PATIENTS AND METHODS

This was a retrospective study of TM patients treated at the Department of Paediatrics, Prince of Wales Hospital. One hundred TM patients (48 males and 52 females) were included. Forty four had undergone allogeneic bone marrow transplantation (BMT) and 56 patients were receiving regular blood transfusions. Median age at diagnosis was 0.7 years, with 75% less than 1 year. Blood transfusions were administered at 3–4 weekly intervals, to maintain a pretransfusion haemoglobin above 95 g/l. Desferrioxamine was given by subcutaneous infusion over 8–10 hours (20–50 mg/kg), generally 5–6 times weekly. Median age of starting chelation was 3.7 years (range 1.1–18 years). The patients had received chelation therapy for a median of 7.7 years (range 1 month to 22 years). They had received regular transfusions for a median of 10.6 years (range 1 month to 25 years). At time of evaluation, the median age was 11.7 years (range 1.5–27 years) (table 1). Two patients were positive for hepatitis B surface antigen. Eight were positive for hepatitis C antibody and for hepatitis C RNA.

Liver biopsy

Liver biopsy had been performed in 37 patients prior to BMT between 1992 and 1999. A total of 36 transfusion dependent (TD) patients underwent liver biopsy to assess their liver iron content (LIC). This LIC study was approved by the institution's ethical committee. Informed consent for liver biopsy was obtained from each patient. A total of 81 biopsies were performed on 73 patients. The biopsy specimens were obtained from the right lobe of the liver under ultrasound guidance using a semiautomated 16 gauge Tru-cut style needle (Temno, Bauer Medical International, Dominican Republic). A core of tissue was sent for LIC analysis by graphite furnace atomic absorption.

Histological study

A core of liver tissue was evaluated histologically and graded semiquantitively for iron storage by a single pathologist. Liver sections were stained with standard haematoxylin and eosin. Iron was shown using the Perls' stain procedure. Grading was based on the method of Scheuler modified by Rowe,⁸ which classified parenchymal and mesenchymal haemosiderin deposition as follows: grade 0, no haemosiderin; grade 1, minimal; grade 2, mild; grade 3, moderate; grade 4, heavy. The presence or absence of fibrosis was recorded. The liver was assessed for evidence of hepatitis.

Biochemical markers

Serum ferritin was measured at six monthly intervals, and the mean serum ferritin in the year before evaluation was recorded. Serum alanine aminotransferase (ALT) levels were routinely measured prior to monthly transfusion, and levels in the two year period before this evaluation were recorded.

Statistics

Continuous variables were expressed as mean and standard deviation. Comparison of factors potentially associated with haemosiderosis, fibrosis, and elevated serum transaminase was performed using Student's t test or analysis of variance as appropriate. These variables included sex, duration of transfusion and chelation, age, serum ferritin, LIC, and hepatitis status. The possible associations between LIC and serum ferritin, and haemosiderosis deposition were explored using Pearson's correlation coefficient or Fisher's exact test. Multivariate analysis by stepwise regression for factors predicting haemosiderosis, hepatic fibrosis, and liver transaminase elevation were performed. The level of significance was set at 0.05 in the analyses and all the statistical testing was two sided.

RESULTS

Histology

All the liver specimens showed at least grade 1 haemosiderosis; grade 3 and 4 occurred in 26% and 18% of the samples respectively (table 2). Serum ferritin, ALT, and LIC were significantly different among the four grades of haemosiderosis (p < 0.001, p = 0.001, and p < 0.001 respectively). The differences remained significant when grade I and II combined were compared grade III and IV. There was no significant difference in age at evaluation, or in the duration of transfusion and chelation. Multivariate analysis revealed serum ferritin as the most significant predictor of moderate to severe haemosiderosis (p = 0.003). However, when multivariate analysis was applied to the 42 in whom LIC was available, LIC appeared to be an even better predictor of moderate to severe haemosiderosis (p < 0.001).

Features of hepatitis were reported in seven of the eight patients positive for hepatitis C, but none of these had features of cirrhosis. The two patients positive for hepatitis B surface antigen did not have evidence of hepatitis. Hepatic fibrosis was present in 22 cases and these patients had significantly higher concentrations of serum ferritin, serum ALT, and LIC (table 2). Hepatic fibrosis was more common in those with hepatitis C (p = 0.008). Multivariate analysis revealed serum ferritin as the most significant predictor of hepatic fibrosis.

