Staging laparoscopy in esophagogastric junction cancer: Systematic review and meta-analysis

Abstract

BACKGROUND

Staging laparoscopy (SL) is a valuable tool for detecting occult peritoneal or hepatic metastases in patients with gastroesophageal junction (GEJ) cancer, especially when imaging suggests resectability. Despite advances in cross-sectional imaging methodologies such as computed tomography, positron emission tomography/computed tomography and endoscopic ultrasound, these techniques may miss low-volume or superficial peritoneal disease. Early identification of metastatic spread through SL can avoid unnecessary surgery, reduce morbidity, and improve treatment planning. However, its use in GEJ tumors - particularly Siewert types I and II - remains debated. Understanding the diagnostic performance (DP) of SL is essential for optimizing staging strategies.

AIM

To evaluate the DP of SL in identifying abdominal metastatic disease in patients with clinically resectable GEJ tumors.

METHODS

Systematic review and meta-analysis in accordance with PRISMA 2020 guidelines. A comprehensive PubMed search was performed up to March 29, 2024, using Medical Subject Headings terms related to GEJ, laparoscopy, and cancer staging. Inclusion criteria: Studies assessing SL in patients with GEJ tumors. Primary outcome was the rate of upstaging to stage IV due to peritoneal or hepatic metastases. Secondary outcomes included details on techniques, patient characteristics, and procedural factors. Risk of bias was evaluated using Risk of Bias in Non-randomised Studies of Interventions and certainty of evidence with Grading of Recommendations, Assessment, Development and Evaluation.

RESULTS

Eighteen studies involving 1591 patients were included. SL upstaged 22% of patients (95% confidence interval: 17-27) to stage IV due to occult metastatic disease. The pooled positivity rates were positive peritoneal malignancy (17.5%), peritoneal carcinomatosis (13%), malignant peritoneal cytology (9%), and hepatic metastases (9.2%). SL avoided unnecessary surgery in 19.8% of cases. Subgroup analysis revealed consistent performance in Siewert II tumors (DP = 13%, I² = 0), while in Siewert I tumors it was more heterogeneous (DP = 18%, I² = 93.7%). Only five studies reported complications, mostly minor, with no procedure-related mortality. No comorbidities, carcinomatosis scoring, conversion to open surgery, complications of follow-up, readmissions, postoperative length of stay, or delay in initiating neoadjuvant therapy were recorded.

CONCLUSION

SL improves staging accuracy in GEJ cancers, especially Siewert II. Despite heterogeneity and limited data stratification, SL may guide therapeutic decisions and help avoid unnecessary or futile surgeries.

Key Words: Esophagogastric junction; Laparoscopy; Neoplasm staging; Staging laparoscopy; Systematic review; Meta-analysis

Core Tip: Staging laparoscopy (SL) upstaged nearly one in five patients with clinically resectable gastroesophageal junction cancer by revealing occult peritoneal or hepatic metastases. This finding underlines the value of SL in refining staging and guiding treatment decisions. Diagnostic performance was especially high in Siewert II tumors, without any significant heterogeneity. However, key procedural and patient-related data were not appropriately reported across studies. These results support the routine use of SL in staging gastroesophageal junction tumors and highlight the urgent need for standardized reporting to improve risk stratification and clinical outcomes in this challenging and heterogeneous cancer subgroup.

INTRODUCTION

Esophagogastric junction (GEJ) tumors have traditionally been defined as those tumors that originate within 5 cm proximal or distal to the gastric cardia. They are subdivided into three types according to the Siewert classification: Siewert 1, originating in the distal esophagus, usually in an area of intestinal metaplasia (Barrett’s esophagus); Siewert 2, arising from the gastric cardia, and originating in the gastric epithelium or intestinal metaplasia; and Siewert 3, arising from gastric tissue distal to the cardia[1]. Thus, Siewert 1 are located between 1 cm and 5 cm above the cardia, Siewert 2 is between 1 cm above and 2 cm below the cardia, and Siewert 3 between 2 cm and 5 cm below the cardia.

Although the Siewert system continues to be used to classify the different areas of the GEJ, the 8^th edition of the American Joint Committee on Cancer (AJCC)-tumor-node-metastasis (TNM) classifies GEJ tumors as those with their epicenter between 2 cm proximal and 2 cm distal to the cardia. The definition includes some Siewert 1 tumors and all Siewert 2 tumors, but excludes Siewert 3 tumors. A tumor with a more proximal epicenter is defined as an esophageal tumor (including some Siewert 1 tumors), and a more distal tumor is defined as a gastric tumor (including all Siewert 3 tumors). This prognostic classification suggests that, theoretically, the approach to Siewert 1 and 2 might be the same as for esophageal cancers, with the possibility of subtotal esophagectomy, and that in Siewert 3 tumors a gastrectomy can be considered, extended to the distal esophagus if necessary[2].

Staging laparoscopy (SL) plays a fundamental role in the treatment of cancers of the gastrointestinal and gynecological tracts. By means of a relatively non-invasive intervention, it allows the detection of positive peritoneal malignancy (PPM), an umbrella term that includes positive peritoneal carcinomatosis or macroscopic carcinomatosis (PPC), ascites with positive peritoneal cytology (PPCyt), and positive peritoneal wash cytology (PPWCyt). SL can also detect hepatic metastatic disease (HMD) not identified by previous diagnostic tests. All these SL findings are considered distant metastases and upstage the tumor to stage IV in the AJCC-TNM classification (8^th edition)[3]. They also change the indication for surgery; unnecessary laparotomies and resections are avoided, and patients are assigned to palliative treatment with chemotherapy and/or radiotherapy.

