PET-CT predictors of mediastinal dissemination in lung cancer

A multicentric study of lung cancer staged by PET-TC and EBUS-NA was performed to identify PET-TC measures with high positive predictive value (PPV) for the identification of stage III disease and to evaluate EBUS-NA sensitivity. A surgical gold standard was used when EBUS-NA was negative. Results: 76/105 patients showed ≥1 hypermetabolic spot in mediastinum (72%). PET-TC SUVmax did not show statistically significant differences between centers. Lung SUVmax had mean of 11.4 (SD 5.3), and hypermetabolic spot in mediastinum median of 3.4 (IQR 0-6.5). EBUS-NA confirmed mediastinal metastasis in 43/54 patients (79.6%). A single hypermetabolic image in mediastinum was not a significant predictor of dissemination (OR 1.99 95%CI 0.75-5.32), but ≥2 hypermetabolic spots showed an increased risk (OR 3.64 95%CI 1.29-10.26). Lung SUVmax was not associated with mediastinal malignancy (OR 0.93 95%CI 0.86-1.01), but mediastinum SUVmax >3.4 was a significant predictor (OR 6,36 95%CI 2,71-14,91), that persists after adjustment for multilevel disease (OR 5.77 95%CI 2.37-14.06) (PPV 74.6%). Relationship between mediastinum SUV max and dissemination was stronger when SUVmax mediastinum/lung > 0.7 (OR 8.10 95%CI 2,54-25,67) and SUVmax mediastinum-lung > -11.7 cut-offs (OR 8.71 95%CI 2.98-25.44) were used (PPV 84.6% and 64.5%, respectively). Sensitivities of EBUS-NA in patients with SUVmax mediastinum/lung ratio over (86.4%) and below (75%) 0.7 did not show a statistical difference. Conclusions: SUVmax in the mediastinum is the main PET-CT predictor of mediastinal malignancy, with higher PPV when a 0.7 SUVmax mediastinum/lung cut-off is used. EBUS-NA sensitivity for the identification of dissemination is independent of PET-CT results.