Application value of indocyanine green fluorescence imaging in guiding sentinel lymph node biopsy diagnosis of gastric cancer: Meta-analysis

Abstract

BACKGROUND

Gastric cancer is a common malignant tumor of the digestive system worldwide, and its early diagnosis is crucial to improve the survival rate of patients. Indocyanine green fluorescence imaging (ICG-FI), as a new imaging technology, has shown potential application prospects in oncology surgery. The meta-analysis to study the application value of ICG-FI in the diagnosis of gastric cancer sentinel lymph node biopsy is helpful to comprehensively evaluate the clinical effect of this technology and provide more reliable guidance for clinical practice.

AIM

To assess the diagnostic efficacy of optical imaging in conjunction with indocyanine green (ICG)-guided sentinel lymph node (SLN) biopsy for gastric cancer.

METHODS

Electronic databases such as PubMed, Embase, Medline, Web of Science, and the Cochrane Library were searched for prospective diagnostic tests of optical imaging combined with ICG-guided SLN biopsy. Stata 12.0 software was used for analysis by combining the "bivariable mixed effect model" with the "midas" command. The true positive value, false positive value, false negative value, true negative value, and other information from the included literature were extracted. A literature quality assessment map was drawn to describe the overall quality of the included literature. A forest plot was used for heterogeneity analysis, and P < 0.01 was considered to indicate statistical significance. A funnel plot was used to assess publication bias, and P < 0.1 was considered to indicate statistical significance. The summary receiver operating characteristic (SROC) curve was used to calculate the area under the curve (AUC) to determine the diagnostic accuracy. If there was interstudy heterogeneity (I² > 50%), meta-regression analysis and subgroup analysis were performed.

RESULTS

Optical imaging involves two methods: Near-infrared (NIR) imaging and fluorescence imaging. A combination of optical imaging and ICG-guided SLN biopsy was useful for diagnosis. The positive likelihood ratio was 30.39 (95%CI: 0.92-1.00), the sensitivity was 0.95 (95%CI: 0.82-0.99), and the specificity was 1.00 (95%CI: 0.92-1.00). The negative likelihood ratio was 0.05 (95%CI: 0.01-0.20), the diagnostic odds ratio was 225.54 (95%CI: 88.81-572.77), and the SROC AUC was 1.00 (95%CI: The crucial values were sensitivity = 0.95 (95%CI: 0.82-0.99) and specificity = 1.00 (95%CI: 0.92-1.00). The Deeks method revealed that the "diagnostic odds ratio" funnel plot of SLN biopsy for gastric cancer was significantly asymmetrical (P = 0.01), suggesting significant publication bias. Further meta-subgroup analysis revealed that, compared with fluorescence imaging, NIR imaging had greater sensitivity (0.98 vs 0.73). Compared with optical imaging immediately after ICG injection, optical imaging after 20 minutes obtained greater sensitivity (0.98 vs 0.70). Compared with that of patients with an average SLN detection number < 4, the sensitivity of patients with a SLN detection number ≥ 4 was greater (0.96 vs 0.68). Compared with hematoxylin-eosin (HE) staining, immunohistochemical (+ HE) staining showed greater sensitivity (0.99 vs 0.84). Compared with subserous injection of ICG, submucosal injection achieved greater sensitivity (0.98 vs 0.40). Compared with 5 g/L ICG, 0.5 and 0.05 g/L ICG had greater sensitivity (0.98 vs 0.83), and cT1 stage had greater sensitivity (0.96 vs 0.72) than cT2 to cT3 clinical stage. Compared with that of patients ≤ 26, the sensitivity of patients > 26 was greater (0.96 vs 0.65). Compared with the literature published before 2010, the sensitivity of the literature published after 2010 was greater (0.97 vs 0.81), and the differences were statistically significant (all P < 0.05).

CONCLUSION

For the diagnosis of stomach cancer, optical imaging in conjunction with ICG-guided SLN biopsy is a therapeutically viable approach, especially for early gastric cancer. The concentration of ICG used in the SLN biopsy of gastric cancer may be too high. Moreover, NIR imaging is better than fluorescence imaging and may obtain higher sensitivity.

