EBV DNA dynamics guide post-local-therapy ICI consolidation in stage II-III pulmonary lymphoepithelioma-like carcinoma.

274

Background: Immune checkpoint inhibitors (ICIs) improve outcomes in pulmonary lymphoepithelioma-like carcinoma (PLELC). Plasma Epstein–Barr virus (EBV) DNA reflects tumor burden, but how on-treatment EBV DNA dynamics can guide post–local-therapy ICI consolidation remains unclear. Methods: We retrospectively enrolled stage II–III PLELC from four centers treated with definitive surgery and/or radiotherapy, with or without immunotherapy, and ≥1 plasma EBV DNA assessment after definitive local therapy. EBV DNA was measured at baseline (t0), after induction therapy (t1), and after definitive local therapy (t2). Within induction-ICI and non–induction-ICI strata, we evaluated consolidation ICI benefit by (i) EBV DNA status at each timepoint, (ii) molecular trajectories (early clearance, delayed clearance, persistent positivity), and (iii) integrated molecular–radiologic response categories. Cox proportional hazards–based prognostic models were built and compared using Harrell’s C-index. Results: Among 208 patients (84.6% stage III), induction ICI and consolidation ICI were given in 36.5% and 36.1%, respectively. In patients without induction ICI, consolidation ICI improved PFS only in the persistent-positivity trajectory (median PFS: not reached vs 19 months; p = 0.037), with no clear benefit in early or delayed clearance. In patients with induction ICI, benefit was confined to the early-clearance group (72 vs 42 months; p = 0.034). Using integrated molecular–radiologic stratification, non–induction-ICI patients benefited mainly among t1 double non-responders (not reached vs 18 months; p = 0.001) and t2 non-responders (not reached vs 46 months; p = 0.024), whereas induction-ICI patients benefited among t1 and t2 double responders (t1: not reached vs 42 months, p = 0.044; t2: not reached vs 47 months, p = 0.036). A PFS model combining clinical factors with t2 EBV DNA and t2 radiologic response showed good discrimination (C-index 0.763, 95% CI 0.734–0.825) and outperformed t2 EBV DNA alone. Conclusions: Peri-treatment plasma EBV DNA trajectories and integrated EBV DNA–radiologic response refine identification of patients most likely to benefit from consolidation immunotherapy after definitive local therapy in stage II–III PLELC, enabling an interpretable biomarker framework and individualized risk assessment to guide consolidation ICI.