The technical aspects of SL have changed since its inception, and even today there is no consensus on the best approach. Currently, in most cases, the procedure consists primarily of two techniques: Macroscopic assessment of metastases with biopsies, and peritoneal lavage with saline solution to confirm cytology. If present, ascitic fluid can be aspirated. Although the areas for macroscopic inspection and saline instillation are not standardized, procedures have emerged in recent years that seek greater diagnostic sensitivity. One of the approaches that has shown the best results is the four-step procedure proposed in 2019 by Liu et al[4].

The European Society for Medical Oncology guidelines recommend performing SL in all patients with stage IB-III gastric cancers considered potentially resectable, including Siewert III, in view of its reported sensitivity of 84.6% and specificity of 100%. These rates of detection are higher than those obtained with other diagnostic techniques such as computed tomography (CT), endoscopic ultrasound, or positron emission tomography-CT scan[5]. According to these same guidelines, in Siewert I and Siewert II, SL should be performed in locally advanced cases (T3-T4) with involvement of the cardia. Approximately 15% are positive for peritoneal metastases[6].

The practical guidelines of the National Comprehensive Cancer Network recommend performance of SL in gastric cancers with T3 and/or N+ involvement (positive lymph nodes) and also in cases of T1B or higher if surgery or preoperative chemotherapy is considered[7].

Various indices can be used to assess the outcomes of carcinomatosis in SL. One of the most useful is the Peritoneal Carcinomatosis Index, which has prognostic and therapeutic value.

The primary objective of this study was to determine the usefulness of SL in GEJ tumors for identifying peritoneal and/or hepatic cancer in patients with clinically resectable disease. The secondary objectives were to record the characteristics of the studies included, country, time period, preoperative tumor staging, indication for intervention, patient characteristics, description of the technique, postoperative assessment, and follow-up.

The research question was developed in accordance with the PICO format: In patients with clinically resectable adenocarcinoma of the GEJ (P), does SL (I), compared to conventional imaging techniques (C), improve the detection of metastatic disease and enable oncological upstaging (O)?

MATERIALS AND METHODS

This study was carried out in accordance with the PRISMA 2020 guidelines[8] and its explanatory guide[9] and complied with the various items included in the checklist. The statistical review of the study was performed by a biomedical statistician.

Search strategy and search terms

The search strategy included the following terms: An unlimited search was conducted in PubMed, updated on March 29, 2024. The following Medical Subject Headings terms were used: Esophagogastric junction; laparoscopy and tumor neoplasm staging, with the addition of the corresponding entry terms.

[(Esophagogastric Junction) OR (Gastroesophageal Junction) OR (Cancer of Esophagus) OR (Cancer of the Esophagus) OR (Esophageal Cancer) OR (Esophagus Cancer) OR (Esophagus Neoplasm) OR (Neoplasms, Esophageal) OR (Siewert)] AND [(Laparoscopy) OR (Laparoscopic Assisted Surgery) OR (Laparoscopic Surgery) OR (Laparoscopic Surgical Procedure) OR (Laparoscopic Surgical Procedures) OR (Peritoneoscopy) OR (Procedure, Laparoscopic Surgical) OR (Procedures, Laparoscopic Surgical) OR (Surgery, Laparoscopic) OR (Surgical Procedure, Laparoscopic) OR (Surgical Procedures, Laparoscopic)] AND [(Neoplasm Staging) OR (Cancer Staging) OR (Staging, Neoplasm) OR (TNM Classification) OR (TNM Staging) OR (TNM Staging System) OR (Tumor Staging) OR (“Staging laparoscopy”)].

The inclusion criterion was the description of patients with GEJ tumors who underwent SL at diagnosis. Exclusion criteria included failure to specify the SL results for the subgroup of patients with GEJ tumors, systematic reviews or meta-analyses, and isolated clinical cases.

Selection process and data extraction

The article selection process was conducted by two researchers and was divided into three parts: (1) Selection of articles based on their titles; (2) Appraisal of abstracts; and (3) Appraisal of full article. In the event of a discrepancy, a third researcher resolved the issue.

For the data collection process, the presence or absence of the study variables was recorded in an Excel spreadsheet. The data to be studied were previously defined, and the spreadsheet was created before the text reading began. Texts in languages other than English were translated using the program UPDF v1.7.

The following variables in the selected articles were analyzed to investigate the primary objective: Diagnostic performance (DP) of SL, defined as the number of patients upstaged to stage IV due to the presence of PPC, PPCyt, and PPWCyt; and HMD, to assess disease dissemination. To evaluate the secondary objectives of the study, the variables shown in Table 1 were analyzed. To determine which data from each study were eligible, the articles were read in full. Variables were recorded as specified or unspecified in each study (Table 2).

Table 1 Variables assessed to investigate the secondary objectives.

