Teresa Pinheiro1, Raquel Silva2, Rita Fleming3, Afonso Gonçalves4, Maria A. Barreiros5, João N. Silva2, Patrice Morlière6, René Santus7 and Paulo Filipe2

1IST-ITN, Instituto Superior Técnico, Universidade Técnica de Lisboa & Centro de Física Nuclear, Sacavém, 2Clínica Universitária de Dermatologia, Hospital de Santa Maria & Faculdade de Medicina, Universidade de Lisboa, 3Serviço Imunohemoterapia, 4Serviço Radiologia, Hospital de Santa Maria, 5LNEG, UES, Lisbon, Portugal, 6INSERM, U1088, and Université de Picardie Jules Verne, Faculté de Médecine et de Pharmacie, Amiens, and 7Département RDDM, Muséum National d’Histoire Naturelle, Paris, France

Measurement of the concentration of iron in the skin, if correlated with total body iron stores, may enable better informed decisions on when to initiate, change or stop
therapy in hereditary heamochromatosis. Naïve
haemochromatosis patients with iron overload and with C282Y and/or H63D HFE mutations were evaluated at the following time-points: disease
diagnosis, end of the therapy programme, and 6 months after the end of therapy. The distribution and concentration of iron in the skin were assessed by quantitative nuclear microscopy methods, in parallel with serum and plasma iron concentration. Iron content in the liver was determined by nuclear magnetic resonance. Iron accumulated in the epidermis; its concentration increased from outer to inner layers, being maximal in the basal layer (7.33 ± 0.98 µmol/g). At all 3 time-points, most of the iron was associated with the extracellular space. During the phlebotomy programme the iron content of the skin and the liver decreased by a factor of 2. These data suggest that measurements of iron concentration in the epidermis, which is a readily accessible tissue, reflect iron overload in the liver. Key words: skin; iron distribution; haemochromatosis; proton microscopy.

Accepted Jan 25, 2013; Epub ahead of print May 27, 2013

Acta Derm Venereol 2013; 93: XX–XX.

Teresa Pinheiro, IST/ITN, Instituto Superior Técnico, Universidade Técnica de Lisboa, E.N. 10, 2685-953 Sacavém, Portugal. E-mail: murmur@itn.pt

Hereditary haemochromatosis (HH) is a disease caused by increased intestinal absorption of iron, resulting in its accumulation in multiple organs, such as the liver, heart, pancreas (1, 2) and skin (3), causing fibrosis and affecting their function. It is generally agreed that the C282Y and H63D mutations of the HFE gene are responsible for hereditary haemochromatosis (1), and both mutations are presently considered in HH screening. The C282Y mutation modifies the structure of the HFE protein, abrogating its association with the transferrin receptor and causing iron storage overload (4). The physiological role of the H63D mutation is less studied, although in vitro studies have shown that the affinity between the transferrin receptors and Fe-transferrin is reduced (4). Large population studies have shown that both the compound heterozygous and homozygous H63D genotypes have much lower penetrance than the C282Y homozygous genotype (1, 2). In the haemochromatosis phenotype, hyper-ferritinaemia, transferrin saturation (1,5) and increased hepatic iron concentration (6) are associated with H63D homo- or hetero-zygosity as observed with C282Y homozygotes or C282/H63D.

The diagnosis of haemochromatosis is based on genetic testing and indirect serological markers (7), which may not accurately reflect tissue overload of iron, as they can be influenced by other clinical conditions (8). Although non-invasive magnetic resonance imaging (MRI) of the liver is often sufficient to assess the iron overload status, some cases require confirmation by invasive hepatic biopsy (9, 10).

However, invasive techniques are not appropriate for serial observations and follow-up during periodic phlebotomies (7), which limits iron availability to tissues by removing iron from the circulation. Thus, new indicators and strategies to provide direct or indirect information on organ iron overload are needed. Because of its accessibility, skin is an organ of choice in this context. Helpful cross-sectional and longitudinal information can be obtained readily regarding the amount of iron deposition during a given time period and the clearance of iron from parenchymatous tissues by therapy intervention. Detailed studies on distribution and quantitation of iron in the skin in patients with HH have however not yet been carried out. There are few reports on iron distribution in normal human skin (11, 12), in patients with thalassaemia (13), and in animal models (14, 15).

