Study characteristics

Patient Sampling

Country: USA

Study design: a prospective, non‐randomised study of women presenting with a preoperative diagnosis of endometrial cancer with clinical stage I disease conducted between March 2006 and August 2008.

Inclusion criteria: preoperative diagnosis of endometrial cancer with clinical stage I disease; performance status of 0, 1, 2, or 3 by the Gynecologic Oncology group criteria; patients agreed to undergo total hysterectomy, removal of both adnexae, and bilateral regional lymphadenectomy via laparotomy or laparoscopy.

Exclusion criteria: not reported

Patient characteristics and setting

Number of patients: 42

Median age: 60.5 years (range, 34‐82 years)

Median body mass index: 29.3 kg/m2 (range, 19‐61 kg/m2)

Histopathological cell type: endometrial endometrioid carcinoma

FIGO stage: IA = 24 (57%), IB = 9 (21%), IC = 1 (2%), IIA = 1 (2%), IIIA = 3 (7%), IIIC = 3 (7%), IVB = 1 (2%)‐ old FIGO staging system used.

Grade on hysterectomy specimen: grade 1 = 32 (76%), grade 2 = 9 (21%), grade 3 = 1 (2%)

Lymphovascular space involvement: present in 7 cases (17%), absent in 35 cases (83%)

Setting: single tertiary centre, namely Memorial Sloan‐Kettering Cancer Centre in New York, USA

Index tests

Type of endometrial sampling: office endometrial biopsy in 17 cases (40%), dilatation and curettage in 25 cases (60%)

Experience of operator: 3 failed cases were performed by surgeons with beginner's experience (1‐2 cases) in SLN mapping early on in the study

Tracer used and amount: 0.1 ‐ 0.5 mci Tc99 in 0.1 ‐ 0.5 mL volume; and 2 cc (n = 21) or 4 cc (n = 21) of isosulfan or methylene blue dye.

Method and timing of application: Tc99 was injected using a 27‐gauge Potocky needle into the stroma of the cervix at 3 and 9 o'clock either on the morning of (n = 14; 33%), or day prior to surgery (n = 28; 67%) and a preoperative lymphoscintigram was taken. Blue dye was injected into the cervix at the beginning of the operation using a spinal needle. In the first half (n = 21) of the cohort study the injection of blue dye was into the cervical stroma at the 3 and 9 o'clock positions; for patients enrolled in the second half of the study cohort, 2 additional blue dye injections into were added that were injected into the anterior mid fundus and posterior mid fundus using a 22‐gauge 36‐cm Nezhat‐Dorsey aspiration/injection laparoscopic needle.

Method of detection: from each paraffin block lacking metastatic carcinoma appreciable in a routine section stained with a haematoxylin and eosin (H&E), 2 adjacent 5‐micron sections were cut at each of two levels 50 microns apart. At each level, one slide was stained with H&E and the other with immunohistochemistry (IHC) using the anti‐cytokeratin AE1:AE3 (Ventana Medical Systems, Inc., Tucson, AZ) for a total of 4 slides per block.

Target condition and reference standard(s)

Type: patients with G1 endometrial cancer all of whom had pelvic and para‐aortic dissection performed to the level of the right ovarian vein/IMA. Twenty‐five cases (60%) were performed by laparoscopic surgery, and 17 (40%) were treated by laparotomy.

Lymph node number and site: total of 145 SLNs were identified; 58 SLNs (40%) located in the right pelvis; 55 SLNs (38%) located in the left pelvis; 4 SLNs (3%) were in right para‐aortic nodes; 1 SLN (0.6%) was in left para‐aortic nodes.

The most common anatomical sites were: internal iliac nodes = 52 (36%), external iliac = 43 (30%), obturator = 34 (23%), common iliac = 11 (8%), para‐aortic = 5 (3%).

Flow and timing

All patients received the index and reference standard within 1 month. All patients received the same reference standard of pelvic and para‐aortic dissection performed to the level of the right ovarian vein/IMA. All patients were included in the analysis.

Comparative

Notes

Study results: preoperative lymphoscintigraphy visualised SLNs in 30 patients (71%); intraoperative localisation of the SLN was possible in 36 (86%). A median of 3 SLNs (range, 1–14) and 14.5 non‐SLNs (range, 4–55) were examined. Six patients (14%) had a failed mapping, with no SLN identified.

In all, 4/36 (11%) had positive SLNs—3 seen on H&E and 1 as cytokeratin‐positive cells on IHC. All node‐positive cases were picked up by the SLN; there were no false‐negative cases. The sensitivity of the SLN procedure in the 36 patients who had an SLN identified was 100%. TP rate was 4. The FN rate was 6. Of the 42 patients, 38 had negative nodes following full dissection, although only 32 were detected by SLNB (TN) and 4 patients had positive nodes (TP). There were no FP nodes.

TP = 4; FP = 0; FN = 0; TN = 32; failed = 6. Type 1 tumours; FIGO Stage IA = 24; Stage Ib+ =18. Cervical injections; bilateral detection rate = not reported.

Adverse reaction from index or reference test: Not reported.

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Unclear

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

Yes

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Low risk

Study characteristics

Patient Sampling

Country: Iran

Study design: cross‐sectional pilot study for patients undergoing staging surgery for stage I and II endometrial or cervical cancer (15 endometrial, 8 cervical) between November 2012 and February 2014

Inclusion criteria: patients with stage I and II endometrial or cervical cancer who were candidate for systematic lymph node dissection

Exclusion criteria: patients with prior radiotherapy

Patient characteristics and setting

Number of patients: 23 (15 endometrial, 8 cervical)

Mean age: 63.47 +/‐ 1.09 years

Mean body mass index: 29.13 +/‐ 0.5 kg/m2

Histopathological cell type: endometrioid carcinoma = 13 (86.7%), clear cell carcinoma = 1 (6.7%), papillary serous carcinoma = 1 (6.7%)

FIGO stage (pre 2009): IA = 7 (46.7%), IB = 3 (20%), IC= 1 (6.7%), IIA = 1 (6.7%), IIB = 2 (13.3%), IIIA = 1 (6.7%). Estimated FIGO stage (post 2009) = IA=10; IB+=5.

Grade on hysterectomy specimen: not reported.

Lymphovascular space involvement: not reported.

Setting: two centres in Isfahan, Iran, namely Shahid Beheshti and Sadoughi Hospitals.

Index tests

Type of endometrial sampling: not reported.

Experience of operator: mention of limitations for access to trained surgeons, though no explicit report of experience of operators.

Tracer used and amount: 4 mL of 1% methylene blue

Method and timing of application: injection into the fundus of the uterus after clamping the fallopian tubes using a 25‐gauge needle. To prevent spillage of dye, fundal gentle pressure on the sites of injection was used.

Method of detection: sections from nodal tissue, including SLN were first stained with haematoxylin and eosin (H&E) and studied for metastatic involvement. SLNs showing metastatic involvement on H&E stained sections were considered as positive for metastatic disease. One extra section from the deeper levels of SLN was then prepared and stained with immunohistochemistry (IHC) technique using anti‐cytokeratin (AE1/AE3) antibody if results from the initial study of SLN for metastatic disease were negative. If the SLN section stained with IHC technique was positive for keratin immunoreactivity, SLN was considered to have metastatic disease.