Characteristics	Classification
Study characteristics	Year
	Country
	Study design
	Study period
Patient characteristics	Number of patients
	Sex
	Mean age
	Reported comorbidities and scoring systems (e.g., ASA, Charlson Comorbidity Index, Clinical Frailty Scale, POSSUM, etc.)
Siewert classification type
Indication for staging laparoscopy (T, N)
Technical description	Number of surgical ports used
	Camera angle employed
	Pneumoperitoneum pressure
	Quantification of carcinomatosis, if present
	Abdominal regions inspected
	Sites of peritoneal lavage sampling
	Timing of lavage
	Lavage volume instilled and aspirated
	Duration of the procedure
	Use of intraoperative ultrasound
	Conversion to laparotomy
Postoperative assessment	Presence of complications
	Sources used to evaluate complications
	Use of complication classification/scoring systems
	Follow-up duration
	Postoperative hospital stay
	Readmission
	Interval between SL and initiation of neoadjuvant therapy or definitive surgery

ASA: American Society of Anesthesiologists; POSSUM: Physiological and Operative Severity Score for the enUmeration of Mortality and Morbidity; SL: Staging laparoscopy.

Table 2 Number of patients included in the studies, with details on whether the different variables analyzed were or were not specified.

Characteristics	Specified		Not specified
	n	Population	n	Population
Comorbidities	0	0	18	1591
Indications for laparoscopy (based on T and N)	1	41	17	1550
TNM version used	5	540	13	1051
Siewert type considered for inclusion	7	668	11	923
Method of tumor localization for Siewert classification	11	911	7	680
Patient position during procedure	2	84	16	1507
Type of access to peritoneal cavity	3	234	15	1357
Number of surgical ports	14	1258	4	333
Type of camera	4	347	14	1244
Pneumoperitoneum pressure	1	43	17	1548
Substance used for peritoneal lavage	8	766	4	418
Volume of lavage fluid	8	766	4	418
Timing of peritoneal lavage during surgery	4	237	8	947
Use of intraoperative ultrasound	7	399	11	1192
Carcinomatosis measured with scores or indices	0	0	18	1591
Study design	14	1385	4	206
Study period specified (months)	15	1255	3	336
Regions inspected	9	567	9	1024
Regions sampled for lavage	2	117	10	1067
Volume of aspirated lavage	1	81	11	1103
Surgical time	4	309	14	1282
Positive macroscopic carcinomatosis	15	1453	3	138
Positive hepatic metastases	6	434	12	1157
Positive ascites	1	43	12	1184
Cytology performed on lavage	9	826	3	358
Conversion to open surgery	0	0	18	1591
Data recording methodology	8	1052	10	539
Evaluation of complications	5	381	13	1210
Use of complication scoring system	0	0	18	1591
Postoperative length of stay	0	0	18	1591
Performed as outpatient surgery	0	0	18	1591
Follow-up duration	0	0	18	1591
Readmissions	0	0	18	1591
Days from surgery to oncological treatment	0	0	18	1591

TNM: Tumor-node-metastasis.

In the event that the SL results were not fully specified, the number of patients who tested positive for the intervention was indicated, but the question of whether the assessment was made macroscopically, in ascitic fluid, or in HMD was not recorded. Patients appeared as unspecified in these subgroups, but they were included in the DP assessment.

Patients who tested positive for both PPC and PPCyt were classified as positive for PPC. This was only necessary in one of the studies. In cases where it was not specified which patients overlapped, all were counted in the respective groups (PPC, PPCyt, PPWCyt, and HMD), and the number of patients in whom the therapeutic approach was modified was recorded under the heading of those no longer defined as resectable after SL. This means that the sum of the different positive sections in the SL may be greater than the final number of patients whose staging was modified. However, this was only necessary in one of the studies.

Overall certainty of the evidence

The overall certainty of the evidence for the main outcomes was assessed using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach. Five domains were considered (risk of bias, inconsistency, indirectness, imprecision, and publication bias), and the certainty of evidence was classified as high, moderate, low, or very low[10].

The main results are shown in Table 3, which indicates the data not specified in each study. Other tables show the number of studies that included/did not include information for specific variables.

Table 3 Main characteristics of the studies and their most relevant results.

Ref.	No. of patients	Study type	Country	Inspection areas	Peritoneal lavage areas	Positive in macroscopic carcinomatosis	Positive in liver metastases	Positive in ascites cytology	Positive in peritoneal lavage	Positive peritoneal malignancy	Patients who became unresectable
Halle-Smith et al[13], 2024	396	R	United Kingdom	NS	NS	29	NS	NS	29	58	58
Mitchell et al[14], 2023	79	R	United States	NS	NS	NS	NS	NS	NS	NS	12
Strandby et al[15], 2020	81	P	Denmark	NS	1, 2	6	NS	NS	4	10	10
Mirza et al[16], 2016	212	R	United Kingdom	2, 3, 4, 9, 10	NS	27	NS	NS	NS	27	27
Strandby et al[17], 2016	171	R	Denmark	NS	NP	9	4	NP	NP	9	13
Simon et al[18], 2016	41	R	France	1, 2, 3, 4, 11, 12, 13	NS	NS	NS	NS	NS	NS	5
Bhatti et al[19], 2014	60	R	Pakistan	NS	NS	5	NS	NS	12	17	17
Munasinghe et al[20], 2013	36	R	United Kingdom	4, 5, 8, 11, 12, 13, 14	1, 2, 3	2	NS	NS	3	5	5
Nath et al[21], 2008	48	P	United Kingdom	NS	NS	3	NS	NS	4	7	7
Sarela et al[22], 2006	105	NS	United States	NS	NS	40	NS	NS	NS	40	40
Clements et al[23], 2004	45	NS	United Kingdom	NS	NP	4	NS	NP	NP	4	4
Menon and Dehn[24], 2003	43	P	United Kingdom	4, 8, 9, 15	NP	6	5	1	NP	7	12
Nieveen van Dikjum et al[25], 1999	36	NS	Netherlands	4, 13, 14	NS	11	NS	NS	NS	11	11
Romijn et al[26], 1998	20	NS	Netherlands	4, 5, 9, 16, 17	NP	4	2	NP	NP	4	6
Stein et al[27], 1997	72	P	Germany	4, 7, 8, 13, 17, 18	NS	8	13	NS	12	20	33
O´brien et al[28], 1995	39	P	Ireland	NS	NP	8	1	NP	NP	8	8
Dagnini et al[29], 1986	89	R	Italy	4, 5, 6, 16, 19	NP	27	15	NP	NP	27	42
Bemelman et al[30], 1995	18	P	Netherlands	4, 8, 17	NS	NS	NS	NS	0	NS	5

