This retrospective study included adult patients (≥18 years) with histologically confirmed, resectable gastric adenocarcinoma (T1-T4a) who underwent laparoscopic gastrectomy at Nanjing First Hospital between January 2020 and December 2021. Patients were eligible if preoperative staging indicated resectable disease, they were scheduled for laparoscopic radical gastrectomy with D2 lymphadenectomy, and had received no prior neoadjuvant chemotherapy or radiotherapy. Exclusion criteria were: (1) distant metastases, (2) severe comorbidities precluding surgery, and (3) allergy to ICG.

Preoperative assessment included enhanced computed tomography (CT) for tumor staging and localization. All patients also underwent diagnostic upper gastrointestinal endoscopy with biopsy to confirm histopathological diagnosis and guide surgical planning.

In the ICG group, a solution of 1.25 mg/mL ICG (Dandong Yichuang Pharmaceutical Co) was prepared using sterile water, and 0.5 mL was endoscopically injected into the submucosal layer at four quadrants around the tumor, totaling 2.5 mg, 12 to 24 hours before surgery. The submucosal injection was confirmed in real time by observing mucosal elevation during the slow injection. All injections were performed under direct endoscopic visualization using a standard 23G endoscopic needle, which allowed precise control of the injection depth and location around the tumor margins. Near-infrared (NIR) fluorescence imaging was utilized intraoperatively to guide tumor resection and LN dissection. The imaging system (NOVADAQ, Stryker, US) enabled seamless switching between visible light and NIR fluorescence to optimize tumor localization and lymphadenectomy. Fluorescence imaging was also used post-dissection to verify the completeness of LN dissection.

All surgeries were performed by a dedicated gastrointestinal oncology team composed of three senior surgeons, each with over 10 years of experience in laparoscopic gastrectomy. All patients underwent laparoscopic gastrectomy with D2 lymphadenectomy following standard oncological principles. The extent of resection (total or subtotal gastrectomy) was determined based on tumor location and size. In the ICG group, the fluorescence signal was used to guide both tumor localization and LN dissection, whereas the non-ICG group relied on conventional intraoperative techniques, including palpation and preoperative imaging. In the non-ICG group, intraoperative tumor localization and lymph node dissection were guided primarily by systematic palpation of the gastric wall and regional lymph node basins. Experienced surgeons palpated the stomach to identify the tumor margins and assessed the consistency and enlargement of lymph nodes manually. This tactile feedback was used to determine the extent of resection and the lymph node stations requiring dissection. Palpation was performed according to a standardized protocol to ensure thorough examination of all relevant anatomical regions.

Postoperative adjuvant chemotherapy was administered according to national clinical guidelines based on the final pathological stage and patient condition (9).

The primary outcomes were the number of LNs retrieved and tumor localization accuracy. The current American Joint Committee on Cancer (AJCC) Staging manual recommends that the removal of ≥30 regional LNs is desirable (10). LN dissection noncompliance was defined as the absence of LNs from more than 1 LN station that should have been excised.

Secondary outcomes included disease-free survival (DFS) and overall survival (OS) at three years. In addition, total operation time and intraoperative blood loss were analyzed. Postoperative follow-up was conducted every 3 months. At each visit, patients underwent clinical examination, routine blood tests including tumor markers, and contrast-enhanced abdominal CT. Upper gastrointestinal endoscopy was performed annually. Any suspected recurrence was confirmed through imaging, histopathological biopsy, or multidisciplinary team (MDT) evaluation. Survival status and recurrence events were collected from the hospital’s database. Patients who were alive at the time of the last follow-up or lost to follow-up before death were considered censored in the survival analysis.

To minimize selection bias, a PSM analysis was performed to balance baseline characteristics between the ICG and non-ICG groups. Propensity scores were calculated using a multivariable logistic regression model based on key covariates, including age, sex, body mass index (BMI), ECOG performance statues, tumor location, lymph vascular invasion, tumor size, tumor stages, and AJCC stage. Nearest-neighbor matching without replacement was conducted using a caliper width of 0.2 standard deviations of the logit of the propensity score.