Liver iron content

Forty three samples were analysed for LIC. The median LIC was 7.5 mg/g dry weight (range 0.8–41.8). Seven samples (16.3%) had an LIC over 15 mg/g, a concentration associated with a high risk for cardiac disease.⁹ The correlation coefficient for LIC and serum ferritin was 0.687 (p < 0.001) (fig 1). Considering those with a serum ferritin greater than 5500 pmol/l (2500 μg/l) (a concentration associated with increased mortality¹⁰) the correlation was only 0.46 (p = 0.11). There were significant differences in LIC in association with the various grades of haemosiderosis (p < 0.001) (fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Relation between serum ferritin and liver iron content.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Relation between grading of haemosidrosis and liver iron content.

Liver enzyme evaluation

There was no correlation between serum ALT concentrations and age, duration of transfusion and chelation, or LIC. Serum ferritin and anti-HCV positivity were the only significant factors associated with increased ALT concentrations (both p < 0.001). Eighteen patients (18%) had persistent elevation of ALT and 10 (10%) had transient elevation. The transiently elevated group had significantly higher serum ferritin than the non-elevated group (p < 0.001).

DISCUSSION

Despite the use of prenatal screening in Hong Kong for more than 10 years,¹¹ there are still two to five new cases of TM annually. These represent either failures of screening or immigrant children from mainland China. Though the number of reported deaths from liver disease in TM is small, severe haemosiderosis and hepatic fibrosis may lead to cirrhosis. Treatment is now available for hepatitis C infection.¹² Thus, early and accurate diagnosis of liver disease followed by prompt intervention may prevent liver disease progression. In our study, we found a prevalence of 44% grade 3–4 haemosiderosis; this is similar to that reported approximately 10 years ago.^5,6 This could reflect the rather unsatisfactory compliance rate with desferrioxamine therapy observed in many of our patients.

Desferrioxamine induced toxicity is relatively common in patients with less severe iron overload.^13,14 Serum ferritin is an acute phase reactant and may be increased as a result of inflammatory disorders such as hepatitis and ascorbate status. Measurement of LIC is an alternative approach to the assessment of body iron content.¹⁵ This has been studied in both TM and hereditary haemochromatosis patients and is found to be a more accurate indicator of overload.^9,16 The safety of liver biopsy has been shown in TM patients.¹⁷ The accurate assessment of body iron stores can help guide chelation therapy, and thus reduce desferrioxamine toxicity. Unfortunately, liver biopsy is an invasive procedure, and patients may not accept frequent biopsy as suggested by some authors.¹⁸ Identifying patients who are at high risk of developing fatal complications from iron overload is an important objective. Continuous central venous administration of desferrioxamine may postpone cardiac death.¹⁹

Although there was a correlation between serum ferritin and LIC in this study, this was less reliable at ferritin concentrations above 5500 pmol/l. Patients with “low” serum ferritin can be monitored without biopsy but LIC should be measured in those with a high serum ferritin concentration. Low serum ferritin should alert the physician to the need for desferrioxamin for dose reduction in order to avoid toxicity.²⁰

Liver histological examination provides information about haemosiderosis and hepatic fibrosis. While LIC measurement may not be available in some centres, the grading of haemosiderosis provides a useful measure of iron overload as there was good correlation with LIC. Grade 4 haemosiderosis was always associated with an LIC of more than 10 mg/g dry weight, and such patients require intensive chelation therapy. The presence of hepatic fibrosis points to a risk of progression to cirrhosis.²¹ The pathogenesis of hepatic fibrosis in iron overload is still not well understood, although severe overload is a known risk factor for developing fibrosis. With effective iron chelation, hepatic fibrosis may be reversible. Hepatic fibrosis is also a known prognostic factor for patient survival following BMT, as reported previously.²²

In this study, we did not observe any patients with liver cirrhosis. The 30% prevalence of hepatic fibrosis was lower than that reported previously.^5,6 In our study, only eight patients were positive for hepatitis C, a much lower prevalence than that reported in Mediterranean TM.²³ This may be a result of the relatively young age of our patients and also the screening of blood products since the early 1990s. The hepatitis C positive patients have entered a trial of combination treatment with interferon and ribavirin.

In conclusion, liver disease is common in TM patients and severe haemochromatosis is still observed in 30–40% of patients despite the use of desferrioxamine chelation. Serum ferritin is useful to monitor treatment in patients who have not yet developed severe iron overload. However patients with high serum ferritin concentrations and those with hepatitis C should be evaluated for LIC, and for hepatic fibrosis and cirrhosis by liver biopsy. In centres where LIC assay is not available, histological grading of haemosiderosis is a useful alternative.