- Email: [email protected]
Serum ferritin iron in iron overload and liver damage: Correlation to body iron stores and diagnostic relevance
Serum ferritin iron in iron overload and liver damage: Correlation to body iron stores and diagnostic relevance PETER NIELSEN, ULRIKE GÜNTHER, MATTHIAS DÜRKEN, ROLAND FISCHER, and JOCHEN DÜLLMANN HAMBURG, GERMANY
The iron content of serum ferritin has been determined in groups of patients with normal or increased iron stores by using a technique of ferritin immunoprecipitation followed by iron quantitation with atomic absorption spectroscopy. The results were correlated to individual liver iron concentrations, measured non-invasively by superconducting quantum interference device (SQUID) biomagnetometry. A close correlation between serum concentrations of ferritin protein and ferritin iron was found (r = 0.92) in all groups of patients. However, the correlation between ferritin iron concentration and individual liver iron concentration was poor in patients with hemochromatosis (r = 0.63) and patients with β-thalassemia major (r = 0.57). The degree of ferritin iron saturation was about 5% in iron-loaded patients, which contrasts with results in two recent studies but confirms older observations. In patients with liver cell damage, the ferritin iron saturation in serum was significantly higher than that found in groups with iron overload disease, probably indicating the release of intracellular iron-rich ferritin into the blood. The monitoring of patients undergoing bone marrow transplantation indicated that the release of iron-rich and iron-poor ferritin occurred during phases of hepatocellular damage and inflammation, respectively. We find the benefits of serum ferritin iron measurement to be marginal in patients with iron overload disease. (J Lab Clin Med 2000;135:413-8)
Abbreviations: AAS = atomic absorption spectroscopy; GvH = graft-versus-host reaction; HFE = hemochromatosis gene on the short arm of chromosome 6; HH = hereditary hemochromatosis; LIC = liver iron concentration; SF-Fe = serum ferritin iron; SQUID = superconducting quantum interference device
S
erum ferritin is the most valuable parameter in the diagnosis and follow-up of iron deficiency and iron-overload diseases.1 However, in
From the Departments of Molecular Cell Biology, Clinical Pediatrics, and Neuroanatomy of the University Hospital of the University of Eppendorf. Submitted for publication September 30, 1998; revision submitted October 19, 1999; accepted November 15, 1999. Reprint requests: Peter Nielsen, MD, PhD, Universitätskrankenhaus Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany. Copyright © 2000 by Mosby, Inc. 0022-2143/2000 $12.00 + 0 5/1/106456 doi:10.1067/mlc.2000.106456
patients with substantial iron overload, the correlation between serum ferritin and the individual iron stores is poor.2 In addition, secondary factors such as infection, inflammation, or the presence of a tumor can also increase serum ferritin values.3,4 Recently the measurement of SF-Fe was reevaluated, and its diagnostic benefit was reported to be superior to that of serum ferritin. 5-8 In these reports SF-Fe was compared only with other blood parameters of iron metabolism and particularly with serum ferritin protein. The aim of the present study, which is centered on patients with either iron overload or liver damage, was to correlate SF-Fe with the individual liver iron concentration, which is the best available parameter for measuring the total body iron stores. 413
J Lab Clin Med May 2000
414 Nielsen et al
METHODS All plastic and glassware used was soaked and rinsed with a solution containing Triton X-100 and nitric acid to remove traces of iron. Ferritin iron was determined in 250 blood samples from different patients with increased serum ferritin according to the method described by ten Kate et al8 with a commercial ferritin immunoassay kit (Quantimune Ferritin IRMA; BioRad, München, Germany). This assay contains ferritin antibodies immobilized on small polyacrylamide beads that can be readily suspended and divided into aliquots. In brief, the immuno beads were separated from the buffer solution containing iodine 125–labeled antibodies by centrifugation and decantation. The remaining pellet was washed three times and finally resuspended in 10 mL of Sörensen buffer. To 300 µL of each respective serum sample (in duplicate) with a known serum ferritin content between 100 and 1000 µg/L was added 300 µL of the suspended immuno beads in Sörensen buffer. After incubation for 30 minutes at room temperature, the beads were separated from the supernatant by centrifugation and washed three times with 300 µL of 0.9% NaCl solution. Finally, 300 µL of distilled water was added, and the resulting suspension was heated at 90°C ± 3°C to remove ferritin from the immobilized antibodies. After centrifugation, iron in the supernatant was determined by using atomic absorption spectroscopy with a graphite furnace technique (HGA 700; Perkin Elmer). Samples were diluted 1:5 with matrix modifier (1% Triton-X-100, 20 mmol/L nitric acid). Serum ferritin was measured with an automated enzyme immunoassay system (Access, Sanofi). Values are expressed as mean values of two serum samples. Ferritin saturation (sat%) was calculated with the assumption of a molecular mass of 450,000 daltons and an iron content of 4500 atoms per ferritin molecule, by the following equation:
sat% =
(Ftn – Fe/55.8) (Ftn/450,000) × 4500
× 100%
Patients. Blood samples were collected from outpatients with primary (n = 27) or secondary (n = 58) hemochromatosis on the occasion of their first diagnostic investigation or as a therapy control, respectively. Serum samples were used immediately or stored until use for up to 24 months at –20°C. Because of the detection limit of the AAS (15 µg Fe per liter), a reliable assessment of ferritin iron was possible only in samples with serum ferritin >200 µg/L. The respective values were correlated with individual liver iron concentrations. Frozen serum samples from 2 patients with leukemia were obtained from the local bone marrow transplantation blood pool bank. The diagnosis of a homozygous hereditary hemochromatosis in the patients was established by the presence of at least three of the following five criteria: (1) transferrin iron saturation >62%, serum ferritin >300 µg/L; (2) LIC of more than 2000 µg Fe per gram wet weight; (3) hepatic iron index (HI [µg/y] = [LIC/age]) greater than 30; (4) grade 3 or 4 stainable iron in the liver; (5) more than 4 g of iron removed by quantitative phlebotomy. All patients of this series showed the C282Y-mutation in the HFE gene in homozygous form and were previously untreated. In addition,
blood samples were collected in HH patients (n = 13) during phlebotomy treatment as well as in obligate heterozygous carriers for HH (n = 11). Patients with β-thalassemia major were undergoing a regular blood transfusion regimen and were treated with either deferoxamine (Desferal; Novartis)(n = 42) or deferiprone (Apotex)(n = 16). A third group of subjects demonstrating increased liver enzymes caused by a known hepatopathy (viral hepatitis, chronic ethanol abuse, etc) was selected that demonstrated increased values of serum iron and serum ferritin. However, a clinically relevant liver siderosis (LIC of more than 1000 µg Fe per gram wet weight) was excluded by SQUID-biomagnetometry of the liver (see below). Liver iron concentrations were measured by using a SQUID biomagnetometer (Ferritometer; BTi, San Diego, CA) as described elsewhere.9 This method is routinely used in our department for the diagnosis of hereditary hemochromatosis and the therapy control of secondary hemosiderosis. Statistics. Means and standard deviations were calculated for all variables. For skewed distributions (serum ferritin, ferritin iron, liver iron), antilogarithmic mean values and standard deviations calculated by the propagation law of errors were applied. Unpaired Student t tests were performed to evaluate ferritin iron saturation between different groups of patients. RESULTS
A close correlation between serum concentrations of ferritin protein and ferritin iron was found for the entire patient group (logarithmic transformation, r = 0.92, P < .001; Fig 1, A). In contrast, the correlation between ferritin iron and individual liver iron concentrations was poor, both in patients with hemochromatosis or patients with β-thalassemia major (Fig 1, B). Ferritin iron saturation varied in the range between 3% and 10%, independent of liver iron concentration (Fig 2). No difference in ferritin iron level or ferritin iron saturation was found between the hemochromatosis group and the βthalassemia major group (Table I). In 13 patients with HH, several serum samples were analyzed for ferritin iron during the progress of phlebotomy treatment. Again, there was a considerable fluctuation in serum ferritin iron saturation, but this parameter seemed to be unchanged in most patients during iron depletion (Fig 3). Only in patients with apparent liver cell damage, as indicated by increased liver enzymes, was the ferritin iron saturation significantly higher than in all other groups (Table I, Fig 1, Fig 2), probably indicating a leakage of iron-rich tissue ferritin from the liver. The occurrence of two different forms of serum ferritin with respect to their iron content was also seen in patients with acute myelocytic leukemia who were undergoing bone marrow transplantation. Because of previous transfusions, serum ferritin is already
J Lab Clin Med Volume 135, Number 5
Nielsen et al
415
A
B Fig 1. A, Correlation of serum ferritin protein and serum ferritin iron. B, Correlation between serum ferritin iron and the individual liver iron concentration. Indicated are patients with homozygous (▲) or heterozygous (crosses) hemochromatosis, β-thalassaemia major (), or hepatopathy (). Solid lines represent, after log. transformation, a linear regression of all groups of patients except patients with hepatopathy.