R: Retrospective; P: Prospective; NP: Not performed; NS: Not specified; 1: Right upper abdominal quadrant; 2: Left upper abdominal quadrant; 3: Pelvis; 4: Liver; 5: Primary tumor region; 6: Douglas pouch; 7: Diaphragmatic hiatus; 8: Lesser sac (omental bursa); 9: Greater omentum; 10: Lesser omentum; 11: Treitz angle; 12: Foramen of Winslow; 13: Hepatoduodenal ligament; 14: Transverse mesocolon; 15: Left gastric artery; 16: Diaphragm; 17: Celiac lymph nodes; 18: Lesser curvature of the stomach; 19: Gastrohepatic ligament.

Risk of bias and quality of DP

The risk of bias in the studies included was formally assessed with the Risk of Bias in Non-randomised Studies of Interventions (ROBINS-I) tool, since all were observational (prospective or retrospective). Domains evaluated included selection of participants, classification of interventions, missing data, measurement of outcomes, and selective reporting[11].

In addition to ROBINS-I and GRADE, the methodological quality of DP was assessed using the quality assessment of diagnostic accuracy studies (QUADAS-2) tool[12]. However, its application to our dataset has limitations, since many of the studies were observational and were not designed as classic studies of diagnostic accuracy. This article was not prospectively registered.

Statistical analysis

Regarding the statistical analysis, the results were presented in tables when looking for associations between certain variables, and in the form of an effect diagram or forest plot for the results of the meta-analysis of the number of patients who tested positive for different outcomes (PPC, PPCyt, PPWCyt, HMD, and DP).

For the analysis of relationships, the χ² test was used to relate two qualitative variables, the nonparametric Mann-Whitney U test to compare a numerical variable between two groups, and the nonparametric Spearman’s Rho correlation coefficient between two numerical variables.

A random-effects meta-analysis was used, as all meta-analyses showed heterogeneity between studies. This heterogeneity was measured using the I² coefficient. A sensitivity analysis was performed to identify factors associated with this heterogeneity, using the Siewert coefficient as a possible cause. Subgroups were therefore created based on this factor and on the two main variables (DP and PPC).

Statistical analyses were performed using STATA/SE v16.0, and P < 0.05 were considered statistically significant. Variables were described using the frequency distribution for qualitative variables and the mean, standard deviation, minimum and maximum values, median, and interquartile range for quantitative variables. For the meta-analysis, the percentage of PPM, PPC, HMD, and DP were used as effect measures. Regarding the certainty or confidence in the body of evidence used, in all the articles selected the results and methodologies were clearly described and allowed for replication. Only articles with sufficient information were considered.

RESULTS

Search process and risk of bias assessment

The PRISMA 2020 flow diagram is shown in Figure 1. In all, 18 studies including 1591 patients were analyzed. SL upstaged 22% of patients to stage IV [95% confidence interval (CI): 17-27].

Figure 1 Preferred reporting items for systematic reviews and meta-analyses statement: Flow diagram of the literature selection.

Risk of bias assessment: Of the 18 studies, 10 (56%) were judged to present a low risk of bias, seven (39%) moderate risk, and one (5%) a high risk according to ROBINS-I. The main concerns were related to the retrospective design and incomplete reporting of variables (Table 4).

Table 4 Evaluation of the risk of bias in the studies included using the Risk of Bias in Non-randomised Studies of Interventions tool.

Ref.	Design	Participant selection	Intervention classification	Missing data	Outcome measurement	Selective reporting	Overall risk
Mitchell et al[14], 2023	Retrospective	Moderate	Moderate	Moderate	Low	Moderate	Moderate
Halle-Smith et al[13], 2024	Prospective	Low	Low	Low	Low	Low	Low
Strandby et al[15], 2020	Prospective	Low	Low	Low	Low	Low	Low
Strandby et al[17], 2016	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Mirza et al[16], 2016	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Simon et al[18], 2016	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Bhatti et al[19], 2014	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Munasinghe et al[20], 2013	Prospective	Low	Low	Low	Low	Low	Low
Nath et al[21], 2008	Prospective	Low	Low	Low	Low	Low	Low
Sarela 2006[22]	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Clements et al[23], 2004	Retrospective	Moderate	Moderate	Moderate	Moderate	Moderate	Moderate
Menon and Dehn[24], 2003	Prospective	Low	Low	Low	Low	Low	Low
Nieveen van Dikjum et al[25], 1999	Prospective	Low	Low	Low	Low	Low	Low
Romijn et al[26], 1998	Prospective	Low	Low	Low	Low	Low	Low
Stein et al[27], 1997	Prospective	Low	Low	Moderate	Low	Low	Low
O´brien et al[28], 1995	Prospective	Low	Moderate	Moderate	Low	Low	Low
Bemelman et al[30], 1995	Prospective	Low	Low	Low	Low	Low	Low
Dagnini et al[29], 1986	Retrospective	High	High	High	Moderate	High	High

QUADAS-2 analysis showed a low risk of bias for the SL index test in all studies. The main concerns arose in patient selection (six studies being recorded as high risk) and flow/timing (unclear also in six studies). Reference standards were frequently unclear due to the lack of explicit reporting (Table 5). A global QUADAS-2 traffic-light plot is provided (Figure 2).

