Serum Ferritin Predicts Liver but Not Cardiac Iron Burden by Noninvasive MRI in Sickle Cell Disease (SCD.

BACKGROUND: Assumptions about iron loading as well as the utility of ferritin to predict transfusional iron overload among individuals with sickle cell disease (SCD) are largely based on extrapolation from data generated in patients with thalassemia major (TM). Yet recent studies suggest the natural history of iron overload in patients with SCD differs significantly from chronically transfused patients with TM. We sought to evaluate the extent of myocardial and hepatic siderosis using noninvasive imaging in chronically transfused patients with SCD and examine its clinical associations, including relationship to long-term trends in serum ferritin, transfusion history, chelation status and markers of hemolysis and inflammation.

METHODS: We evaluated 17 subjects (mean age 15±3.6 yrs, range 9 to 20). The mean transfusion duration was 7.3±3.6 yrs (range 2 to 15). Thirteen (76%) patients were on chelation with deferasirox at the time of screening; 4 were not on chelation Rx. MRI T2*/R2* of the heart and liver using a multiple gradient echo sequence was performed on a single 1.5T GE scanner. Hepatic iron concentration (HIC) values were predicted from liver R2* values. RESULTS: Mean HIC in subjects was 9.9±6.7 mg/gm liver dry weight (range 2.5 to 20.8) and was ≥15 mg/gm in 6/17 (35%) subjects. The mean long-term serum ferritin (past 5 yrs, or duration of transfusion if < 5yrs) was 2318±1122 ng/mL (range 541 to 4225). Using Pearson’s correlation coefficient, we observed a significant relationship between HIC and ferritin (r=0.765, p=<0.001). We generated a receiver operator characteristic (ROC) curve to assess the utility of ferritin as a predictor of elevated HIC, using a threshold HIC thought to predict serious iron-related complications. A ferritin cut-off value ≥2164 ng/mL correctly identified 80% of cases of HIC ≥15 mg/gm (AUC 0.96, p=0.003) in our subjects with 83% sensitivity and 73% specificity. Despite markedly elevated HIC and ferritin values in some subjects, none had myocardial siderosis. All 17 subjects had cardiac MRI T2* values in the normal range > 25 ms. Cardiac iron load measured by T2* did not correlate with HIC or serum ferritin. We examined C-reactive protein (CRP) and B-type natriuretic peptide (BNP) as markers for inflammation and myocardial strain, respectively, in our subjects but neither demonstrated a significant relationship to ferritin or MRI findings. BNP, however, did correlate modestly with both age (r=−0.574, p=0.013) and left ventricular ejection fraction on cardiac MRI (r=0.510, p=0.036). A subset of subjects (n=8) had histologic iron measurements by percutaneous liver biopsy (LBx) within 6 months of MRI. While liver iron content by LBx correlated significantly with HIC by MRI (r=0.759, p=0.03), liver iron content by LBx did not correlate with ferritin (r=0.312, p=0.452). CONCLUSION: We found that serum ferritin is a good predictor of liver iron by MRI R2*, and that long term ferritin values ≥2164 ng/mL predict significant hepatic iron overload as assessed by this noninvasive method. We did not observe appreciable cardiac iron loading in our subjects with SCD, which otherwise might have been predicted by elevated HIC alone, as in individuals with TM. These data suggest that reliable, long term surveillance of transfusion-induced iron overload in SCD may be achieved using serum ferritin and HIC by MRI R2* as surrogate markers of hepatic siderosis rather than relying on liver iron content measured invasively by LBx. Also, previously determined thresholds for significant cardiac iron loading in TM, based on degree of hepatic siderosis, may not be applicable in SCD. Further investigation into alternative mechanisms of iron loading or distribution in these related but distinct disorders is warranted.