Application of apparent diffusion coefficient of extraocular muscles from diffusion tensor imaging scanning in the assessment of disease activity of thyroid eye disease.

Purpose: To evaluate the utility of apparent diffusion coefficient (ADC) values of extraocular muscles (EOMs) in differentiating activity of thyroid eye disease (TED).

Method: Forty-two TED patients who underwent diffusion tensor imaging(DTI) were retrospectively enrolled in this study, including 29 patients in analysis group and 13 patients in validation group. The mean, maximum and minimum ADC value of each EOM were regarded as ADCmean, ADCmax and ADCmin. The difference between ADCmax and ADCmin was regarded as △ADC. The correlations between ADCmean or △ADC of each EOM and clinical activity score (CAS) were assessed.

Results: In analysis group, ADCmean differed between active and inactive eyes and positively correlated with CAS in IR (P < 0.05), not in SR,LR and MR(all p > 0.05). While △ADC differed between two groups and negatively correlated with CAS in all EOMs (all P < 0.05). ADCmean predicted active disease at cut-off value of 1259.3 × 10^-6mm²s^-1 with sensitivity of 66.7% and specificity of 71.4% in IR[area under curve = 0.667, P < 0.05]. △ADC predicted disease activity in all EOMs [area under curve 0.658-0.746,all P < 0.05]. The cut-off values of △ADC were 382, 823,520 and 572 × 10^-6mm²s^-1 with sensitivity of 80.0%, 50.0%, 43.3%, 83.3% and specificity of 67.9%, 85.7%, 89.3%, 60.7% in SR, IR, MR, and LR respectively. There were no significant differences in the predictive efficacy among all cut-off values.

Conclusions: Our results showed that DTI is a valuable tool in the assessment of disease activity of TED. Both ADCmean of IR and △ADC of all four EOMs can be used in discriminating disease activity with the same predictive power.