Table I. Serum ferritin iron in different groups of patients n
Age
Liver iron* [mg Fe/g]
Serum ferritin* (µg/L)
Ferritin-Fe* (µg/L)
Untreated hemochromatosis Homozygote Heterozygote
27 11
51 ± 10 48 ± 14
2.65 ± 1.3 0.63 ± 0.2
1641 ± 1374 460 ± 258
42 ± 43 8 ± 11
4.9 ± 2.0a 4.2 ± 2.3
37 ± 21 18 ± 10
β-Thalassemia major All HCV negative HCV positive
58 8 50
20 ± 6 11 ± 5 21 ± 5
1.98 ± 1.6 1.60 ± 1.2 2.10 ± 1.7
2046 ± 1723 1321 ± 748 2198 ± 1896
45 ± 39 23 ± 20 50 ± 41
4.3 ± 1.7b 3.5 ± 1.0c 4.4 ± 1.8d
44 ± 37 16 ± 5 46 ± 37
Hepathopathy
39
53 ± 12
0.47 ± 0.3
845 ± 581
29 ± 20
6.4 ± 2.4e
94 ± 125
Group
Ferritin-Fe-sat. (%)
Ferritin-Fe-sat., Ferritin-Fe saturation; HCV, seropositive for hepatitis C virus. *Anti-logarithmic mean values. Probability values: a versus b, not significant, P > .05; a versus e, b versus e, significant, P < .01; c versus d, not significant, P > .05.
GPT (IU/L)
416 Nielsen et al
J Lab Clin Med May 2000
Fig 2. Correlation between serum ferritin iron saturation and the individual liver iron concentration in patients with homozygous (▲) or heterozygous (crosses) hemochromatosis, β-thalassaemia major (), or hepatopathy ().
Fig 3. Changes in serum ferritin iron saturation during phlebotomy treatment in 13 patients with hemochromatosis. Storage iron was calculated from the amount of hemoglobin iron removed. The two solid lines indicate two examples with progression of serum ferritin saturation during phlebotomy treatment.
increased before transplantation (Fig 4). In the posttransplantation period, ferritin protein and ferritin iron are sometimes asynchronously increased when hepatotoxicity or infection occurs. DISCUSSION
The origin of plasma ferritin is unclear. 10 Unlike tissue ferritin, a substantial proportion of serum ferritin is glycosylated, which suggests that plasma ferritin is actively secreted from reticuloendothelial11 or parenchymal cells12 rather than resulting from nonspecific leakage from cells.13 Earlier studies indicated
that serum ferritin, in contrast to tissue ferritin, has a low iron content even in iron-loaded patients (ferritin saturation, 2% to 11%).13-16 It was therefore assumed that serum ferritin does not provide a major source of hepatic iron either in normal individuals or in patients with iron-overload diseases.12 Recently Herbert et al5-7 and ten Kate et al8 have described novel methods for the direct measurement of serum ferritin iron. Preliminary data from a small number of patients indicated that a much higher ferritin saturation (up to 50%) was observed in patients with iron overload. In contrast, using the analytic method described by ten Kate et al,8
J Lab Clin Med Volume 135, Number 5
Nielsen et al
417
Fig 4. Progress of serum ferritin iron levels before and after allogenous bone marrow transplantation (BMT) in a patient with acute myelocytic leukemia.
we have found a relatively low ferritin iron saturation both in patients with untreated hemochromatosis (4.9%) and in patients with β-thalassemia major (4.3%, Table I). For the typical patient with a medium grade liver siderosis, the amount of serum ferritin iron would account for about 5% of the total plasma iron. This fraction is small but certainly not negligible as far as patients with hyperferritinemia are concerned. It can be speculated that this iron fraction may play a role in the recirculation of iron. For example, heterotopic iron deposits in the sinus lining endothelia of bone marrow, liver, and spleen are frequently observed in patients with iron overload diseases. The endothelial cells store the iron almost exclusively in the form of ferritin-containing and hemosiderin-containing lysosomes and not as free cytosolic ferritin molecules, indicating endogenous ferritin synthesis.17 The uptake of iron in the form of serum ferritin by micropinocytosis and its transfer to the lysosomal cell compartment could thus be a reasonable explanation for endothelial iron deposits, which are of diagnostic significance in iron overload. Another question is the relevance of serum ferritin iron measurement as a diagnostic parameter. Herbert et al5-7 have stated that serum ferritin iron is a new and better indicator than serum ferritin for measuring human body iron stores because serum ferritin iron is unconfounded by inflammation. As far as iron deficiency and normal iron stores are concerned, this seems to be very doubtful, simply for technical reasons. The sensitivity of the iron detection by AAS is too low, even when larger volumes of serum (1 mL) are available. In our hands, it was impossible to deter-