Key Words: Gastric neoplasms, Sentinel lymph nodes, Near infrared imaging, Fluorescence imaging, Indocyanine green, Meta-analysis

Core Tip: To explore Indocyanine green fluorescence imaging (ICG-FI) in guiding the diagnosis of gastric cancer sentinel lymph node biopsy through meta-analysis. We will collect and analyze relevant literature to systematically evaluate the clinical manifestations of ICG-FI in gastric cancer patients and compare its diagnostic effectiveness with that of traditional methods. This study will provide clinicians with a more accurate intraoperative evaluation method for gastric cancer, which will help guide surgical decision making and improve patient prognosis.

INTRODUCTION

At present, the main treatment for gastric cancer is surgical resection. If lymph node metastasis can be evaluated accurately and in real time during surgery and then lymph node dissection can be performed, operative complications can be reduced, and postoperative quality of life can be improved[1]. According to the different reagents used, the traditional methods for sentinel lymph node (SLN) biopsy can be divided into the biological dye method, the isotope method, and the double reagent method. Although these methods have made some progress in research, they also present their own disadvantages: The deposition of biological dyes at lymph node sites is poor, and the isotopes are characterized by background scattering around the injection site[2].

Gastric cancer is one of the most common malignant tumors in the world, and SLN biopsy, as a means of early diagnosis, is of great value for accurate treatment[3]. Indocyanine green fluorescence imaging (ICG-FI), a new imaging guidance technology, has been widely used in cancer surgery in recent years. Its unique biocompatibility and high selectivity make it an ideal tool to guide SLN biopsy. The purpose of this study was to systematically evaluate the diagnostic efficacy of ICG-FI in guiding SLN biopsy for gastric cancer through meta-analysis and to provide a scientific basis for its application in clinical practice[4]. The comprehensive evaluation of the application of ICG-FI in SLN biopsy of gastric cancer not only provides clinicians with more accurate and reliable intraoperative navigation, improves surgical accuracy, and reduces surgical risk but is also expected to promote further innovation in SLN biopsy technology and provide new ideas for the early diagnosis and treatment of gastric cancer[5]. This study has far-reaching clinical significance for promoting the wide application of ICG-FI technology in the field of gastric cancer and improving the level of surgical treatment.

Since the beginning of the 21^st century, many scholars have reported the application of ICG to guide SLN biopsy for gastric cancer patients. These studies have shown good application prospects, but the results reported by each study are inconsistent: The sensitivity and specificity fluctuate between 0.40 and 1.00 and between 0.60 and 1.00, respectively. In view of the large numerical fluctuation range of effect indicators, we conducted a meta-analysis on the diagnostic value of SLN biopsy guided by optical imaging combined with ICG to provide guidance for the treatment of gastric cancer.

MATERIALS AND METHODS

Retrieval method

The search strategy followed the Cochrane Handbook of Systematic Reviews.

The search terms "gastric/stomach" and "sentinel lymph node" and the subject terms "near-infrared/NIR or fluorescent imaging" and "indocyanine green/ICG" (as well as all free words, synonyms, and MeSH words) were used. Electronic databases such as PubMed, Embase, Medline, Web of Science, and the Cochrane Library were searched for relevant literature in any language, the search period was from the establishment of the database to the present, and the search was expanded according to the corresponding references. If the original data in the literature were incomplete or missing, the original author was contacted by email to request the original or missing data.

Literature inclusion and exclusion criteria

The inclusion criteria for patients were as follows: (1) Had surgically resectable gastric cancer [clinical T stage of the tumor (cT) 0-3]; (2) had a clinical stage of the tumor determined by at least two imaging examinations; (3) had a diagnostic accuracy test of SLN biopsy guided by ICG combined with optical imaging (near infrared imaging and fluorescence imaging); (4) had prospective studies to predict lymph node metastasis in gastric cancer; (5) had intraoperative or postoperative pathological biopsies performed on all lymph nodes removed during surgery; and (6) had > 10 patients statistically analyzed in the literature.

