Natural history of juvenile haemochromatosis


Clara Camaschella, MD, Dipartimento Scienze Cliniche e Biologiche, Università di Torino, Azienda Ospedaliera S. Luigi, 10043 Orbassano, Torino, Italy. E-mail: clara.camaschella@


Summary. Juvenile haemochromatosis or haemochromatosis type 2 is a rare autosomal recessive disorder which causes iron overload at a young age, affects both sexes equally and is characterized by a prevalence of hypogonadism and cardiopathy. Patients with haemochromatosis type 2 have been reported in different ethnic groups. Linkage to chromosome 1q has been established recently, but the gene remains unknown. We report the analysis of the phenotype of 29 patients from 20 families of different ethnic origin with a juvenile 1q-associated disease. We also compared the clinical expression of 26 juvenile haemochromatosis patients with that of 93 C282Y homozygous males and of 11 subjects with haemochromatosis type 3. Patients with haemochromatosis type 2 were statistically younger at presentation and had a more severe iron burden than C282Y homozygotes and haemochromatosis type 3 patients. They were more frequently affected by cardiopathy, hypogonadism and reduced glucose tolerance. In contrast cirrhosis was not statistically different among the three groups. These data suggest that the rapid iron accumulation in haemochromatosis type 2 causes preferential tissue damage. Our results clarify the natural history of the disease and are compatible with the hypothesis that the HFE2 gene has greater influence on iron absorption than other haemochromatosis-associated genes.

Different inherited disorders of iron metabolism which lead to iron overload have been recognized. The most common form in Caucasians, hereditary haemochromatosis, is associated with mutations in the HFE gene (Feder et al, 1996), which encodes for a protein that interacts with transferrin receptor in the regulation of intestinal iron absorption (Lebron et al, 1998). Most haemochromatosis patients are homozygous for C282Y substitution in HFE, a minority are C282Y/H63D compound heterozygotes whereas other genotypes are infrequent (for review see Camaschella et al, 2000a). Haemochromatosis occurs rarely at birth (neonatal haemochromatosis) (Knisely, 1992; Kelly, 1998) and in infancy (Kaikov et al, 1992). The ‘juvenile form’, now renamed haemochromatosis type 2 (MIM #602390), has been characterized as a disease distinct from classic type haemochromatosis and was first recognized in 1978 (Lamon et al, 1978). It affects both sexes and the most prominent symptoms in the sporadic cases reported were abdominal pain, hypogonadism and cardiac disease (Cazzola et al, 1983; Camaschella, 1998). In HFE-associated haemochromatosis intestinal iron uptake is only modestly increased. As a consequence, iron overload requires several decades to become clinically manifest. In haemochromatosis type 2 iron overload is severe and organ failure occurs before 30 years. Haemochromatosis type 2 is unrelated to HFE and the corresponding locus (named HFE2) maps to the long arm of chromosome 1, at 1q21 (Roetto et al, 1999, 2000). Sporadic reports of patients and families are present in the literature (Lamon et al, 1978; Cazzola et al, 1983, 1998; Haddy et al, 1988; Jensen et al, 1993; Farina et al, 1995; Camaschella et al, 1997; Kelly et al, 1998; Roetto et al, 1999, 2000; Papanikolau et al, 2000; Rivard et al, 2000; Varkonyi et al, 2000), but it is unknown whether all the reported cases are related to mutations in a single gene or in different genes. Recently, other types of haemochromatosis have been characterized. Haemochromatosis type 3 is caused by mutations in transferrin receptor 2 (TFR2) gene (Camaschella et al, 2000b; Roetto et al, 2001). Haemochromatosis type 4 is an atypical form of haemochromatosis characterized by dominant inheritance, increased serum ferritin, but normal transferrin saturation and iron overload in macrophages and phagocytic cells (Montosi et al, 2001; Njajou et al, 2001) due to mutation in the ferroportin 1 (FP1) gene.

In this study, we describe the clinical data of 29 patients of different age, from 20 unrelated families of various ethnic origin. We compare the clinical phenotype of these patients with that of HFE-related haemochromatosis and of haemochromatosis type 3. Finally, we report the phenotype of 37 relatives heterozygous for haemochromatosis type 2. Our data define the natural history of juvenile haemochromatosis, provide further molecular evidence of the genetic heterogeneity of iron overload disorders and suggest that the product of the HFE2 gene has a major role in the regulation of iron absorption.