Target condition and reference standard(s)

Type: patients with stage I or II endometrial cancer all underwent systematic lymph node dissection. All patients were treated by laparotomy The level of lymph node dissection is not reported.

Lymph node number and site: a total of 15 SLNs were identified. The most common anatomical sites were: obturator = 8 (53.3%), internal iliac = 6 (40%), para‐aortic = 1 (6.7%).

Flow and timing

All patients received the index and reference standard within 1 month. It is unclear whether all patients received the same reference standard, as the level of the lymph node dissection is not reported. All patients were included in the analysis.

Comparative

Notes

Study results: the median SLN count was 3 (range, 1‐4). Sentinel LN detection rate was 80%. Sensitivity was 100%, specificity was 25%; PPV was 25%; NPV was 100%. There were no FP nodes.

Adverse reaction from index or reference test: uneventful light blue urine was made in most patients that resolved by 24 hours postoperatively.

TP = 3; FP = O; FN = 0 ; TN = 9; failed = 3. Type 1 tumours. FIGO Stage IA = 10; FIGO stage IB+ = 5. Subserosal uterine injections; bilateral detection rate = not reported.

Operating time: not reported.

Other intraoperative complications: no perioperative complication was observed.

Other postoperative complications: no perioperative complication was observed.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes

Could the selection of patients have introduced bias?

Low risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

No

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

Unclear

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: Brazil

Study design: a study of patients treated for endometrial cancer from June 2007 to February 2017 at AC Camargo Cancer Center, Brazil, by the same gynaecologic oncology team. Patients were grouped into the non‐sentinel lymph node group (N‐SLN) if they had only undergone systematic lymphadenectomy as part of the surgical staging procedure without SLN‐mapping, or the sentinel lymph node group (SLN) if they had undergone both systematic lymphadenectomy and SLN‐mapping. The SLN group was recruited prospectively, while the N‐SLN group was retrospectively analysed.

Inclusion criteria: patients with high risk endometrial cancers (defined as requiring one of the following: high‐grade tumour (endometrioid grade 3 and non‐endometrioid histologies: serous, clear cell, or carcinosarcoma), deep myometrial invasion (greater than or equal to 50%, or the presence of LVSI).

Exclusion criteria: suspicious lymph node involvement on preoperative imaging or found during surgery; patients who did not undergo lymph node dissection together with SLN‐mapping; low‐risk endometrial cancers were excluded (endometrioid grades 1 or 2, < 50% myometrial invasion, absence of LVSI)

Patient characteristics and setting

Number of patients: N‐SLN group = 161 patients, SLN group = 75 patients.

Median age: 61 years (range, 41‐83 years)

Median body mass index: 27.2 kg/m2 (range, 17.9 ‐ 43.7 kg/m²)

Histopathological cell type: endometrioid carcinoma = 107 (66.5%) in N‐SLN group & 48 (64%) in SLN group, serous carcinoma = 22 (13.6%) in N‐SLN group & 10 (13.3%) in SLN group, clear cell carcinoma = 13 (8.1%) in N‐SLN group & 8 (10.7%) in SLN group, carcinosarcoma = 14 (8.7%) in N‐SLN group & 7 (9.3%) in SLN group, serous + clear cell carcinoma = 3 (1.9%) in N‐SLN group & 2 (2.7%) in SLN group, undifferentiated = 2 (1.2%) in N‐SLN group & 0 in SLN group.

FIGO stage: not reported.

Grade on hysterectomy specimen: grade I = 21 (13%) in N‐SLN group & 7 (9.3%) in SLN group, grade II = 19 (11.8%) in N‐SLN group & 23 (30.7%) in SLN group, grade IIIa = 121 (75.2%) in N‐SLN group & 45 (60%) in SLN group.

Lymphovascular space involvement: presence of LVSI in 25 cases (15.5%) in N‐SLN group & 32 cases (42.7%) in SLN group.

Setting: single centre in Brazil, namely AC Camargo Cancer Center.

Index tests

Type of endometrial sampling: not reported.

Experience of operator: not reported.

Tracer used and amount: 4 mL of patent blue dye

Method and timing of application: a total of 4 mL of patent blue dye (1 mL superficial and 1 mL at 1 cm depth) was injected into the cervix at 3 and 9 o'clock.

Method of detection: the SLNs were examined by IHC when the haematoxylin and eosin (H&E) stain was negative. The SLNs were serially sectioned every 2 mm and stained with H&E at three levels of the tissue block. If the sample was negative, a pan‐cytokeratin stain was performed at each of the three levels. The SLNs were classified as macrometastasis (tumour > 2.0 mm), micrometastases (tumour cell aggregates between 0.2 mm and 2 mm), isolated tumour cells (ITCs) (individual tumour cells or aggregates </= 0.2 mm), or negative. All lymph nodes with macroscopic, microscopic, and isolated tumour cells were considered to be positive. Non‐ sentinel lymph nodes were reported as positive or negative for metastases based on routine sectioning and examination of a single H&E‐stained slide per a standard protocol.

Target condition and reference standard(s)

Type: patients with high‐risk endometrial cancer all underwent pelvic +/‐ para‐aortic lymph node dissection. In the N‐SLN group, 19 patients (11.8%) had pelvic lymph node dissection alone and 142 patients (88.2%) had pelvic and para‐aortic lymph node dissection. In the SLN group, 23 patients (30.7%) had pelvic lymph node dissection alone and 52 patients (69.3%) had pelvic and para‐aortic lymph node dissection. In the N‐SLN group 5 patients (3.1%) had minimally invasive surgery; in the SLN group 51 patients (68%) had minimally invasive surgery, of which 35 (46.7%) were conventional laparoscopies and 16 (21.3%) were robotic assisted laparoscopies.

Lymph node number and site: median SLN detected was 2 (range, 1‐5). Of all lymph node metastases, 20 (31.3%) were pelvic and 6 (12%) were para‐aortic.

Flow and timing

All patients received the index and reference standard within 1 month. The patients did not all receive the same reference standard, as only 88.2% and 69.3% received full pelvic and para‐aortic lymph node dissection in the N‐SLN and SLN groups, respectively. The remaining 11.8% and 30.7% received only pelvic lymph node dissections in the N‐SLN and SLN groups, respectively. All patients were included in the analysis.

Comparative

Notes

Study results: the median SLN count was 2 (range, 1‐5). The median positive SLN count was 1.5 (range, 1‐4). The SLN detection rate was 85.3%, and bilateral SLNs were observed in 60% of patients. Sensitivity was 90%, NPV was 95.7%, false‐negative predictive value was 4.3%, and the false negative rate was 10%. Of 20 positive SLNs, 8 (40%) were detected only after immunohistochemistry (IHC).

Blue dye alone: TP = 20; FN = 0; FN = 2; TN= 42; failed = 11 (2 ITCs moved from TP to TN group c.f. published paper as defined as TN for the review if ITC only). Type 1 tumours; FIGO stage = not clear; Subserosal uterine injections; 45 bilateral LN detection.

Adverse reaction from index or reference test: not reported.

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Unclear

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: France

Study design: a retrospective study of 133 women with early stage endometrial cancer who were consecutively enrolled in nine gynaecological oncology centres in France from July 2007 and August 2009.

