Comparison of clinical characteristics between groups

A total of 122 locally advanced gastric cancer patients were screened, of which 43 patients did not meet the inclusion criteria and 8 patients met the exclusion criteria. Thus, 71 locally advanced gastric cancer patients who received bevacizumab plus chemotherapy (bevacizumab plus chemo group, n = 23) or chemotherapy alone (chemo group, n = 48) were retrospectively included. In the bevacizumab plus chemo group, there were 7 (30.4%) females and 16 (69.6%) males with a mean age of 52.5 ± 11.6 years. In the chemo group, there were 19 (39.6%) females and 29 (60.4%) males with a mean age of 56.1 ± 9.8 years. No discrepancy of clinical characteristics was observed between the bevacizumab plus chemo group and the chemo group, including age, sex, Eastern Cooperative Oncology Group Performance Status (ECOG PS) score, tumor site, or histological grade (all P > 0.05). More detailed characteristics of patients in the two groups are shown in Table 1.

Table 1 Clinical characteristicsComparison of RECIST 1.1 response between groups

The RECIST 1.1 response did not vary between groups (P = 0.109). In detail, in the bevacizumab plus chemo group, the rates of complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD) were 4.3%, 47.8%, 39.1%, and 4.3%, respectively. There was one (4.3%) patient who was not evaluable. In the chemo group, the rates of CR, PR, SD, and PD were 2.1%, 33.3%, 45.8%, and 14.6%, respectively. Two (4.2%) patients were not evaluable. There was no statistical significance in the objective response rate (ORR) (52.2% vs. 35.4%, P = 0.179) or disease control rate (DCR) (91.3% vs. 81.3%, P = 0.484) between the two groups; however, ORR and DCR tended to be increased numerically in the bevacizumab plus chemo group versus the chemo group (Table 2).

Table 2 RECIST 1.1 response Comparison of surgery information and pathological response between groups

The surgical resection rate (95.7% vs. 85.4%, P = 0.261) and R0 resection rate (87.0% vs. 75.0%, P = 0.356) did not reach statistical significance between the two groups, but they showed an increased trend numerically in the bevacizumab plus chemo group versus the chemo group. Regarding pathological response, the proportions of patients with TRG 0, 1, 2, and 3 were 13.6%, 18.2%, 54.5%, and 13.6% in the bevacizumab plus chemo group, and 4.9%, 12.2%, 51.2%, and 31.7% in the chemo group (P = 0.066). There was no statistical significance in the proportion of patients with TRG 0–1 between the two groups, while it tended to be elevated numerically in the bevacizumab plus chemo group versus the chemo group (31.8% vs. 17.1%, P = 0.213) (Table 3).

Table 3 Surgery and pathological response informationComparison of PFS and OS between groups

The 12-, 24-, 36-, and 48-month PFS rates were 85.8%, 72.8%, 58.3%, and 58.3% in the bevacizumab plus chemo group. In the chemo group, the 12-, 24-, 36-, and 48-month PFS rates were 69.4%, 47.0%, 33.4%, and 33.4%, respectively. There was no statistical difference in PFS between groups, but it tended to ascend numerically in the bevacizumab plus chemo group versus the chemo group (P = 0.078) (Fig. 1A).

Fig. 1

PFS and OS between groups. Comparison of PFS (A) and OS (B) between bevacizumab plus chemo group and chemo group

Regarding OS, the 12-, 24-, 36-, and 48-month OS rates were 94.7%, 86.8%, 65.1%, and 65.1% in the bevacizumab plus chemo group. The 12-, 24-, 36-, and 48-month OS rates were 90.4%, 76.2%, 46.5%, and 46.5% in the chemo group. The OS was not different between the bevacizumab plus chemo group and the chemo group (P = 0.224) (Fig. 1B).

Subgroup analyses for PFS and OS according to different clinical features

Subgroup analyses were performed based on age, sex, ECOG PS score, tumor site, histological grade, T stage, and N stage, respectively. The univariate Cox proportional hazard regression analysis revealed that bevacizumab plus chemo (vs. chemo) was not related to PFS or OS in any subgroup (all P > 0.05) (Table 4).

Table 4 Subgroup comparisons of prognosis between patients with different neoadjuvant regimensIndependent factors related to PFS and OS

Enter multivariate Cox proportional hazard regression analysis showed that bevacizumab plus chemo (vs. chemo) was independently linked with longer PFS in locally advanced gastric cancer patients [hazard ratio (HR) = 0.263, P = 0.015]. However, higher histological grade (HR = 2.480, P = 0.009) and higher tumor (T) stage (HR = 2.816, P = 0.033) were independently related to shortened PFS in locally advanced gastric cancer patients (Fig. 2A). In terms of OS, bevacizumab plus chemo (vs. chemo) was not independently linked with OS in locally advanced gastric cancer patients (HR = 0.207, P = 0.056). Other factors were not independently related to OS in locally advanced gastric cancer patients, either (all P > 0.05) (Fig. 2B).

Fig. 2

Enter multivariate Cox proportional hazard regression analysis for PFS and OS. The multivariate analysis of independent factors associated with PFS (A) and OS (B) in locally advanced gastric cancer patients

Comparison of grade 3–4 adverse events between groups

The grade 3–4 adverse events in the bevacizumab plus chemo group included nausea and vomiting (8.7%), anemia (4.3%), leukopenia (4.3%), neutropenia (4.3%), fatigue (4.3%), and diarrhea (4.3%). In the chemo group, the grade 3–4 adverse events included nausea and vomiting (6.3%), anemia (4.2%), leukopenia (4.2%), neutropenia (2.1%), fatigue (2.1%), liver dysfunction (2.1%), and lethargy (2.1%). No difference was found in the incidences of grade 3–4 adverse events between groups (all P > 0.05). Notably, there were no grade 3–4 hypertension or hemorrhage in either group (Table 5).