Fetal and infantile hemochromatosis

Authors

  • Peter F. Whitington

    Corresponding author
    1. Department of Pediatrics, Northwestern University Feinberg Medical School, Children's Memorial Hospital, Chicago, IL
    • Department of Pediatrics, Northwestern University Feinberg Medical School, Children's Memorial Hospital, 2300 Children's Plaza, Box 57, Chicago, IL
    Search for more papers by this author
    • fax: 773-975-8671


  • Potential conflict of interest: Nothing to report.

Hemochromatosis is the term coined by Von Recklinghausen in the late 19th century to describe the association of endocrine dysfunction (i.e., diabetes), cirrhosis, and abnormally bronze-colored skin. Observed mainly in adult men,1 it was later determined that abnormal iron deposition in diseased tissues was involved in its pathogenesis and that it was a familial condition. It was not until 1996 that the hemochromatosis gene HFE was discovered and that the homozygous mutation C282Y was determined to be the cause of the vast majority of cases of hereditary hemochromatosis.2 In the intervening decade, mutations of other genes whose products are involved in the regulation of the body's iron status have been identified in patients expressing the phenotype of hemochromatosis, resulting in several disorders that now constitute “hereditary hemochromatosis.” There are few patients with the classic hemochromatosis phenotype in whom no gene defect can be identified.

Children—particularly young children—have never fit into the classic hemochromatosis phenotype. Clinical observation has shown that many years are required for patients with hereditary hemochromatosis to accrue the amount of excess iron that causes disease. Thus, decades before the discovery of HFE, rare patients with iron overload and organ failure or dysfunction beginning in childhood were classified differently. Children presenting signs of the disorder some years after birth were diagnosed with “juvenile hemochromatosis” while newborns simply had “neonatal hemochromatosis”. Mutations of newly described genes whose products are important in iron homeostasis have been determined to cause most cases of juvenile onset hereditary hemochromatosis. These findings have helped unravel the complexities of the system that regulates iron metabolism. Yet, the clinical and pathological findings in these children who rapidly accrue iron, perhaps even during intrauterine life, have shed some doubt as to whether neonatal hemochromatosis could possibly result from iron overload.

This review covers the clinical gamut of hemochromatosis as it affects fetuses and infants. The distinct clinical and pathological findings of neonatal hemochromatosis are the main focus of discussion. The subjects of hereditary hemochromatosis and iron homeostasis have been ably reviewed by Pietrangelo1 and Fleming and Bacon.3 Herein, brief overviews of these subjects will set the stage for discussing emerging concepts related to the pathogenesis of neonatal hemochromatosis.

Iron Homeostasis

Controlling the accrual of dietary iron involves a complex system for sensing iron status and signaling whether increased or decreased enterocyte iron export is indicated—a kind of “ferro-stat.” The 2 central figures in the ferro-stat are hepcidin, the sensor signal, and ferroportin, the regulator of enterocyte iron export. Hepcidin is synthesized by the liver and acts on ferroportin to cause its cellular internalization and destruction.4, 5 Ferroportin is essential for the export of iron from any cell, including enterocytes, hepatocytes, and macrophages.6 In times of iron sufficiency, the liver limits the further uptake of dietary iron by increasing the synthesis and export of hepcidin, which in turn reduces the amount of ferroportin available for export of iron from enterocytes. Hepcidin similarly acts on macrophages to reduce the amount of systemic iron available from recycling. Hemojuvelin appears to be upstream of hepcidin in the hepatic system of sensing cellular iron stores and regulating available iron and is critically involved in signaling the upregulation of hepcidin synthesis in conditions of iron sufficiency.7, 8 The function of HFE in the regulation of iron homeostasis is not clearly understood. In conditions of iron sufficiency, it may act directly by programming enterocytes to not export iron to the circulation or indirectly by regulating hepcidin synthesis and export by hepatocytes. Serum transferrin receptor 2 may function along with HFE to modulate hepatic hepcidin expression.9