Figure 2 A global traffic-light plot of quality assessment of diagnostic accuracy studies. QUADAS-2: Quality assessment of diagnostic accuracy studies-2.

Table 5 Quality assessment of studies’ diagnostic accuracy (quality assessment of diagnostic accuracy studies-2).

Ref.	Patient selection	Index test (laparoscopy)	Reference standard	Flow and timing
Mitchell et al[14], 2023	Low	Low	Low	Low
Halle-Smith et al[13], 2024	Low	Low	Low	Low
Strandby et al[15], 2020	Low	Low	Low	Low
Strandby et al[17], 2016	High	Low	Unclear	Unclear
Mirza et al[16], 2016	High	Low	Unclear	Unclear
Simon et al[18], 2016	Low	Low	Low	Low
Bhatti et al[19], 2014	High	Low	Low	Unclear
Munasinghe et al[20], 2013	Low	Low	Low	Low
Nath et al[21], 2008	Low	Low	Unclear	Unclear
Sarela et al[22], 2006	High	Low	Unclear	Unclear
Clements et al[23], 2004	High	Low	Unclear	Unclear
Menon and Dehn[24], 2003	Low	Low	Low	Low
Nieveen van Dikjum et al[25], 1999	Low	Low	Low	Low
Romijn et al[26], 1998	Low	Low	Low	Low
Stein et al[27], 1997	Low	Low	Low	Low
O´brien et al[28], 1995	Low	Low	Low	Low
Bemelman et al[30], 1995	Low	Low	Low	Low
Dagnini et al[29], 1986	High	Low	Unclear	Unclear

Characteristics and results of the studies

The characteristics and most relevant results from the studies included are shown in Table 3[13-30].

A total of 1591 patients underwent SL. Of these, 315 (19.8%) were no longer considered resectable after SL, avoiding further surgical interventions. Regarding positive results for PPC, PPCyt, PPWCyt, and HMD, only a few studies provided separate data (Table 6). These studies reported malignancy, as follows: PPM in 17.5%, PPC in 13%, PPWCyt in 9%, and HMD in 9.2% (Table 6).

Table 6 Number of articles, patients, and stage iv confirmation in cancers of the gastroesophageal junction.

Description	Number of studies	Patients (n)	Positive		Negative
			n	%	n	%
Total patients with staging laparoscopy identifying disseminated disease	18	1591	315	19.8	1276	80.2
Studies specifying positive peritoneal metastases	15	1453	254	17.5	1199	82.5
Studies specifying positive peritoneal carcinomatosis	15	1453	189	13	1264	87
Studies specifying positive peritoneal washing cytology	7	711	64	9	647	91
Studies specifying hepatic metastases detected	6	434	40	9.2	394	90.8

Only nine articles reported the areas inspected in SL[16,18,20,24-27,29,30]. The most frequently explored was the liver, present in all nine articles, followed by the hepatoduodenal ligament, in four articles[18,20,25,27]. In general, the areas inspected by the studies varied widely (Table 3).

The data that the selected articles specified for the variables studied in this meta-analysis appear in Table 2. The variables that were not specified in any of the articles were: Comorbidities, carcinomatosis measured with scores or indices, conversion to open surgery, use of a complication scoring system, postoperative length of stay, performance as outpatient surgery, follow-up duration and days from surgery to oncological treatment (Table 2).

Most of the articles did not mention subgroups by age or sex, and few specified whether cancers were gastric, esophageal, or GEJ. In the four articles that did define these data, 156 patients were male and 51 females, with mean ages of 55 years and 66 years respectively.

Regarding the methodology applied for data collection in different studies, only eight of the articles specified the source of the data: Six obtained data from databases, and two from retrospective reviews of medical records. As regards the Siewert classification, only seven articles specified the Siewert type they took as the basis for defining a GEJ tumor. None of the articles considered Siewert 3 as GEJ, but many of them included Siewert 3 in the group of gastric tumors.

Focusing now on specific aspects of the procedure, eight articles reported the performance of peritoneal lavage and described some of its characteristics. In all cases, physiological saline solution was used as the lavage agent, the most common volume being 500 mL and the range from 150 mL to 1000 mL. Surgical time was specified in four of the articles, with mean times of 32 minutes, 32 minutes, 34 minutes, and 60 minutes.

Seven studies referred to intraoperative ultrasound: Four performed it, and three did not. Only five studies mentioned complications post-intervention. Two reported no associated morbidity or mortality, one no major complication, and the other two reported a complication rate of 4% (without specifying their nature) and a surgical site infection rate of 3.5%. None mentioned any specific scales for assessing complications.

Statistical analysis

In the analysis of the relationship between the main outcomes (DP, PPM, PPC, HMD) and the main variables under study (Siewert type, number of areas examined, peritoneal lavage, lavage volume, ultrasound), none of the results obtained were statistically significant, mainly due to the small number of articles that mentioned most of the variables used (Tables 7, 8, 9, 10, and 11).

Table 7 Relationship between staging laparoscopy results and Siewert type (non-parametric Mann-Whitney U test).

