undergoing ICG-SLN mapping (Papadia 2016), FIRES trial confirmed the diagnostic accuracy of this technique with an overall detection rates of 86% and a sensitivity of 97.2% (Rossi 2017). Due to the high bilateral detection rate, in our study, there was little need to adopt the MSKCC SLN mapping algorithm performing a side-specific lymphadenectomy when the detection of a SLN failed on one side (Barlin 2012). Our excellent overall and bilateral detection rates confirm the high quality of the NIR-ICG SLN mapping technique that has consistently proven to be better than the combination of blue dye and Tc-99 in uterine malignancies (Ruscito 2016; Imboden 2015; Buda 2016; Papadia 2018). The detection rate is still high even if the concentration of ICG injected is lower than that of other series in the literature. In our institution, we inject a volume of 8 ml of ICG solution at a concentration of 0.25 mg/ml. This concentration is significantly lower than the dose used by other authors, who typically inject a total of 4 ml at a concentration of 1.25 mg/ml (Ruscito 2016).

Most of the SLNs of our series mapped in the pelvis and only eight patients had a SLN biopsy in the paraaortic region, for a total of nine SLNs removed, none of them with metastasis. Cormier et al. found that the pattern of SLN mapping is influenced by the injection site: in his series, the para-aortic SLN detection rate was 39% after corporeal injection, 2% after cervical injection and 17% after deeper cervical injection (Cormier 2011). In the GOG protocol 33, the incidence of isolated paraaortic metastases was only 2% (Creasman 1987).

We wonder if a different anatomic site of tracer injection such as the uterine cavity or the fundus would have led to the identification of metastatic para-aortic NSLNs as SLNs. The hysteroscopic injection of ICG for the SLN mapping is feasible and has been associated with higher number of SLNs identified in the para-aortic region (Ditto 2015).

Coherent with the available literature, micrometastases and ITC account for approximately 50% of the SLN metastases (Papadia 2018). The real clinical meaning of micrometastases and ITC is still unknown: Plante suggests that the presence of ITC does not impair survival and should not be considered an indication to adjuvant treatment (Plante 2017). In the future, the information derived from genomic profiling and its integration with the pathologic data may clarify the biologic significance of small volume lymph nodal disease.

Touhami et al. demonstrated that the risk of harboring metastatic NSLNs correlates to the size of the metastases to the SLNs (Touhami 2017). This is negligible for low volume disease and is as high as 17.7% when macrometastases are detected in the SLNs. This has been confirmed by the series of Monterossi et al. in which none of the patients with micrometastases in the SLNs had positive  NSLNs, whereas 100% of patients with SLNs macrometastases had at least one positive NSLN (Monterossi 2019). Consequently, the patients we want to identify the most are those with macrometastases, for whom our methodology worked out well.

Frozen section has intrinsic limitations in giving more detailed information on the architectural and cytological characteristics of the lesions. Time constrains, limitation in number of sections as compared to ultra-staging and the need to leave the tissue for definitive diagnosis may lead to erroneous interpretations and inadequate sampling (Papadia 2019).

Recently, Fanfani et al. reported that one step nucleic amplification assay (OSNA) technique, compared with standard ultra-staging, could be an alternative and a more accurate technique for the detection of nodal metastases leading to an upstage in 20% of patients (Fanfani 2018).

In the era of SLN mapping in EC, there are nowadays three possible strategies:

1. SLN mapping without intraoperative evaluation of the SLNs. If metastatic disease to the SLNs is detected at final pathological analysis, the adjuvant strategy will be based on available information.

2. SLN mapping without intraoperative evaluation of the SLNs. If metastatic disease to the SLNs is detected at final pathological analysis, a second surgery to complete the lymph nodal staging is performed. In our study, if this approach had been performed, eight patients (13.8%) would have needed a second surgery for a complete LN staging.

3. SLN mapping with intraoperative evaluation of the SLNs. If metastatic disease to the SLNs is detected a full lymphadenectomy to complete the lymph nodal staging is performed.

