Ratio of lymph node to primary tumor SUVmax multiplied by maximal tumor diameter on positron emission tomography/integrated computed tomography may be a predictor of mediastinal lymph node malignancy in lung cancer

Positron emission tomography/integrated computed tomography (PET/CT) provides the most accurate imaging modality for preoperative lung cancer staging. However, the diagnostic accuracy of maximum standardized uptake value (SUVmax) for mediastinal (N2) lymph nodes (LN) is unclear. We compared SUVmax, the ratio of LN to primary tumor SUVmax (SUVn/t), and SUVn/t multiplied by maximal tumor diameter (SUVindex) in terms of their abilities to predict mediastinal LN malignancy.We retrospectively analyzed 170 mediastinal LN stations from 73 consecutive patients who underwent systemic LN resection and PET/CT within 27 days. The SUVmax of the primary tumors was >2.0 and the SUVmax of the mediastinal LN stations ranged from 2.0 to 7.0 on PET/CT. Receiver-operating characteristic curves (ROCs) of SUVmax, SUVn/t, and SUVindex were calculated separately and the areas under the curves (AUCs) were used to assess the abilities of the parameters to predict LN malignancy. The optimal cutoff values were calculated from each ROC curve and the diagnostic abilities were also compared. The diagnostic accuracies of the 3 methods were also assessed separately in smoking and nonsmoking patients.Twenty-eight LN stations were malignancy-positive and the remaining 142 were malignancy-negative. The AUCs for SUVindex, SUVn/t, and SUVmax were 0.709, 0.590, and 0.673, respectively, and the optimal cutoff values for SUVindex, SUVn/t, and SUVmax were 1.11, 0.34, and 3.6, respectively. The differences between SUVindex and SUVn/t were significant, but there was no significant difference between SUVindex and SUVmax. There were no significant differences between smokers and nonsmokers in the AUCs for any of the methods for predicting LN malignancy (P values >0.05).SUVindex may be a predictor of mediastinal LN malignancy in lung cancer patients.