Continuous variables were compared using the Student’s t-test or Mann-Whitney U test, while categorical variables were analyzed using the chi-square test or Fisher’s exact test. Kaplan-Meier survival analysis with the log-rank test was used to compare DFS and OS between the ICG and control groups. Multivariate Cox proportional hazards regression analysis was performed to identify independent prognostic factors for DFS and OS, adjusting for potential confounders, including age, sex, tumor stage, and lymph node involvement. A P value < 0.05 was considered statistically significant. Data were analyzed using SPSS version 22.0 (IBM Corp., Armonk, NY, USA).

A total of 847 patients diagnosed with gastric cancer were initially enrolled. After excluding 129 cases, 718 patients remained, with 256 (35.7%) receiving ICG. Propensity score matching yielded 168 matched pairs (336 patients) for subsequent analysis ( Figure 1 ). The baseline characteristics of the matched cohorts are summarized in Table 1 , and no significant differences were observed in age, sex, BMI, ECOG performance status, tumor location, lymphovascular invasion, tumor size, tumor stage, or AJCC stage (all P>0.05).

As shown in Table 2 , the ICG group had a significantly higher mean LN yield (46.4 ± 8.5) compared to the non-ICG group (42.6 ± 11.5, P<0.01). Notably, all patients (100%) in the ICG group achieved the recommended LN yield (≥30 nodes), whereas only 87.5% (n=147) of the non-ICG group met this threshold (P<0.01). When stratified by anatomical region, the number of LNs retrieved was significantly higher in both the D1 (25.5 ± 4.7 vs. 23.5 ± 6.3, P<0.01) and D2 (20.8 ± 3.8 vs. 19.1 ± 5.1, P<0.01) stations in the ICG group. In terms of lymph node dissection accuracy, the ICG group showed a significantly lower noncompliance rate than the non-ICG group (31.0% vs. 49.4%, P<0.01). These results suggest that preoperative ICG marking contributes not only to a higher total lymph node yield, but also to a more standardized and comprehensive lymphadenectomy.

Operative efficiency was also improved: the total operation time was significantly shorter in the ICG group (200.0 ± 11.4 minutes vs. 210.4 ± 11.6 minutes, P<0.01) and intraoperative blood loss was reduced (26.9 ± 8.7 mL vs. 31.3 ± 9.2 mL, P<0.01). Postoperative complication rates were similar between the groups (22.0% vs. 19.0%, P=0.50).

ICG fluorescence marking was associated with significantly better long-term outcomes. The 3-year recurrence rate was lower in the ICG group (13.7% vs. 25.0%, P<0.01), and mortality was reduced (19.0% vs. 32.7%, P<0.01). Kaplan-Meier curves ( Figure 2 ) demonstrated significantly better three-year DFS (HR=0.53, 95% CI: 0.34-0.80; log-rank P<0.01) and OS (HR=0.51, 95% CI: 0.31-0.84; log-rank P<0.01) in the ICG group. Multivariable Cox regression analysis ( Table 3 ) confirmed that ICG use was independently associated with improved DFS (HR=0.44, 95% CI: 0.28–0.71, P<0.01) and OS (HR=0.44, 95% CI: 0.25–0.76, P<0.01) after controlling for potential confounding factors ( Table 3 ). Additionally, advanced AJCC stage (II and III) was a significant negative prognostic factor for both DFS and OS (both P<0.01).

Accurate intraoperative tumor localization remains a challenge in totally laparoscopic procedures, particularly for early-stage or non-palpable lesions. Our results show that preoperative submucosal ICG injection significantly enhanced visualization of tumor margins, facilitating precise proximal margin determination and surgical planning. This aligns with the retrospective study by Yoon and Lee (11), which demonstrated that ICG marking not only secured an oncologically safe proximal resection margin during totally laparoscopic distal gastrectomy but also reduced operative time by approximately 34 minutes. However, several challenges persist regarding tumor localization accuracy. These include variability in the depth and volume of ICG injection, timing of administration relative to surgery, and heterogeneity in fluorescence signal intensity due to factors such as tissue thickness and tumor characteristics (12, 13). Moreover, subjective interpretation of fluorescence images may contribute to inconsistency in surgical decision-making. Previous studies have highlighted these issues as significant barriers to widespread clinical adoption (14). To address these challenges, future research should focus on developing standardized protocols for ICG administration, including precise injection techniques, optimal dosing, and timing schedules. In addition, the integration of objective quantification methods for fluorescence intensity and real-time imaging guidance could improve localization accuracy and reproducibility. Multicenter studies employing such standardized approaches would help validate the clinical utility of ICG fluorescence marking and facilitate its broader implementation in clinical practice.