Inclusion criteria: invasive endometrial carcinoma (FIGO stages I and II according to the 1988 FIGO classification) confirmed by biopsy, patients older than age 18 years affiliated with the French Health Care System and able to speak and read French, and intention of surgical staging.

Exclusion criteria: preoperative FIGO stage III and IV, previous lymphadenectomy or surgery that could change the uterine lymphatic drainage (conisation, myomectomy), and pregnancy.

Patient characteristics and setting

Number of patients: 133; 8 were excluded for protocol deviations.

Median age: 63 years (range, 38‐100)

Median body mass index: 27 kg/m2 (range, 18 ‐ 54 kg/m2)

Histopathological cell type: type 1 tumours = 94 (84.7%), type 2 tumours = 17 (15.3%)

Assumed FIGO 2009 by preoperative MRI for 125 included patients: Stage 1A = 82; 1B+ =43.

Stage and grade on hysterectomy specimen: not reported.

Lymphovascular space involvement: 6 of 9 (66.7%) true‐positive cases had LVSI reported as present; 3 of 7 false‐negative cases had LVSI reported as present.

Setting: nine specialist gynaecological oncology centres in France, namely Tenon University Hospital (Paris), Croix Rousse University Hospital (Lyon), Georges Pompidou University Hospital (Paris), Poissy‐Saint Germain en Laye University Hospital (Poissy), Edouard Herriot University Hospital (Lyon), Bretonneau University Hospital (Tours), Claudius Regaud Comprehensive Cancer Center (Toulouse), Centre Hospitalier Lyon Sud (Lyon), and Centre Oscar Labret (Lille).

Index tests

Type of endometrial sampling: diagnosis confirmed by biopsy.

Experience of operator: not reported.

Tracer used and amount: 0.2 mL of unfiltered technetium sulphur colloid, 2 mL of patent blue dye

Method and timing of application: four cervical injections of 0.2 mL of unfiltered technetium sulphur colloid were given with a 25‐gauge spinal needle, at 3, 6, 9, and 12 o'clock positions the day of or morning before surgery. Patent blue was injected intracervically through a 25‐gauge spinal needle at 3 and 9 o'clock positions.

Method of detection: ultrastaging was done for SLNs only. Normal‐appearing SLNs were cut perpendicular to the long axis. Air‐dried cytological smears were prepared by scraping the cut surfaces and staining with a rapid May‐Grünwald‐Giemsa method. Each half‐SLN was sectioned at 3‐mm intervals. Each 3‐mm section was analysed at four additional levels of 200 µm and four parallel sections: one was used for H&E staining, and H&E‐negative sections were examined by IHC with an anticytokeratin antibody cocktail (cytokeratins AE1‐AE3; Dako Corporation, Glostrup, Denmark). Non‐SLNs were submitted totally and blocked individually after 3‐mm sectioning and H&E staining. Isolated tumour cells (ITCs) were defined as cells or masses of cells with a size </= 0.2 mm, micrometastases as tumours with a size >0.2 mm but </=2 mm, and macrometastases as tumours of a size >2 mm. SLNs were recorded as positive when they contained macrometastases, micrometastases, or ITCs.

Target condition and reference standard(s)

Type: patients with stage I‐II endometrial cancer had pelvic +/‐ para‐aortic lymph node dissection. Patients had surgery performed either by laparoscopy of by an open approach, but the ratios are not reported.

Lymph node number and site: sentinel lymph nodes were detected in 111 patients. 87 of these 111 patients (78.4%) had an intraoperative examination: 30 by frozen section (34.5%) and 57 by imprint cytology (65.5%). The anatomical location was not reported.

Flow and timing

All patients received the index and reference standard within 1 month. It is unclear whether all received the same reference standard (pelvic +/‐ para‐aortic lymph node dissection) as the ratios are not reported. All patients in whom SLNs were detected were included in the analysis, but not all patients who were recruited were included in the analysis.

Comparative

Notes

Study results: SN detection rate was 111/125 (89%). NPV 100% for hemipelvis, sensitivity 100% for hemipelvis. For per patient analysis NPV 97%, sensitivity 84%. 3 false negative cases occurred.

TP = 16; FP = 0; FN = 3; TN = 92; failed = 14. FIGO stage from preoperative MRI: IA = 82; Stage IB+ = 43. Type 1 and 2 tumours. Cervical injections; bilateral detection 77 patients.

Adverse reaction from index or reference test: not reported

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Yes

Could the selection of patients have introduced bias?

Low risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

Unclear

Were all patients included in the analysis?

Unclear

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: France

Study design: a prospective study of 33 consecutive patients with endometrial cancer from July 2002 to October 2005.

Inclusion criteria: biopsy‐confirmed endometrial cancer of apparent stage I or II according to the preoperative criteria of the FIGO.

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 33

Median age: 66.1 years (range, 46‐84 years)

Median body mass index: 25.8 kg/m2 (range, 17.8 ‐ 36.9 kg/m2)

Histopathological cell type: endometrioid adenocarcinoma = 29 (87.9%), adenosquamous carcinoma = 1 (3%), clear cell carcinoma = 2 (6.1%), malignant mixed Muellerian tumour = 1 (3%)

1988 FIGO stage: Ia = 4 (12.1%), Ib = 13 (39.4%), Ic = 13 (39.4%), II = 3 (9.1%).

Grade on hysterectomy specimen: for endometrial adenocarcinoma, grade 1 = 18 (54.5%), grade 2 = 9 (27.3%), and grade 3 = 2 (6.1%)

Lymphovascular space involvement: not reported.

Index tests

Type of endometrial sampling: diagnosis confirmed by biopsy.

Experience of operator: not reported

Tracer used and amount: 0.2 mL of unfiltered technetium sulfur colloid, 2 mL of patent blue

Method and timing of application: all patients received four pericervical injections (1.5 cm depth) of 0.2 mL (10 MBq each) of unfiltered technetium sulfur colloid (Nanocis; CIS Bio International, Saclay, France) administered with a 25‐G spinal needle on the day before surgery. All patients received Patent blue dye injected pericervically (1mL per injection, 1.5 cm deep) with a 25‐G spinal needle at 3 and 9 o’ clock. Grossly metastatic nodes were sectioned. Normal‐appearing SNs were cut perpendicular to the long axis. All SNs were examined intraoperatively by imprint cytology for the pathologist learning curve. Air‐dried cytologic smears were prepared by scraping the cut surfaces and were stained by using a rapid May‐Grünwald‐Giemsa method. Each half‐SN was sectioned at 3‐mm intervals. Each 3‐mm section was analysed by 4 additional levels of 150 µm and 4 parallel sections; 1 section was used for haematoxylin and eosin (H&E) staining, and H&E‐negative sections were then examined by immunohistochemistry with an anticytokeratin antibody cocktail (Cytokeratin AE1‐AE3; Dako Corporation, Glostrup, Denmark). Non‐SNs were totally submitted and blocked individually following 3‐mm distances and H&E staining.

