Indocyanine Green Tracer Using in Laparoscopic Radical Gastrectomy for Locally Advanced Gastric Cancer

1. Research purpose Patients with locally advanced gastric adenocarcinoma (cT2-4a, N0/+, M0) were selected as study subjects to investigate the safety, efficacy, and feasibility of ICG near-infrared imaging tracing in guiding laparoscopic D2 LND for GC by comparing an injection ICG group and a non-injection ICG group.
2. Research design Prospective, multicenter, randomized controlled, open-control, parallel assignment, superiority test. ClinicalTrials.gov: NCT04593615. IRB number: 2020YF028-01 2.1 multicenter This study was conducted in 15 centers, including Fujian Medical University Union Hospital; Peking University Cancer Hospital and Institute; Zhongshan Hospital, Fudan University; Teaching Hospital of Putian First Hospital of Fujian Medical University; Zhangzhou Affiliated Hospital of Fujian Medical University; The First Hospital of Jilin University; Southern Medical University Nanfang Hospital; The First Affiliated Hospital of Nanjing Medical University; Affiliated Hospital of Qinghai University; Qilu Hospital of Shandong University; Ruijin Hospital, Shanghai Jiao Tong University School of Medicine; Tangdu Hospital, Fourth Military Medical University; Tianjin Medical University Cancer Institute and Hospital; Cancer Hospital, Chinese Academy of Medical Sciences; and Sun Yat-Sen University Cancer Center. 2.2 Case Group Group A (Study Group, 535 cases): Laparoscopic gastrectomy group with the use of near-infrared imaging (ICG group) Group B (Control Group, 535 cases): Laparoscopic gastrectomy group without the use of near-infrared imaging (Non-ICG group) 2.3 Estimate Sample Size The three-year disease-free survival (DFS) was this study's main effectiveness evaluation index. The study implemented a superiority test (unilateral), assuming that the study group's three-year DFS would be better than that of the control group,23, according to previous research results of laparoscopic surgery for locally advanced GC in which the three-year DFS was 65.2%; therefore, the 3-year DFS of the control group would be 65.2%, assuming that the 3-year DFS of the experimental group could be increased by 9% to 74.2%. With an inspection level of 0.025 (unilateral) and an inspection efficiency of 0.9 and using PASS 11 log-rank tests (Lakatos) [Proportion surviving] the calculated sample size was N=428; namely, each group needed 428 people, considering cases of possible exclusion and loss to follow-up (20% drop out rate). The final sample size for each group was 535 cases, for 1070 cases. Approximately 62-63 cases were initially assigned to each center. 2.4 randomization After laparoscopic exploration and confirmation that the cases complied with the standards, they were included in the randomized groups in this study. We adopted the central dynamic layered segment randomization method, with control factors including age, gender, tumor location, clinical staging, research center of the given seed number, and segment length. SAS 9.2 produced a serial number ranging from 0001 - 1070 corresponding to the treatment allocation, which was reserved in the data center and research center. Each center had a specialist who sent the patients' information (age, sex, tumor location, clinical staging, and research center) to the randomization implementation department of the data center via email, telephone call, or text message, and the central randomization department analyzed this information to determine the case enrollment and inform the research center. 2.5 Blinding Method: This research adopts an open design. 2.6 Research cycle: Estimated enrollment cycle: complete enrollment within 2 years. Follow-up period: Enrolling the first case was the starting point for follow-up. and the enrollment of the last case to the postoperative pathology report (generally 2 weeks after surgery) was the endpoint of follow-up for secondary outcomes. The follow-up endpoint for the main outcomes was three years after the last case was enrolled. Estimated time: 2020.11-2022.10 (complete enrollment) to 2025.10 (complete follow-up)
3. Study Objects All patients who meet the inclusion criteria and do not conform to the exclusion criteria are qualified for this study.
4. Interim analysis: Safety and effectiveness of the interim analyses in the present study will be conducted when all of the patients had been enrolled, but there were no planned interim analyses for 3-year DFS rate.

