Streamlining liver iron assessment: accuracy and limitations of 3D qDixon MRI for liver iron overload quantification.

OBJECTIVE

To prospectively evaluate the accuracy and limitations of the 1.5 T 3D-Dixon sequence (qDixon) in quantifying liver iron concentration (LIC), using FerriScan technology-determined LIC and the 1.5 T multi-echo GRE sequence (ME-GRE) as reference standards.

METHODS

A total of 161 chronic transfusion-dependent patients, 83 men and 43 women with a mean age of 21.04 years, were scanned using ME-GRE and qDixon sequences; among them, 67 were scanned using FerriScan technology. R2* values (ME-GRE-R2* and qDixon-R2*) and LIC were obtained for both sequences. Comparative analysis was conducted using concordance correlation coefficients (CCC), Bland-Altman plots, and linear regression. A piecewise regression model was constructed to determine the upper limit for quantifying LIC using the qDixon.

RESULTS

The qDixon-LIC and ME-GRE-LIC showed mean differences of 0.22, 0.01, -0.27, and 6.8 mg/g dry weight (dw) in normal, mild, moderate, and severe liver iron overload with corresponding CCC values of 0.986, 0.982, 0.956, and 0.216, respectively. The piecewise regression model established a 28.47 mg/g dw threshold for accurate liver iron quantification using qDixon, beyond which qDixon-R2* reached 892.86 s. Below this threshold, qDixon demonstrated a strong linear correlation with FerriScan in estimating LIC (r = 0.82, p < 0.001), with a negligible mean difference of 1.68 mg/g dw and a CCC of 0.984.

CONCLUSION

The 1.5 T qDixon sequence can be used to quantitatively assess LIC within a specific range (LIC < 28.47 mg/g dw), simplifying clinical workflow procedures.