作者
Luca Marinelli,Maria Chiara Brunese,Vittorio Patanè,Fiammetta Cappabianca,Nicola Maria Tarantino,Nicoletta Giordano,Valerio Nardone,Alfonso Reginelli,Salvatore Cappabianca
摘要
Background In non-small cell lung cancer (NSCLC), image-guided biopsy must provide tissue for both diagnosis and biomarker-driven treatment selection. We systematically evaluated molecular adequacy of image-guided biopsy techniques for NSCLC biomarker profiling. Methods PubMed/MEDLINE, Embase, Scopus, Web of Science, CENTRAL, and CDSR were searched from inception to 30 May 2026, supplemented by reference screening and forward citation tracking. Eligible studies reported extractable numerator-denominator data for molecular/genotyping adequacy, NGS/genotyping adequacy, PD-L1 adequacy, diagnostic yield, or procedure-related complications. Random-effects meta -analyses of logit-transformed proportions used restricted maximum-likelihood estimation, with I 2 , τ 2 , and 95% prediction intervals. Results Twenty-six studies were included; 25 contributed to the main quantitative evidence base. The primary analysis included 23 study arms and yielded pooled molecular/genotyping adequacy of 89.5% (95% CI, 84.3–93.2%; I 2 = 91.2%; τ 2 = 1.070), with a 95% prediction interval of 48.4–98.7%. Pooled NGS/genotyping adequacy was 88.4% (95% CI, 80.9–93.1%; I 2 = 91.6%), PD-L1 adequacy 93.9% (95% CI, 91.5–95.6%; I 2 = 62.9%), and diagnostic yield 91.7% (95% CI, 86.3–95.1%). Sensitivity analyses did not materially alter the primary estimate. Conclusions Image-guided biopsy provides high average molecular and biomarker adequacy in NSCLC, but performance varies substantially across clinical, procedural, and laboratory settings. Pooled estimates should be interpreted within study-specific testing denominators and workflows rather than as unconditional probabilities of successful biomarker profiling. Diagnostic success and molecular readiness should be reported as distinct outcomes.