Variable	Siewert 1								Siewert 2								P value
	n	mean	SD	Min	Max	p50	p25	p75	n	mean	SD	Min	Max	p50	p25	p75
Macroscopic carcinomatosis	3	14.67	10.69	8	27	9	8	27	2	3.5	0.71	3	4	3.5	3	4	0.200
Hepatic metastases	2	8.5	6.36	4	13	8.5	4	13	0	-	-	-	-	-	-	-	-
Peritoneal malignancy	3	12.67	12.9	2	27	9	2	27	2	5.5	2.12	4	7	5.5	4	7	0.800
Staging changes	4	19.5	12.79	5	33	20	9	30	3	7.67	4.04	4	12	7	4	12	0.228

Min: Minimum; Max: Maximum.

Table 8 Relationship between laparoscopy findings and number of inspected areas (non-parametric correlation coefficient: Spearman’s Rho).

Variable	Number of inspected areas
	Correlation coefficient	P value
Staging changes	0.316	0.684
Peritoneal malignancy	-0.632	0.135
Macroscopic carcinomatosis	0.316	0.684
Hepatic metastases	0.316	0.684

Table 9 Relationship between peritoneal malignancy results and the use of peritoneal lavage, non-parametric Mann-Whitney U test.

Peritoneal lavage performed	n	mean	SD	Min	Max	p50	p25	p75	P value
Not performed	6	9.83	8.66	4	27	7.5	4	9	0.999
Performed or not specified	9	14.67	18.24	1	58	7	4	17

Min: Minimum; Max: Maximum.

Table 10 Relationship between laparoscopy findings and volume of lavage used.

Variable	Lavage volume
	Correlation coefficient	P value
Peritoneal malignancy	0.559	0.248
Staging changes	0.067	0.886

Non-parametric correlation coefficient, Spearman’s Rho.

Table 11 Relationship between laparoscopy findings and the use of intraoperative ultrasound, non-parametric Mann-Whitney U Test.

Ultrasound performed	n	Mean	SD	Min	Max	p50	p25	p75	P value
Yes	3	5.67	4.73	2	11	4	2	11	0.700
No	3	8	1	7	9	8	7	9

Min: Minimum; Max: Maximum.

The meta-analysis recorded the following results. Regarding the number of patients who tested positive for PPM, the overall effect size (ES) suggested that 19% of patients had PPC (ES = 0.19, 95%CI: 0.14-0.24), with high heterogeneity (I² = 83.6%, P < 0.001, Figure 3A). As for PPC, 13% of patients tested positive (ES = 0.13, 95%CI: 0.10-0.17), again with high heterogeneity (I² = 82.8%, P < 0.001, Figure 3B). Nine per cent of patients tested positive for HMD (ES = 0.09, 95%CI: 0.04-0.14), with high heterogeneity (I² = 81.7%, P < 0.001, Figure 3C).

Figure 3 Peritoneal and hepatic malignancy spectrum visualization insights into cancer progression. A: Patients positive for peritoneal malignancy; B: Patients presenting macroscopic peritoneal carcinomatosis; C: Patients presenting hepatic metastatic disease. ES: Effect size; CI: Confidence interval.

Finally, with regard to the DP, the overall ES suggested that 22% of patients were upstaged (ES = 0.22, 95%CI: 0.17-0.27), with high heterogeneity (I² = 88.3%, P < 0.00, Figure 4).

Figure 4 Diagnostic performance of staging laparoscopy. ES: Effect size; CI: Confidence interval.

For the sensitivity analysis, a stratified study was performed according to Siewert type for PPM and for DP. For DP, in Siewert I GEJ cancers, 18% of patients experience a stage change (ES = 0.18, 95%CI: 0.08-0.28), with high heterogeneity (I² = 93.7%, P < 0.001, Figure 5A). In Siewert II GEJ cancers, 13% of patients experience a stage change (ES = 0.13, 95%CI: 0.08-0.18), with no significant heterogeneity (I² = 0.00%, P = 0.512, Figure 5B).

Figure 5 Stratified sensitivity analysis. A: Stratified sensitivity analysis of the diagnostic performance of staging laparoscopy in patients Siewert type I gastroesophageal junction cancer; B: Stratified sensitivity analysis of the diagnostic performance of staging laparoscopy in patients Siewert type II gastroesophageal junction cancer; C: Stratified sensitivity analysis for positive peritoneal malignancy in studies of patients with Siewert type I gastroesophageal junction cancer; D: Stratified sensitivity analysis for positive peritoneal malignancy in studies of patients with Siewert type II gastroesophageal junction cancer. ES: Effect size; CI: Confidence interval.

The same sensitivity analysis was performed for PPM. Seven per cent of patients with Siewert I GEJ cancers were upstaged (ES = 0.07, 95%CI: 0.01-0.12), with high heterogeneity (I² = 84.4%, P < 0.002, Figure 5C), as were 11% of patients with Siewert II GEJ cancers, (ES = 0.11, 95%CI: 0.05-0.18), with no significant heterogeneity (I² = 84.4%, P = 0.389, Figure 5D). Table 12 summarizes the findings of the meta-analysis and sensitivity analysis of the studies.

Table 12 Summary of meta-analysis and sensitivity analysis.