The fetus parasitizes the mother to insure adequate iron for the growth and oxygen-carrying capacity of the fetus and newborn.10 The placenta acts as an active interface between the mother's huge iron pool and the highly controlled and relatively small fetal iron pool, insuring adequate iron supply to the fetus and protecting against potentially toxic iron overload.11 Concepts regarding how the flux of iron from mother to fetus is regulated are currently unfolding. It appears that the ferro-stat that functions after birth to control accrual of dietary iron also functions during fetal life, with the placental trophoblast functioning in analogy to the duodenal mucosa.12 Ferroportin is highly expressed and colocalizes with HFE in placental trophoblast cells.13, 14 Ferroportin expression increases with gestational age in parallel with the fetus' increasing demand for iron.13 Fetal hepcidin evidently regulates fetal iron stores. Disruption of the gene encoding for the transcription factor upstream stimulatory factor 2 (Usf2) in the mouse results in complete absence of hepcidin expression and a hemochromatosis phenotype.15Usf2−/− mice have massive iron overload in the liver that is significantly greater than that of wild-type mice as early as 2.5 months of age, suggesting fetal iron overload.15 In contrast, mice exhibiting an overexpression of transgenic hepcidin are born profoundly anemic and iron-deficient.16

Abbreviations

NH, neonatal hemochromatosis; AFP, alpha-fetoprotein; IgG, immunoglobulin G.

Hereditary Hemochromatosis

The vast majority of hemochromatosis results from homozygous C282Y mutation of HFE located on chromosome 6 (OMIM 235200; hereditary hemochromatosis type 1).1 In homozygous affected individuals, defective HFE function results in poor control of dietary iron assimilation, with approximately 3 mg/d of excess iron being assimilated. When accumulating in critical excess, the iron produces tissue damage and organ dysfunction, particularly involving the liver, endocrine glands, and heart. The age of onset and expression of clinical disease are highly variable but are usually in the fifth and sixth decades of life. The most common clinical presentation is with liver disease. HFE-associated hemochromatosis is one of the most common hereditary diseases; the rate for the homozygous C282y mutation is approximately 5 per 1,000 persons in populations of European descent.

Hemochromatosis with relatively early onset of clinical disease in adolescence and early adulthood has been termed juvenile hemochromatosis.1 Juvenile onset hereditary hemochromatosis is exceedingly rare, with at most a few dozen cases recorded. The mutations of 2 genes have been shown to produce juvenile hemochromatosis (OMIM 602390, hereditary hemochromatosis type 2): HJV, located on chromosome 1,17, 18 whose product is hemojuvelin (hereditary hemochromatosis type 2A), and HAPM, located on chromosome 19,19 whose product is hepcidin (hereditary hemochromatosis type 2B). The majority of cases are due to HJV mutations. The rate of iron accretion in juvenile hemochromatosis has not been determined, but is presumed to be much greater than in HFE-associated hemochromatosis. The clinical features are dictated by the rapidly accumulating iron overload. Iron parameters and tissue iron distribution in adolescents are similar to middle-aged adults with HFE-associated hemochromatosis. Endocrine manifestations (particularly hypogonadism) and cardiac disease are the most common presenting clinical features. The liver is involved as well, but significant liver dysfunction or cirrhosis is rarely seen, presumably because the other organ dysfunction dictates the natural history. It is important with regard to the potential role of iron overload in the pathogenesis of neonatal hemochromatosis to note the lack of liver injury in human juvenile hemochromatosis (Fig. 1). Furthermore, the hepcidin-deficient Usf2−/− mouse shows virtually no hepatic injury at 19 months of age despite having a liver iron content of 1,000 μg/g dry tissue, which is 10 times the wild-type value.15 It seems that it may take considerable time for acquired excess iron to produce liver injury. This may contribute to the lack of liver disease at clinical presentation of juvenile hemochromatosis compared with HFE-associated hemochromatosis.

