This study retrospectively analyzed the data of 101 patients with locally advanced gastric cancer who received NCI in the Department of Gastric Surgery, Fujian Medical University Union Hospital from January 2019 to April 2021. The inclusion criteria were as follows: age 18–75 years; with primary gastric adenocarcinoma confirmed by histopathology, clinical stage: cT2–4, lymph node N0~N3, and no distant metastasis (M0); received no chemotherapy (radiotherapy) or other antineoplastic therapy within 6 months; with computed tomography (CT) scans available during diagnosis and before operation; and without evidence of distant metastasis, such as liver metastasis or peritoneal implantation metastasis after laparoscopic exploratory surgery. Meanwhile, the exclusion criteria were as follows: with cancer complicated with malignant diseases of other organs; with evidence of peritoneal dissemination or distant metastasis (including intraoperative exploration after neoadjuvant therapy); and with history of gastrectomy or endoscopic submucosal dissection. In total, 11 cases without operation, six cases with incomplete CT data, and three cases with abdominal implant metastasis were excluded. Finally, 81 patients were included in the analysis ( Figure S1 ). The study was reviewed and approved by the Ethics Committee of Fujian Medical University Union Hospital.

Body components comprise adipose and non-adipose tissues, the former including subcutaneous adipose, visceral adipose, and intermuscular adipose tissues, and the latter including the muscles, bones, and internal organs (13). A single CT image of the third lumbar vertebra (L3) was selected to quantify muscle and adipose features, as the anatomical location was closely related to body volume (13). According to the standard Hounsfield unit (HU) range, skeletal muscle cross-sectional area (SMT, −29–150 HU), visceral adipose tissue (VAT, −15–50 HU), and subcutaneous adipose tissue (SAT, −190–30 HU) were quantified. A researcher (L.J.X.) tackled how to accurately capture the image in the middle of L3 and segment muscle and adipose tissue. All CT images without any patient information were then analyzed using the SliceOmatic version 5.0 (TomoVision) (14). The measured value of each body component (square centimeter) divided by the square meter of height was converted into an index (SMI, visceral adipose index [VAI], and subcutaneous adipose index [SAI]) (15, 16),. Body mass index (BMI) was calculated by dividing the weight by height squared. In the analysis, the patients were divided into groups according to BMI as follows: low BMI (<25 kg/m²) group and high BMI (≥25 kg/m²) group. According to the results of Martin (17), the male patients with BMI < 25 kg/m² and SMI < 43 cm/m² (or BMI ≥ 25 kg/m² and SMI < 53 cm/m²) were considered L-SMI, while the female patients with SMI < 41 cm/m² were considered L-SMI regardless of BMI. According to the relationship between the SAI and incidence of irAEs, the patients were divided into two groups as follows: high subcutaneous adipose group (H-SAI) and low subcutaneous adipose group (L-SAI). Additionally, we used the median to classify the VAI because no threshold for VAI has been clinically established. The △SMI, △VAI, and △SAI represent the changes of SMI, VAI, and SAI before and after neoadjuvant therapy, respectively.

The neoadjuvant immunotherapy regimen is a fluorouracil-based chemotherapy combined with ICBs. The SOX/XELOX regimen generally comprised 2–4 cycles of SOX/XELOX regimen (18)(S-1 40–60 mg/m² or capecitabine 1000 mg/m², twice a day, days 1–14, and oxaliplatin 130 mg/m² intravenous injection on the first day). The FOLFOX4 regimen comprised 2–4 cycles (19)(day 1: oxaliplatin 85 mg/m², calcium folinate 200 mg intravenous drip for 2 h, fluorouracil 400 mg/m² intravenous drip, 22 h intravenous drip of fluorouracil 600 mg/m²). ICBs were administered intravenously along with the chemotherapy cycle on the first day of chemotherapy (the drug dose was determined according to the patient’s body surface area, and the dose was reduced appropriately for patients with severe blood toxicity or non-blood toxicity). The next cycle of chemotherapy was repeated on the 22nd day. According to the criteria described by the Japan Gastric Cancer Association (JGCA), a whole abdominal CT scan was performed every 6–8 weeks to evaluate the response to neoadjuvant therapy, and improve the results of the relevant laboratory tests (including blood routine, liver and kidney functions, and tumor markers.) (20). The operation was performed at least 3 weeks after the completion of the neoadjuvant therapy. All surgical operations, including the extent of lymph node dissection, were performed in accordance with the guidelines of the JGCA (21), while staging was performed according to the tumor–node–metastasis classification (American Joint Committee on Cancer staging, 8th edition) (22). The Becker regression criteria were used to quantify the pathological reaction after treatment. The standard is based on the estimation of the percentage of living tumor cells relative to the tumor bed that can be recognized by the naked eye, and includes the following categories: TRG1a (no residual tumor cells), TRG1b (<10% residual tumor cells), TRG2 (10%–50% residual tumor cells), and TRG3 (>50% residual tumor cells) (23). In this study, TRG grade 1a/1b/2 was considered tumor remission (TR), and the TRG3 grade was considered non-tumor remission (non-TR).