Renz et al. recently demonstrated that immediate SLN mapping with FS has a high sensitivity and negative predictive value of 83.3% and 98.7%, respectively. Nevertheless, their protocol did not include IHC and ultrastaging at permanent section (Renz 2020). Metastatic lymph nodes in EC determine an upgrade to FIGO stage III C. Stage III C is further categorized as III C1 (indicating positive pelvic nodes) and III C2 (indicating positive para-aortic nodes with or without positive pelvic nodes) (Creasman 2009). A variety of different adjuvant treatments have been proposed for these patients, with the most commonly adopted being concurrent chemo-radiotherapy (De Boer 2019). However, a debate is still ongoing on which treatment offers the best oncologic outcome. In GOG#258, an increase of locoregional failure with use of chemotherapy alone has been registered in 813 women with stage III–IV EC (Matei 2019). The results of the PORTEC-3 trial did not show any benefit of a combined adjuvant chemo-radiotherapy as compared to radiotherapy alone in node negative high-risk EC patients (De Boer 2019).

A complete macroscopic cytoreduction and comprehensive staging is recommended in advanced EC (Lambrou 2004) and a LND is suggested for patients with high risk tumors (grade 3 with deep myometrial invasion > 50%) (Colombo 2016). However, for patients with early-stage disease, who had undergone systematic PLND, two randomized trials showed no improvement in disease-free or overall survival (Benedetti Panici 2008; ASTEC study group 2009). Nevertheless, the absence of impact on survival of a systematic LND has been questioned by the SEPAL study (Todo 2010), demonstrating that women with intermediate or high-risk EC had a longer survival if a complete PPALND had been performed instead of a PLND only.

Although a systematic LND does not seem to have a direct effect on survival, it provides important pathologic information that define prognosis and can help to tailor the most appropriate adjuvant treatment (Gasparri 2019). Interestingly, in a large multicenter retrospective study on patients with high-intermediate and high-risk EC, the worst survival was recorded for those patients in whom the lymph nodal status was unknown (Ouldamer 2017). Patients with pathologically negative lymph nodes are less often subjected to adjuvant radiotherapy as compared to patients in whom the lymph nodal status is unknown (Bogani 2014). There is retrospective evidence that a SLN biopsy may safely replace a full lymphadenectomy without affecting survival (Bogani 2020; Buda 2018). However, at present, there are no series focusing only on lymph node positive patients and the previously reported results might be biased by the large number of lymph node negative patients.

Given the controversial results reported from the PORTEC 3 trial and the GOG 258 with regards to the best adjuvant treatment for advanced stage endometrial cancer patients, depending on the scheme adopted for lymph node positive patients (combined chemo-radiotherapy as proposed by the PORTEC 3 trial or chemotherapy based on GOG 258), knowledge of the extension of the lymph nodal involvement may be relevant (DeBoer 2019; Matei 2019). In case of involvement of para-aortic LNs, an extension of the radiation fields to the paraaortic region may be indicated and may provide a survival improvement. Moreover, setting a precise standard for the adjuvant therapy, tailored on the extend of the LN disease, may provide precise data for designing future trials aimed to define the best adjuvant therapy.

Frozen section of SLNs can help in giving intraoperative pathological information facilitating the decision whether a complete PPALND is required. A strategy based on intraoperative evaluation of the SLNs results in a shorter operative time as compared to a PPALND and a smaller morbidity. On the other hand, the need of the intra-operative pathology and the presence of a pathologist for an instant analysis have to be considered. Furthermore, due to the low overall incidence of lymph nodal metastases in EC, many SLNs need to be tested to identify the metastatic ones.

Strengths of the study include the standardized use of ICG tracer and laparoscopic performance of a single center. Moreover, all the SLNs and NSLNs pathological analysis was centralized in one institution, reducing the bias between different FS and PS readings. This can explain the consistency of the pathological analysis that well correlates with those already reported in the literature. However, some limitations of the present study have to be underlined, including the limited sample size of patients and the retrospective nature of the analysis.

In conclusion, although an intraoperative evaluation of the SLNs in EC patients has a low sensitivity in the detection of micrometastases and ITCs, it accurately identifies all the patients with macrometastases. These patients are at the highest risk of harboring additional lymph nodal metastases and represent the cohort who might benefit the most of a complete PPALND. Additional studies addressing the cost effectiveness of this approach and the oncologic outcome in this setting are needed.

Author contributions

SB: data curation, formal analysis, writing—original draft preparation, editing. MB: data curation, formal analysis, writing—original draft preparation. MLG: supervision, review. LC: data curation, Software. PM: resources. LM: resources. AP: conceptualization, project, review.

Funding

No external funding was obtained for the purpose of the research.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethics approval

The study was approved by ethics committee (2019-01376-CE 3506).

Consent for publication

Authors provide consent for publication.