Congenital dyserythropoietic anaemia (CDA), type I, is an uncommon inherited disorder which is characterized by macrocytes in the blood film, certain dysplastic changes in the erythroblasts, gross ineffectiveness of erythropoiesis and an autosomal recessive inheritance ( Wickramasinghe, 1998). The plasma bilirubin level is elevated to different extents in different patients. The same is true for the serum ferritin level and the diagnosis of CDA type I is occasionally made in untransfused patients who present with the manifestations of iron overload. In the present investigation we have studied various biochemical parameters of iron overload in the sera of seven patients with CDA type I. In addition we have investigated factors that might account for differences in the severity of the iron overload, including co-inheritance of two mutations of the HFE gene, namely C282Y (the most commonly found mutation in hereditary haemochromatosis) and H63D. The possibility that the different degrees of hyperbilirubinaemia may be related to the presence of the A[TA]7TAA variant in the TATA box of the promoter of the uridine diphosphate glucuronosyltransferase (UGT-1A) gene was also studied.
Seven untransfused patients with congenital dyserythropoietic anaemia type I were investigated to assess the determinants of both iron overload and serum bilirubin levels. The serum ferritin concentration was increased in all patients and non-transferrin-bound iron (NTBI) was increased in all but one patient. None of the patients showed the C282Y mutation in the hereditary haemochromatosis gene, HFE. One patient was homozygous for the H63D mutation in this gene. The data indicated that differences in the extent of iron overload were not mediated by co-inheritance of the C282Y mutation in the HFE gene but could largely be explained by differences in the severity of anaemia and ineffective erythropoiesis, and in the age of the patient. In one patient an unusually high plasma bilirubin level was associated with the variant A[TA]7TAA configuration in the TATA box of the uridine diphosphate glucuronosyltransferase (UGT-1A) gene promoter, the mutation found in most patients with mild Gilbert's syndrome.
MATERIALS AND METHODS
The diagnosis of CDA type I was based on conventional criteria including the presence of internuclear chromatin strands joining a small proportion of incompletely separated erythroblasts and a spongy (‘Swiss cheese’) appearance of the erythroblast heterochromatin in a substantial proportion of cells. The main clinical and some laboratory findings in the seven patients studied are summarized in 1 Table I in which the cases are divided into two groups, A and B, according to the severity of the anaemia. Urine was tested for haemosiderin in six patients and only case 5 had haemosiderinuria (case 3 was not tested).
None of the patients had been transfused except for case 5 who had undergone an exchange-transfusion at birth. Cases 3 and 6 have been reported previously ( Wickramasinghe & Pippard, 1986). When case 6 was re-assessed at the age of 56 years the serum ferritin was 3695 μg/l, the transferrin saturation was 98%, the liver biopsy showed cirrhosis, and the liver iron was 465 μmol/g dry tissue. He was subsequently treated with infusions of desferrioxamine and regular phlebotomy for 2 years followed by regular phlebotomy for 7 years until 1994. At the age of 34 years case 3 had a serum ferritin of 3869 μg/l, a transferrin saturation of 87.5%, and liver iron of 537 μmol/g dry tissue. He was treated with infusions of desferrioxamine and venesection over 8 months in 1985. Case 2 was found to have a high serum ferritin and hepatic siderosis at the age of 17 years and she received subcutaneous infusions of desferrioxamine intermittently over the next year. Case 5 received subcutaneous infusions of desferrioxamine four times a week between the ages of 3 and 6 years but not since then. The other three cases had not been venesected or been treated with desferrioxamine; case 4 was previously reported by Lewis et al (1972 ).
Serum thymidine kinase (sTK) and serum transferrin receptor (sTfR) assays
sTK activity was measured using the Profilogen® TK-REA radioenzyme assay (Sangtec Medical, Bromma, Sweden) and sTfR levels using the QuantikineTM IVDTM Soluble Transferrin Receptor ELISA (R & D Systems Inc., Minneapolis, U.S.A.).
Polymerase chain reaction (PCR) amplification and analysis of the UGT-1A gene promoter
A DNA fragment (286–288 bp) encompassing the 5′ promoter region of UGT-1A gene was amplified by PCR using the following primers; F: 5′-CTG AAA GTG AAC TCC CTG CTA CC-3′ and R: 5′-CCC CAA GCA TGC TCA GCC AGT G-3′. Amplifications were carried out under standard conditions using 200 ng genomic DNA and 1.25 units of Taq polymerase (Cetus). PCR was carried out for 35 cycles at 95°C for 30 s, 52°C for 45 s and 72°C for 45 s with an initial extended denaturation time of 3 min and a final extension at 72°C for 3 min. The products were then directly sequenced using either of the amplification primers as sequencing primers, and an ABI 377 sequencer (Applied Biosystems).
