Abstract

Iron overload is an inevitable consequence of multiple transfusions and occurs in many hematological diseases including sickle cell anemia (SCA) and beta thalassemia (β-thal). Liver biopsy provides quantification of iron content in the liver, but is not without risks such as bleeding, pain, and infection. MRI R2* has the advantage of quantifying liver iron without the risks of invasive procedures, however, this technique has not been fully investigated or validated. Furthermore, the variability introduced by multiple MRI readers has not been investigated to date. Patients with hematological diseases were selected to participate in this prospective IRB-approved study if they received ≥ 18 transfusions or had a serum ferritin ≥ 1000 ng/mL. All study participants completed 1.5 Tesla MRI R2* testing (Siemens Symphony), serum ferritin, and liver biopsy with quantification of liver iron content (LIC) within 30 days. Regions of interest (ROI) were drawn on R2* maps in a homogeneous area of the right hepatic lobe, avoiding blood vessels and obvious bile ducts. Three independent reviewers, blinded to the patients’ clinical status and the other 2 reviewers’ results, performed the ROI analysis. The correlation between LIC and liver R2* was calculated using the Spearman’s Rank-Order Correlation Coefficient. Due to possible outliers in the data, robust simple linear regression methods were used to fit a regression line to scatter plots. All liver biopsy samples were sent to Mayo Laboratories for LIC quantification. The agreement among the 3 raters was assessed using the interclass correlation coefficient (ICC). Forty-seven patients, median age 14 years (range 7 – 37) participated; 24 were female. Thirty-five of them had SCA, 8 had β-thal (major or intermedia), and 4 had bone marrow failure syndromes. Total table time for R2* MRI testing was between 20 to 30 minutes. All patients tolerated the liver biopsy without complications. The mean (±1SD) ferritin was 2917ng/mL (±2239), mean LIC was 12.139 mg/ 100g of dry weight liver (±8.269), and mean liver R2* ranged from 425 to 432 Hz (±257 to 249 Hz). All 3 raters produced R2* values strongly associated with LIC, with correlation coefficients from 0.93 to 0.95 (p<0.00005). There was a significant positive association between serum ferritin and R2* liver values (correlation among the 3 reviewers ranged from 0.39 to 0.50 with all p<0.009). The agreement among the 3 raters was 0.98. To summarize: 1) 1.5 T MRI R2* liver values are highly associated with LIC in patients with iron overload. This is the largest sample of MRI R2* liver measurements correlated with LIC obtained by liver biopsy; 2) The three raters had excellent agreement which suggests that, in our study, R2* liver values do not differ greatly among qualified readers; 3) R2* measurements of the liver were significantly associated with serum ferritin, however, the correlation coefficients were relatively low (0.39 to 0.50), documenting a weak relationship; and 4) MRI R2* of the liver is a feasible and valid technique to assess LIC non-invasively, and appears to be reproducible when performed by qualified reviewers. We conclude that liver MRI R2* can be incorporated into clinical research protocols for safe, painless, and accurate liver iron quantitation.

Author notes

Disclosure:Research Funding: NIH.