Target condition and reference standard(s)

Type: patients with stage I‐II endometrial cancer all underwent pelvic +/‐ para‐aortic lymphadenectomy performed to the level of the aortic bifurcation. Patients with clinical stage I disease underwent laparoscopic treatment, including peritoneal washing, bilateral salpingo‐oophorectomy, the SN procedure, systematic pelvic lymphadenectomy, and laparoscopically assisted vaginal hysterectomy. Patients with apparent stage II disease underwent a peritoneal washing, bilateral salpingo‐oophorectomy, SN procedure followed by systematic pelvic lymphadenectomy and laparoscopic radical hysterectomy.

Lymph node number and site: At least 1 SN was identified in only 27 patients (81.8%). The mean number of SNs was 2.5 per patient (range, 1‐5). A total of 71 SNs was removed. 18 patients (54.5%) had an identified bilateral SN. The most common site of the SNs was the medial external iliac region (67.6%). Other sites included intermediate external iliac region (2.8%), obturator fossa (5.6%), interiliac area (19.7%), common iliac region (2.8%), and aortic bifurcation (1.4%).

Flow and timing

All patients received the index and reference standard within 1 month. Patients did not all receive the same reference standard, as 90.9% received laparoscopy‐assisted vaginal hysterectomy and 9.1% received laparoscopic radical hysterectomy. All patients underwent systemic pelvic lymph node dissection. Para‐aortic lymph node dissection was only performed if a para‐aortic SN was detected. None of the patients required para‐aortic lymph node dissection, as no para‐aortic SN was detected. All patients were included in the analysis.

Comparative

Notes

Study results: Total number of SNs removed was 71. SN were identified in only 27 patients (81.8%). Mean number of SNs was 2.5/patient. 54.5% had bilateral SNs detected. No false‐negative SLNs were reported. Fourteen SNs (19.7%) from 8 patients were found to be metastatic at the final histological assessment. Sensitivity, specificity and true positives are not reported.

TP = 8; FP = 0; FN = 0; TN ‐ 19; Failed = 6. FIGO Stage (converted to 2009 classification) Ia = 17; IB+ = 16. Type 1 & 2 tumours. Cervical injections; 18 patients with bilateral detection.

Adverse reaction from index or reference test: no anaphylactic reactions to patent blue occurred.

Operating time: median of 160.2 minutes (range, 120‐280).

Other intraoperative complications: No intraoperative complications occurred.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: France

Study design: a prospective study of patients with clinical stage I endometrial cancer of any histological type from January 2002 to March 2006.

Inclusion criteria: clinical stage I endometrial cancer of any histological type scheduled for primary surgery

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 43

Mean age: 67.8 +/‐ 10.4 years (range, 40‐90 years)

Median/mean body mass index: not reported.

Histopathological cell type: endometrioid adenocarcinoma = 35 (81.4%), papillary serous carcinoma = 4 (9.3%), clear cell carcinoma = 1 (2.3%), carcinosarcoma = 3 (7.5%)

1988 FIGO stage: IA = 3 (7.0%), IB = 14 (32.6%), IC = 9 (20.9%), IIA =3 (7.0%), IIB = 3 (7.0%), IIIA = 1 (2.3%), and IIIC = 10 (23.2%). FIGO 2009 conversion: stage IA = 17; Stage 1B+ = 26; Type 1 & 2 tumours.

Grade on hysterectomy specimen: not reported.

Lymphovascular space involvement: not reported.

Setting: single centre in France, namely The Hôpital Européen Georges‐Pompidou, Paris

Index tests

Type of endometrial sampling: not reported.

Experience of operator: not reported, though mention that SLN detection rate was not influenced by team experience (P = 0.1).

Tracer used and amount: 120 MBq of Colloidal rhenium sulfate labelled with technetium‐99m (Nanocis, Schering, CIS BIO International, 91192, Gif‐sur‐Yvette, France), 2 mL of 50% patent blue dye

Method and timing of application: colloidal rhenium sulfate labelled with technetium‐99m was injected into the cervix in the nuclear medicine department, on the day before surgery. A 25‐gauge needle was used to inject a total dose of 120 MBq, 30 MBq at each of four sites, at the 12, 3, 6, and 9 o’clock positions on the cervix. Lymphoscintigraphy was performed 12 hours later using a dual‐head camera (Axis 2000, Philips Medical Systems, Cleveland, OH). During surgery, 2 mL of 50% patent blue dye was injected into the cervix at the 12, 3, 6 and 9 o'clock positions. SLNs were formalin‐fixed, mostly bisected and paraffin embedded. Both parts of SLNs were sampled by five step sectioning at 250 µm intervals. Four sections were stained with haematoxylin‐eosin‐saffron (HES) and one section was used for immunohistochemistry with the broad‐spectrum monoclonal antibody AE1/AE3 (Dako, Trappes, France). Non‐sentinel lymph nodes were assessed according to routine HES staining.

Target condition and reference standard(s)

Type: patients with stage I endometrial cancer who had pelvic +/‐ para‐aortic lymphadenectomy performed. Laparoscopy was used in 36 (83.7%) patients. The remaining 7 (16.3%) patients required laparotomy either because of marked uterine enlargement or anaesthesia‐related contraindications to laparoscopy. Para‐aortic lymph node dissection was performed in 9 (20.9%) patients. 34 patients (79.1%) had pelvic lymph node dissection only.

Lymph node number and site: 86 SLNs were detected in total from 30 patients. The mean number of SLNs per patient was 2.9 overall, 1.4 on the right side, and 1.3 on the left side, with the remaining SLNs being located on or near the midline. Bilateral SLNs were detected in 16 (53.3%) of the 30 patients. The interiliac area was the main site of SLN detection, with 28 (93.3%) patients and 71 of the total 86 SLNs (82.6%). Of these 37 (43%) were right‐sided and 34 (39.5%) were left‐sided. 5 (5.8%) were located in the common iliac area on the right side, 4 (4.7%) in the common iliac area on the left side, and 6 (7.0%) were found at the promontery. None of the patients had inguinal or para‐aortic SLNs without interiliac SLNs.

Flow and timing

All patients received the index and reference standard within 1 month. The patients did not all receive the same reference standard, as only 20.9% received full pelvic and para‐aortic lymph node dissection; the remaining 79.1% received pelvic lymph node dissection only. All patients were included in the analysis.

Comparative

Notes

Study results: 86 SLNs were identified in 30 patients (69.8%). The prevalence of node involvement was 23.2% (10/43 patients). Two of these 10 patients had a failure of the SLN detection, and 8 had involvement of the SLNs; thus the prevalence of SLN involvement was 27% (8/30). In six of the eight patients with SLN involvement, six had negative nonsentinel nodes. All the metastatic SLNs were located in the interiliac area. There were no false‐negative results; therefore, the negative predictive value of SLN biopsy was 100%. The sensitivity for detecting lymph nodes disease was 100%. Para‐aortic node involve‐ ment was found in a single patient, who also had positive pelvic nodes; in this patient, no SLNs were detected.

TP = 8; FP = 0; FN = 0; TN = 22; failed =13. FIGO 2009 (converted) stage IA = 17; Stage 1B+ = 26; Type 1 & 2 tumours. Cervical injections; bilateral detection = 16 patients.

Adverse reaction from index or reference test: no adverse events related to SLN detection and biopsy were recorded. More specifically, no cases of blue dye sensitisation or blue dye‐related desaturation during surgery occurred. There were no port‐site metastases (follow‐up = 15 months).