mine reliably the SF-Fe content in samples from irondeficient patients. In heavily iron-loaded patients we found more of a decrease in the saturation than an increase as speculated by Pootrakul15 and Herbert et al.5-7 This is also supported by the data from patients undergoing phlebotomy treatment. Similar to the data from Zuyderhoudt et al,18 the serum ferritin saturation was rather stable under the iron depletion therapy. It can still be argued that inflammation results in falsely increased ferritin values. Because serum ferritin is an acute phase reactant, it was hypothesized that serum ferritin protein generated in response to inflammation would contain much less iron than would “normal” ferritin. This aspect is only partly included in our study—in patients with β-thalassemia, in whom a subgroup of patients was HCVpositive (Table I). Also, some of the patients in the hepatopathy group were affected with chronic viral hepatitis. In both groups it was not possible to demonstrate a lower ferritin iron content in these subjects as compared with other patients without HCV. Interestingly, in patients undergoing bone marrow transplantation we could find SF-Fe saturation levels that fluctuated by as much as 20% within 3 days. Correlation with parameters of hepatic toxicity and inflammation indicated the release of iron-rich ferritin during liver cell damage and iron-poor ferritin during inflammation. Whether SF-Fe measurements may help to better differentiate hepatic complications in the early post-transplantation period after bone marrow transplantation (ie, drug-induced hepatotoxicity, infections
418 Nielsen et al
affecting the liver, veno-occlusive disease, or graft-versus-host disease of the liver) awaits further study. We conclude that the measurement of serum ferritin iron can routinely be measured in serum samples by using a combination of immunoprecipitation and atomic absorption spectroscopy. However, when compared with the more-sensitive serum ferritin protein determination, the additional diagnostic value of this new parameter in iron overload diseases seems to be limited. This article contains essential data included in the medical dissertation of Ulrike Günther. We thank Robert Hider, Kings College London, for critically reading this article.
REFERENCES
1. Worwood M. Laboratory determination of iron status. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 449-76. 2. Powell LW. Primary iron overload. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 227-70. 3. Birgegard G, Hallgren R, Killander A, Stromberg A, Venge P, Wide L. Serum ferritin during infection. Scand J Hematol 1978;21:333-40. 4. Jacobs A. Serum ferritin and malignant tumors. Med Oncol Tumor Pharmacother 1984;1:149-56. 5. Herbert V, Shaw S, Jayatilleke E. Serum ferritin-iron (holoferritin): the first reliable (not confounded by inflammation) serum measurement of body iron stores. J Invest Med 1995;43(suppl 1):198a. 6. Herbert V, Shaw S, Jayatilleke E. High serum ferritin protein does not distinguish iron overload from inflammation, but a new assay, high serum-ferritin-iron, does. Am J Clin Nutr 1995;61:911. 7. Herbert V, Jayatilleke E, Shaw S, Rosman AS, Giardina P, Grady RW, et al. Serum ferritin iron, a new test, measures human body iron stores unconfounded by inflammation. Stem Cells 1997;15:291-6.
J Lab Clin Med May 2000
8. ten Kate J, Wolthuis A, Westerhuis B, van Deursen C. The iron content of serum ferritin: physiological importance and diagnostic value. Eur J Clin Chem Clin Biochem 1997;35:536. 9. Nielsen P, Fischer R, Engelhardt R, Tondüry P, Gabbe EE, Janka GE. Liver iron stores in patients with secondary haemosiderosis under treatment with deferoxamine or deferiprone. Br J Haematol 1995;91:827-33. 10. Lindner MC, Schaffer KJ, Hazeh-Azam M, Zhou CYJ, Tran TN, Nagel GM. Serum ferritin: does it differ from tissue ferritin? J Gastroenterol Hepatol 1996;11:1033-6. 11. Cragg SJ, Wagstaff M, Worwood M. Detection of a glycosylated subunit in human serum ferritin. Biochem J 1981;199:565-71. 12. Worwood M, Dawkins S, Wagstaff M, Jacobs A. The purification and properties of ferritin from human serum. Biochem J 1976;157:97-103. 13. Arosio P, Yokota M, Drysdale JW. Characterisation of serum ferritin in iron overload: possible identity to natural apoferritin. Br J Haematol 1977;36:199-207. 14. Cragg SJ, Wagstaff M, Woorwood M. Sialic acid and microheterogeneity of human serum ferritin. Clin Sci 1980;58:25962. 15. Pootrakul P, Josephson B, Huebers H, Finch C. Quantitation of ferritin in plasma, an explanation for non-transferrin iron. Blood 1988;71:1120-3. 16. Halliday JW. Cellular iron processing and storage. The role of ferritin. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 97-121. 17. Düllmann J, Wulfhekel U, Hausmann K. Differences between Kupffer cells and endothelial cells of human liver in iron storage and release under normal conditions, after i.v. administration and in iron overload. In: Wisse E, Knook DL, editors. Kupffer cells and other liver sinusoidal cells. Amsterdam: Elsevier/North Holland Biomedical Press; 1977. p. 233-41. 18. Zuyderhoudt FMJ, Lindhorst C, Hengeveld P. On the iron content of human serum ferritin, especially in acute viral hepatitis and iron overload. Clin Chim Acta 1978;90:93-9.
