Chemical-shift and diffusion-weighted magnetic resonance imaging of thymus in myasthenia gravis: usefulness of quantitative assessment.

OBJECTIVES

The objective of this study was to prospectively investigate the usefulness of chemical-shift and diffusion-weighted (DW) magnetic resonance imaging (MRI) in patients with myasthenia gravis (MG) for distinguishing thymic lymphoid hyperplasia (TLH), normal thymus (NT), and thymoma (THY) by using the signal intensity index (SII) and the apparent diffusion coefficient (ADC).

MATERIALS AND METHODS

We examined 87 subjects (44 males, 43 females; range, 15-71 years) with generalized MG and antibodies to the acetylcholine receptor seropositivity who underwent surgery. They were divided into a TLH/NT group (A, 64 patients; TLH, 49; NT, 15) and a THY group (B, 24 patients; nonadvanced THY, 15; advanced THY, 9) on the basis of histological findings. One patient with contemporary findings of TLH and nonadvanced THY at histology was listed in both groups (87 subjects, 88 findings). Chemical-shift MRI (CS-MRI) was performed with dual-echo acquisition, and the SII was measured for each subject. Diffusion-weighted MRI was performed at b values of 0, 150, 500, and 800 s/mm, and the ADC value was obtained on the ADC map after excluding the 0-s/mm b value diffusion weighting. All measures were performed independently by 2 radiologists, and interreader agreement was assessed by calculating the intraclass correlation coefficient. Differences on SII and ADC levels between the groups and subgroups were tested using the Student t test. Logistic regression models were estimated, and discrimination abilities were individuated according to the area under the receiver operating characteristic curve (AUROC). The optimal cut points for the differentiation of the groups and subgroups were obtained by using the Youden index.

RESULTS

The interreader agreement was excellent (intraclass correlation coefficient: SII, 0.998; ADC, 0.944). For CS-MRI, the mean (SD) SII value was significantly different between the groups (A, 36.37% [12.60%]; B, -0.06% [3.85%]; P < 0.001). No overlap in indexes was found with sensitivity, specificity, and cut point of 100%, 100%, and 6.37%, respectively. Conversely, the mean SII value was not different between the subgroups of each group (A, P = 0.607; B, P = 0.252). For DW-MRI, the mean (SD) ADC values were significantly different between the groups (A, 1.92 [0.21] × 10·mm/s; B, 1.36 [0.33] × 10 mm/s; P < 0.001) and between the subgroups of group A (TLH, 1.86 [0.17] × 10 mm/s; NT, 2.10 [0.23] × 10 mm/s; P = 0.002), although overlapped values were found. The AUROC of ADC in discriminating TLH/NT from THY was 0.931 (95% confidence interval, 0.863-0.998), and the optimal cut point for this distinction was 1.625 × 10 mm/s (Youden index, J = 0.760) with sensitivity of 96.8% and specificity of 79.2%. For the subgroups of group A, the AUROC of ADC in discriminating NT from TLH was 0.794 (95% confidence interval, 0.666-0.923), and the optimal cut point for this distinction was 2.01 × 10 mm/s (Youden index, J = 0.458) with sensitivity of 66.7% and specificity of 79.2%.

CONCLUSIONS

CS-MRI and DW-MRI are both useful tools for examining patients with MG. The SII is more accurate than the ADC to differentiate TLH and NT from THY (AUROC, 1.000 and 0.931, respectively). Furthermore, the ADC is a noninvasive parameter that could be used for distinguishing TLH from NT, which is useful in selecting patients for surgery because, for nonthymomatous MG, acceptable rates of complete stable remission after thymectomy are found in TLH but not in NT.