Patient characteristics
At first, 298 patients were eligible, but 28 were excluded because of the presence of residual tumors, palliative surgery, death within 1 month of surgery, primary or secondary tumor history, neoadjuvant chemotherapy, adjuvant radiotherapy, or incomplete medical record. The remaining 270 patients were included in the analysis; 169 and 101 were classified in the surgery and chemo groups, respectively (Figure 1).
Table 1 presents the characteristics of the patients. The median age in the surgery and chemo groups was 70 years (69-75 years) and 69 years (66-72 years), respectively (P=0.001). The male-to-female ratio was 2.1:1 in the surgery group and 3.2:1 in the chemo group (P=0.152). Most patients in the two groups had an ECOG score and a Charlson comorbidity score of <1 (ECOG: 88.2% and 89.1%; Charlson score: 94.7% and 96.0%; all P>0.05). The number of CTCs in the surgery group was 5 (2-16), while 4 (1-7) in the chemo group (P=0.328). In the surgery group, there were 56 patients with AJCC stage II and 113 with AJCC stage III, while in the chemo group, there were 25 patients with AJCC stage II and 76 with AJCC stage III (P=0.147). In the chemo group, six patients received monotherapy, and 95 patients received platinum-based doublet chemotherapy therapy; 57 patients received adjuvant chemotherapy for less than 3 months, while 44 patients received 3-6 months of adjuvant chemotherapy. The median follow-up in the surgery and chemo groups was 25 (IQR 13-44) months, and 22 (IQR 11-54.5) months, respectively (P=0.452).
Factors associated with OS and DFS in all patients
The univariable and multivariable Cox proportional hazards models for all patients are shown in Table 2. In the multivariable analysis, age >70 years (HR=1.640, 95% CI: 1.119-2.403, P=0.011) and stage III GC (HR=2.738, 95% CI: 1.677-4.471, P<0.001) were independently associated with OS. Surgery plus adjuvant chemotherapy (HR=0.511, 95% CI: 0.322-0.811, P=0.004), surgery performed in 2015-2018 (HR=0.586, 95% CI: 0.376-0.912, P=0.018), and stage III GC (HR=2.345, 95% CI: 1.466-3.751, P<0.001) were independently associated with DFS. Therefore, stage III GC was independently associated with both OS and DFS.
Overall survival
Unadjusted Kaplan-Meier survival curves were constructed for all patients in the two groups. The 1-, 3-, and 5-year OS rates of the surgery group were 72.9%, 51.8%, and 48.3%, compared with 90.1%, 66.4%, and 48.6% for the chemo group, respectively (HR=0.61, 95% CI: 0.42-0.92, P=0.018) (Figure2A). In the stage II cohort, the 1-, 3-, and 5-year OS rates of the surgery group were 96.3%, 80.4%, and 72.7%, compared with 90.5%, 73.8%, and 50.9% for the chemo group, respectively (HR=1.26, 95% CI: 0.483-3.29, P=0.637) (Figure 2B). For stage III patients, the 1-, 3-, and 5-year OS rates of the surgery group were 83.7%, 40.7%, and 28.7%, compared with 89.9%, 61.2%, and 43.6% for the chemo group, respectively (HR=0.58, 95% CI: 0.38-0.90, P=0.015) (Figure 2C).
Disease-free survival
The 1-, 3-, and 5-year DFS rates of the surgery group were 72.9%, 51.8%, and 48.4%, compared with 81.3%, 65.1%, and 53.6% for the chemo group, respectively (HR=0.682, 95% CI: 0.463-1.005, P=0.053) (Figure 2D). In the stage II cohort, the 1-, 3-, and 5-year DFS rates of the surg group were 85.3%, 78.5%, and 74.2%, compared with 91.8%, 64.4%, and 64.4% for the chemo group, respectively (HR=1.02, 95% CI: 0.416-2.54, P=0.9500 (Figure 2E). In stage III patients, the 1-, 3-, and 5-year DFS rates of the surg group were 66.5%, 37.7%, and 34.8%, compared with 77.9%, 60.0%, and 49.0% for the chemo group, respectively (HR=0.55, 95% CI: 0.36-0.83, P=0.007) (Figure 2F). The OS and DFS in the stage III subgroup were significantly different between the surgery and Chemo groups.
Subgroup survival analysis in stage III patients
The univariable and multivariable Cox proportional hazards models in stage III patients are shown in Table 3. Surgery plus adjuvant chemotherapy (HR=0.568, 95% CI: 0.357-0.903, P=0.017) and age >70 years (HR=1.573, 95% CI: 1.029-2.405, P=0.036) were independently associated with OS. Surgery plus adjuvant chemotherapy (HR=0.511, 95% CI: 0.322-0.811, P=0.004) and surgery performed in 2015-2018, HR=0.586, 95% CI: 0.376-0.912, P=0.018) were independently associated with DFS.
Chemotherapy regimens, duration, and toxicity
In stage II patients, platinum-based doublet chemotherapy group had the similar OS and DFS compared to the monotherapy group (P=0.473 and P=0.499) (Supplementary Figure S1A,C). There were no differences in OS and DFS when considering the duration of adjuvant chemotherapy (P=0.531 and P=0.531) (Supplementary Figure S1B,D). In stage III patients, the platinum-based doublet chemotherapy led to better OS and DFS compared with monotherapy (OS: P=0.037; DFS: P=0.013) (Figure 3A,C). No significant differences in OS and DFS were observed in relation to chemotherapy duration (P=0.430 and P=0.418) (Figure 3B,D). There were ten patients with grade 3 or above non-hematological toxicity adverse events, and six with grade 3 hematological toxicity adverse events (neutropenia) (Table 1).
Analysis in patients with available CTC data
An analysis was performed in patients who had a CTC count before surgery. There were 43 patients (13 stage II patients and 30 stage III patients) who were evaluated for CTCs, and 40 were positive. There were no significant differences in OS and DFS between the surgery and chemo groups among CTC-tested patients (OS: P=0.344; DFS: P=0.259) (Figure 4A,C). There were no significant differences in OS and DFS between the surgery and chemo groups among CTC-positive patients (OS: P=0.092; DFS: P=0.095) (Figure 4B,D).