Figure 1.

A 2-year-old girl was discovered to have increased serum iron indices (serum ferritin 3,600 ng/mL; transferrin saturation 79%) and elevated alanine aminotransferase levels during a work-up of failure to thrive. (A) Perl's Prussian blue stain of the liver biopsy showed marked iron overload (original magnification ×100). (B) This overload primarily involved hepatocytes, but it affected Kupffer cells as well (arrows) (original magnification ×400). A clinical diagnosis of juvenile hemochromatosis was made and phlebotomy therapy was instituted. An evaluation of the possible gene defects involved is underway. Examination of standard histological preparations showed little if any disturbance of hepatic architecture, and hepatocytes showed no evidence of cellular injury despite the massive iron overload.

Neonatal Hemochromatosis

Neonatal hemochromatosis (NH) (OMIM 231100) has been clinically defined as severe neonatal liver disease in association with extrahepatic siderosis in a distribution similar to that seen in hereditary hemochromatosis.20 Considerable evidence exists to suggest that it is a gestational disease in which fetal liver injury is the dominant feature.21 It has also been called “congenital hemochromatosis.”

NH is associated with severe fetal liver injury, and one of its most common presentations is late second and third trimester fetal loss as evidenced by the gestational histories of women who have had an infant diagnosed with NH.22 We have diagnosed NH in necropsies of stillborns as early as 22 weeks' gestation. Most affected live-born infants show evidence of fetal insult (i.e., intrauterine growth restriction and oligohydramnios), and premature birth is common. Liver disease is generally apparent within hours of birth, though in rare cases the liver disease takes a subacute course and manifests days to weeks after birth. There is a spectrum of disease, and some infants recover with supportive care.23, 24 Indeed, it may be that some “affected” infants have no clinical disease. We have shown that twins may have disparate clinical findings, with one severely affected and the other minimally so.25 It is one of the most commonly recognized causes of liver failure in the neonate.26

The presenting findings are those of acute liver failure and usually multiorgan failure. Affected infants are frequently diagnosed as having overwhelming sepsis of the newborn even with negative cultures. Hypoglycemia, marked coagulopathy, hypoalbuminemia and edema with or without ascites, and oliguria are prominent features. Jaundice develops during the first few days after birth. Most cases exhibit significant elevations of both conjugated and nonconjugated bilirubin. Serum aminotranferase concentrations are disproportionately low for the degree of hepatic injury, whereas circulating concentrations of alpha-fetoprotein (AFP) are characteristically very high, usually 100,000 to 600,000 ng/mL. Elevated concentrations of AFP in NH could be due to a failure to downregulate the synthesis of AFP, and perhaps both high AFP and low aminotransferase activities reflect a failure of the diseased liver to switch from fetal to neonatal metabolic patterns. Studies of iron status often show hypersaturation of available transferrin, with hypotransferrinemia and hyperferritinemia (>800 ng/mL), which though characteristic in NH are nonspecific in liver disease of the newborn infant.20 The low transferrin levels probably reflect severe liver disease. Notably, several patients in our experience have received the diagnosis of tyrosinemia based on elevated serum tyrosine levels, which are reflective of failed hepatic metabolic function. Urinary succinylacetone is absent.

The prognosis in severe NH is generally very poor. The average life expectancy of a severely affected infant is days to a few weeks with good care. A cocktail of antioxidants and an iron chelator has been used as a treatment. Success rates with medical treatment have been reported to be 10% to 20%, though it is not clear what factors determine a successful outcome.23, 24 No controlled trial of the use of this cocktail has been undertaken. Within the spectrum of severe NH, it may be that patients with more hepatic reserve profit from it, while more severely affected patients fail to respond.23 It is clear that patients with clinical liver failure due to NH can fully recover with no residual liver disease.23 NH is a frequent indication for liver transplantation in the neonate.26, 27 However, both patient and allograft survival are poor in neonatal liver transplantation, and neurological function is often impaired in survivors.27 When performed for NH, the difficulties attendant to transplanting newborns are frequently compounded by prematurity, small size for gestational age, and multiorgan failure.