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

No

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: Istanbul

Study design: a retrospective study of patients with a preoperative histopathologically‐proven endometrioid endometrial cancer from September 2012 to December 2015.

Inclusion criteria: preoperative histopathologically‐proven endometrioid endometrial cancer

Exclusion criteria: inability to perform the optimal procedure (defined as abdominal or laparoscopic total hysterectomy, bilateral salpingo‐oophorectomy, and SLN mapping with PLA, with or without para‐aortic lymphadenectomy (PALA) (because of morbidly obese patients, patients with serious co‐morbidities limiting the time of surgery); PET/CT findings from another clinic; history of neoadjuvant treatment; diagnosis of any other malignancy

Patient characteristics and setting

Number of patients: 95

Median age: 58.9 years (range, 33‐82 years)

Median body mass index: not reported

Histopathological cell type: endometrioid endometrial cancer

FIGO stage: IA = 61 (64.2%), IB = 16 (16.8%), II = 11 (11.6%), IIIA = 1 (1.1%), IIIC1 = 4 (4.2%), IVB = 2 (2.1%)

Grade on hysterectomy specimen: grade 1 = 39 (41.1%), grade 2 = 49 (51.6%), grade 3 = 7 (7.3%)

Lymphovascular space involvement: present in 27 (28.4%), absent in 68 (71.6%)

Setting: single centre in Turkey, namely the Cerrahpasa Medical Faculty, Division of Gynecologic Oncology, Department of Gynecology and Obstetrics, Istanbul University

Index tests

Type of endometrial sampling: not reported.

Experience of operator: study performed by only 3 surgeons (TB, V, and NT); experience of operators not reported explicitly

Tracer used and amount: 4 mL of 1% methylene blue dye

Method and timing of application: methylene blue dye (1%) was injected to the cervix at 3‐ and 9‐o’clock positions for a total of 4 mL (1 mL to the superficial surface [depth of 2‐3 mm] and 1 mL deep in the tissue [depth of 1 cm ‐ 2 cm]). All the SLNs were sectioned into 2‐mm sections, and in the first step, the SLNs were examined by imprint cytology and frozen sections. Afterwards, the lymph nodes were completely embedded in paraffin; then, the lymph nodes were first examined by a routine histological technique, staining with haematoxylin and eosin (H&E). If the SLNs were negative upon the initial H&E staining, the ultrastaging pathology protocol was applied for the SLNs. Ultrastaging consists of 2 adjacent 5 µm sections cut from each paraffin block at 2 levels, 50 µm apart, for a total of 4 slides per block. At each level, 1 slide was stained with H&E, and the other was stained with immunohistochemistry, using anticytokeratin AE1:AE3 (Ventana Medical Systems, Inc, Tucson, AZ).

Target condition and reference standard(s)

Type: patients with grade I, II, or III endometrial cancer who had pelvic +/‐ para‐aortic dissection performed to the level of the left renal artery. Thirty‐nine cases (41.1%) were performed by laparoscopic surgery, and 56 (58.9%) were treated by laparotomy. Pelvic lymph node dissection only was performed in 68 patients (71.6%); pelvic and para‐aortic lymph node dissection was performed in 27 patients (28.4%).

Lymph node number and site: the total number of SLNs removed was 227. The mean number of SLNs removed per patient was 2.95 (range, 1‐9). A SLN was detected in 77 patients (81.1%) and was absent in 18 patients (18.9%). 53% of patients had bilateral SLNs removed; 47% of patients had unilateral SLNs removed. The most common anatomical site for SLNs was: right obturator region 25.1%, right external iliac 17.6%, right common iliac 5.7%, vena cava 2.2%, right internal iliac 0.4%, left obturator region 22.5%, left external iliac 19.4%, left common iliac 4.9%, left internal iliac 2.2%.

Flow and timing

All patients receive the index and reference standard within 1 month. The patients did not all receive the same reference standard (pelvic +/‐ para‐aortic lymphadenectomy), as only 28.4% received full pelvic and para‐aortic lymph node dissection, the remaining 71.6% received pelvic lymph node dissection only. All patients were included in the analysis.

Comparative

Notes

Study results: FDG PET/CT study was performed preoperatively in all patients. 227 SLNs were identified in 77 patients (81.1%). The mean number of SLNs removed was 2.95 per patient (range, 1‐9). Lymph node metastases were found in 4 (4.2%) of 95 patients. Results were analysed on a per‐patient basis and as a combination of preoperative PET/CT and SLN frozen section results, and showed a sensitivity of 33%, specificity of 100%, PPV of 100%, and NPV of 97.1%.

TP = 2; FP = 0; FN = 2; TN = 73; failed = 18. Type 1 tumours. FIGO stage IA = 61; Stage 1B= = 34. Cervical injection; bilateral detection 41 patients.

Adverse reaction from index or reference test: not reported.

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: Canada

Study design: a prospective study of patients with biopsy‐proven endometrial cancer from February 2014 to December 2015.

Inclusion criteria: biopsy‐proven endometrial cancer

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 119

Median age: 65.5 years (range, 49‐91 years)

Median body mass index: 31.0 kg/m2 (range, 18‐56 kg/m2)

Histopathological cell type: endometrioid = 101 (85%), serous = 8 (7%), dedifferentiated = 2 (2%), clear cell = 4 (3%), carcinosarcoma = 4 (3%)

FIGO stage: IA = 69 (58%), IB = 30 (25%), II = 3 (3%), IIIA = 4 (3%), IIIC1 = 8 (7%), IIIC2 = 4 (3%), IVB = 1 (1%)

Grade on hysterectomy specimen: grade 1 = 1 (51%), grade 2 = 33 (28%), grade 3 = 25 (21%)

Lymphovascular space involvement: present in 38 (32%), absent in 81 (68%)

Setting: single centre in Canada, namely L'Hotel‐Dieu de Québec hospital

Index tests

Type of endometrial sampling: not reported.

Experience of operator: mention of the effect of the surgeon's learning curve on SLN detection, but experience of operator not explicitly defined.

Tracer used and amount: 4 mL of ICG.

Method and timing of application: ICG was injected by intracervical injection with a 25‐gauge spinal needle. 1 mL of ICG was injected superficially and 1 mL was injected deeper into the cervical stroma at the 3 and 9 o'clock position immediately before the surgery. SLNs were not routinely submitted for frozen section unless there was a suspicion of metastatic disease on gross inspection. For ultrastaging, SLNs were cut perpendicular to the long axis at 3 mm intervals and embedded in paraffin. Six 4 μm sections at 40 μm intervals were cut from the paraffin blocks and stained with haematoxylin and eosin (H&E). An additional section was taken adjacent to the 3rd H&E cut and used for immunohistochemistry using anti‐cytokeratin AE1/ AE3 (DAKO, Agilent, CA). The non‐sentinel lymph nodes (non‐SLNs) were bisected parallel to the long axis and examined with routine H&E on one level.

Target condition and reference standard(s)

Type: 119 patients with grade I, II, or III endometrial cancer all of whom had pelvic +/‐ para‐aortic dissection at the surgeon's discretion. Thirty‐one (26%) cases were performed by laparoscopic surgery, 59 (50%) were performed by robotic surgery, and 29 (24%) were treated by laparotomy. Pelvic lymph node dissection only was performed in 100 patients (84%); pelvic and para‐aortic lymph node dissection was performed in 19 patients (16%).

