Background: Pathologic complete response (pCR) is utilized more frequently as an endpoint in neoadjuvant immunotherapy trials for resectable non-small cell lung cancer (NSCLC).Its utility as a surrogate for long-term survival, however, remains debated.We conducted a systematic review to evaluate the association between pCR and survival outcomes in this setting. Methods:We performed a systematic search of PubMed, Embase and Cochrane Library, according to PRISMA guidelines.We included prospective trials and retrospective cohort studies that either with or without chemotherapy reporting pathologic complete response (pCR) after neoadjuvant immunotherapy in resectable nonsmall cell Lung cancer (NSCLC) with survival outcomes.We extracted data on definitions of pCR, major pathologic response (MPR), disease-free survival (DFS), eventfree survival (EFS), overall survival (OS), and patterns of recurrence.We used ROBINS-I to assess risk of bias.Results: A total of 23 studies with 3,487 patients were included.The reported pCR rates were between 6% and 38%, with higher and consistent pCR rates after chemo-immunotherapy regimens.Patients achieving pCR demonstrated improved survival outcomes, with pooled 2-year DFS rates of 85-92%, compared with 55-70% in non-pCR patients.Despite favorable prognosis, recurrence after pCR was reported in 8-15% of cases, predominantly as distant relapse.Correlation between pCR and OS was limited by short follow-up in most studies.Notably, several studies reported a similar or stronger association between MPR and survival compared with pCR.Substantial heterogeneity was observed in pathological assessment methods and pCR definitions.Conclusions: pCR is associated with improved survival following neoadjuvant immunotherapy in resectable NSCLC, supporting its prognostic relevance.Despite pCR, recurrence occurs, and the heterogeneity of pathologic evaluation limits the validity of pCR as a solitary surrogate endpoint.The combination of pCR with MPR and other clinical or molecular variables may thus offer a more solid framework for outcome prediction.
Zhang et al. (Tue,) studied this question.