Limited incremental value of quantitative [18F]FDG PET features beyond CT tumor diameter for pathological extension in resected nonsmall cell lung cancer presenting as solitary pulmonary nodules

Objective The study aimed to assess whether quantitative fluorine-18 fluorodeoxyglucose ([ 18 F]FDG) PET features provide stable incremental predictive information beyond maximum axial computed tomography (CT) tumor diameter for pathological T category ≥2 in resected nonsmall cell lung cancer (NSCLC) presenting as solitary pulmonary nodules. Methods We retrospectively analyzed 180 consecutive resected solitary pulmonary nodules. The primary cohort comprised 118 NSCLCs with pathological T category, including 33 pathological T category ≥ 2 cases. PET features from 41% maximum standardized uptake value and fixed standardized uptake value 2.5 segmentations were evaluated alongside CT diameter. CT interobserver reproducibility and pathological determinants of pathological T category ≥ 2 were assessed. Performance was evaluated using bootstrap confidence intervals (CIs), repeated five-fold cross-validation, calibration, Brier score, and nested ridge regression. Results CT diameter was associated with pathological T category ≥ 2 (odds ratio per 5 mm = 1.95; 95% CI = 1.31–2.93; area under the receiver operating characteristic curve (AUC) = 0.684) and showed excellent reproducibility (intraclass correlation coefficient = 0.966; 95% CI = 0.953–0.976). Twenty-five of 33 events (75.8%) were nonsize-driven, including 22 with visceral pleural invasion. CT diameter and metabolic tumor volume (MTV) at the 41% threshold were strongly but not redundantly correlated ( ρ = 0.783). CT+MTV41% yielded apparent AUC 0.687 (ΔAUC 0.003; 95% CI = −0.004 to 0.010) and cross-validated AUC 0.656 versus 0.672 for CT alone. CT + mean standardized uptake value 2.5 achieved cross-validated AUC 0.677. Penalized models likewise showed no stable improvement after internal validation. Conclusion Quantitative PET features were associated with pathological extension, but no stable incremental predictive value beyond CT tumor diameter was demonstrated in this cohort. Modest additional value cannot be excluded, and external validation is warranted.