Patients and methods

Patients.  This study included 26 patients (10 males and 16 females) with juvenile presentation and three young siblings (two girls and one boy), previously described as examples of haemochromatosis in the first decade of life (Kaikov et al, 1992). The patients were from 20 unrelated families, 11 Italian, five Greek, three from the UK and one from Canada. All the available unaffected relatives were also examined. Diagnosis of ‘juvenile’ haemochromatosis was according to the following criteria (Camaschella, 1998): age at presentation < 30 years (at least in one affected member of the family), increased transferrin saturation and serum ferritin in the absence of causal HFE mutations and of known causes of secondary iron overload. Clinical data of patients from 18 families were previously described (Kaikov et al, 1992; Camaschella et al, 1997; Cazzola et al, 1998; Kelly et al, 1998; Roetto et al, 1999; Papanikolaou et al, 2000). Chromosome 1q linkage was already known in 14 families (Roetto et al, 1999; Papanikolaou et al, 2001) and was established in this study in the other six families.

Obligate HFE2 heterozygous were parents (19 subjects) and children (three subjects) of the patients. In addition, linkage analysis allowed the identification of 15 heterozygous siblings.

Ninety-three Italian C282Y homozygous males (Piperno et al, 2000) and 11 Italian patients with haemochromatosis type 3 (seven males and four females) (Roetto et al, 2001) were studied for comparison. All parameters were expressed as mean ± standard deviation. Differences between groups of patients were evaluated using the non-parametric Mann–Whitney test and Fisher's exact test. They were calculated using the statistical program graphpad prism 3·0 (GraphPad Software, San Diego, CA, USA).

Informed consent was obtained for molecular studies according to the different institutional guidelines.

Clinical studies.  All patients had a complete haematological evaluation. Transferrin saturation and serum ferritin were measured by standard methods. Clinical complications of iron overload were assessed as previously described (Piperno et al, 1996). Liver biopsy, performed to determine the extent of liver damage, was available in 22 patients. Liver iron concentration (LIC) was determined by atomic absorption spectrophotometry as previously described (Piperno et al, 1998). Hepatic iron index (HII) was calculated as the ratio between LIC and age. Finally the total iron removed (IR), calculated on the basis of number and volume of phlebotomies, and IR/age were determined in 14 patients who were treated by phlebotomy at regular weekly intervals.

Molecular studies.  EDTA blood samples were collected for genetic studies. DNA was prepared from peripheral blood buffy coats or lymphoblastoid cell lines by standard phenol-chlorophorm extraction (Sambrook et al, 1989). The C282Y and H63D HFE mutations were studied on genomic DNA, using PCR-based tests and restriction enzyme digestion with RsaI and MboI (New England Biolabs, Berkeley, MA, USA), respectively, as described (Carella et al, 1997). Linkage analysis to chromosome 1q was performed in six families (F10, F11, F12, F18, F19 and F20) by studying the intrafamilial segregation of D1S442, D1S2344, D1S498 (Roetto et al, 1999) D1S1556 and GATA13C08 (Roetto et al, 2000) markers, mapping within the critical region. Haplotypes were constructed manually.


Molecular studies

In the six new families studied, the segregation of marker alleles was consistent with 1q linkage (data not shown). Considering the whole series, the interaction with mutations in HFE was observed in eight patients (Table I). None was C282Y homozygous. Two children (F20 II-1 and II-3) were C282Y/H63D compound heterozygotes; two patients (F11 II-1, F12 II-3) were C282Y heterozygotes and four (F4 II-2, F9 II-1, F18 II-2 and F20 II-2) were H63D heterozygotes.

Table I.  Clinical data and HFE mutations in patients with haemochromatosis type 2.
PatientsSex Age*
CardiopathyHypogonadism ↓GTLiver biopsySkin
ArthropathyHFE mutationsIR/Age
  • *

    Age at presentation.

  • Cardiac death.

  • Cardiac trasplantation.

  • § β

    -thalassaemia trait.

  • TS, transferrin saturation; SF, serum ferritin; GT, reduced glucose tolerance; IR, iron removed; C, cirrhosis; F, fibrosis; N, normal; /, not done.

