highly significantP < 0.001
Importantly, our multivariate analysis already adjusted for all major clinicopathological confounders (age, gender, smoking status, clinical stage, PD-L1), and the association between Ki-67 and MPR/pCR remained highly significant (adjusted OR per 1% increase = 1.069, P < 0.001), suggesting that the observed predictive value is robust against measured confounding.