Liver injury in NH is very severe and out of proportion to that seen in hereditary hemochromatosis (Fig. 2). Autopsy materials from stillbirths and newborns have provided most of the pathological descriptions of the disease. Cirrhosis is universally described. The liver is generally small and deeply bile-stained. Its contours may be irregular as a result of hepatocellular loss, collapse of stroma, and varying degrees of regeneration. Fibrosis is pronounced, particularly in the lobule and around the central vein. Regenerative nodules may be present. In some instances, almost no hepatocytes remain. The residual and/or regenerating hepatocytes may exhibit either giant cell or pseudoacinar transformation with canalicular bile plugs. Considerable acute and chronic inflammation is usually evident. The pattern of injury is strongly reminiscent of subacute or acute hepatic failure in postnatal life.

Figure 2.

Liver failure was noted at birth in a 2,900-g, 38 weeks' gestation male infant. Decreased fetal movement and oligohydramnios had been noted at 36 weeks' gestation. Refractory hypoglycemia and severe coagulopathy were the most prominent signs of liver failure. NH was suspected. On magnetic resonance imaging performed at 8 days of age, the liver (left arrow) and pancreas (right arrow) had a markedly reduced T2 signal intensity (dark area) relative to the spleen. Serum iron indices were not measured. Supportive therapy failed, and the infant expired on day 9. The composite image shows the most prominent histological findings on the postmortem examination. (a) Severe liver damage. A trichrome stain shows a fibrotic portal tract on the upper left. The hepatic parenchyma has been replaced by necrotic and fibrous tissue (blue-stained areas) and ductules; there were very few preserved hepatocytes (original magnification ×100). Perl's Prussian blue stain showed abnormal iron deposition in the (b) liver (original magnification ×200), (c) pancreas (original magnification ×200), (d) myocardium (original magnification ×400), (e) Hassall's corpuscles of the thymus (original magnification ×400), and (f) thyroid follicles (original magnification ×400). The liver iron content was modestly elevated at 4,559 μg/g wet weight (reference normal 200–2,000). R, right; L, left.

Hepatocytes often—though not always—show siderosis, while Kupffer cells are spared. The siderosis is coarsely granular, which is in contrast to the hazy iron staining of normal newborn liver. Siderosis may affect any of several tissues outside the liver (Fig. 2).20, 28 The most consistently affected are the acinar epithelium of the exocrine pancreas, myocardium, epithelia of thyroid follicles, and the mucosal (“minor salivary”) glands of the oronasopharynx and respiratory tree. Less affected are the gastric and Brunner glands, parathyroid glands, choroid plexus, thymus (Hassall corpuscles), pancreatic islets, adenohypophysis, and chondrocytes in hyaline cartilage. The spleen, lymph nodes, and bone marrow contain comparatively trivial quantities of stainable iron. There appears to be little tissue damage associated with the extrahepatic siderosis (Fig. 2), and no organ dysfunction (i.e., endocrine or cardiac disease) has been reported to result specifically from it. These findings make it seem nearly impossible that the universally observed severe liver injury could be the result of iron overload taking place over a period of weeks.

Occasionally, severely affected infants exhibit renal hypoplasia, with dysgenesis of proximal tubules and paucity of peripheral glomeruli.29 Correlation with the process of normal renal development dates this arrest of renal development to about 24 weeks' gestation. It is believed that this final stage of renal development is dependent upon liver function; therefore, the disordered development dates liver failure to the late second and early third trimester.