F1 II-1M231004585+++C4+ – / – – / – /
F1 II-3F15  88  615+F3+ – / – – / – /
F2 II-1F20  821400+C4+ – / – – / – 0·32
F3 II-1F21  843500+F3+ – / – – / – /
F4 II-2M20 / 2000+ / / + – / – + –0·55
F5 II-1§F20  983768+++C4+ – / – – / – /
F6 II-1M30  773800+++C4+ – / – – / – 0·64
F7 II-1F26  982000+F4+ – / – – / – 0·67
F7 II-2F21  752300+++F3+ – / – – / – /
F8 II-5F14  923280++F4+ – / – – / – /
F8 II-6M17  894400++F4+ – / – – / – 1·02
F9 II-1M30  931036++F4+ – / – + –1·24
F9 II-2F241002130+C4+ – / – – / – /
F10 II-1F311003000+C4 – / – – / – /
F10 II-2M211002850++C4 – / – – / – 0·38
F11 II-1F19  853313+++ / / ++/– – / – 0·7
F12 II-3M24  754150++ / / +/– – / – /
F13 II-1M21  862283+F4++ – / – – / – /
F14 II-1F39  904217+F4++ – / – – / – 0·48
F15 II-1F32  653553+++F4++ – / – – / – 0·22
F15 II-2F31 / 4328+ / / + – / – – / – 0·45
F16 II-2M25  852500+++F4++ – / – – / – 0·30
F17 II-1F19  85 / ++ / / + – / – – / – 1·06
F17 II-2F20  79 / ++ / / ++ – / – – / – 1·11
F18 II-1F201004840+++ / / – / – – /+ /
F19 II-1M291005665+C4+ – / – – / – /
F20 II-1F  7  94  339F3++/– – /+ /
F20 II-2M  6  90  148F3 – / – – /+ /
F20 II-3F  4  59  187N3+/– – /+ /

Clinical data in patients

The patient's clinical and biochemical profiles are shown in Table I. The three affected children (F20) (Kaikov et al, 1992) did not show clinical complications: however, all three had heavy parenchymal iron deposition detected at the age of 7, 6 and 4 years; their HII was 21·1, 19·6 and 20·6 respectively (Kaikov et al, 1992). On account of their young age and of the absence of clinical complications, they were not included in the statistical analysis.

In all other patients the mean age at presentation was 23·5 ± 5·9 years (Table II). Three women (F14 II-1, F15 II-1 and F15 II-2) had a rather late presentation, respectively, at the age of 39, 32 and 31 years (Papanikolaou et al, 2000), whereas F10 II-1 had symptoms at 31 but did not receive any treatment until she was 36 years (Camaschella et al, 1999). The mean values of transferrin saturation and serum ferritin were 88·6 ± 9·7% and 3146 ± 1270 µg/l respectively (Table II). All patients, except a young female aged 15 years detected through families studies (F1 II-3), had hypogonadotropic hypogonadism. Cardiac complications were recorded in nine of 26 patients (34·6%) and were the cause of death in two. One patient with severe iron accumulation in the heart underwent cardiac transplantation (Kelly et al, 1998) and three cardiac deaths secondary to haemochromatosis were recorded in two families (not shown). Reduced glucose tolerance was found in 15/26 (57·7%) patients. Eight of the 19 (42·1%) adult patients who underwent liver biopsy had cirrhosis and the remaining 11 had fibrosis; siderosis was grade 3–4 in all cases. The mean IR, available in 14 patients, was 14·0 ± 5·2 g and the mean IR/age was 0·65 ± 0·3. Transferrin saturation, serum ferritin levels and the occurrence of clinical complications were not statistically different among males and females (data not shown). As for the severity of the clinical picture, no difference was observed between patients who were also carriers of HFE mutations and patients with wild-type HFE. Also, the co-inheritance of a β-thalassaemia trait, observed in a single patient (F5 II-1), did not aggravate the clinical phenotype. In this case, iron depletion was achieved through a combined treatment of recombinant human erythropoietin and phlebotomies (De Gobbi et al, 2000).

Table II.  Statistical comparison of clinical data in haemochromatosis type 2 with C282Y homozygotes and type 3 patients.
 HFE (n = 93)PType 2 (n = 26)PType 3 (n = 11)
  1. TS, transferrin saturation; SF, serum ferritin; IR, iron removed (data available in 14*and in 6†patients respectively); NS, not significant (P > 0·05).