The exclusion criteria for patients were as follows: (1) Had a history of drug allergy or chemoradiotherapy; (2) had previously undergone endoscopic mucosal resection or endoscopic submucosal dissection; (3) had multiple digestive tract neoplasms; (4) had case reports, conference abstracts, clinical guidelines, editorials, reviews, meta-analyses or letters; (5) had previously undergone in vitro and animal experiments; and (6) had insufficient diagnostic efficiency data.

Literature screening and data extraction

The literature screening was carried out independently by two literature evaluators, and the final inclusion of the literature was decided after discussion by the two scholars. When there was disagreement, a third senior researcher participated in the discussion to make a decision. After the qualified literature was selected, the author, publication year, country, number of patients, age, sex, clinical T stage of the tumor, tumor diameter, ICG concentration, injection site, surgical method, lymph node dissection method, optical imaging equipment type, and pathological staining method were extracted [immunohistochemical (IHC)], hematoxylin-eosin (HE) staining, SLN detection method, sensitivity, specificity, and other information.

Literature quality evaluation

The quality of diagnostic tests was evaluated via the QUADAS-2. The evaluation indicators included the following: (1) Case selection: Whether cases were consecutively included and whether case-control design was avoided; (2) test to be evaluated: Whether to interpret the test results to be evaluated without knowing the gold standard test results (blind method); (3) gold standard test: Whether the gold standard test results are interpreted without knowing the test results to be evaluated (blind method) and whether the gold standard test can correctly diagnose the target disease; and (4) trial process and progress: Whether the interval between the trial to be evaluated and the gold standard test is appropriate, whether all subjects receive the same gold standard test, and whether all subjects are included in the statistical analysis.

Statistical methods analysis

Stata 12.0 software was used for the analysis with the "bivariable mixed effect model" combined with the "midas" command. The true positive value, false positive value, false negative value, true negative value, and other information from the included literature were extracted. A literature quality assessment map was drawn to describe the overall quality of the included literature. The heterogeneity of the forest map was analyzed, and P < 0.01 was considered to indicate statistical significance. A funnel plot was used to assess publication bias, and P < 0.1 was considered to indicate statistical significance. The area under the curve (AUC) calculated by the integrated receiver operating curve method (SROC) describes the diagnostic accuracy. The closer the AUC is to 1, the greater the diagnostic accuracy is; that is, the greater the diagnostic value of SLN biopsy guided by optical imaging combined with ICG is. If there was interstudy heterogeneity (I² > 50%), meta-regression analysis and subgroup analysis were performed, and P < 0.05 was considered to indicate statistical significance. The effect measures used in this meta-analysis included sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, combined AUC value (SROC), and 95%CI.

RESULTS

Included literature and literature quality evaluation

Fifteen studies[6-20] were included in this study, and a total of 1020 patients were included. The literature screening process is shown in Figure 1. Optical imaging equipment includes two methods: Near-infrared (NIR) imaging and fluorescence imaging. The tumor stage included cT1-3. The pathological staining used was IHC + HE or HE alone. The basic characteristics of the included studies are shown in Table 1. A literature quality evaluation revealed that among the 15 included studies, only 3 explicitly stated "continuously included cases"[16,19,20]. Only 2 articles explicitly stated that "interpretation results were interpreted blindly"[6,18], which reduced the overall quality of the included literature and presented a high risk of bias. The literature quality assessment is shown in Figure 2.

Figure 1 Literature screening flow chart.

Figure 2 Document quality assessment map. NIR: Near-infrared; ICG: Indocyanine green; IHC: Immunohistochemical; HE: Hematoxylin-eosin staining.

Table 1 Basic characteristics of 15 included literature.

