Diffusion-weighted MRI in differentiating malignant from benign thyroid nodules: a meta-analysis

Objectives: To perform a meta-analysis to evaluate the diagnostic efficacy of diffusion-weighted imaging (DWI) in differentiating malignant from benign thyroid nodules. Design: A meta-analysis. Data sources and study selection: Medical and scientific literature databases were searched for original articles published up to August 2015. Studies were selected if they (1) included diagnostic DWI for differentiating malignant from benign thyroid lesions, (2) included patients who later underwent biopsy and (3) presented sufficient data to enable the construction of contingency tables. Data synthesis: For each study, the true-positive, false-positive, true-negative and false-negative values were extracted or derived, and 2x2 contingency tables were constructed. Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) instrument. The heterogeneity test, threshold effect test, subgroup analyses and publication bias analyses were performed. Results: From the 113 identified search results, 15 studies, representing a total of 765 lesions, were included in the meta-analysis. We detected heterogeneity between studies but found no evidence of publication bias. The methodological quality was moderate. The pooled weighted sensitivity was 0.90 (95% CI 0.85 to 0.93); the specificity was 0.95 (95% CI 0.88 to 0.98); the positive likelihood ratio was 16.49 (95% CI 7.37 to 36.86); the negative likelihood ratio was 0.11 (95% CI 0.08 to 0.16); and the diagnostic OR was 150.73 (95% CI 64.96 to 349.75). The area under the receiver operator characteristic curve was 0.95 (95% CI 0.93 to 0.97). Conclusions: Quantitative DWI may be a non-invasive, non-radiative and accurate method of distinguishing malignant from benign thyroid nodules. Nevertheless, large-scale trials are necessary to assess its clinical value and to establish standards regarding b values and cut-off values for DWI-based diagnosis.