Variable	Number of studies	Global effect size (%)	Heterogeneity (I2)
Overall diagnostic performance of the staging laparoscopy	18	22 (95%CI: 17-27)	88.3%
Positive peritoneal malignancy	15	19 (95%CI: 14-24)	83.6%
Positive peritoneal carcinomatosis	18	13 (95%CI: 10-17)	82.8%
Hepatic metastatic disease	6	9 (95%CI: 4-14)	81.7%
Diagnostic performance - Siewert I	4	18 (95%CI: 8-28)	93.7%
Diagnostic performance - Siewert II	3	13 (95%CI: 8-18)	0%
Positive peritoneal malignancy - Siewert I	3	7 (95%CI: 1-12)	84.4%
Positive peritoneal malignancy - Siewert II	2	11 (95%CI: 5-18)	0%

CI: Confidence interval.

GRADE approach

Using the GRADE methodology, the primary outcomes (PPC, HMD, PPM, DP, and stratified analysis according to Siewert subtype) presented low certainty in most cases, due to high heterogeneity and the predominance of observational studies with moderate risk of bias. Only the DP and the PPM in in Siewert II inabsence of heterogeneity (Table 13). A summary of the tumor classification systems (Siewert and/or AJCC) used in the studies is presented in Table 14, highlighting notable heterogeneity in reporting.

Table 13 Certainty of evidence assessment (Grading of Recommendations, Assessment, Development and Evaluation).

Outcome	Number of studies	Patients (n)	Estimated effect	Certainty of evidence (GRADE)	Comments
Positive peritoneal carcinomatosis	18	1591	13% (95%CI: 10-17)	Low	High heterogeneity (I2 > 80%), moderate risk of bias
Hepatic metastatic disease	6	434	9% (95%CI: 4-14)	Low	High heterogeneity, small number of studies
Positive peritoneal malignancy	15	1453	19% (95%CI: 14-24)	Low	Variable results, wide confidence intervals, potential publication bias
Overall diagnostic performance of staging laparoscopy	18	1591	22% (95%CI: 17-27)	Low to moderate	High I2, majority observational studies, moderate risk of bias
Diagnostic performance - Siewert I	4	-	18% (95%CI: 8-28)	Low	High heterogeneity, one outlier study
Diagnostic performance - Siewert II	3	-	13% (95%CI: 8-18)	Moderate	Consistent results across studies, no heterogeneity
Positive peritoneal malignancy Siewert I	3	-	7% (95%CI: 1-12)	Low	Very heterogeneous, low frequency of events, wide CI
Positive peritoneal malignancy Siewert II	2	-	11% (95%CI: 5-18)	Moderate	Few studies, but consistent and homogeneous results

CI: Confidence interval; GRADE: Grading of Recommendations, Assessment, Development and Evaluation.

Table 14 Tumor classification systems used in included studies.

Ref.	Siewert classification	AJCC classification
Halle-Smith et al[13], 2024	No	No
Mitchell et al[14], 2023	Yes	Yes
Strandby et al[15], 2020	No	Yes
Mirza et al[16], 2016	yes	Yes
Strandby et al[17], 2016	Yes	Yes
Simon et al[18], 2016	Yes	No
Bhatti et al[19], 2014	No	Yes
Munasinghe et al[20], 2013	No	No
Nath et al[21], 2008	Yes	No
Sarela et al[22], 2006	No	No
Clements et al[23], 2004	Yes	No
Menon and Dehn[24], 2003	No	No
Nieveen van Dikjum et al[25], 1999	No	No
Romijn et al[26], 1998	No	No
Stein et al[27], 1997	Yes	yes
O´brien et al[28], 1995	No	No
Dagnini et al[29], 1986	No	No
Bemelman et al[30], 1995	No	No

Siewert classification and/or the American Joint Committee on Cancer. AJCC: American Joint Committee on Cancer.

DISCUSSION

In this study, the results of the SL were similar to those obtained in previous reviews of gastric cancer, which have reported a greater efficiency in detecting peritoneal metastases than other diagnostic techniques habitually considered as the gold-standard[31].

In the descriptive analysis, the results of the studies showed that up to 19.8% of patients defined as resectable by diagnostic tests such as CT or positron emission tomography-CT were shown by SL to have metastatic disease and were upstaged to stage IV, thus ruling out the possibility of surgery.

Although data can be extracted for specific variables of macroscopic metastasis, peritoneal lavage, or HMD, these were only specified in some of the articles, and so the results are less robust. A third of the articles specified that they did not perform peritoneal lavage[23,24,26,28,29], and therefore may not have been able to diagnose existing peritoneal metastases. The statistical analysis did not establish any significant relationships between the variables studied (DP, PPM, PPC, HMD) and the Siewert type, the number of areas examined, the performance of peritoneal lavage, the lavage volume, or the performance of intraoperative ultrasound (Tables 6, 7, 8, 9, 10, and 11). This is largely due to the small number of studies that specify these variables for the subgroup of GEJ cancers.

In the meta-analysis, the results show that SL is useful for detecting cases of metastatic disease in GEJ cancers, since the rates of detection were statistically significant for PPC, HMD and PPM, and for staging changes after laparoscopy. The meta-analysis also revealed significant heterogeneity in the results. As a consequence, sensitivity analyses were performed, which showed homogeneity in the SL results for Siewert type II with regard to PPM (11%) and DP (13%). However, these results must be considered with caution, given that only seven studies specified the Siewert type. A more exhaustive analysis is needed to examine this heterogeneity, and, if confirmed, to establish its possible causes.

In the assessment of our secondary objectives, we were surprised to find that most studies did not specify complications related to the procedure or delays in initiation of oncological medical therapy.

