This single-arm, phase 2 trial was conducted at Henan Cancer Hospital. The trial was approved by the Ethics Committee of Henan Cancer Hospital (NO. 2019211) and conducted in adherence with the Declaration of Helsinki and Good Clinical Practice guidelines. Written informed consent was obtained from all patients before enrollment. The trial was registered at ClinicalTrials.gov with the identifier NCT03950271.

Eligible patients were aged 18–75 years; had pathologically confirmed G/GEJ adenocarcinoma with cT4 and/or N + M0 per American Joint Committee on Cancer (AJCC) staging manual 8th edition; resectable as determined by computed tomography (CT) or magnetic resonance imaging (MRI) for distant metastases, bone scan for suspected bone metastasis, laparoscopy for suspected peritoneal metastasis and CT or MRI for suspected brain metastasis; CT confirmed absence of peritoneal metastasis; a multidisciplinary team (MDT) believed that perioperative treatment was needed; HER2 positive status in tumors (immunohistochemistry (IHC) 3 + or IHC 2 + with fluorescence in situ hybridization positivity); an Eastern Cooperative Oncology Group performance status of 0–1; and no prior chemotherapy, targeted therapy, or local tumor resection. Exclusion criteria included prior immunotherapies (anti-CTLA4, anti-PD-1, or anti-PD-L1 antibodies); and a history of or active autoimmune diseases.

Eligible patients received four cycles of neoadjuvant therapy with camrelizumab (200 mg, iv, day 1, q3w), trastuzumab (initially 8 mg/kg, followed by 6 mg/kg, iv, day 1, q3w) and CapOx (oxaliplatin 130 mg/m2, iv, day 2, and capecitabine 1000 mg/m2, orally bid, days 1–14, q3w). When this trial was designed, no data was available on the appropriate interval between the last dose of camrelizumab and surgery. Thus, the administration of the fourth cycle of camrelizumab was at the investigator’s discretion to minimize safety concerns. Before surgery, the MDT reassessed the clinical stage and the resectability of the tumor. Patients deemed resectable underwent gastrectomy and D2 lymphadenectomy within two to six weeks following neoadjuvant therapy. Those who prematurely discontinued neoadjuvant therapy, but were still considered resectable, proceeded to surgery. Adjuvant therapy with four cycles of CapOx started within three to eight weeks after surgery. Treatment was continued until the completion of the prescribed cycles, disease progression, unacceptable toxicity, or consent withdrawal.

Imaging assessments were performed by CT or MRI every six weeks during neoadjuvant and adjuvant treatment, then every three months for up to three years, and every six to eight months for up to five years. Gastroscopy was repeated annually. Survival status was followed up every three months for the first three years and every six months thereafter. Radiologic response was evaluated based on RECIST v1.1 criteria. Pathologic response, including ypTxNxMx, tumor regression grade (TRG), and R0 resection, were evaluated based on the AJCC 8th edition. TRG 0, no viable cancer cells (complete response); TRG 1, single cells or rare small clusters of cancer cells (near-complete response); TRG 2, residual cancer cells with obvious tumor regression, but more than single cells or rare small groups of cancer (partial response); TRG 3, substantial residual cancer with no obvious tumor regression (poor or no response). Adverse events (AEs) were graded based on the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0. Surgical complications were assessed based on the Clavien-Dindo classification.

The primary endpoint was pCR (ypT0N0) rate, defined as the proportion of patients with no residual tumor cells in both the resected tumor tissue and lymph nodes. Secondary endpoints included near pCR (ypT0) rate (defined as the proportion of patients with no residual tumor cells in the resected tumor tissue), TRG rate, R0 resection, downstaging, intraoperative blood loss, operative duration, postoperative hospital stay, postoperative 90-day mortality, secondary surgery, objective response rate (ORR), disease-free survival (DFS, defined as the duration from surgical procedure to the first occurrence of disease recurrence or death), event-free survival (EFS, defined as the duration from the initiation of neoadjuvant treatment to the first occurrence of disease recurrence or death), overall survival (OS, defined as the duration from the initiation of neoadjuvant treatment to death from any cause) and safety.

The pCR (ypT0N0) rate of neoadjuvant chemotherapy was 5% [15]. Assuming neoadjuvant camrelizumab plus trastuzumab, oxaliplatin and capecitabine could achieve a pCR rate of 20%, a sample size of 22 would provide at least 80% power at an overall one-sided α = 0.05 to demonstrate the efficacy of the test neoadjuvant treatment was promising. Considering 10% drop-out, 25 patients were required to be enrolled.

All patients who received at least one dose of the study drug were included in the full analysis set (FAS) and safety analysis set (SAS). The surgery set (SS) comprised all patients who underwent gastrectomy. Efficacy endpoints were analyzed in FAS and SS. Neoadjuvant safety and surgical complications were evaluated in SAS and SS, respectively. Adjuvant safety was evaluated in patients who received at least one dose of CapOx during adjuvant phase. Continuous data were described as median (range), and categorical data were described as number (percentage). The pCR rate, near pCR rate, TRG rate, and ORR were calculated and the corresponding 95% confidence intervals (CI) were calculated by the Clopper-Pearson method. The median and two-year or three-year probability of DFS, EFS, and OS were estimated by the Kaplan–Meier method, and the 95% CI of two/three-year survival probability were calculated using Greenwood's formula. Subgroups analysis of survival outcomes were grouped by pathological responses (pCR vs. non-pCR). Statistical analyses were executed with SAS software (version 9.4).