This study is a retrospective cohort study. This study was conducted in Fujian Union Hospital from March 2019 to December 2020 retrospectively. The clinicopathological data of patients with serosal invasion gastric cancer (cT4NxM0) who received nab-paclitaxel + S-1(SAP) or camrelizumab + nab-paclitaxel + S-1 (C-SAP) regimen and underwent laparoscopy assisted gastrectomy were retrospectively analyzed.

Inclusion criteria: 1. gastric adenocarcinoma confirmed by gastroscopy and pathology before operation; 2. the clinical staging of CT and other imaging evaluation included: cT4; 3. lymph nodes N1 to N3; 4. patients with M0 without distant metastasis from liver, lung, peritoneum and other places evaluated by preoperative CT imaging were included; 5. The ECOG score was 0-2, and the blood indexes, liver and kidney function, cardiopulmonary function could tolerate chemotherapy or surgery. Exclusion criteria: 1. incomplete pathological diagnosis data; 2. patients with gastric stump cancer; 3. gastric cancer patients undergoing emergency surgery; 4. combined with other malignant tumors. Finally, 128 patients were included. Flow diagram was described in Supplementary Figure 1 .

The primary end-point was the 2-years overall survival rate, and the secondary end-points included 2-year recurrence free survival rate, initial recurrence time, pCR, MPR and safety.

We divided the patients into two groups according to the different neoadjuvant drug treatments: the SAP group (nab-paclitaxel + S-1), and C-SAP group (camrelizumab + nab-paclitaxel + S-1). The specific scheme was as follows:

The cycle of Nab-paclitaxel + S-1 chemotherapy consisted of the following: Day 1: Intravenous Nab-paclitaxel 260 mg/m² over 30 min. Dose reductions (220 mg/m², 180 mg/m², or 150 mg/m²) were permitted in patients with severe haematological or non-haematological toxicity. Day 1–14: S-1 at 120 mg/day for surface area ≥ 1.5m², 100 mg/day for surface area between 1.25 and 1.5m², and 80 mg/day for surface area < 1.25 m² were given 2 times daily. The next chemotherapy was repeated on the 22nd day.

The cycle of Camrelizumab consisted of the following: Day 1: Intravenous Camrelizumab 200mg.

Tumor regression grade according to Becker criteria (6, 7) included “Grade 1a” (Complete tumor regression i.e., 0% residual tumor per tumor bed), “Grade 1b” (Subtotal tumor regression i.e., <10% residual tumor per tumor bed) “Grade 2” (Partial tumor regression i.e., 10–50% residual tumor per tumor bed), “Grade 3” (Minimal or no tumor regression i.e., >50% residual tumor per tumor bed). MPR is defined as: TRG1a+TRG1b. Pathological complete response: pCR is defined as no invasive disease within an entirely submitted and evaluated gross lesion and histologically negative nodes based on central review.

TNM staging was performed according to the 8th edition of AJCC staging standard in 2016. Methods of lymph node treatment: after the specimens were isolated, the lymph nodes of each group were collected and subpackaged, fixed with 10% formalin solution, sent for pathological examination, and sorted by experienced pathologists. The depth of tumor invasion was divided into T1 (invasion of lamina propria, muscularis or submucosa), T2 (invasion of muscularis propria), T3 (invasion of subserosa, but not invasion of visceral peritoneum or adjacent organ), T4a (invasion of serosa) and T4b (invasion of adjacent organ). The staging of lymph node metastasis was divided into N0 (no regional lymph node metastasis), N1 (1 or 2 regional lymph node metastasis), N2 (3 to 6 regional lymph node metastasis), N3a (7 to 15 regional lymph node metastasis), and N3b (≥16 regional lymph node metastasis) according to the 8th edition AJCC staging of gastric cancer. Similarly, ypTNM staging was divided into stage I, stage II, stage III and IV according to the 8th Edition AJCC staging of gastric cancer. If the patient is diagnosed as cT4b before suegery and the postoperative pathology indicates ypT3 or below, it is considered as R0 resection.