Serum iron and unsaturated iron binding capacity were measured using a Guanidine/FerroZine® method with COBAS® INTEGRA Roche diagnostic reagents. Serum ferritin was assayed using a Microparticle Enzyme Immunoassay with the AxSYM Ferritin reagent kit. Serum non-transferrin-bound iron (NTBI) was quantitated by the HPLC-based method described by Porter et al (1996 ).
Analysis of the HFE gene
The HFE gene was analysed for the C282Y and H63D mutations by PCR amplification followed by restriction enzyme digestion ( Merryweather-Clarke et al, 1997 ). The C282Y mutation creates a new Rsa I site and the H63D mutation abolishes an Mbo I restriction site.
One of the seven patients (case 6) was homozygous for the A[TA]7TAA variant in the UGT-1A gene promoter ( 1 Table I).
Four of the patients had markedly increased percentage transferrin saturation values and all of the patients had increased serum ferritin levels ( 2 Table II). NTBI values were elevated in all except one of the patients (case 6) who had received prolonged treatment for iron overload. Serum LDH and sTfR levels were elevated in all patients and sTK activity was increased in all of the patients tested.
None of the patients had the C282Y mutation in the HFE gene and one of the patients was homozygous for the H63D mutation.
In CDA type I high NTBI levels were found in patients with ferritin levels both above and below 1000 μg/l ( 2 Table II). In view of the potential cytotoxicity of NTBI these data suggest that some form of chelation therapy may be desirable in this disorder. However, case 4, a very active 72-year-old who had never been treated with desferrioxamine or venesections and who currently has a ferritin level of 766 μg/l and an NTBI concentration of 0.85 μmol/l, illustrates the fact that some untreated mildly anaemic patients survive without clinical manifestations of iron overload until old age.
The majority of patients with hereditary haemochromatosis are homozygous for a missense mutation in codon 282 of the HFE gene which substitutes Tyr for Cys in this position (C282Y) ( Worwood, 1998). This mutation was not found in any of the cases of CDA type I studied. Evidently, the greater degree of iron overload in some of our cases of CDA type I did not result from heterozygosity for the C282Y mutation. The H63D mutation in the HFE gene was found in one of the seven patients (case 7) and this patient was homozygous for the mutation. Whether the H63D mutation contributed to the moderate degree of iron overload in case 6 is uncertain, especially as the role of this mutation in hereditary haemochromatosis also remains unclear ( Worwood, 1998). Nevertheless, it is of interest that a moderate degree of iron overload in a recently reported patient with a previously undescribed form of congenital dyserythropoiesis was also associated with homozygosity for the H63D mutation ( Wickramasinghe et al, 1998 ).
The extent of ineffective erythropoiesis was not directly measured in the present study but may be reasonably considered to be proportional to the severity of the anaemia. Examination of our data suggests that severity of ineffective erythropoiesis (as judged by anaemia) and age can explain most of the variation in serum ferritin levels. Thus, the three patients with Hb levels < 10 g/dl ( 2 Table II, group A) had the highest serum ferritin and NTBI values, and among these the youngest (case 1) had the lowest serum ferritin level. In addition, among the patients with Hb > 10 g/dl ( Table II, group B), the youngest (case 5) had the lowest serum ferritin level. However, comparisons within group B are difficult, as the iron status of two of the patients had been modified by treatment with desferrioxamine or desferrioxamine and phlebotomy for 3 and 9 years, respectively. Haemosiderinuria is known to limit iron overload ( Wickramasinghe, 1998), but was only found in one case.
Inheritance of the TA insertion in the TATA box of the UGT-1A gene results in reduced expression of the gene ( Bosma et al, 1995 ), and homozygosity for the [TA]7 variant is associated with a mild form of Gilbert's syndrome ( Monaghan et al, 1996 ). The [TA]7 allele is also involved in the increased bilirubin levels observed in individuals with β-thalassaemia ( Galanello et al, 1997 , 1999; Sampietro et al, 1997 ). One of the seven patients with CDA type I (case 6) was homozygous for the variant A[TA]7TAA genotype and this patient also had the second highest plasma bilirubin level even though he had the smallest increase of ineffective erythropoiesis as judged by the Hb level, plasma LDH activity and sTK activity, and the smallest increase in total erythropoietic activity as judged by the sTfR concentration. In this patient it is likely that the elevation in the plasma bilirubin was partly caused by co-inheritance of Gilbert's syndrome.
We are grateful to Drs D. A. Chidwick, C. Hatton, D. Lewis and A. J. Steed for allowing us to investigate patients in their care.