Biexponential R2* relaxometry for estimation of liver iron concentration in children: A better fit for high liver iron states

Background: R2* relaxometry's capacity to calculate liver iron concentration (LIC) is limited in patients with severe overload. Hemosiderin increases in these patients, which exhibits a non-monoexponential decay that renders a failed R2* analysis.

Purpose/hypothesis: To evaluate a biexponential R2* relaxometry model in children with different ranges of iron overload.

Study type: Retrospective.

Population: In all, 181 children with different conditions associated with iron overload.

Field strength/sequence: 1.5T, T₂ *-weighted gradient echo sequence.

Assessment: Bi- and monoexponential R2* relaxometry were measured in the liver using two regions of interest (ROIs) using a nonproprietary software: one encompassing the whole liver parenchyma (ROI-1) and the other only the periphery (ROI-2). These were drawn by a single trained observer. The residuals for each fitting model were estimated. A ratio between the residuals of the mono- and biexponential models was calculated to identify the best fitting model. Patients with 1) residual ratio ≥1.5 and 2) R2*_fast ≥R2*_slow were considered as having a predominant biexponential behavior.

Statistical tests: Nonparametric tests, Bland-Altman plots, linear correlation, intraclass correlation coefficient. Patients were divided according to their LIC into stable (n = 23), mild (n = 58), moderate (n = 61), and severe (n = 39).

Results: The biexponential model was more suitable for patients with severe iron overload when compared with the other three LIC categories (P < 0.001) for both ROIs. For ROI-1, 37 subjects met criteria for a predominant biexponential behavior. The slow component (5.7%) had a lower fraction than the fast component (94.2%). For ROI-2, 22 subjects met criteria for a predominant biexponential behavior. The slow component (4.7%) had a lower fraction than the fast component (95.2%). The intraobserver variability between both ROIs was excellent.

Data conclusion: The biexponential R2* relaxometry model is more suitable in children with severe iron overload.

Level of evidence: 3 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019;50:1191-1198.