Lymph node number and site: the total number of SLNs removed was 267. The median number of SLNs identified per patient was 2 (range, 0‐7). SLNs were located primarily in the external iliac node basin (70%), usually on or just under the medial aspect of the external iliac vein near the bifurcation, followed by the obturator area (23%). Unusual locations (6%) were the common iliac area, the parametrium, the para‐aortic or the presacral area.

Flow and timing

All patients receive the index and reference standard within 1 month. The patients did not all receive the same reference standard (pelvic +/‐ para‐aortic lymphadenectomy) as only 16% receive full pelvic and para‐aortic lymph node dissection, the remaining 84% received pelvic lymph node dissection only. All patients were included in the analysis.

Comparative

Notes

Study results: SLNs were detected bilaterally in 89 patients (74%) and unilaterally in 22 patients (19%). No SLNs were identified in 8 patients (6%). The overall SLN detection rate was 93%. In terms of hemipelvis, sentinel lymph nodes were detected in 84% (200/238) of hemipelvis, and not detected in 16% (38/238). When calculated on a per‐patient basis, sensitivity and negative predictive value were both 100% in patients with bilateral detection; 95% and 99% respectively in patients with at least unilateral detection. The side‐specific sensitivity and negative predictive value were respectively, 92.5% and 98.8%. Failed SLN detection rate 37%. 1 false negative at unilateral level, so the patient was not counted to be a false negative case at a patient level.

Per patient results:

TP = 21; FP = 0; FN = 1; TN = 89; failed = 8. FIGO stage IA= 69; Stage IB+ = 50. Type 1 and 2 tumours. Cervical injections; bilateral detection =89 patients.

Per hemi‐pelvis results:

230 hemi‐pelvices total. TP = 25; FP = 0 ; FN = 2; TP = 173; failed = 30 (8 bilateral/ 22 unilateral)

Adverse reaction from index or reference test: not reported.

Operating time: not reported.

Other intraoperative complications: not reported.

Other postoperative complications: not reported.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

No

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: Italy

Study design: a retrospective study of patients with preoperative stage I endometrial cancer and stage I (1A2‐1B1) cervical cancer from October 2010 to May 2015. Patients were split into three groups:

Group 1 received technetium‐99 and blue dye (n = 77);

Group 2 received blue dye alone (n = 38);

Group 3 received ICG (n = 48).

Inclusion criteria: Patients with preoperative stage I endometrial cancer and stage I (1A2‐1B1) cervical cancer

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 163; 118 (72.4%) patients with endometrial cancer and 45 (27.6%) patients with cervical cancer.

Median age: group 1 = 61 +/‐ 13.4 years (range, 26‐86), group 2 = 62 +/‐ 13.2 years (range, 29‐86), group 3 = 59 +/‐ 14.5 years (range, 29‐86)

Median body mass index: group 1 = 23 +/‐ 4.9 kg/m2 (range, 18‐50), group 2 = 26 +/‐ 5.3 kg/m2 (range, 18‐40), group 3 = 25 +/‐ 6.6 kg/m2 (range, 15‐50)

Histopathological cell type: EIN = 2 (5.1%), endometrioid = 101 (85.6%), serous papillary = 6 (5.1%), other = 9 (7.6%)

FIGO stage: EIN = 2 (5.1%), IA = 67 (56.8%), IB = 21 (17.8%), II = 5 (4.2%), IIIA = 1 (0.8%), IIIC1 = 19 (16.1%), IIIC2 = 1 (0.8%), IV = 2 (1.7%)

Grade on hysterectomy specimen: for cervical and endometrial cancer combined grade 1 = 44 (27%), grade 2 = 67 (41.1%), grade 3 = 48 (29.4%), NA = 4 (2.5%)

Lymphovascular space involvement: for cervical and endometrial cancer combined present in 55 (33%) of patients, absent in 108 (66.3%) of patients

Setting: single centre in Italy, namely the Gynecology Oncology Surgical Unit of the San Gerardo Hospital, Italy

Index tests

Type of endometrial sampling: not reported.

Experience of operator: authors state that they began SLN mapping in 2010. After the completion of the learning curve and initial experience, [they] moved in the direction of an injection of blue dye alone into the cervix before ICG injection, which started in 2014.

Tracer used and amount: 4 mL of blue dye, 4 mL to 5 mL of ICG solution (1.25 mg/mL), technetium‐99 (amount not reported)

Method and timing of application: a total of 4 mL of blue dye (2 mL per injection for each side) was injected at the 3‐ and 9 o’clock positions. The ICG concentration used was 1.25 mg/mL. For each patient, a 25 mg vial with ICG powder was diluted in 20 mL of aqueous sterile water. A total of 4 mL to 5 mL of this ICG solution were injected into the cervix alone, divided into the 3‐ and 9‐ o’clock positions. One millilitre of ICG solution was injected with penetration to a depth of 1 cm into the stroma, and 1 mL was injected into the submucosal layer on the right and the left of the cervix, usually after initial laparoscopic exploration to evaluate the feasibility of the surgical procedure. For open surgical cases performed with the Vitom II exoscope, the cervical injection was performed once the laparotomy was completed. LNs with macroscopic metastases were sectioned, and SLNs that appeared normal were cut perpendicular to the long axis. Two adjacent 5 µm sections were cut at each of 2 levels 50 µm apart from each block lacking metastatic carcinoma, detected by means of a section routinely stained with haematoxylin and eosin (H&E). At each level, one slide was stained with H&E and the other with immunohistochemistry using AE1/AE3, an anti‐cytokeratin antibody (Dako, Glostrup, Denmark), as well as one other negative control slide, for a total of five slides per block. All other non‐SLNs were examined only by routine H&E. Micrometastasis was defined as a metastatic deposit within the LNs ranging from 0.2 mm to no more than 2 mm in size. Isolated tumour cells were defined as single tumour cells or as clusters of malignant epithelial cells less than 0.2 mm in size.

Target condition and reference standard(s)

Type: 118 patients with endometrial cancer all of whom had pelvic +/‐ para‐aortic lymph node dissection performed. In the women with endometrial cancer, aortic lymphadenectomy was performed in 14 % of cases (17 of 118). For cervical and endometrial cancer combined, in 70 % of the cases (115 of 163), the surgical procedure was laparoscopy.

Lymph node number and site: for cervical and endometrial cancer combined, the total number of SLNs was 384 and the median number of SLNs per hemipelvis was 2. SLNs were most frequently located in the external iliac region (71 %), internal iliac (2 %), obturator fossa (10 %), common iliac (12 %) and para‐aortic (2 %), sacral (2 %), and parametrial (1 %).

Flow and timing

All patients received the index and reference standard within 1 month. The patients did not all receive the same reference standard. Para‐aortic lymph node dissection was only performed in patients with endometrial cancer with a positive preoperative positron emission tomography/computed tomography (PET/CT) scan and in the absence of SLN mapping or unilateral mapping. In the women with endometrial cancer, aortic lymphadenectomy was performed in 14% of cases (17 of 118), while the remaining 101 (86%) had only pelvic lymph node dissection. All patients were included in the analysis.