The iron content of serum ferritin has been determined in groups of patients with normal or increased iron stores by using a technique of ferritin immunoprecipitation followed by iron quantitation with atomic absorption spectroscopy. The results were correlated to individual liver iron concentrations, measured non-invasively by superconducting quantum interference device (SQUID) biomagnetometry. A close correlation between serum concentrations of ferritin protein and ferritin iron was found (r = 0.92) in all groups of patients. However, the correlation between ferritin iron concentration and individual liver iron concentration was poor in patients with hemochromatosis (r = 0.63) and patients with β-thalassemia major (r = 0.57). The degree of ferritin iron saturation was about 5% in iron-loaded patients, which contrasts with results in two recent studies but confirms older observations. In patients with liver cell damage, the ferritin iron saturation in serum was significantly higher than that found in groups with iron overload disease, probably indicating the release of intracellular iron-rich ferritin into the blood. The monitoring of patients undergoing bone marrow transplantation indicated that the release of iron-rich and iron-poor ferritin occurred during phases of hepatocellular damage and inflammation, respectively. We find the benefits of serum ferritin iron measurement to be marginal in patients with iron overload disease. (J Lab Clin Med 2000;135:413-8)
Abbreviations: AAS = atomic absorption spectroscopy; GvH = graft-versus-host reaction; HFE = hemochromatosis gene on the short arm of chromosome 6; HH = hereditary hemochromatosis; LIC = liver iron concentration; SF-Fe = serum ferritin iron; SQUID = superconducting quantum interference device
S
erum ferritin is the most valuable parameter in the diagnosis and follow-up of iron deficiency and iron-overload diseases.1 However, in
From the Departments of Molecular Cell Biology, Clinical Pediatrics, and Neuroanatomy of the University Hospital of the University of Eppendorf. Submitted for publication September 30, 1998; revision submitted October 19, 1999; accepted November 15, 1999. Reprint requests: Peter Nielsen, MD, PhD, Universitätskrankenhaus Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany. Copyright © 2000 by Mosby, Inc. 0022-2143/2000 $12.00 + 0 5/1/106456 doi:10.1067/mlc.2000.106456
patients with substantial iron overload, the correlation between serum ferritin and the individual iron stores is poor.2 In addition, secondary factors such as infection, inflammation, or the presence of a tumor can also increase serum ferritin values.3,4 Recently the measurement of SF-Fe was reevaluated, and its diagnostic benefit was reported to be superior to that of serum ferritin. 5-8 In these reports SF-Fe was compared only with other blood parameters of iron metabolism and particularly with serum ferritin protein. The aim of the present study, which is centered on patients with either iron overload or liver damage, was to correlate SF-Fe with the individual liver iron concentration, which is the best available parameter for measuring the total body iron stores. 413
J Lab Clin Med May 2000
414 Nielsen et al
METHODS All plastic and glassware used was soaked and rinsed with a solution containing Triton X-100 and nitric acid to remove traces of iron. Ferritin iron was determined in 250 blood samples from different patients with increased serum ferritin according to the method described by ten Kate et al8 with a commercial ferritin immunoassay kit (Quantimune Ferritin IRMA; BioRad, München, Germany). This assay contains ferritin antibodies immobilized on small polyacrylamide beads that can be readily suspended and divided into aliquots. In brief, the immuno beads were separated from the buffer solution containing iodine 125–labeled antibodies by centrifugation and decantation. The remaining pellet was washed three times and finally resuspended in 10 mL of Sörensen buffer. To 300 µL of each respective serum sample (in duplicate) with a known serum ferritin content between 100 and 1000 µg/L was added 300 µL of the suspended immuno beads in Sörensen buffer. After incubation for 30 minutes at room temperature, the beads were separated from the supernatant by centrifugation and washed three times with 300 µL of 0.9% NaCl solution. Finally, 300 µL of distilled water was added, and the resulting suspension was heated at 90°C ± 3°C to remove ferritin from the immobilized antibodies. After centrifugation, iron in the supernatant was determined by using atomic absorption spectroscopy with a graphite furnace technique (HGA 700; Perkin Elmer). Samples were diluted 1:5 with matrix modifier (1% Triton-X-100, 20 mmol/L nitric acid). Serum ferritin was measured with an automated enzyme immunoassay system (Access, Sanofi). Values are expressed as mean values of two serum samples. Ferritin saturation (sat%) was calculated with the assumption of a molecular mass of 450,000 daltons and an iron content of 4500 atoms per ferritin molecule, by the following equation:
sat% =
(Ftn – Fe/55.8) (Ftn/450,000) × 4500
× 100%
Patients. Blood samples were collected from outpatients with primary (n = 27) or secondary (n = 58) hemochromatosis on the occasion of their first diagnostic investigation or as a therapy control, respectively. Serum samples were used immediately or stored until use for up to 24 months at –20°C. Because of the detection limit of the AAS (15 µg Fe per liter), a reliable assessment of ferritin iron was possible only in samples with serum ferritin >200 µg/L. The respective values were correlated with individual liver iron concentrations. Frozen serum samples from 2 patients with leukemia were obtained from the local bone marrow transplantation blood pool bank. The diagnosis of a homozygous hereditary hemochromatosis in the patients was established by the presence of at least three of the following five criteria: (1) transferrin iron saturation >62%, serum ferritin >300 µg/L; (2) LIC of more than 2000 µg Fe per gram wet weight; (3) hepatic iron index (HI [µg/y] = [LIC/age]) greater than 30; (4) grade 3 or 4 stainable iron in the liver; (5) more than 4 g of iron removed by quantitative phlebotomy. All patients of this series showed the C282Y-mutation in the HFE gene in homozygous form and were previously untreated. In addition,
blood samples were collected in HH patients (n = 13) during phlebotomy treatment as well as in obligate heterozygous carriers for HH (n = 11). Patients with β-thalassemia major were undergoing a regular blood transfusion regimen and were treated with either deferoxamine (Desferal; Novartis)(n = 42) or deferiprone (Apotex)(n = 16). A third group of subjects demonstrating increased liver enzymes caused by a known hepatopathy (viral hepatitis, chronic ethanol abuse, etc) was selected that demonstrated increased values of serum iron and serum ferritin. However, a clinically relevant liver siderosis (LIC of more than 1000 µg Fe per gram wet weight) was excluded by SQUID-biomagnetometry of the liver (see below). Liver iron concentrations were measured by using a SQUID biomagnetometer (Ferritometer; BTi, San Diego, CA) as described elsewhere.9 This method is routinely used in our department for the diagnosis of hereditary hemochromatosis and the therapy control of secondary hemosiderosis. Statistics. Means and standard deviations were calculated for all variables. For skewed distributions (serum ferritin, ferritin iron, liver iron), antilogarithmic mean values and standard deviations calculated by the propagation law of errors were applied. Unpaired Student t tests were performed to evaluate ferritin iron saturation between different groups of patients. RESULTS
A close correlation between serum concentrations of ferritin protein and ferritin iron was found for the entire patient group (logarithmic transformation, r = 0.92, P < .001; Fig 1, A). In contrast, the correlation between ferritin iron and individual liver iron concentrations was poor, both in patients with hemochromatosis or patients with β-thalassemia major (Fig 1, B). Ferritin iron saturation varied in the range between 3% and 10%, independent of liver iron concentration (Fig 2). No difference in ferritin iron level or ferritin iron saturation was found between the hemochromatosis group and the βthalassemia major group (Table I). In 13 patients with HH, several serum samples were analyzed for ferritin iron during the progress of phlebotomy treatment. Again, there was a considerable fluctuation in serum ferritin iron saturation, but this parameter seemed to be unchanged in most patients during iron depletion (Fig 3). Only in patients with apparent liver cell damage, as indicated by increased liver enzymes, was the ferritin iron saturation significantly higher than in all other groups (Table I, Fig 1, Fig 2), probably indicating a leakage of iron-rich tissue ferritin from the liver. The occurrence of two different forms of serum ferritin with respect to their iron content was also seen in patients with acute myelocytic leukemia who were undergoing bone marrow transplantation. Because of previous transfusions, serum ferritin is already
J Lab Clin Med Volume 135, Number 5
Nielsen et al
415
A
B Fig 1. A, Correlation of serum ferritin protein and serum ferritin iron. B, Correlation between serum ferritin iron and the individual liver iron concentration. Indicated are patients with homozygous (▲) or heterozygous (crosses) hemochromatosis, β-thalassaemia major (), or hepatopathy (). Solid lines represent, after log. transformation, a linear regression of all groups of patients except patients with hepatopathy.