Tumor response was assessed response evaluation criteria in solid tumors (RECIST), version 1.1 (8): target lesion evaluation criteria (1) CR: all (non lymph node) target lesions diSAPpeared, emphasizing that the short diameter of all original pathological lymph nodes (including target lesions and non target lesions) was<10mm after treatment. (2) PR: the total length and diameter of all target lesions decreased by 30% or more. (3) SD: the change is between PR and PD. (4) PD: the total length and diameter of all target lesions increased by at least 20%, and it was emphasized that the absolute value of the total length and diameter increase was more than 5mm; Or new lesions appear. When lymph nodes were evaluated as target lesions, it was judged that the sum of the long diameters of CR target lesions did not include lymph nodes, as long as the short diameters of all lymph nodes were<10mm; When judging PR, SD and PD, the short diameter of lymph nodes was added to the long diameter of all other target lesions for comparison before and after treatment.

After neoadjuvant therapy every 2 cycles, all patients need to review abdominal enhanced CT. Fasting for 6-12 hours before the examination. In addition, the patient drank 600-1000 ml of water to expand the stomach before CT examination. Generally, after 4-6 cycles of neoadjuvant therapy, the operation plan is formulated after multidisciplinary consultation according to the tumor regression grade.

The scope of gastric resection was selected according to the Japanese “Regulations on the treatment of gastric cancer” and the lymph node dissection around the stomach was performed. According to the Japanese Gastric Cancer Treatment Guidelines (5th edition) (9), we perform D2 lymph node dissection. For the distal stomach, we perform lymph node dissection for No.1,3,4sb,5,6,7,8a,9,11p,12a. For the entire stomach, we perform lymph node dissection for No.1-7,8a,9,11p,12a. Standard lymphadenectomy sequences and resection methods were performed as described in the Laparoscopic Gastrectomy for Gastric Cancer (10). LN dissection at station 10 was performed as a selective dissection: (1) When the primary tumor is located in the upper or middle part of the stomach and invades the greater curvature, and (2) When preoperative imaging or intraoperative findings show enlarged lymph nodes in the splenic hilum area, a lymph node dissection for the No.10 region is performed. When No. 14v nodes were highly suspicious for tumor involvement, a lymph node dissection for the No.14v region is performed. This is also a selective dissection.

The incidence of surgical complications is based on the number of patients who received surgical treatment as the denominator, and the number of patients with any of the following intraoperative/postoperative complications is the numerator ratio. The standard of intraoperative/postoperative complications refers to the early and late surgical complications mentioned in the intraoperative and postoperative observation items. The severity of complications was graded according to Clavien–Dindo (11) complication scoring system. IIIa and above were serious complications, Supplementary Table 1 .

The adverse Events of chemotherapy and immunotherapy are classified into grade 0-IV according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 (12), Supplementary Table 2 .

The endpoints of this study include OS and RFS, where OS is defined as the time from surgery to death from any cause. The outcomes include tumor related death, non tumor related death, loss of follow-up, and survival. RFS is defined as the time from the beginning of surgery to the recurrence or death of gastric cancer. According to the follow-up strategy of the Japanese gastric cancer treatment guidelines, patients are followed up every 3 months for the first 2 years and every 6 months for the following 3-5 years. The follow-up routine examination items include physical examination, laboratory examination (CA19-9, CEA, CA72-4), chest X-ray, abdominal ultrasound, CT, and annual gastroscopy examination. The follow-up methods include outpatient follow-up, phone calls, letters, and doorstep visits. The last follow-up date is due to January 2023.

The human research involved in this study has obtained approval from the Ethics Committee of ethics committee of Fujian Union Hospital (Ethics registration number: 2020YF013-01). The study was conducted in accordance with local laws, regulations, and institutional requirements. Detailed information about medication was provided to the patients, and treatment informed consent forms were signed and provided before administration.

All of the data were analyzed by SPSS software (SPSS, Chicago, IL, USA), version 25.0. The chi-square test or Fisher’s exact test was used for comparisons of categorical variables. The independent sample t test or the Mann-Whitney U test was used for comparisons of continuous variables. Apply COX regression analysis to analyze factors that affect overall survival rate. RFS and OS were estimated using the Kaplan Meier method, RFS and OS were compared using a log-rank test. When P<0.05, it indicates that the difference is statistically significant.