Regarding technical aspects of the SL, although some areas such as the liver were examined systematically during laparoscopy, there was considerable heterogeneity in the remaining areas. This finding indicates that there is still no established common methodology for macroscopic examination. There seems to be greater consensus about other aspects of laparoscopy: For instance, the use of saline solution for peritoneal lavage, which was specified in all the studies that mentioned it. However, the amount applied varied widely between 150 mL and 1000 mL, with a preference for 500 mL.

Regarding the limitations of the evidence included in this review, using the GRADE rating[10] most outcomes presented low to moderate certainty due to the heterogeneity of the studies and the risk of bias identified, indicating that the estimated effects might change if studies of higher methodological quality were available. The most robust outcome was the DP in Siewert II tumors, which showed moderate certainty. The high heterogeneity in the overall results and the retrospective design of many of the studies limit the reliability of the conclusions and draw attention to the need for multicenter prospective studies that apply standardized methodologies.

The results of the QUADAS-2 tool[12] in this article should be interpreted with caution, as not all the studies included met its requirements for diagnostic test accuracy. Nevertheless, it provided a structured framework for identifying highlight domains with a high risk of bias, complementing ROBINS-I[11] and GRADE assessments.

Certain issues highlighted by the study deserve mention. Some of the variables were not specified in any of the articles (Table 2); others were not specified in a large number of articles, making it difficult to obtain robust results. Many articles did not report these variables according to group, a circumstance that prevented the use of some of these data because they were not specified for the GEJ subgroup (Table 3). In this situation, it is difficult to carry out an in-depth study of the characteristics associated with the SL that goes any further than recording the number of patients who were upstaged as a result. Therefore, the evidence, especially for secondary endpoints, may be limited.

As the SL was performed by different surgeons, the results may be influenced by individual experience and skill. However, this is not a major limitation; in all likelihood, the use of an acceptable number of articles with different surgeons means that the results reflect a level of surgical skill that is quite close to the average.

Determining the association between the patients’ stage prior to the intervention and the results obtained might allow us to define the situations in which SL is indicated. Unfortunately, few studies specify the T and N values taken as a threshold for performing SL: They only stated that the tumors were resectable and that there was no metastasis. Commonly used guidelines such as European Society for Medical Oncology[6] and National Comprehensive Cancer Network[7] apply the indications of T3 or higher, N+, or stages IB-III, but these data were not specified in the studies included here. As for the site of the tumor, only seven articles specified which Siewert type is taken as the basis for defining a GEJ tumor, and none of them considered Siewert type 3 as GEJ; many included type 3 tumors under the heading of gastric cancers.

Other limitations of the study include the fact that only eight articles specify the sources of information and the use of a single database (PubMed) for the literature search, even though the search was conducted without any language or publication status restrictions. While this may reduce the overall comprehensiveness of the search, PubMed offers high sensitivity and relevance in gastrointestinal oncology literature. In addition, our search strategy was broad and included manual screening of reference lists to capture potentially missing studies.

A further limitation of this systematic review is that it was not prospectively registered in any public registry, such as the international prospective register of systematic reviews[32]. Although the review adhered to PRISMA 2020 guidelines and followed a predefined protocol, the absence of registration may limit transparency and reproducibility.

Finally, substantial heterogeneity observed in several pooled estimates (I² > 80%). While high heterogeneity typically warrants subgroup analyses, we found that key study-level characteristics - such as tumor classification, use of peritoneal cytology, or preoperative imaging techniques - were inconsistently reported across the included studies. This lack of uniform data prevented the reliable definition of clinically meaningful subgroups. Nevertheless, the direction of the effect was consistent across studies, supporting the robustness of the main findings.

This study highlights the need for future studies of GEJ tumors that properly disaggregate all the data for this subgroup. A Delphi study would be desirable to determine the key variables. For the moment, the results obtained corroborate the usefulness of SL in GEJ cancer for diagnosing metastatic disease in patients in whom it has not been previously detected. They also demonstrate the value of peritoneal lavage when performing SL, given that it increases the number of positive results by up to 9% in comparison with the macroscopic examination performed previously.

However, due to the lack of key data provided in the studies - for instance, the T and N grades on which the indication for laparoscopy is based, and the results stratified according to pre-SL stage or Siewert type - it is difficult to offer guidance regarding the situations in which SL is indicated or would be most effective. It is also difficult to determine the specific morbidity and mortality associated with this technique, as most studies did not report relevant data. Another factor that was not specified is the time to initiate cancer treatment, an issue that is of particular importance since a delay could impact survival.

Although there is consensus on the usefulness of SL in cases of gastric cancer, further studies are needed to assess its value in the subgroup of GEJ tumors, considering not only its performance in detecting disseminated disease based on Siewert type, T, and N, but also its surgical outcomes, such as possible associated morbidity, extended length of stay, and its effect on oncological outcomes. Finally, future studies should include the different aspects of patient experience among their objectives, something that this study has not addressed.

CONCLUSION

The findings confirm the usefulness of SL in the diagnosis of metastatic disease (about 22%) in GEJ tumors previously considered resectable, particularly in Siewert type II. However, the strength of the evidence is limited by the high proportion of observational studies in the sample and the notable methodological heterogeneity. Significant heterogeneity was also observed in the meta-analysis, particularly among Siewert type I. Future multicenter prospective studies should incorporate standardized reporting of technical details and should present results according to Siewert type, indication of SL based on Siewert type, and a formal assessment of complications and oncological outcomes.