Table I. Serum ferritin iron in different groups of patients n
Age
Liver iron* [mg Fe/g]
Serum ferritin* (µg/L)
Ferritin-Fe* (µg/L)
Untreated hemochromatosis Homozygote Heterozygote
27 11
51 ± 10 48 ± 14
2.65 ± 1.3 0.63 ± 0.2
1641 ± 1374 460 ± 258
42 ± 43 8 ± 11
4.9 ± 2.0a 4.2 ± 2.3
37 ± 21 18 ± 10
β-Thalassemia major All HCV negative HCV positive
58 8 50
20 ± 6 11 ± 5 21 ± 5
1.98 ± 1.6 1.60 ± 1.2 2.10 ± 1.7
2046 ± 1723 1321 ± 748 2198 ± 1896
45 ± 39 23 ± 20 50 ± 41
4.3 ± 1.7b 3.5 ± 1.0c 4.4 ± 1.8d
44 ± 37 16 ± 5 46 ± 37
Hepathopathy
39
53 ± 12
0.47 ± 0.3
845 ± 581
29 ± 20
6.4 ± 2.4e
94 ± 125
Group
Ferritin-Fe-sat. (%)
Ferritin-Fe-sat., Ferritin-Fe saturation; HCV, seropositive for hepatitis C virus. *Anti-logarithmic mean values. Probability values: a versus b, not significant, P > .05; a versus e, b versus e, significant, P < .01; c versus d, not significant, P > .05.
GPT (IU/L)
416 Nielsen et al
J Lab Clin Med May 2000
Fig 2. Correlation between serum ferritin iron saturation and the individual liver iron concentration in patients with homozygous (▲) or heterozygous (crosses) hemochromatosis, β-thalassaemia major (), or hepatopathy ().
Fig 3. Changes in serum ferritin iron saturation during phlebotomy treatment in 13 patients with hemochromatosis. Storage iron was calculated from the amount of hemoglobin iron removed. The two solid lines indicate two examples with progression of serum ferritin saturation during phlebotomy treatment.
increased before transplantation (Fig 4). In the posttransplantation period, ferritin protein and ferritin iron are sometimes asynchronously increased when hepatotoxicity or infection occurs. DISCUSSION
The origin of plasma ferritin is unclear. 10 Unlike tissue ferritin, a substantial proportion of serum ferritin is glycosylated, which suggests that plasma ferritin is actively secreted from reticuloendothelial11 or parenchymal cells12 rather than resulting from nonspecific leakage from cells.13 Earlier studies indicated
that serum ferritin, in contrast to tissue ferritin, has a low iron content even in iron-loaded patients (ferritin saturation, 2% to 11%).13-16 It was therefore assumed that serum ferritin does not provide a major source of hepatic iron either in normal individuals or in patients with iron-overload diseases.12 Recently Herbert et al5-7 and ten Kate et al8 have described novel methods for the direct measurement of serum ferritin iron. Preliminary data from a small number of patients indicated that a much higher ferritin saturation (up to 50%) was observed in patients with iron overload. In contrast, using the analytic method described by ten Kate et al,8
J Lab Clin Med Volume 135, Number 5
Nielsen et al
417
Fig 4. Progress of serum ferritin iron levels before and after allogenous bone marrow transplantation (BMT) in a patient with acute myelocytic leukemia.
we have found a relatively low ferritin iron saturation both in patients with untreated hemochromatosis (4.9%) and in patients with β-thalassemia major (4.3%, Table I). For the typical patient with a medium grade liver siderosis, the amount of serum ferritin iron would account for about 5% of the total plasma iron. This fraction is small but certainly not negligible as far as patients with hyperferritinemia are concerned. It can be speculated that this iron fraction may play a role in the recirculation of iron. For example, heterotopic iron deposits in the sinus lining endothelia of bone marrow, liver, and spleen are frequently observed in patients with iron overload diseases. The endothelial cells store the iron almost exclusively in the form of ferritin-containing and hemosiderin-containing lysosomes and not as free cytosolic ferritin molecules, indicating endogenous ferritin synthesis.17 The uptake of iron in the form of serum ferritin by micropinocytosis and its transfer to the lysosomal cell compartment could thus be a reasonable explanation for endothelial iron deposits, which are of diagnostic significance in iron overload. Another question is the relevance of serum ferritin iron measurement as a diagnostic parameter. Herbert et al5-7 have stated that serum ferritin iron is a new and better indicator than serum ferritin for measuring human body iron stores because serum ferritin iron is unconfounded by inflammation. As far as iron deficiency and normal iron stores are concerned, this seems to be very doubtful, simply for technical reasons. The sensitivity of the iron detection by AAS is too low, even when larger volumes of serum (1 mL) are available. In our hands, it was impossible to deter-
mine reliably the SF-Fe content in samples from irondeficient patients. In heavily iron-loaded patients we found more of a decrease in the saturation than an increase as speculated by Pootrakul15 and Herbert et al.5-7 This is also supported by the data from patients undergoing phlebotomy treatment. Similar to the data from Zuyderhoudt et al,18 the serum ferritin saturation was rather stable under the iron depletion therapy. It can still be argued that inflammation results in falsely increased ferritin values. Because serum ferritin is an acute phase reactant, it was hypothesized that serum ferritin protein generated in response to inflammation would contain much less iron than would “normal” ferritin. This aspect is only partly included in our study—in patients with β-thalassemia, in whom a subgroup of patients was HCVpositive (Table I). Also, some of the patients in the hepatopathy group were affected with chronic viral hepatitis. In both groups it was not possible to demonstrate a lower ferritin iron content in these subjects as compared with other patients without HCV. Interestingly, in patients undergoing bone marrow transplantation we could find SF-Fe saturation levels that fluctuated by as much as 20% within 3 days. Correlation with parameters of hepatic toxicity and inflammation indicated the release of iron-rich ferritin during liver cell damage and iron-poor ferritin during inflammation. Whether SF-Fe measurements may help to better differentiate hepatic complications in the early post-transplantation period after bone marrow transplantation (ie, drug-induced hepatotoxicity, infections
418 Nielsen et al
affecting the liver, veno-occlusive disease, or graft-versus-host disease of the liver) awaits further study. We conclude that the measurement of serum ferritin iron can routinely be measured in serum samples by using a combination of immunoprecipitation and atomic absorption spectroscopy. However, when compared with the more-sensitive serum ferritin protein determination, the additional diagnostic value of this new parameter in iron overload diseases seems to be limited. This article contains essential data included in the medical dissertation of Ulrike Günther. We thank Robert Hider, Kings College London, for critically reading this article.