Ref.	Year	Country	Cases	Average age (years)	Sex (male/female)	CT staging	Average tumor diameter (mm)	ICC concentration (g/L)	Average number of SLN detections	Sensitivity (%)	Specific (%)
Nimura et al[6]	2004	Japan	84	_	_	T1-2	_	5	10.5	100	59.7
Ishikawa et al[7]	2007	Japan	16	57.0	8/8	T1-2	21.0	5	2.9	50.0	100
Ohdaira et al[8]	2007	Japan	52	59.7	37/15	T1	_	5	_	100	100
Kusano et al[9]	2008	Japan	22	67.7	9/13	T1-3	_	5	3.6	40.0	100
Koyama et al[10]	2009	Japan	14	56.9	9/5	T1	26.0	5	7.1	100	100
Ohdaira et al[11]	2009	Japan	30	62.9	23/7	T1-2	42.6	5	4.8	100	100
Tajima et al[12]	2009	Japan	56	68.4	30/26	T1-3		5	7.2	64.7	100
Kelder et al[13]	2010	Japan	212	60.0	159/53	T1	30.0	5	6.0	97.0	100
Tajima et al[14]	2010	Japan	38	64.3	25/13	T1-2	33.8	5	7.9	75.0	100
Yano et al[15]	2012	Japan	130	_	_	T1-2	_	0.5	_	100	86.8
Kinami et al[16]	2016	Japan	72	69.3	44/28	T1-3	27.6	0.05	6.0	90.9	100
Takahashi et al[17]	2016	Japan	36	_	_	T1-2		5	9.2	100	100
Takahashi et al[18]	2017	Japan	44	60.9	3519	T1	24.8	5	7.9	100	100
Kim et al[19]	2019	South Korea	28	56.8	16/12	T1	16.0	_	_	100	92.3
Okubo et al[20]	2018	Japan	17	_	_	T1-2	19.6	_	4.5	100	100

ICC: Intraclass correlation coefficient; SLN: Sentinel lymph node; ICG: Indocyanine green; CT: Clinical T stage of the tumor.

Main index results and heterogeneity test

The combined sensitivity of all studies was 0.95 (95%CI: 0.82-0.99), the specificity was 1.00 (95%CI: 0.92-1.00), the positive likelihood ratio was 30.39 (95%CI: 9.14-101.06), and the negative likelihood ratio was 0.05 (95%CI: 0.92-0.99), with a diagnostic odds ratio of 225.54 (95%CI: 88.81-572.77). A heterogeneity test of sensitivity and specificity (Q test) was performed, and the heterogeneity among the included studies was statistically significant (P < 0.001). The sensitivity (I²) was 89.41%, and the specificity (I²) was 97.82%, indicating significant heterogeneity (Figure 3).

Figure 3 Forest map. A: Included document sensitivity forest map; B: Forest plot with reference specificity.

Diagnostic accuracy and publication bias

The plotted SROC curve did not show an "arm and arm" distribution, indicating that there was no significant threshold effect (actual threshold difference among the included studies). The AUC was 1.00 (95%CI: 0.99-1.00), and the critical values were a sensitivity of 0.95 (95%CI: 0.82-0.99) and a specificity of 1.00 (95%CI: 0.92-1.00). The "diagnostic odds ratio" of SLN biopsy for gastric cancer patients was analyzed with a funnel plot, the asymmetry test of the funnel plot was performed via the Deeks method, and the asymmetry of the funnel plot was found to be significant (P = 0.01), indicating significant publication bias (Figure 4).

Figure 4 Funnel plot of the diagnostic odds ratio for gastric cancer patients who underwent sentinel lymph node biopsy. OR: Odds ratio.

Meta-regression and subgroup analysis

Surgical techniques for SLN biopsy of gastric cancer: Compared with fluorescence imaging, NIR imaging can achieve greater sensitivity (0.98 vs 0.73). Compared with optical imaging immediately after ICG injection, optical imaging after 20 minutes obtained greater sensitivity (0.98 vs 0.70). Compared with that of patients with an average SLN detection number < 4, the sensitivity of patients with a SLN detection number ≥ 4 was greater (0.96 vs 0.68). Compared with HE staining, IHC + HE staining had greater sensitivity (0.99 vs 0.84). Compared with subserous injection of ICG, submucosal injection achieved greater sensitivity (0.98 vs 0.40). Compared with 5 g/L ICG, 0.5 and 0.05 g/L ICG had significantly greater sensitivities (0.98 vs 0.83) (P < 0.05). However, there was no significant difference in sensitivity between laparotomy and laparoscopy, lymph node regional dissection, or dissection (all P > 0.05).