A total of 128 patients were included in this study and divided into two groups: S-1 + nab-paclixatel (SAP), n = 90; and camrelizumab-SAP (C-SAP), n = 38. The median follow-up was 22 months (range, 1–39 months). There were no statistical differences in age, sex, Eastern Cooperative Oncology Group (i.e., “ECOG”) score, Borrmann type, tumor location, preoperative adjuvant cycles, or tumor differentiation type between the two groups (P > 0.05) ( Table 1 ).

The proportion of ypT0 patients in the C-SAP group was 21.1%, which was significantly higher than that in the SAP group (5.6%) (P = 0.008). The proportion of ypN0 in the C-SAP group (63.2%) was significantly higher than that in the SAP group (38.9%), and the difference was statistically significant (P = 0.012). The proportion of those with pathological complete response (pCR) in the C-SAP group (18.4%) was statistically greater than that in SAP group (5.6%) (P = 0.03). The proportion of patients with TRG grade 1a+1b in the C-SAP group was 39.5%, which was similar to that in the SAP group (25.6%) (P = 0.115). The proportion of nerve invasion in the C-SAP and SAP groups was 28.9% and 42.2%, respectively (P = 0.158). The proportion of vascular invasion in the C-SAP and SAP groups was 26.3% and 36.7%, respectively (P = 0.241). The mean (± SD) number of harvested lymph nodes in the C-SAP and SAP groups was 41.6 ± 18.2 and 43.3 ± 14.1, respectively (P = 0.58). The number of positive lymph nodes in the C-SAP and SAP groups 2.3 ± 5.1 and 4.2 ± 3.1, respectively (P = 0.13). Finally, the proportion of those achieving partial response (PR) in the SAP and C-SAP groups was 92.2% and 90.9% (P = 0.982) ( Table 2 ).

The mean estimated blood loss in the C-SAP and SAP groups was 70.3 ± 120.9 ml and 41.9 ± 21.6 ml, a difference that was statistically significant (P = 0.008). The proportion of total gastrectomy in the C-SAP and SAP groups was 89.5% and 86.7%, respectively, with no statistical difference (P = 0.661). Two (5.3%) patients in the C-SAP group were converted to open laparotomy, with no conversion to open surgery in the SAP group (P = 0.109). In the SAP group, 1 (1.1%) patient underwent left partial hepatectomy. The R0 resection rate in the C-SAP and SAP groups was 97.4% and 98.9%, respectively (P = 0.58). Operative duration in the C-SAP and SAP groups was 201.3 ± 62.6 min and 208.6 ± 63.5 min, respectively (P = 0.867). The mean time to start liquid diet in the C-SAP and SAP groups was 3.5 ± 1.2 and 3.6 ± 1.0 days, respectively, (P=0.867). The mean time to start semifluid in the C-SAP and SAP groups was 5.1 ± 0.9 and 5.3 ± 0.7 days, respectively (P = 0.851). The mean length of postoperative hospital stay in the C-SAP and SAP groups was 9.0 ± 5.0 and 9.2 ± 11.9 days, respectively (P = 0.794) ( Table 3 ).

The proportions of overall postoperative complications were 24.2% and 22.1% in the C-SAP and SAP groups, respectively, with no statistically significant difference (P=0.801). The proportion of grade I-II complications in the C-SAP and SAP groups was 18.2% and 18.9%, respectively (P = 0.923). The proportion of grade III complications in the C-SAP and SAP groups was 6.1% and 3.2%, respectively (P = 0.826). Grades IV and V complications did not occur in either group. There was no statistically significant difference in the incidence of complications such as pneumonia, abdominal infection, postoperative bleeding, and anastomotic leakage between the two groups (P > 0.05) ( Supplementary Table 3 ).

