Differentiation of rebound and lymphoid thymic hyperplasia from anterior mediastinal tumors with dual-echo chemical-shift MR imaging in adulthood: reliability of the chemical-shift ratio and signal intensity index.

Abstract

To prospectively evaluate (a) effectiveness and limits of dual-echo chemical-shift magnetic resonance (MR) imaging for distinguishing hyperplastic thymus from anterior mediastinal tumors in adulthood by using chemical-shift ratio ( CSR chemical-shift ratio ) and signal intensity index ( SII signal intensity index ), with proposal of optimal threshold value for each, and (b) whether age affects these indexes. Study was institutional review board approved, with informed consent obtained. Ninety-two subjects (53 men, 39 women; age range, 18-84 years) were divided into a rebound and lymphoid hyperplasia group (group A, 30 patients) and a tumor group (group B, 62 patients). MR images were assessed; interrater reliability was evaluated. Differences in CSR chemical-shift ratio and SII signal intensity index were tested with the Mann-Whitney U test and the Kruskal-Wallis test. Discrimination abilities of CSR chemical-shift ratio and SII signal intensity index were evaluated with logistic regression models, and optimal cutoff points were proposed. Quantitative parameters were correlated with age by using Pearson correlation coefficients. Interreader agreement was excellent (intraclass correlation coefficient: CSR chemical-shift ratio , 0.893; SII signal intensity index , 0.898). Mean CSR chemical-shift ratio and SII signal intensity index ± standard deviation were 0.545 ± 0.162 and 46.29% ± 18.41 for group A and 1.045 ± 0.094 and -0.06% ± 4.89 for group B, respectively, with significant differences for both indexes between groups (P < .0001). No overlap was found for SII signal intensity index between groups; CSR chemical-shift ratio values overlapped in a few younger adults. Distinguishing hyperplastic thymus from tumors was better with SII signal intensity index than CSR chemical-shift ratio . Respective sensitivity, specificity, and cutoff points were 100%, 100%, and 8.92% for SII signal intensity index and 100%, 96.7%, and 0.849 for CSR chemical-shift ratio . Significant correlation was found for CSR chemical-shift ratio (r = -0.761) and SII signal intensity index (r = 0.821) with age in group A (P < .001). For group B, significant correlation with age was seen for CSR chemical-shift ratio (r = 0.702, P < .001) but not SII signal intensity index (r = -0.196, P = .127). All subjects but one in group A and none in group B had signal intensity decrease at chemical-shift MR imaging. With dual-echo chemical-shift MR imaging, SII signal intensity index and CSR chemical-shift ratio have high accuracy to distinguish thymic hyperplasia from tumors, although overlapped CSR chemical-shift ratio values can occur in early adulthood.