Patient characteristics: Compared with CT2-3 patients, cT1 patients achieved greater sensitivity (0.96 vs 0.72, P = 0.02), while there was no significant difference in sensitivity between patients with a tumor diameter ≤ 30 mm and patients with a tumor diameter > 30 mm (P > 0.05). Number of patients and reference year: Compared with the sample size of ≤ 26 patients, the sensitivity of > 26 patients was greater (0.96 vs 0.65). Compared with that in the literature published before 2010, the sensitivity in the literature published after 2010 was greater (0.97 vs 0.81), and the differences were statistically significant (all P < 0.05; Table 2).

Table 2 Meta-subgroup analysis of the sensitivity-based bivariable mixed effects model.

Influencing factor	Number	Sensitivity (95%CI)	P value
Optical imaging			< 0.001
Near infrared imaging	10	0.98 (0.96-1.00)
Fluorescence imaging	7	0.73 (0.56-0.91)
Image time			< 0.001
20 minutes after injection	10	0.98 (0.96-1.00)
Immediately after injection	5	0.70 (0.54-0.86)
Average number of SLNS detected			0.010
≥ 4	12	0.96 (0.90-1.00)
< 4	3	0.68 (0.20-1.00)
Pathological staining			< 0.001
IHC + HE	7	0.99 (0.98-1.00)
HE	10	0.84 (0.70-0.98)
Injection site			< 0.001
Submucosal injection	14	0.98 (0.95-1.00)
Subserous injection	2	0.40 (0.18-0.62)
ICG concentration (g/L)			0.010
5	12	0.83 (0.80-0.89)
0.5	3	0.98 (0.93-1.00)
Operation for gastric cancer			0.460
Laparotomy	8	0.95 (0.86-1.00)
Laparoscopic surgery	9	0.96 (0.88-1.00)
Cleaning method			0.200
Lymph node regional dissection	14	0.96 (0.89-1.00)
Lymph node dissection	3	0.96 (0.87-1.00)
Clinical stage of tumor			0.200
cT1	13	0.96 (0.92-1.00)
cT2-3	4	0.72 (0.41-1.00)
Tumor diameter (mm)			0.160
≤ 30	8	0.96 (0.91-1.00)
> 30	3	0.81 (0.64-0.98)

SLN: Sentinel lymph node; ICG: Indocyanine green; IHC: Immunohistochemical; HE: Hematoxylin-eosin staining; cT: Clinical T stage of the tumor.

DISCUSSION

Since the beginning of the 21^st century, personalized minimally invasive surgery to preserve the function of the stomach has not only been proposed as a new surgical strategy for gastric cancer but also as an urgent requirement for accurate preoperative staging of gastric cancer[21]. Following the traditional SLN biopsy technique for gastric cancer, optical imaging combined with ICG-guided SLN biopsy of gastric cancer has gradually been included in clinical studies, so it is necessary to objectively evaluate its diagnostic accuracy[22-24]. In this meta-analysis, 1020 patients were included, and a systematic review revealed that the clinical feasibility of this diagnostic technique was good. However, there were differences in several operational technical criteria in the included literature, and the heterogeneity of sensitivity and specificity was significant among the included studies. Therefore, meta-regression and subgroup analysis were used to further explore the factors that may affect diagnostic accuracy.

Optical imaging technology uses near-infrared fluorescence at 700-900 nm, which has the advantages of real-time, low intrinsic fluorescence, and high tissue penetration[25-27]. According to Cabrera et al[28], fluorescence imaging may be superior to infrared imaging because the tissue structure image it produces is clearer[28]. It was also mentioned in the included literature that even if individual lymph nodes were not stained, fluorescence technology could be used for effective imaging. However, meta-subgroup analysis revealed that NIR imaging has greater sensitivity than fluorescence imaging (0.98 vs 0.73). The reason may be the background scattering of fluorescence imaging, leading to a high false negative rate[29]. Therefore, we recommend that NIR imaging be used more in clinical studies[30-32]. However, it is worth noting that with the development of optical imaging technology, fluorescence molecular imaging and intraoperative multimodal imaging technology may be more commonly applied in the clinic in the future. In the included literature, to obtain a good deposition effect, surgeons mostly adopt optical imaging 20 minutes after ICG injection with high sensitivity. Subgroup analysis also revealed that imaging 20 minutes after injection yielded greater sensitivity than imaging immediately after injection[33]. However, if the interval is longer than 20 min, the diagnostic accuracy will be reduced. Therefore, to obtain good diagnostic results, NIR imaging should be performed 20 minutes after ICG injection during surgery[34].