Comparative

Notes

Tc99 & blue dye

three arms to study:

(A) TC 99 & Blue dye:

TP=10; FP=0; FN=0; TN=32; failed=2. Type 1 tumours. FIGO stage IA=23; Stage IB+ = 26. Cervicalx injections; bilateral detection = 22.

(B) Blue dye only:

TP=5; FP=0; FN=0; TN=19; failed =6. Type 1 tumours. FIGO stage IA= 17; Stage IB+ =26. Cervical injections; bilateral detection = 14.

(C) ICG only

TP=4; FP=0; FN=0; TN=28; failed=0. Type 1 tumours. FIGO stage IA = 29; Stage IB+ =10. Cervical injections; bilateral detection = 32

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

No

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Unclear risk

Study characteristics

Patient Sampling

Country: Switzerland

Study design: a prospective study of patients with histologically‐proven early endometrial carcinoma from July 2001 to June 2005.

Inclusion criteria: histologically‐proven early endometrial cancer

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 60 (note although stated 60 patient in study, staging given for 61 patients)

Median age: 65 years (range, 43‐87 years)

Median body mass index: 27 kg/m2 (range, 21 ‐ 45 kg/m2)

Histopathological cell type: endometrioid = 54 (90%), serous papillary = 4 (7%), carcinosarcoma = 2 (3%)

1988 FIGO stage: IA = 12 (20%), IB = 22 (36%), IC = 5 (8%), IIA = 2 (3%), IIB = 5 (8%), IIIA = 6 (10%), IIIC = 9 (15%) Based on Table 1 FIGO 1A = 34 and FIGO 1B+ is 28 (note amounts to 601, not 60),

Grade on hysterectomy specimen: grade 1 = 22 (37%), grade 2 = 25 (42%), grade 3 = 13 (21%)

Lymphovascular space involvement: not reported.

Setting: single centre in Italy, namely the Department of Gynecology and Obstetrics, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland

Index tests

Type of endometrial sampling: not reported.

Experience of operator: states that the staging accuracy of the SN procedure depends on the surgeon's proficiency, but the experience of the operator is not explicitly reported.

Tracer used and amount: 2 mL of patent blue dye and technetium‐99 m‐colloidal albumin

Method and timing of application: In all patients but one [polyallergic], 2 mL of patent blue dye and then technetium‐99 m‐colloidal albumin (Nanocoll®, Amershaw Health AG, 8869 Wädenswil, Switzerland) were injected into the sub‐endometrial layer beneath the tumour using a Cook needle. The lymph nodes were analysed without freezing. SN were fixed in neutral buffered formaldehyde for 24 to 72 hours, then cut into 0.2 cm or less thick slices, and embedded in a paraffin block per node. Multiple sections were prepared from each block. A set of three 4 μm thick sections was cut every 250 μm. One section was stained by haematoxylin–eosin (H&E). When negative, then immunohistochemistry (IHC) was performed, using the cytokeratin C‐11 antibody (Readysysteme, AG, Bad Zurzach, CH). Detection of tumour cells defined a positive SN. Nonsentinel nodes were totally processed without serial sections or immunohistochemistry.

Target condition and reference standard(s)

Type: 60 patients with grade I, II, or III endometrial cancer all of whom had pelvic and para‐aortic lymphadenectomy to the level of the left renal vein performed. Thirteen patients (21.7%) had laparoscopic surgery, and 47 (78.3%) were treated by laparotomy. All patients had pelvic and para‐aortic lymph node dissections.

Lymph node number and site: Sentinel nodes were identified in 49 of 60 patients (82%). The mean number of SN retrieved was 3.7 per patient (range, 1 to 8). Sixteen patients (33%) had SN in both pelvic and para‐aortic areas. No patient had SN only at the para‐aortic level. The anatomical sites were as follows: internal iliac = 81 (12%), external iliac = 229 (34%), obturator = 187 (28%), common iliac = 111 (17%), para‐aortic = 63 (9%).

Flow and timing

All patients receive the index and reference standard within 1 month. The patients all received the same reference standard. All patients were included in the analysis.

Comparative

Notes

Study results: sentinel nodes were identified by radiotracer alone in 15 patients, by blue dye alone in 1 patient, and by both tracers in 33 patients, for a total of 49 patients (82%). No SN was identified in 11 patients (18.3%). Twenty‐seven patients had unilateral SN only and 22 patients had bilateral SN. More than one SN was identified in 30 of 49 patients (61%). Mean number of nodes was 3.7 per patient. Metastases were found in nine patients (15%), and in SN of eight of them. In seven of these eight patients, metastases were confined to the SN.

Adverse reaction from index or reference test: no SN was identified in 11 patients, because of hazardous injection of tracers due to insufficient intrauterine pressure (4 patients), or injection into the bloodstream (5 patients), through a damaged endometrium (1 patient) or through the myometrium (1 patient). This 18% failure rate corresponds to the main author's learning curve.

TP = 8; FN = 0; FN = 1; TN = 40; failed = 11. re‐calculated 2009 FIGO stage IA = 34; stage IB+ = 28; Type 1 and type 2; subserosal uterine injections; bilateral node detection = 22 patients.

Operating time: not reported.

Other intraoperative complications: no morbidity was related to the procedure.

Other postoperative complications: no morbidity was related to the procedure.

Of note, the stage numbers for Delaloye et al add up to 61 patients in table 1, but only 60 included in the study.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

No

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

Yes

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Low risk

Study characteristics

Patient Sampling

Country: Turkey

Study design: a prospective, randomised‐controlled trial of patients with histologically‐confirmed endometrial cancer from July 2018 to June 2019. Patients were randomised to two groups: the cervical group (n = 40) and thee endometrial group (n = 41) based on the site of tracer injection.

Inclusion criteria: histologically‐confirmed endometrial cancer

Exclusion criteria: patients with suspected lymph node disease and metastatic disease in preoperative imaging, those who were treated with radiotherapy or chemotherapy, those who underwent surgery without lymph node dissection, patients who had uterine sarcoma histology, co‐existent ovarian malignancies.

Patient characteristics and setting

Number of patients: 81; cervical group = 40, endometrial group = 41

Median age: cervical group = 61 years (range, 53.2‐68.7 years), endometrial group = 62 years (range, 55‐65 years)

Median body mass index: cervical group = 32 kg/m² (range, 31.2‐32.6 kg/m²), endometrial group = 32.7 kg/m² (range, 32‐33.4 kg/m²)

Histopathological cell type:

Cervical group: endometrioid = 34 (85%), serous = 2 (5%), endometrioid + serous = 1 (2.5%), clear cell = 1 (2.5%), endometrioid + clear cell = 1 (2.5%), undifferentiated = 1 (2.5%)

Endometrial group: endometrioid = 33 (80.5%), serous = 5 (12.2%), endometrioid + serous = 1 (2.4%), clear cell = 1 (2.4%), mucinous = 1 (2.4%)

FIGO stage:

Cervical group: IA = 21 (52.5%), IB = 9 (22.5%), II = 3 (7.5%), III = 5 (12.5%), IV = 2 (5%)

Endometrial group: IA = 27 (65.9%), IB = 4 (9.8%), II = 3 (7.3%), III = 4 (9.8%), IV = 3 (7.3%)

Grade on hysterectomy specimen:

Cervical group: Grade 1 = 19 (51.4%), grade II = 11 (29.7%), grade III = 7 (18.9%)

Endometrial group: Grade 1 = 18 (51.4%), grade II = 14 (40%), grade III = 3 (8.6%)

Lymphovascular space involvement:

Cervical group: present in 4 (10%), absent in 36 (90%)

Endometrial group: present in 7 (17.1%), absent in 34 (82.9%)

Setting: Single centre in Turkey, namely the Kocaeli University Hospital, Kocaeli, Turkey

Index tests

Type of endometrial sampling: not reported.