Age (years)44·8 ± 10·7< 0·000123·3 ± 6·2< 0·000139·4 ± 7·1
TS (%)87·7 ± 11·5NS88·6 ± 9·7NS92·9 ± 11·5
SF (µg/l)2830 ± 2239NS3146 ± 12700·0032023 ± 1245
Hypogonadism18·4%< 0·000196·1%< 0·000127·3%
Cardiopathy6·5%< 0·000134·6%NS9·1%
Reduced glucose tolerance26·9%0·00357·7%0·0049·1%
IR (g)14·2 ± 8·9NS14·0 ± 5·2*NS14·4 ± 8·9
IR/Age0·32 ± 0·2< 0·0010·65 ± 0·3*NS0·41 ± 0·3

Statistical analysis of the difference in iron parameters and clinical complications of patients with haemochromatosis type 2 as compared with both C282Y homozygotes and subjects with haemochromatosis type 3 are reported in Table II. Patients with haemochromatosis type 2 were statistically younger at presentation and had higher levels of serum ferritin (statistically significant only vs haemochromatosis type 3); transferrin saturation was not different. Moreover, patients with haemochromatosis type 2 showed higher prevalence of hypogonadism, reduced glucose tolerance and cardiopathy (the difference in cardiopathy was not statistically significant vs type 3). Hepatic cirrhosis and arthropathy did not differ among the three groups. IR also did not differ, but IR/age was greater in type 2 (statistically significant only vs C282Y homozygotes). The lack of a statistically significant difference of cardiopathy and especially of IR/age vs type 3 patients was probably due to the limited data in the last series.

Clinical data in carriers

The HFE2 heterozygous were 12 males and 25 females. Iron parameters were normal in all subjects but four. The mean values of haemoglobin, transferrin saturation and serum ferritin, except in three subjects with secondary iron overload and in one iron-deficient patient, are reported in Table III. In a 60-year-old woman from family F3, who had received blood transfusions and parenteral iron treatment for anaemia several years before, transferrin saturation was 66% and serum ferritin was 6745 µg/l. Moreover, two men (age 49 and 67 years) had increased transferrin saturation (51% and 45%) and hyperferritinaemia (3840 µg/l and 620 µg/l): the first was a heavy alcohol drinker (Cazzola et al, 1983) and the other was affected by porphyria cutanea tarda (Kelly et al, 1998) that may partially explain the high serum ferritin levels. The iron parameters were normal also in one carrier of C282Y, in five H63D heterozygotes and in one H63D homozygote. One C282Y-H63D heterozygous compound female was iron deficient (serum ferritin 3 µg/l). β-thalassaemia trait was present in four young HFE2 carriers: their iron parameters were comparable to those of the others subjects (data not shown).

Table III.  Iron parameters in HFE2 carriers.
 Age (years)Hb (g/dl)TS (%)SF (µg/l)
  1. Hb, haemoglobin; TS, transferrin saturation; SF, serum ferritin.

Males (n = 10)49·2 ± 16·214·9 ± 1·031·7 ± 6·0124·30 ± 55·9
Females (n = 23)32·3 ± 21·112·8 ± 1·426·0 ± 9·830·4 ± 17·7


This is the first report that outlines the natural history of haemochromatosis type 2 from the analysis of the largest published series of unrelated patients, with a disorder in linkage with chromosome 1q. The observation that kindred of different ethnic background have a disorder linked to the same locus clearly indicates that juvenile haemochromatosis results from mutations at the same gene. Also in juvenile haemochromatosis patients from Saguenay-Lac-Saint-Jean, an isolated region of Quebec, Canada, linkage analysis recently showed that the HFE2 locus is located on 1q, confirming that juvenile haemochromatosis is a homogeneous genetic disorder (Rivard et al, 2001).

Data concerning the disease expression in the first decade of life were previously unreported. Patients of family F20 are the first example of this description. Compound heterozygosity C282Y/H63D in F20 II-1 and II-3 does not explain the severe iron overload, as this genotype is usually mild (Piperno et al, 1998; Moirand et al, 1999). In addition, F20 II-2, who shows a phenotype identical to his sisters, has only H63D heterozygosity, a condition common in normal subjects. Although no clinical complications were present in F20 II-1, II-2 and II-3, the huge amounts of liver iron (Kaikov et al, 1992) indicate that iron accumulates within the first years of life.

Considering the other cases the mean age at presentation was 23·5 years. The observation that the disease appeared later in life in four women suggests the existence of modifiers (genetic or acquired) even in this severe disease. The late appearance of juvenile haemochromatosis makes the age at onset an insufficient criterion for diagnosis, as it can be confounded with a classic form of early onset (Adams et al, 1997). The commonest symptom at presentation was hypogonadism, which, at the end of the second decade, was present in all cases. Although hypogonadism can be the presenting symptom also in HFE (Adams & Valberg, 1996) and in haemochromatosis type 3 (Camaschella et al, 2000b), its appearance at a young age should raise the HFE2 suspicion.

