Liver Histology, Liver Iron Concentration (LIC), and Serum Ferritin in a Large Cohort of Chronically Transfused Children with Sickle Cell Anemia: Limitations of LIC As a Marker for Hepatic Injury and Ferritin as An Indicator for Chelation Initiation

Abstract ²¹⁴²

Chronic blood transfusion therapy reduces clinical events and prevents recurrent brain injury in children with sickle cell anemia. The benefit of administering chronic transfusion is weighed against the risk of an increased iron burden leading to chronic organ injury. The gold standard technique for evaluating the adverse effect of iron is to perform a liver biopsy for quantification of hepatic iron content and evaluation of liver pathology. Surrogate evaluations for iron overload include monitoring liver enzymes, serum ferritin and performing r2* MRI of the liver. In order to evaluate the role for utilizing surrogate markers to monitor liver injury, we conducted a retrospective review of 262 liver biopsies in 109 children with sickle cell anemia on chronic transfusion therapy over a nine year period at a single center. Ninety one patients had HbSS, 17 had HbSB0 thalassemia, and one patient had HbSD. Chronic transfusion therapy was performed by either simple transfusion (65%) or erythrocytapheresis (35%) primarily for stroke prevention (n=236), with a few for other indications (n=26) including lung injury and acute vascular necrosis. Patients were initiated on chronic transfusion at a mean age of 6.2 ± 3.6 yrs (0.75–17yrs) with initial biopsy obtained at a mean age of 14.6 ± 5.3 yrs (3–34 yrs). Chelation with deferoxamine or deferasirox was determined by the physician with a practice standard of initiation of chelation once ferritin increased to > 1000ng/mL. All patients at the time of liver biopsy were treated with chelation therapy with either deferoxamine (30%) or deferasirox (70%). Two pathologists reviewed the biopsies and utilized a standardized hepatic scoring system to evaluate the degree of portal/periportal and lobular inflammation and hepatic fibrosis (0= none, 1= mild, 2= moderate, 3= severe). Portal/periportal inflammation was scored 0–3 respectively in 132, 89, 38, and 0 patients and lobular inflammation in 29, 226, 4, and 0 patients. Fibrosis was scored 0–3 respectively in 23, 107, 103, and 26 patients. Ferritin and ALT were recorded prior (median of 3 days) to the liver biopsy. Seven biopsies performed as part of a therapeutic clinical trial were excluded from this analysis. Results show that the mean (± SD) serum ferritin, liver iron concentration (LIC), and ALT were 3509 ± 2617ng/mL, 17.12 ± 13.0 mg Fe/gm dry weight, and 40.2 ± 40.2 IU/L. With respect to histology, ferritin and LIC levels were significantly increased with higher periportal inflammation score (F= 21, p<0.001. F= 20, p<0.001) and severe fibrosis (score 3) (F=36, p<0.001, F=10.4, p<0.001), but not for lower fibrosis scores (0–2), or lobular inflammation score (p=0.20). Despite this significant histologic correlation with surrogate markers, individual overlap exists between ferritin, LIC and liver pathology. A strong linear correlation exists between ferritin and LIC (r=0.74, p<0.001) but with a spread in LIC (R²=55%). With respect to ferritin as a predictor of LIC, all patients with ferritin >1000ng/mL, a standard value for initiation of chelation therapy, had abnormally high LICs, and surprisingly 11 patients were identified with an abnormal LIC despite a ferritin <1000ng/ml. Furthermore, patients with a high LIC (≥ 7 mg Fe/gm dry weight) demonstrate a significantly higher ferritin as compared to patients with lower LIC< 7 (p<0.001) and this positive relationship between LIC and ferritin was replicated in a population with a higher LIC (LIC ≥ 30mg Fe/g dry weight vs. <30) (p<0.001). ROC curves demonstrate an AUC of.88 ± 0.02 (p<0.001) utilizing a LIC of ≥7 or <7 and 0.91± 0.02 (p<0.001) utilizing a LIC of ≥30 or <30. A weak association was noted between ferritin and alanine aminotransferase (ALT) (r=0.27, p<0.001, R2=8%). The results show that although strong statistical correlations exist between liver histology and ferritin or LIC, variability exists. Additionally, a ferritin >1000ng/mL always predicts abnormal LIC, but is inadequate as an indicator for initiation of chelation. The results suggest caution when using surrogate markers alone to predict histological changes in the liver and to initiate chelation therapy in individual patients on chronic blood transfusion therapy.