Adequate LN dissection is critical for accurate staging and improved disease control in gastric cancer. In our analysis, the ICG group yielded a significantly higher number of retrieved LNs compared to the non-ICG group. Notably, this increase in LN yield was observed uniformly across both D1 and D2 nodal stations, indicating that ICG fluorescence guidance enhances retrieval from anatomically diverse and potentially challenging regions. This observation aligns with the findings of a phase III randomized clinical trial by Chen et al. (15), where the mean LN count was substantially higher in the ICG group (50.5 vs. 42.0, P<0.001). Moreover, their study reported superior three-year overall survival (OS) and disease-free survival (DFS) in the ICG group, along with a lower recurrence rate (17.8% vs. 31.0%). These results strongly support the prognostic value of ICG-guided lymphadenectomy, suggesting that a more extensive and precise LN dissection can contribute to improved long-term outcomes. It is noted that the more comprehensive LN retrieval in the ICG group may contribute to potential “upstaging” by detecting additional metastatic nodes. Such upstaging could partly explain the improved survival outcomes observed, as more accurate staging facilitates tailored postoperative treatment and surveillance. Importantly, the addition of ICG did not alter the predefined anatomical clearance range. All patients in both groups underwent D2 lymphadenectomy according to standardized oncologic guidelines. ICG fluorescence imaging served as an intraoperative navigational aid, improving visualization and identification of lymphatic tissue, but did not expand or reduce the intended dissection field. Therefore, the observed increase in lymph node yield in the ICG group likely reflects enhanced precision and completeness of dissection rather than a modification of surgical extent.

In addition, a prospective cohort study by Wei et al. (16) further validated the oncologic safety and feasibility of ICG fluorescence imaging in laparoscopic radical gastrectomy. Patients in the ICG-assisted group not only exhibited a higher LN yield but also benefited from shorter operation and dissection times and reduced intraoperative blood loss. Although the 2-year OS and DFS did not differ significantly between groups in Wei et al.’s study, the observed procedural improvements mirror our findings and suggest that ICG guidance enhances surgical quality without compromising oncologic safety.

The potential mechanism behind the improved outcomes with ICG may be twofold. First, real-time fluorescence imaging allows for more accurate identification of tumor margins and lymphatic drainage pathways, facilitating a more comprehensive resection. Second, a more thorough LN dissection may reduce the residual micrometastatic burden, thus contributing to a lower recurrence rate and improved survival. Our findings that LN yield increased uniformly across nodal stations and the potential for upstaging underscore that ICG fluorescence imaging not only enhances surgical precision but also may improve pathological staging accuracy, which in turn guides optimal postoperative management. Although these mechanisms remain to be fully elucidated, our results, in conjunction with previous studies, support the notion that ICG fluorescence imaging can translate into meaningful oncologic benefits beyond technical facilitation.

Despite the promising results, this study has several limitations. As a single-center retrospective analysis, our findings may be influenced by selection bias and center-specific surgical expertise. Although propensity score matching was employed to mitigate confounding, unmeasured variables might still affect the outcomes. Additionally, as detailed earlier, variability in injection depth, volume, timing, and subjective interpretation of fluorescence signals remains a major challenge. Future work should prioritize developing standardized ICG administration protocols and objective imaging quantification, supported by multicenter validation studies to facilitate clinical adoption. Finally, while our follow-up duration was sufficient to detect early recurrences, longer-term survival outcomes require further evaluation.