The side effects of neoadjuvant therapy were also analyzed. The most common (grade 3, 4) side effects included declines in neutrophil and white blood cell counts, and increased plasma aspartate aminotransferase (AST)/alanine aminotransferase (ALT) levels. The proportion of patients exhibiting a decrease in white blood cell count (grade 3, 4) in the C-SAP and SAP groups was 20.0% and 21.1%, respectively (P = 0.984). The proportion of patients exhibiting a decrease in neutrophil count (grade 3, 4) was 22.9% in the C-SAP group and 23.3% in the SAP group (P = 0.955). The proportion of anemia (grade 3, 4) in the C-SAP and SAP groups was 3.0% and 2.1% respectively P = 1.000). The proportion of thrombocytopenia (grade 3, 4) in the C-SAP and SAP groups was 5.7% and 3.3%, respectively (P = 0.619). The proportion of patients exhibiting an increase in plasma AST/ALT level in the C-SAP group was 25.7% and 13.3% in the SAP group (P = 0.096). The proportion of febrile neutropenia in the C-SAP and SAP groups was 2.9% and 6.7 (P = 0.649) ( Supplementary Table 4 ).

Analysis revealed that the two-year overall survival (OS) rate in the C-SAP group (80.7%) was higher than that of the SAP group (67.8%) (P = 0.112) ( Figure 1A ). Further stratified analysis revealed that, among M0 patients, the two-year OS rate in the C-SAP group (79.9%) was similar to that of the SAP group (71.9%) (P = 0.703) ( Figure 1B ). Among M1 patients, the two-year OS rate in the C-SAP group (85.7%) was significantly better than that of the SAP group (0%) (P = 0.002) ( Figure 1C ).

The effect of risk factors on the prognosis of the population is illustrated in Supplementary Figure 1 . The OS rate of ypT0 patients (100.0%) was significantly higher than that of ypT1-T4 patients (68.5%) (P = 0.044) ( Supplementary Figure 2A ). The OS rate of ypN0 patients (74.6%) was higher than that of ypN1-N3b patients (66.2%) (P = 0.21) ( Supplementary Figure 2B ). The OS rate of M0 patients (79.1%) was significantly higher than that of M1 patients (34.9%) (P < 0.001) ( Supplementary Figure 1C ). The OS rate of patients with TRG 1a+1b (76.2%) was higher than that of patients with TRG 2+3 (70.8%) (P = 0.788) ( Supplementary Figure 2D ). The OS rate of patients who underwent ≤ 3 preoperative chemotherapy cycles (75.0%) was higher than that of those who underwent ≥ 4 cycles (70.4%) (P = 0.765) ( Supplementary Figure 2E ). The OS rate of patients who underwent ≤ 3 postoperative chemotherapy cycles (64.0%) was lower than that of those who underwent ≥ 4 cycles (76.2%), a difference that was statistically significant (P = 0.042) ( Supplementary Figure 2F ).

Univariate Cox regression analysis revealed that that ypT0, ≥ 4 postoperative chemotherapy cycles, and M0 were closely associated with patient prognosis. Further multivariate Cox analysis revealed that ≥ 4 postoperative chemotherapy cycles (hazard ratio [HR] 0.418 [95% confidence interval (CI) 0.207–0.891]; P = 0.023) and M1 (HR 5.304 [95% CI 2.464–11.417]; P < 0.001) were risk factors for long-term survival ( Supplementary Table 5 ).

The two-year recurrence free survival (RFS) rate was higher in the C-SAP group (62.0 %) than in the SAP group (46.2%) (P = 0.361) ( Figure 2 ).

There was no significant difference in the proportion of recurrence between the C-SAP and SAP groups after 2 years (44.3% versus [vs.] 55.8%, respectively; P = 0.294). However, the average time to recurrence in the C-SAP group was longer (18.9 months) than that in the SAP group (13.1 months) (P = 0.004) ( Table 4 ).

The two-year recurrence patterns of the C-SAP and SAP groups are detailed in Supplementary Table 6 , with distant metastasis being the most common, followed by peritoneal metastasis, and local recurrence being the least common.

There was no statistically significant difference in OS after recurrence between the two groups (P = 0.097) ( Supplementary Figure 3 ).

Computed tomography scans for continued immunotherapy after recurrence in the C-SAP group are shown in Supplementary Figure 4 .