Treatment-related adverse events (TRAEs) were assessed according to the National Cancer Institute-General terminology Standard for adverse events (AEs) version 4.0 (24). TRAEs included events reported between the first administration and the last study 30 days after treatment. For further analysis, the toxicity was classified into grades I–II and III–IV. irAEs are defined as AEs associated with immunosuppressant exposure and in accordance with immune-related phenomena (25). TRAEs include general AEs and irAEs. Any delayed dose or early cessation of treatment recorded the result of significant toxicity (III–IV), which was defined as dose limited toxicity in this study.

As in previous studies, all the patients received nutritional risk screening using the Nutritional Risk Screening 2002 (NRS 2002), as recommended by the European Society for Clinical Nutrition and Metabolism (26), and developed personalized nutritional support therapy. Patients with an NRS score ≥ 3 were routinely provided oral nutritional supplements. For patients who were unable to meet their energy needs through oral feeding, enteral tube feeding and/or parenteral nutrition were provided (27). Additionally, all patients received a nutritional assessment every 2 weeks to adjust their nutritional support treatment until 1 week preoperatively.

The main endpoint was pathological reaction. The secondary endpoints included TRAEs and irAEs. Normally distributed variables are described as the absolute number and percentage, mean, and standard deviation, and nonparametric variables are described as median and interquartile range. The classified variables were analyzed by double X² test or Fisher’s exact test, the continuous variables were compared by Student’s t test, and the paired t-test was used before and after the comparison. The correlation among the parameters was analyzed by pearson correlation coefficient, and the cutoff point of △SMI was intercepted according to the maximum area under the receiver operating characteristic curve. According to the smooth curve, the relationship between SAI and irAEs is explored, and the potential confounding factors are adjusted ( Figure 1 ). The experimental method was used to determine the relationship between the incidence of irAEs and SAI levels, move the test inflection point along predefined intervals, and detect the inflection point of the maximum model possibility. We further apply applied a two-stage linear regression model to test the threshold effect of SAI on irAEs according to the smoothing curve ( Table S2 ). A logistic regression model was used for the univariate and multivariate analyses. Significance was set at p<0.05. All statistical analyses were conducted using the SPSS software version 22.0 and Empower Stats 2.0.

Altogether, 81 patients who received NCI and underwent gastrectomy in the Fujian Medical University Union Hospital from January 2019 to April 2021 were included. Among them, 63 patients were male (77.8%), and 18 were female (22.2%). The median age of the patients was 62 years (57–67). The median preoperative neoadjuvant therapy cycle was three (3–4). Total gastrectomy and distal subtotal gastrectomy were performed in 64 (79%) and 17 (21%) cases, respectively. Postoperative complications occurred in 16 cases (19.7%), including pulmonary infection in 10 cases (12.3%), abdominal infection in four cases (4.9%), and anastomotic leakage in two cases (2.5%). Postoperative pathological stages included PCR in 12 cases (14.8%), ypI stage in 11 cases (13.6%), ypII stage in 20 cases (24.7%), and ypIII stage in 38 cases (46.9%). According to the Lauren classification, the intestinal type was recorded in 68 cases (84%) and the diffuse and mixed type in 13 cases (16%). After neoadjuvant therapy, 51 cases (62.9%) had TR, and 30 cases (37.1%) had non-TR. According to the set threshold, the body components were divided into the L-SMI group with 56 cases (69.1%), high SMI (H-SMI) group with 25 cases (30.8%), L-SAI group with 47 cases (57.1%), H-SAI group with 34 cases (41.9%), low VAI (L-VAI) group with 40 cases (49.4%), and high VAI (H-VAI) group with 41 cases (50.6%) ( Table 1 ).

