The patients enrolled in this study were at a relatively more advanced stage, with 30.6% classified as N3 and 37.4% as stage IIIC. These proportions were higher than those reported in previous studies, which were around 16–18 and 15%, respectively [14, 15]. The enrollment criteria for neoadjuvant gastric cancer research vary slightly between Eastern and Western studies. Western studies usually included patients with cT2, while Asian studies predominantly enrolled more advanced patients with T3–4 or those with Borrmann type 4, large type 3, or bulky N2 tumors. In this randomized, phase 2 clinical trial, preoperative DOS significantly improved the MPR rate compared to the SOX regimen (25.4 vs. 11.8%, P = 0.04) in patients with LAG/GEJ cancer. Furthermore, the DOS regimen showed a notable 17.1% increase in the R0 resection rate. The preoperative DOS triplet chemotherapy exhibited higher efficacy in shrinking tumors and showed a promising trend in translating this efficacy into long-term survival benefits. Compared to the SOX regimen, perioperative DOS demonstrated a 33.3% reduction in the risk of progression with numerical improvement in the 3-year PFS rate (Log-rank P = 0.07). Furthermore, although distant metastasis remained the main pattern of recurrence, the DOS regimen exhibited a relative advantage over SOX in reducing local recurrence and controlling distant metastasis. The local recurrence rate was as low as 7%, and the distant metastasis rate was decreased by 16.7%.

Although the DOS regimen demonstrated a higher antitumor activity compared to the SOX regimen, both groups exhibited relatively lower rates of pCR (7 vs. 3.9%) and 3-year PFS (52.3 vs. 35%). In this study, the pCR rate was defined as the proportion of patients with ypT0N0M0 in the mITT population. In the FLOT4 study, which enrolled patients with cT2, pCR rate (defined as ypT0) in mITT population was 16% with FLOT regimen [16]. While in the Asian PRODIGY and RESOLVE studies primarily included patients with cT3–4, the pCR rates (defined as ypT0N0) in surgery population for the DOS and SOX regimens were 10.4 and 5.6%, and the 3-year PFS/DFS rates were 66.3 and 59.4%, respectively [14, 15]. The recent MATTERHORN study also reported an incidence of 7% with ypT0N0 in the FLOT plus placebo arm [17]. The undesirable outcome in this study could be partly due to the enrollment of patients with more locally advanced diseases and a relatively higher proportion of patients who did not undergo surgery. Additionally, as laparoscopic exploration was not mandatory at baseline, some patients with peritoneal metastasis were included.

In addition, a significant improvement in the survival of patients who achieved MPR was observed in this study, which was consistent with previous meta-analysis findings. Gastric cancer patients with residual tumor cells <10% after neoadjuvant chemotherapy experienced better survival outcomes, with a 54% reduction in the risk of death (P < 0.001) [18]. Our results also provided support for the use of MPR as the surrogate primary endpoint in phase 2 studies dealing with neoadjuvant treatment in gastric cancer.

In metastatic GC, modified SOX with a reduced dose of oxaliplatin (100 mg/m2) had a similar efficacy compared to standard dose of CS (cisplatin plus S-1) in the first-line treatment (ORR 55.7 vs. 52.2%) [19]. Based on the finding, we designed the DOS regimen with a reduced dose of oxaliplatin (100 mg/m2) in this study. Actually, the completion rates of 4 cycles of preoperative chemotherapy (76.1 vs. 71.1%) and 6 or more cycles of perioperative chemotherapy (56.1 vs. 60.8%) were similar in both the DOS and SOX groups. Interestingly, the DOS regimen exhibited favorable safety without increasing toxicities compared with SOX. A lower incidence of thrombocytopenia in the DOS group was also observed in this study. The reduced dose of oxaliplatin might partly contribute to the favorable tolerance of the triplet DOS regimen. Immune-mediated reaction is one of the mechanisms of thrombocytopenia induced by oxaliplatin [20]. Steroid was reported to play a role in managing oxaliplatin-induced thrombocytopenia [21]. Therefore, premedication with dexamethasone before docetaxel might help to reduce the occurrence of thrombocytopenia to some extent. The DOS regimen did not increase the risk of postoperative complications or perioperative mortality. When compared with the FLOT4 study, the incidence of some adverse events of DOS in our study was numerically lower, mainly including grade 3–4 neutropenia and diarrhea [9]. This may be associated with the lower dose intensity of docetaxel (20 vs. 25 mg/m2/week) and oxaliplatin (33.3 vs. 42.5 mg/m2/week) [9].

Although FLOT is recommended as one of the perioperative chemotherapy regimens in China, its efficacy and safety have not been fully confirmed in Chinese population. And the inconvenience of intravenous infusion of 5-FU also contributes to the limited widespread use. However, the DOS regimen, with the oral administration of S1, is more convenient to promote. In recent years, several studies have focused on incorporating ICIs into the perioperative treatment for locally advanced GC [17, 22,23,24,25]. In the KEYNOTE-585 phase 3 trial, doublet chemotherapy plus pembrolizumab revealed a significant increase in pCR rate (12.9 vs. 2%, P < 0.0001) compared with chemotherapy plus placebo, while without a significant improvement in event-free survival (EFS) [26]. In the MATTERHORN phase 3 study, combining FLOT with durvalumab significantly increased the pCR rate (19 vs. 7%, P < 0.00001), and survival outcomes have not been reported yet [17]. In view of the favorable safety profile, it was worth further investigating DOS as a partner for ICIs in preoperative treatment.

There are several limitations in the present study. First, it was a single-center study, so the survival advantage needs to be further clarified in a multicenter phase 3 clinical trial. Second, routine diagnostic laparoscopy (DSL) was not mandatory according to the study design. Only CT scans were used to exclude peritoneal metastasis. DSL should be recommended for patients undergoing neoadjuvant treatment. Third, the fact that a relatively high proportion of patients did not receive surgery might bring bias in interpretating the data.