Background: Preoperative chemotherapy has been widely recommended in patients with locally advanced gastric or gastro-esophageal junction cancer (GC/GEJC), and the addition of programmed cell death 1 (PD-1) inhibitor is under investigation. We conducted a single-arm phase II trial (Stimulate-01) to evaluate the efficacy and safe of the anti-PD-1 antibody nivolumab plus SOX regimen in patients with locally advanced GC/GEJC and to explore the biomarker associated with therapy response. Methods: Patients with locally advanced GC/GEJC (cT3-4N+M0) were enrolled and received either 3 preoperative and 3 postoperative cycles of nivolumab (360 mg, IV, d1, Q21d) plus SOX regimen (oxaliplatin 130 mg/m², IV, d1 with oral S-1 40-60mg, bid, d1-d14, Q21d) therapy, followed by 11 cycles of nivolumab monotherapy. The primary endpoint was the pathological complete response (pCR) rate, while the major pathological response rate (mPR), 3-year-DFS and 3-year-OS as the second endpoint. This clinical trial was registered at Clinicaltrial.gov (NCT05739045). In addition, tissue sample before and after treatment from patients were performed scRNA-seq. Results: Forty-six patients were enrolled from November 2022 to March 2023, with a median age of 66 years (range, 34-74), and 38 were male. The pCR rate was 28.26% [95% confidence interval (CI), 15.2%-41.3%], and the major pathologic response (mPR) [tumor regression grade (TRG)0/1] rate was 41.3% (95% CI, 27.1%-55.5%). All the patients underwent R0 resection. The mPR rate in patients with PD-L1 CPS ≥5 and patients with PD-L1 CPS<5 was 42.86% vs 38.89% (P= 0.790). In addition, This clinical study period spans from epidemic control to full liberalization in China. The mPR rate in patients infected with COVID-19 was 44.44% (12/27), while the mPR rate in patients who have not been infected with COVID-19 was 36.84% (7/19) (P=0.146). There were no severe complications or death related to the surgery. The scRNA-seq data revealed that the composition of MHC-II (HLA-DRA, HLA-DPA1, HLA-DRB1, HLA-DPB1 and HLA-DQB1) were significantly over-expressed in cancer cell of treatment-sensitive patients (TRG0/1) compared to that of treatment-insensitive patients (TRG2/3). Then the MHC-II expression in pre-treatment biopsy samples from 240 GC patients who received preoperative immunotherapy plus chemotherapy were performed by immunohistochemistry. The results showed that the mPR rate in patients with MHC-II IHC ≥2 and patients with MHC-II IHC<2 was 58.87% (83/141) vs 21.21% (21/99) (P<0.001). Conclusions: Nivolumab plus SOX as neoadjuvant therapy showed an encouraging pCR rate, mPR rate, and manageable safety. The long-term efficacy of this combination of regimens will be continued to follow up. In addition, the MHC-II expression may be a potential biomarker for predicting the effectiveness of immunotherapy combined with chemotherapy for GC/GEJC. Clinical trial information: NCT05739045.