Figure 2 presents a representative L3 plane CT image segmentation legend with the patient’s baseline state. Panel A depicts the representative segmentation of L-SMI, panel B indicates the representative segmentation of H-SAI, and panel C demonstrates the representative segmentation of H-VAI. After neoadjuvant therapy, the BMI (from 23 to 22.6 kg/m², p=0.042), SAI (from 34.7 to 32.9 cm²/m², p=0.01), and VAI (from 32.4 to 26.8 cm²/m², p=0.005) were significantly lower than those before treatment, while the SMI had no significant change (44.7 vs 42.5 cm²/m², p=0.278)( Figure 3 ). We also analyzed the relationship between the body components of patients at baseline and serum nutritional markers. The median of albumin (ALB) levels was 38g/L (33–41) before treatment and 37g/L (34–42) after treatment. The difference between the two groups before and after treatment was significant (p=0.047) ( Table 2 ). Before neoadjuvant therapy, the SMI was positively correlated with ALB levels (Pearson’s=0.3, p=0.009), but not with SAI and VAI (SAI: Pearson’s=−0.151, p=0.626; VAI: Pearson’s=0.119, p=0.856).

Compared with the TR group, the patients in the non-TR group were younger (63 years vs 60 years, p=0.029) and had more patients in stage III (27.5% vs 80%, p=0.001). No significant differences in sex, interval between diagnosis and operation, neoadjuvant treatment cycle, surgical methods, postoperative complications, and AEs were observed between the two groups. In terms of body composition, the non-TR group had more patients with L-SMI (60.8% vs 83.3%, p=0.046), and no significant differences in SAI, VAI, and BMI were noted between the two groups (p>0.05). During the neoadjuvant therapy, the SMI of patients in the non-TR group decreased more (△SMI: 0.5 vs −1.5, p=0.05), although the SAI and VAI did not change significantly (p>0.05) ( Table 3 ).

Table 4 summarizes the results of the univariate and multivariate logistic analyses of body composition and their changes in patients with tumor regression before neoadjuvant therapy. Among them, the L-SMI (odds ratio [OR]: 3.45, 95% confidence interval [CI]: 2.06–6.81, p=0.041), and SMI ≥ 1.8 (OR: 1.38, 95%CI: 1.09–2.89, p=0.037) were risk factors for non-TR. After adjusting for age, Eastern Cooperative Oncology Group score, cT, cN and other clinical-related factors, the multivariate logistic analysis revealed that L-SMI (OR: 3.23, 95%CI: 1.06–9.81, p=0.047), SMI ≥ 1.8 (OR: 1.45, 95%CI: 1.20–3.48, p=0.048), and clinical node positivity (cN+) (OR: 6.99, 95%CI: 2.35–20.82, p=0.001) were independent risk factors for non-TR.

In this study, 50 cases (61.7%) of TRAEs were recorded, of which 32 cases (39%) were grade III–IV TRAEs. Meanwhile, 15 cases (18.5%) had irAEs, of which seven cases (8.9%) were grade III–IV irAEs, including three cases of abnormal liver function, one case of interstitial pneumonia, two cases of maculopapular rash, and one case of immune colitis,irAEs are recorded separately ( Table S1 ). No grade V AEs and drug withdrawal due to AEs were recorded.

Logistics analysis found that SMI(OR: 3.54,95%CI: 0.66-6.45,p=0.891), SAI (OR: 2.24,95%CI: 0.79-2.53, p= 0.119), VAI (OR: 0.56,95%CI: 0.46-1.64, p= 0.215) had no significant correlation with TRAEs ( Table S2 ). However, when analyzing the relationship among SMI, SAI, VAI and irAEs, we found a nonlinear relationship between the incidence of SAI and irAEs.( Figure 1 ). As SAI reaches a turning point (28.5 cm²/m²), the risk of irAEs increases ( Table S3 ). The incidence of irAEs in the H-SAI group (SAI ≥ 28.5 cm²/m²) was 29.8%. The incidence of irAEs in the L-SAI (SAI < 28.5 cm²/m²) group was 3%. Table 5 presents the results of the univariate and multivariate logistics analyses of body composition and its changes before neoadjuvant therapy. The H-SAI, H-BMI, and cN+ were related to the occurrence of irAEs. Further multivariate analysis revealed H-SAI as an independent risk factor for irAEs (OR: 14, 95%CI: 1.73–112.7; p=0.013). Among them, the incidence of abnormal liver function increased mainly (aspartate aminotransferase increased at 29.6% vs 8.9%, p=0.043; alanine aminotransferase increased at 25.9% vs 2.8%, p=0.009) ( Figure S2 ).