FDG PET as a Predictive Factor of Spread Through Air Spaces in Non-Small Cell Lung Cancer

1037 Introduction: Spread Through Air Spaces (STAS) in non-small cell lung cancer (NSCLC) is a newly defined form of cancer invasion and a prognostic factor independent of TNM stages. STAS could be confirmed by histology only postoperatively. We aimed to identify preoperative predictors of STAS using FDG PET. Materials and Methods: Three-hundred seventy patients who underwent a surgical procedure and had FDG PET data between October 2019 and August 2020 were retrospectively collected as a retrospective single-center study. The interval between surgery and FDG PET was within 4 months. FDG uptake was measured by the maximum standardized uptake value (SUVmax). The SUVmax of tumors were compared according to the presence of STAS. To investigate whether FDG uptake was associated with STAS in a specific subtype of NSCLC, the comparison of FDG uptake according to the presence of STAS was analyzed in a subgroup defined by histology and stage. A predictive value of SUVmax for the presence of STAS was evaluated by a logistic regression model. Results: SUVmax of NSCLC with STAS was significantly higher than those without STAS (STAS+ 8.33±6.34 vs. STAS- 5.40±5.80, t = 4.6, p = 0.001). In Adenocarcinoma (ADC) group, SUVmax showed a significant difference, according to the positivity of STAS (STAS+ 7.15±4.99 vs STAS- 4.06±4.72, t = 5.7, p = 0.001). Squamous cell carcinoma (SqCC) group also showed significant difference (STAS+ 16.75±8.48 vs STAS- 12.48±5.9, t = 2.1, p = 0.036). In stage I NSCLC, the ADC group showed a significant difference in SUVmax according to STAS, however, the SqCC group did not. (“stage I ADC” t = 4.2, p = 0.001, “stage I SCC” t = 0.7, p = 0.47). In stage I ADC, area under the curve (AUC) was 0.72 and the best cutoff value of SUVmax to discriminate STAS+ ADC was 2.08 (Sensitivity 86%, Specificity 50%). In the total NSCLC group, there are 4 independent predictors, “SUVmax”, “pN2”, “pT≥2b”, “subtype=ADC”. In both subgroups of stage I NSCLC and stage I ADC, SUVmax was the only independent predictor according to the logistic regression model (p = 0.0010, p = 0.0013, respectively). Conclusions: SUVmax is significantly associated with the presence of STAS in the NSCLC. Compared with SqCC, lung ADC showed a significant association between FDG uptake and STAS. The results suggested SUVmax might be used as a factor to preoperative predict STAS, particularly in early-stage lung ADC.