NH should be suspected in infants who manifest liver disease antenatally or very shortly after birth.30 Demonstration of extrahepatic siderosis is necessary for diagnosis. Finding siderosis in the liver is not diagnostic, and its absence does not exclude the diagnosis. NH is often diagnosed at autopsy, where siderosis of many tissues can be demonstrated. Demonstration of extrahepatic siderosis in living infants can be done via tissue biopsy or magnetic resonance imaging. Biopsy of the oral mucosa is a clinically useful approach to obtain glandular tissue in which to demonstrate siderosis.31 Differences in magnetic susceptibility between iron-laden and normal tissues on T2-weighted magnetic resonance imaging can document siderosis of various tissues, particularly the pancreas and liver (Fig. 2), though the diagnostic value of these approaches has never been formally evaluated. Oral mucosal biopsy often fails because specimens are either inadequate or do not contain submucosal glands. In our experience, adequate biopsies contain some stainable iron (any amount is abnormal) in approximately two thirds of cases ultimately proven to have NH, whereas magnetic resonance imaging demonstrates abnormal iron distribution in approximately 90% of proven cases.

The etiology of NH is not completely understood. Until just recently it was thought to be the phenotypic consequence of any severe fetal liver injury (i.e., due to intrauterine infection)32–34 or possibly a genetically inherited disease.35, 36 However, NH has an unusual pattern and a high rate of recurrence in the progeny of affected women that cannot be explained by these proposed mechanisms. The sporadic nature of NH and the effect an affected infant has on a woman's consideration for further pregnancies make it difficult to determine an exact recurrence rate. However, it is generally believed that after the index case there is an approximately 80% probability that each subsequent infant born to that mother will be affected.37 In our series, more than 80% of women have had more than one affected infant or an unexplained fetal loss.22 There are several documented instances of a woman giving birth to affected infants with different male parentage,22, 35 but not vice versa. There have been no reports of female siblings of women having an infant with NH themselves having affected infants. In summary, NH appears to be congenital and familial, but not hereditary.

We have hypothesized and have mounting supportive evidence that recurrent NH is an alloimmune gestational disease that results from maternal exposure during pregnancy to a common human liver antigen that is expressed mainly in fetal life.38 The pattern of recurrence, which is so much like that of known alloimmune diseases and so incompatible with hereditary disease, led us to this hypothesis. Maternal antibodies of the immunoglobulin G (IgG) class (and only IgG) are actively transported across the placenta to the fetus from approximately 18 weeks' gestation. The principle function of this process is to provide humoral immunity for the fetus and newborn against a broad array of microbiological antigens to which there has been no exposure. The principle of alloimmunity involves exposure of a woman to a fetal antigen (in NH we believe it to be a liver antigen) that she fails to recognize as “self,” which results in sensitization and production of specific immunoglobulin of the IgG class capable of recognizing and binding to the antigen. In NH-sensitized mothers, transplacental passage of maternal IgG to the fetus (in this and subsequent pregnancies) is accompanied by movement of antifetal liver antigen IgG to the fetal circulation, where it binds the in situ liver antigen and results in immune injury of the fetal liver. Liver injury results in the mishandling of iron that normally fluxes through the liver, which leads to siderosis and the NH phenotype. In this model, alloimmune injury of the fetal liver leads to the accumulation of iron, as opposed to iron overload leading to liver damage. It has been suggested from clinical observations that liver disease precedes iron deposition in NH,32, 34 which fits well with this model.

In this model, the hemochromatosis phenotype results from secondary iron deposition in extrahepatic tissues. It should be noted that it is not known if newborns with NH are actually iron overloaded, which is to say that total body iron content has never been determined. There are 2 possibilities with regard to the mechanism of siderosis in NH. One possibility is that the infants are actually iron-overloaded as a result of poor control of iron flux across the placenta. As stated before, fetal hepcidin is involved in regulating placental ferroportin function, which is intimately involved in regulating maternal–fetal iron flux. It is reasonable to speculate that reduced hepcidin synthesis by the severely injured liver could lead to poor regulation of ferroportin in the placenta and to iron overload. The failure to regulate ferroportin could also contribute to the lack of macrophage and Kupffer cell iron. The other possibility is that the infants are not actually iron-overloaded; rather, they have an abnormal but discrete deposition of iron in certain tissues. With severe liver injury, the central repository and distribution apparatus for iron in the fetus is inoperable. Iron bound to fetal transferrin, the concentration of which is usually low in NH, in the placenta must find a tissue to use or store it. In addition, transferrin saturation is high in NH, so free (nontransferrin bound) iron may also be in excess. The tissues showing siderosis are those capable of taking up iron from plasma. These mechanisms and possibly others deserve further study.

