| Abstract: |
Introduction/Background Among non-small-cell lung cancers with appreciable functional activity, positron emission tomography/computed tomography (PET/CT) is the most accurate imaging modality for clinical staging. However, lymph nodes (LN) with marginally elevated standardized uptake value (SUV) present a diagnostic challenge. In this retrospective study, we hypothesized that normalizing the LN SUV by using the ratio of the LN to primary tumor SUVmax (SUVN/T) may be a better predictor of nodal malignancy than using SUVmax alone for nodes with low to intermediate SUV. Patients and Methods We identified 172 patients with newly diagnosed non-small-cell lung cancer who underwent pathologic LN staging and PET/CT within 31 days before biopsy. Receiver operating characteristic curves with area under the curve (AUC) calculations were used to evaluate SUVmax and SUVN/T for their ability to predict nodal malignancy for both the entire cohort of 504 LNs and a subset of 132 LNs from 85 patients who had both primary tumor SUVmax > 2.5 and LN SUVmax 2.0 to 6.0. Results In patients with primary tumor SUVmax > 2.5 and LN SUVmax 2.0 to 6.0, SUVN/T was significantly more accurate in predicting nodal malignancy (AUC, 0.846; 95% confidence interval, 0.775-0.917) than SUVmax (AUC, 0.653; 95% confidence interval, 0.548-0.759). The optimal cutoff value of SUVN/T to predict nodal malignancy was 0.28 (90% sensitivity, 68% specificity). Sensitivity was > 95% for SUVN/T < 0.21, whereas specificity was > 95% for SUVN/T > 0.50. Conclusion The ratio of LN SUV to primary tumor SUV on PET/CT is more accurate than SUVmax when assessing nodes of low to intermediate SUV. © 2015 Elsevier Inc. All rights reserved. |