REFERENCES
1. Worwood M. Laboratory determination of iron status. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 449-76. 2. Powell LW. Primary iron overload. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 227-70. 3. Birgegard G, Hallgren R, Killander A, Stromberg A, Venge P, Wide L. Serum ferritin during infection. Scand J Hematol 1978;21:333-40. 4. Jacobs A. Serum ferritin and malignant tumors. Med Oncol Tumor Pharmacother 1984;1:149-56. 5. Herbert V, Shaw S, Jayatilleke E. Serum ferritin-iron (holoferritin): the first reliable (not confounded by inflammation) serum measurement of body iron stores. J Invest Med 1995;43(suppl 1):198a. 6. Herbert V, Shaw S, Jayatilleke E. High serum ferritin protein does not distinguish iron overload from inflammation, but a new assay, high serum-ferritin-iron, does. Am J Clin Nutr 1995;61:911. 7. Herbert V, Jayatilleke E, Shaw S, Rosman AS, Giardina P, Grady RW, et al. Serum ferritin iron, a new test, measures human body iron stores unconfounded by inflammation. Stem Cells 1997;15:291-6.
J Lab Clin Med May 2000
8. ten Kate J, Wolthuis A, Westerhuis B, van Deursen C. The iron content of serum ferritin: physiological importance and diagnostic value. Eur J Clin Chem Clin Biochem 1997;35:536. 9. Nielsen P, Fischer R, Engelhardt R, Tondüry P, Gabbe EE, Janka GE. Liver iron stores in patients with secondary haemosiderosis under treatment with deferoxamine or deferiprone. Br J Haematol 1995;91:827-33. 10. Lindner MC, Schaffer KJ, Hazeh-Azam M, Zhou CYJ, Tran TN, Nagel GM. Serum ferritin: does it differ from tissue ferritin? J Gastroenterol Hepatol 1996;11:1033-6. 11. Cragg SJ, Wagstaff M, Worwood M. Detection of a glycosylated subunit in human serum ferritin. Biochem J 1981;199:565-71. 12. Worwood M, Dawkins S, Wagstaff M, Jacobs A. The purification and properties of ferritin from human serum. Biochem J 1976;157:97-103. 13. Arosio P, Yokota M, Drysdale JW. Characterisation of serum ferritin in iron overload: possible identity to natural apoferritin. Br J Haematol 1977;36:199-207. 14. Cragg SJ, Wagstaff M, Woorwood M. Sialic acid and microheterogeneity of human serum ferritin. Clin Sci 1980;58:25962. 15. Pootrakul P, Josephson B, Huebers H, Finch C. Quantitation of ferritin in plasma, an explanation for non-transferrin iron. Blood 1988;71:1120-3. 16. Halliday JW. Cellular iron processing and storage. The role of ferritin. In: Brock JH, Halliday JW, Pippard MJ, Powell LW, editors. Iron metabolism in health and disease. London: WB Saunders; 1994. p. 97-121. 17. Düllmann J, Wulfhekel U, Hausmann K. Differences between Kupffer cells and endothelial cells of human liver in iron storage and release under normal conditions, after i.v. administration and in iron overload. In: Wisse E, Knook DL, editors. Kupffer cells and other liver sinusoidal cells. Amsterdam: Elsevier/North Holland Biomedical Press; 1977. p. 233-41. 18. Zuyderhoudt FMJ, Lindhorst C, Hengeveld P. On the iron content of human serum ferritin, especially in acute viral hepatitis and iron overload. Clin Chim Acta 1978;90:93-9.