If more SLN is obtained during SLN biopsy, the false negative rate of SLN biopsy for gastric cancer may be reduced. Meta-subgroup analysis suggested that a higher sensitivity could be obtained with an SLN detection number ≥ 4. This result indirectly reflects the complexity of the lymphatic system in gastric cancer; that is, multiple lymphatic drainage events can lead to multiple SLNS[35]. HE and IHC are currently the most widely used histopathological staining reagents. However, HE frequently misses lymph node micrometastasis[36]. Subgroup analysis revealed that IHC (+ HE) had greater sensitivity than HE staining. With increasing attention given to the concept of micrometastasis of isolated tumor cells, the advantages of continuous section technology and molecular diagnostic technology, which may become more reliable methods for intraoperative detection of lymph node metastasis and micrometastasis, are gradually becoming more prominent[37]. Yin et al's study[38] showed that there was no significant difference between submucosal ICG injection and subserosal injection. However, the meta-subgroup analysis in this paper suggested that submucosal injection achieves greater sensitivity[39]. Therefore, this study suggested that submucosal injection may be more advantageous and may be more suitable for early gastric cancer. The recommended concentration for SLN biopsy in breast cancer patients is 0.625 g/L[40]. For gastric cancer, the ICG concentrations used in the included studies were 5 g/L, 0.5 g/L, and 0.05 g/L. Meta-subgroup analysis indicated that higher sensitivity could be obtained at 0.5 or 0.05 g/L ICG than at 5 g/L. These results suggest that high ICG concentrations may reduce optical image recognition. Moreover, excessive concentrations of ICG may have been used in recent SLN biopsy studies of gastric cancer.

In this meta-analysis, there was no statistically significant difference between open and laparoscopic SLN biopsy for gastric cancer (P = 0.460). This may be due to the many differences between the application, operation techniques. In this meta-analysis, there was no significant difference between lymph node dissection and lymph node extraction (P = 0.200). Therefore, it is not yet possible to determine the optimal gastric cancer surgery and lymph node dissection in an SLN biopsy of gastric cancer patients. The diagnostic accuracy of a reaction-guided SLN biopsy for gastric cancer was negatively correlated with cT. The clinical stages of gastric cancer included in the literature included in this paper were cT1-3, and subgroup analysis also suggested that early gastric cancer can be more sensitive. This may be because advanced gastric cancer cells are more likely to block lymphatic vessels, and new lymphatic vessels also increase the complexity of the lymphatic system. Therefore, the SLN biopsy technique for gastric cancer may be more suitable for early gastric cancer. It is generally believed that the greater the diameter of a gastric cancer tumor is, the greater the degree of invasion and the later the clinical T stage. The average tumor diameter reported in the included literature was mostly < 50 mm, so 30 mm was used as the critical point for grouping in this paper[41]. However, subgroup analysis indicated that there was no statistically significant difference in SLN biopsy sensitivity between the two groups with different tumor diameters. Therefore, it is uncertain whether tumor diameter affects diagnostic accuracy.

There are some limitations to this paper. First, the Deeks funnel plot suggested obvious publication bias. Second, most of the eligible literature was submitted by Japanese scholars, which showed regional bias. Finally, the original data (true positive value, false positive value, false negative value, and true negative value) were mainly presented by the number of patients in the included literature, and the data on the number of lymph nodes could not be effectively extracted, so there was result-reporting bias. In addition, factors such as the small number of included studies and the lack of large-scale multicenter studies may also affect the quality of the evidence in this meta-analysis.

CONCLUSION

In summary, according to the satisfactory results of systematic evaluation of sensitivity and specificity, optical imaging combined with an ICG-guided SLN biopsy of gastric cancer is a feasible clinical diagnostic method, especially for early gastric cancer. Current studies on SLN biopsy in gastric cancer patients may use excessively high concentrations of ICG. Near-infrared imaging may be superior to fluorescence imaging for obtaining higher sensitivity. However, more effective and trustworthy large-scale multicenter studies are still necessary to confirm these findings.