Experience of operator: not reported.

Tracer used and amount: 4 mL (1 Mci) of 99m‐technetium labelled with nano colloid albumin particles

Method and timing of application: In the cervical group, injections of 99mTc (4 mL, 1 Mci) were performed at the 3 and 9 o’clock positions of the uterine cervix using a 25‐gauge hypodermic needle. In the endometrial group, 99mTc (4 mL, 1 Mci) was injected into the fundal endometrium using a 21 gauge, 20 cm sheathed transcervical catheter. Both tracer injections were performed 3 hours before surgery. Thirty minutes after the tracer injection the patients were evaluated using lymphoscintigraphy. The histopathologic examination for lymph nodes consisted of haematoxylin and eosin (H&E) staining after being fixed in 10% buffered formalin and embedded in paraffin wax. All SLNs went through a pathologic ultrastaging process. For ultrastaging, the SLNs were sectioned along the longitudinal axis into 2 mm thick sections and stained with H&E. In the event that the first H&E stained section was negative for metastasis, three additional sections at 50 μm intervals were obtained for H&E. Anti‐cytokeratin antibody immunohistochemistry (AE1/AE3 and PCK26, Ventana Medical Systems, Tucson, Arizona, USA) was used to detect low‐volume metastasis on one ultra‐section specimen. After ultrastaging, metastases of ≤0.2 mm were identified as isolated tumour cells, those >0.2 mm and ≤ 2 were micro‐metastases, and those >2 mm were considered as macro‐metastases.

Target condition and reference standard(s)

Type: 81 patients with grade I, II, or III endometrial cancer all of whom had pelvic and para‐aortic lymphadenectomy. All surgeries were performed by laparotomy.

Lymph node number and site: the median number of pelvic SLNs for each group was 2. The median number of para‐aortic SLNs for each group was 0. The bilateral detection rate was higher in the endometrial group than in the cervical group, but the difference was not statistically significant (69% vs 43%, P = 0.085).

Flow and timing

All patients receive the index and reference standard within 1 month. The patients all received the same reference standard. All patients were included in the analysis.

Comparative

Notes

Study results: The rate of detection of at least one SLN, sensitivity, and negative predictive value were 80%, 66.6%, 96.6% for the cervical group and 85%, 66.6%, and 96.9% for the endometrial group, respectively. The false‐negative rate for both groups was 33%. The pelvic bilaterality rates with cervical and endometrial injections were 43% and 69%, respectively.

Arm A‐ subserosal

TP = 2, FP = 0, FN = 1, TN = 32; failed SLN = 6. FIG0 stage 1A = 27; Stage Ib+ = 14; Type 1 tumours. Subserosal uterine injection site; bilateral detection = 20.

Arm B ‐ cervical injection

TP = 2, FP = 0, FN = 1, TN = 29; failed SLN = 8. FIG0 stage 1A = 21; Stage Ib+ = 19; Type 1 tumours. Cervical injection site; bilateral detection = 14.

Adverse reaction from index or reference test: none.

Operating time: not reported.

Other intraoperative complications: none.

Other postoperative complications: none.

Methodological quality

Item

Authors' judgement

Risk of bias

Applicability concerns

DOMAIN 1: Patient Selection

Was a consecutive or random sample of patients enrolled?

Yes

Was a case‐control design avoided?

Yes

Did the study avoid inappropriate exclusions?

Unclear

Could the selection of patients have introduced bias?

Unclear risk

Are there concerns that the included patients and setting do not match the review question?

Low concern

DOMAIN 2: Index Test (All tests)

Were the index test results interpreted without knowledge of the results of the reference standard?

Unclear

If a threshold was used, was it pre‐specified?

Yes

Could the conduct or interpretation of the index test have introduced bias?

Unclear risk

Are there concerns that the index test, its conduct, or interpretation differ from the review question?

Low concern

DOMAIN 3: Reference Standard

Is the reference standards likely to correctly classify the target condition?

Yes

Were the reference standard results interpreted without knowledge of the results of the index tests?

Unclear

Could the reference standard, its conduct, or its interpretation have introduced bias?

Unclear risk

Are there concerns that the target condition as defined by the reference standard does not match the question?

Low concern

DOMAIN 4: Flow and Timing

Was there an appropriate interval between index test and reference standard?

Yes

Did all patients receive the same reference standard?

Yes

Were all patients included in the analysis?

Yes

Could the patient flow have introduced bias?

Low risk

Study characteristics

Patient Sampling

Country: USA

Study design: a retrospective study of patients with endometrial cancer from May 2011 to September 2011.

Inclusion criteria: not reported.

Exclusion criteria: not reported.

Patient characteristics and setting

Number of patients: 35

Mean age: 63.4 +/‐ 10.4 years (range, 35‐88)

Mean body mass index: 33.1 +/‐ 9.3 (range, 18‐56)

Histopathological cell type: Mayo Clinic criteria "Low‐risk" = 9 (25.7%), Mayo Clinic criteria "High‐risk" = 26 (74.3%)

FIGO stage: not reported.

Grade on hysterectomy specimen: grade 1 = 13 (37.1%), grade 2 = 14 (40%), grade 3 = 8 (22.9%)

Lymphovascular space involvement: present in 13 (37%), absent in 22 (63%)

Setting: single centre in USA, namely Florida Hospital Cancer Institute and the Global Robotics Institute, Florida, USA

Index tests

Type of endometrial sampling: not reported.

Experience of operator: not reported.

Tracer used and amount: 4 mL of isosulfan blue, 2 mL of diluted ICG dye (25mg of ICG in 20ml normal saline [1.25 mg/mL])

Method and timing of application: immediately prior to insertion of the uterine manipulator, 1 mL of ISB was injected sub‐mucosally in four quadrants of the cervix (1 to 2 mm with no blood return prior to injection). Immediately prior to docking, 0.5 mL of diluted ICG dye (25 mg of ICG in 20 mL normal saline [1.25 mg/mL]) was injected in each cervical quadrant immediately prior to the placement of a uterine manipulator. The pelvic dissections were initiated within approximately 10 minute of the cervical injections. Pathologists recorded initial H&E impressions of ultra‐sectioned SLN (at least 6 serial sections 4 μm thick at 40 μm intervals) prior to IHC analysis, and then amended their reports after IHC if groups of malignant cells were subsequently identified in H&E slides with the aid of IHC. An additional 4 μm section was cut between the third and fourth levels and immuno‐stained with mouse monoclonal anti‐AE1/AE‐3 cytokeratin (Dako, Carpentaria, CA). “Isolated cytokeratin tumour cells” were less than 0.2 mm; and micro‐metastases were defined as 0.2 mm to 2 mm of tumour. Non‐sentinel lymph nodes underwent one section with routine H&E staining.