Understanding the association between clinical staging accuracy, treatment response, and survival among gastric cancer patients through Bayesian analysis

Background: Neoadjuvant therapy (NAT) improves survival among patients with locally advanced gastric cancer (GC), but it remains unclear whether its benefit is contingent on treatment response. Methods: This is a national cohort study of stage Ib-III GC patients in the National Cancer Data Base (2006–2015) treated with upfront resection or NAT followed by surgery. Bayesian analysis was used for NAT patients to ascertain staging concordance and to account for down-staging. We used multivariable Cox regression to evaluate the association between staging concordance, treatment, response to NAT, and survival. Results: The cohort included 13 340 patients treated at 1124 hospitals. Staging concordance ranged from 86.1% for cT3-4N+ to 34.7% for cT2N0 patients. Relative to accurately staged patients treated with upfront surgery, NAT was associated with a decreased risk of death if there was disease down-staging among those with cT1-2N+ (hazard ratio [HR]: 0.43 [0.30–0.61]), cT3-4N0 (HR: 0.69 [0.54–0.88]), and cT3-4N+ (HR: 0.51 [0.48–0.58]) tumors, and in the absence of down-staging among cT3-4N+ patients (HR: 0.83 [0.74–0.92]). Conversely, NAT without down-staging increased the risk of death among those with intermediate-stage disease. Conclusions: NAT is associated with improved survival for GC, but it seems to be contingent on treatment response among patients with intermediate-stage disease.