Statistical Analysis Plan

1. Statistical software: A database will be established, and data will be entered into the database using Epidata3.0. Statistical analyses will be performed using SAS9.2 statistical software.
2. Basic principle: Except for the primary endpoint which is tested using superiority testing, other analyses employ methods of difference testing. All statistical tests were conducted on the cases that were randomized into groups. These individuals underwent surgical treatment either with or without indocyanine green fluorescent imaging (ICG-FI)-guided laparoscopic GC lymphadenectomy, and safety evaluation data post-intervention was available. These cases constituted the safety analysis population in this study. Safety indicators and the incidence of adverse reactions were statistically described and analyzed using a two-sided test. A test level with statistical significance was set at 0.05, and the confidence interval of the parameter was estimated using the 95% confidence interval.
3. Shedding analysis: The total shedding rate and shedding rate due to adverse events will be analyzed with Pearson χ2 test.
4. Statistical analysis of population division: The superiority test analysis of the primary endpoint, 3-year DFS, will be conducted based on the per-protocol (PP) population, and baseline data and efficiency analysis will also be conducted on a PP and Intention-to-treat (ITT) populations. The primary endpoint will also be analyzed based on the ITT basis, with the PP analysis results prevailing. The main efficacy indicators will be the incidence of adverse reactions, and SAP will be used as the denominator.
5. Balanced analysis for baseline characteristics: Demographic characteristics, general conditions before intervention, and baseline characteristics that affect prognosis will be included. Quantitative data will be analyzed using t tests or t' tests (for variance that is not homogeneous), qualitative data will be analyzed using the Pearson 2 test, and grade data will be analyzed using the Wilcoxon rank test.
6. Method for the determination of outliers: Any observed value that is lower or higher than the lowest (P25) or highest (P75) interquartile range will be considered an outlier. In the process of analysis, a sensitivity analysis of the outlier data will be performed, and the effects of retention and the elimination of outliers will be analyzed; in the case of no contradictions, the data will be retained; in the case of any contradictions, a decision will be made depending on individual cases.
7. Descriptive statistics: The measurement variables are the mean, standard deviation and confidence interval, and if necessary, the minimum, maximum, P25, median and P75; the paired measurement variables also consist of the mean and standard deviation of the difference; when using a nonparametric method, the median and average rank should be shown. Counting variables are the frequency distribution and the corresponding percentage. Ranked variables are the frequency distribution and the corresponding percentage as well as the median and average rank. Qualitative variables are the positive rate, the number of positives, and the denominator cases. Survival time data are the number of events and censored numbers, the median survival time, and the survival rate.
8. Frequencies of causes of first recurrence and death within 3 years after surgery in ICG and non-ICG groups were compared with Pearson 2 test, then P for chi-square was calculated.
9. Subgroup analysis: Prognostic factors will be analyzed according to the specific circumstances of the data.
10. Missing value processing: This study will not fill in missing values.
11. Effectiveness analysis: Survival data (time and rate) will be analyzed using a log-rank test. Multivariable and central effect analysis will be conducted using Cox's proportional hazards model. The analysis of center effects and subgroup analysis will be conducted based on specific situations. Logistic regression for qualitative indicators and Cox's proportional hazards model for survival data will be used to assess the effects of baseline, treatment, center, and treatment-by-center interactions. Quantitative variables will be analyzed using Student's t-test or another t-test (when the variance is not uniform), and categorical variables will be analyzed with the Pearson 2 test. Ranked variables will be analyzed with the Wilcoxon rank sum test. The central effect analysis and subgroup analysis are conducted according to the specific situation.
12. Safety analysis: The incidence of adverse reactions and incidence of adverse events will be analyzed with the Pearson 2 test, and a list of adverse events occurring in this study will be collected; laboratory test results will be described either before or after the study to identify normal/abnormal changes or the relationships between abnormal changes and drugs in the study, and the "normal/abnormal" changes that occurred in this study will be listed; the results of laboratory tests before and after the study of normal/abnormal changes and identification of the relationship between the drug and these results when abnormal changes occurred will be recorded. More detailed statistical analysis descriptions can be found in the statistical analysis plan.
13. Interim analysis: Safety and effectiveness of the interim analyses in the present study will be conducted when all of the patients had been enrolled, but there were no planned interim analyses for 3-year DFS rate.
14. The 3-year disease-free survival and overall survival rates were calculated using the Kaplan-Meier method, and the log-rank test was used to determine significance. The hazard ratios (HRs) comparing the ICG and non-ICG groups were estimated using Cox regression after confirmation of the proportional hazards assumption. Multivariate Cox regression analyses were performed to evaluate the effect of surgery type on survival, after adjustment for clinicopathologic covariates that were significantly associated with the outcome in univariate analyses. All-cause mortality was treated as a competing event for recurrence. The cumulative incidence in the presence of competing risks was calculated, and competing-risk survival regression was used as an alternative to Cox regression.
15. LND noncompliance was defined as the absence of LNs that should have been excised from more than 1 LN station. Major LN noncompliance was defined as more than 2 intended LN stations that were not removed25-26.
16. Current guidelines suggested that at least 16 regional LNs should be removed pathologically, and the removal of 30 or more nodes was desirable. The reference numbers 16 and 30 were used.
17. Subgroup analysis: Subgroup analysis is to find the factors that may affect prognostic according to the specific circumstances of the data. For example, Subgroup analyses, using log-rank tests, were conducted for disease-free and overall survival stratified by LN noncompliance (ie. compliant vs. noncompliant lymphadenectomy), and number of LN dissection ((ie. ≥30 retrieved LNs vs. <30 retrieved LNs).
18. Subgroup analysis and interaction tests: Use the P-value for an interaction term to test its significance.
19. When performing multiple comparisons, Benjamini-Hochberg method was used to control the false discovery rate.

20.95%CI was calculated by the Wilson procedure with a correction for continuity.