Comparison with C282Y homozygotes and with patients with haemochromatosis type 3 clearly indicates that haemochromatosis type 2 is the most severe disease in terms of iron overload. Levels of serum ferritin are significantly higher in juvenile patients, especially when data were corrected for age (not shown). Clinical complications (cardiopathy, hypogonadism and reduced glucose tolerance) were more frequently recorded in type 2 patients than in C282Y homozygotes. IR/age, which reflects total iron burden, was statistically significantly greater in haemochromatosis type 2 than in C282Y patients. However, cirrhosis was not statistically more common. Also in the series of juvenile patients reported from Saguenay-Lac-Saint-Jean, cirrhosis was present only in 15·4% of the cases, whereas 53·8% of the cases had cardiopathy and 76·9% had hypogonadism (Rivard et al, 2000). The onset of hypogonadism or cardiopathy before cirrhosis suggests that pituitary hypophysis and heart have a particular susceptibility to iron toxicity at a young age, as also observed in the secondary iron overload of thalassaemia-major patients (Olivieri & Brittenham, 1997). It is plausible that this selective damage may be related to non-transferrin-bound plasma iron (NTBI), which is high in thalassaemia patients (Hershko et al, 1998). Although NTBI was not measured in our cases, some of them, as shown in Table I, had fully saturated serum transferrin. Alternatively, it is possible that the rapid iron accumulation that occurs is more dangerous to these tissues than to the liver, which is a natural iron store; or that cirrhosis requires more time to develop. The proportion of patients with arthropathy was not statistically different among the three groups, confirming that arthropathy is the single major haemochromatosis symptom that is not directly related to iron overload (Adams & Valberg, 1996). Comparison of the disease expression with haemochromatosis type 3 is more difficult, in view of the limited number of cases identified to date with this disorder.

At present, combined criteria based on (a) early onset of clinical symptoms, (b) severity of iron overload, (c) wild-type HFE genotype and, when family members are available, (d) linkage to markers of chromosome 1q are needed to establish a diagnosis of juvenile haemochromatosis. Early diagnosis will require the identification of the responsible gene.

The phenotype in juvenile haemochromatosis heterozygous has not been described previously. No evidence of iron overload is present in most cases and one female was even iron deficient, in spite of the co-existence of an HFE genotype at mild risk for iron overload. The clinical features of the heterozygotes are normal also in association with heterozygous or homozygous H63D mutations or in association with β-thalassaemia in young subjects. However, considering the limited series studied, we cannot exclude the possibility that in association with inherited or acquired conditions able to cause secondary iron overload, overt iron loading may occur, as observed in the three oldest patients.

The severity and the early onset of clinical complications suggest that the HFE2 gene encodes a protein that has greater influence over iron absorption than other haemochromatosis-associated protein. Based on phlebotomy requirements to maintain iron balance once the patient had achieved iron depletion, it has been calculated that iron absorption in haemochromatosis type 2 is 3–4 times greater than in HFE-associated disease (Cazzola et al, 1998). A comparable degree of iron overload, resulting from increased intestinal absorption, is observed only in congenital anaemias characterized by a high degree of ineffective erythropoiesis (Cazzola et al, 1999). We hypothesize that haemochromatosis type 2 is the result of a derangement of the duodenal pathway that increases iron absorption according to the erythropoietic needs (the so called ‘erythroid regulator’) (Finch, 1994; Andrews, 1999). Indeed the ‘erythroid regulator’ has a stronger effect on iron absorption than HFE inactivation. In animal models, hypotransferrinaemic mice, which have an iron-deficient erythron, show higher iron absorption and greater iron loading than HFE-deficient mice (Trenor et al, 2000). In humans, the addition of a minimal erythropoietic expansion, as in β-thalassaemia trait, significantly aggravates the iron burden of C282Y homozygotes (Piperno et al, 2000). Although the data are limited to a single patient, β-thalassaemia trait does not aggravate the clinical expression of juvenile haemochromatsosis. If our hypothesis is correct, the identification of the HFE2 gene will be important not only for more precise molecular diagnosis of haemochromatosis and to improve our knowledge of iron absorption, but also to gain insights into the regulation of erythropoiesis-mediated iron absorption.


This work was partially supported by EC Contract QLK6-1999–02237; Telethon GP0255Y01, Italy; the Italian Ministry of University, IRCCS Pavia.