Comparison of LIC Obtained from Biopsy, BLS and R2-MRI in Iron Overloaded Patients with β-Thalassemia, Treated with Deferasirox (Exjade®, ICL670)

Chelation therapy for patients with transfusional hemosiderosis is guided by the degree of iron overload and ongoing transfusion requirements; liver iron concentration (LIC) is one tool for measuring current iron burden. During the clinical development of the novel oral iron chelator deferasirox, LIC was measured in subcutaneous liver biopsy samples, but also non-invasively by biomagnetic liver susceptometry (SQUID-BLS) and magnetic resonance imaging (R2-MRI: FerriScan®). These methods were compared in a substudy of a large, open-label, randomized Phase III study in 586 β-thalassemia patients with transfusional hemosiderosis; the study evaluated the efficacy and safety of long-term deferasirox and deferoxamine treatment. 48 patients (aged 17–35 yrs) gave informed consent for adding BLS and MRI assessments (baseline, 6 and 12 months) to the biopsy assessments (baseline and 12 months) of the main study. At baseline, 47 patients were assessed by BLS and biopsy, 45 by R2-MRI; at 6 months, 42 by BLS and R2-MRI; at 12 months, 41 by BLS and R2-MRI, 39 by biopsy. BLS was performed and evaluated locally at 3 centers in Hamburg, Turin and Oakland. MR images were recorded at 4 centers in Hamburg, Turin, Stanford and Los Angeles and were analyzed centrally. Data from BLS and R2-MRI remained blinded until study completion. LIC from biopsies was determined by AAS from extracted paraffin blocks in a central laboratory and used as the standardized reference method. LIC was expressed as mg/g dry weight for all methods: for BLS, a wet-to-dry weight conversion factor of 3.3 was applied, while MRI was calibrated against fresh freeze-dried liver biopsies. The relationship between BLS or R2-MRI with biopsy measurements was modeled by linear regression after log-transformation of the data. To account for the correlation between baseline and post-baseline measurements of the same patient, mixed-effects models were fitted. On average, the LIC data obtained from BLS and biopsy were related by a factor of 0.46 (95% CI: 0.32, 0.67). With R2-MRI assessment of LIC vs biopsy, the factor was 0.72 (95% CI: 0.54, 0.97). BLS and R2-MRI measurements could not be related appropriately by a factor. Overall, LIC from biopsy was generally larger than that obtained from BLS; R2-MRI measurements fell in-between biopsy and BLS. Differences are caused in part by the variability in the wet-to-dry weight ratio relating in vivo with in vitro methods. In conclusion, LIC data from 3 independent methods were correlated but not equivalent. It is therefore important that for an individual patient, a single method is consistently used under standardized conditions.