Prognostic dynamics of pathological complete response and ctDNA clearance after neoadjuvant/perioperative immunotherapy in cancer: a reconstructed individual patient analysis.

内容摘要

Pathological complete response (pCR) and circulating tumor DNA (ctDNA) clearance are increasingly used as intermediate endpoints in neoadjuvant and perioperative immunotherapy (IO) trials. However, whether complete pathological and molecular responders share comparable residual risk across tumor types remains uncertain. We performed a post-hoc pan-tumor analysis using reconstructed individual patient data from published Kaplan-Meier curves of phase II/III trials. All included trials contained a neoadjuvant immunotherapy component; trials that additionally incorporated postoperative adjuvant immunotherapy were classified as perioperative. We evaluated associations of pCR and ctDNA clearance, defined as conversion from baseline ctDNA positivity to undetectable ctDNA after neoadjuvant therapy, with event-free survival (EFS) and tested whether these biomarkers modified the association between treatment timing (neoadjuvant-only versus perioperative) and EFS. We reconstructed survival data for 1,867 patients achieving pCR and 352 patients achieving ctDNA clearance across 8 tumor types. Among patients with pCR, 3-year EFS varied substantially by tumor type (from 78.6% to 100%). In tumor types where such a comparison was possible, postoperative adjuvant IO did not improve EFS among patients achieving pCR. In contrast, ctDNA clearance showed heterogeneous prognostic dynamics. In within-tumor analyses restricted to trials reporting both endpoints, pCR was associated with superior EFS. pCR remains the most reproducible prognostic indicator across tumor types following neoadjuvant or perioperative IO. In cross-trial comparisons, no EFS benefit from postoperative adjuvant IO was observed among patients with pCR. ctDNA clearance showed substantial variability across tumors, likely reflecting both biological heterogeneity in tumor shedding and inter-trial assay differences.