We have devised a treatment for women to prevent recurrent lethal NH during gestation based on the hypothesis that it is an alloimmune disease. The treatment consists of intravenous immunoglobulin derived from pooled sera of multiple donors administered weekly at a dose of 1 g/kg body weight from 18 weeks' gestation until the end of term. This is a treatment commonly used to modify the severity of other gestational alloimmune diseases. We have chosen for treatment women whose most recent gestation was affected with proven NH in lieu of any other marker for high risk of recurrence. To date, 30 infants have been born to mothers receiving gestational treatment according to our protocols. No intrauterine growth restriction, fetal liver disease, or other evidence of fetal distress have been detected in any case. This is in stark contrast to the typical progression of NH, in which severe intrauterine growth restriction and oligohydramnios are essentially universal. Only 6 infants showed significant clinical evidence of liver disease. However, there was biochemical evidence that 24 infants (80%) were affected with NH, as evidenced by elevated serum AFP (range 100,820–670,000 ng/mL) and/or elevated serum ferritin (range 1,250–15,948 ng/mL). Liver biopsies were obtained from 4 infants: 2 showed severe hepatitis with extensive necrosis and intense iron deposition, and 2 demonstrated well-preserved hepatic architecture and modest numbers of giant cells with siderosis. We have reported the results of 15 women treated through 16 pregnancies.22 When analyzed on a per-mother basis comparing outcomes of treated gestations to randomly selected previous affected gestations, gestational intravenous immunoglobulin therapy was associated with improved infant survival (P = .0009). We conclude from this experience that treatment with high-dose intravenous immunoglobulin during gestation appears to have modified recurrent NH so that it was not lethal to the fetus or newborn. These results provide additional strong evidence for an alloimmune mechanism for recurrent NH.

At present, we cannot say if all NH is alloimmune. Hepatic siderosis has been observed in cases of fetal and neonatal liver injury due to genetic/metabolic disease39, 40 and perinatal infection.33 However, extrahepatic siderosis has not been recorded in these cases. There continues to be a search for a gene locus associated with NH with no success to date.35, 36

In summary, there is little evidence to suggest that more than a small fraction of cases result from the aforementioned causes. There is reason to believe that the majority represents a humoral alloimmune fetal liver injury, though a reliable serological test for NH-associated antibody is needed to prove this. Such a test should be able to determine if all cases of NH represent a single disease—alloimmune NH—or if are there some cases that are secondary to another, perhaps sporadic fetal liver injury. This might explain the 20% lack of recurrence. If all or most NH is alloimmune, measuring maternal antibody could become a powerful diagnostic tool when dealing with a sick newborn in which liver failure is a component. The ability to measure antibody titer would allow preventative treatment trials aimed at streamlining and reducing the costs of the current intravenous immunoglobulin protocol. It should also provide promise in improving the care of affected newborns; the presence of high-titer antibody would be cause to remove maternal IgG from the infant via exchange transfusion or plasmaphoresis, as is performed for other severe fetal alloimmune and passive autoimmune diseases. Finally, it might be a useful screening test in pregnant women, because at present the only way to know if a woman is at risk is for her to have an affected infant. Our data show that severe NH is preventable, so screening would have significant value.

Acknowledgements

The figures were contributed by Hector Melin-Aldana M.D., Department of Pathology, Children's Memorial Hospital, Chicago: HMelinAldana@childrensmemorial.org

Ancillary