The main objective of the present study was therefore to assess the appropriateness of skin as an indicator of organ iron deposition, and of therapy efficacy, by measuring iron concentration in the skin, in combination with conventional blood tests, plasma and serum iron concentrations, and the concentration of iron in the liver assessed by MRI.

Materials and Methods

Patients

A total of 28 patients, over 18 years of age, with elevated iron levels (18 men and 10 women; mean age ± SD 50 ± 14 years) diagnosed with HH for the first time were enrolled in the study (demographic, biological and resonance data from patients at diagnosis are shown in Table SI; available from: http://www.medicaljournals.se/acta/content/?doi=10.2340/00015555-1601). All patients were diagnosed on the basis of the usual clinical and biochemical data and the most frequent mutations on the HFE gene. Recommended biochemical blood tests for screening of haemochromatosis were performed, including serum ferritin (Ferr), transferrin saturation (TS) and total iron-binding capacity (TIBC). Blood cells, platelets counts, glycaemia and liver function evaluation were also performed. Liver iron content (LIC) was measured by quantitative MRI. The patients were either symptomatic or asymptomatic. They underwent phlebotomy therapy weekly (as tolerated) along with therapy for other relevant co-morbidities, such as cirrhosis, hepatocarcinoma, hepatitis C virus infection, diabetes mellitus, porphyria cutanea tarda, and steatohepatitis (Table SI). The number of phlebotomies, volume of blood withdrawn per course, treatment duration, and total amount of iron removed during the therapy programme, varied according to the patient’s condition. During treatment, patients underwent dietary adjustment. The clinical results were evaluated by following the evolution of blood iron indicators, with the aim of reaching ferritin levels of 50–100 ng/ml.

Study design

Patients were assessed at 3 time-points: Phase 1: at disease diagnosis, prior to phlebotomy therapy; Phase 2: at the end of the phlebotomy therapy programme; and Phase 3: 6 months after stopping the therapy programme.

Biochemical blood tests, LIC and determinations of iron content in the skin were performed in all patients before starting the therapy programme, at the end of the therapy programme, and during follow-up. Measurements performed during the pre-therapeutic and subsequent phases were synchronized. Informed consent was obtained for all subjects, in accordance with the Declaration of Helsinki, and the study was approved by the local ethics committee.

Sampling and methods

Blood samples were collected by venipuncture. Serum and plasma were separated by centrifugation (2,500 rpm, 10 min) and stored at –80ºC.

Skin was obtained by a 3-mm diameter punch biopsy from the lumbar-sacral region. Biopsies were quench-frozen in liquid nitrogen and stored at –80ºC until processing. Only one biopsy at a time was taken from each patient. Sections 14-µm thick were cut from the frozen biopsy in a cryostat at –25ºC (Cryotome 620E, Thermo Shandon, Cheshire, UK), dried and mounted in specific frames, as described elsewhere (16).

Serum iron concentration (S-Fe) was determined by the standard colorimetric method (together with TIBC determination), and plasma iron concentration (P-Fe) was determined by total reflection X-ray fluorescence (17).

Distribution and concentration profiles of iron in transverse sections of cryopreserved skin samples were compiled by nuclear microscopy using a combination of 3 techniques (16): (i) particle-induced X-ray emission (PIXE), (ii) Rutherford backscattering spectroscopy (RBS), and (iii) scanning transmission ion microscopy (STIM). This method has the unique capability of providing real-time morphological images and maps of multiple elements above sodium in the periodic table and, simultaneously, to quantify them down to the level of parts per million.

Nuclear microscopy was performed in the “Ion Beam Laboratory” of the IST/ITN. A focused 2.0 MeV proton beam with a spot size of 2–3 µm was used to scan across a selected area of interest of the skin cryosection. Data from the 3 techniques were obtained simultaneously. Images of the sample density were obtained by STIM and used to identify skin morphology. The concentration of iron was determined by PIXE with high quantitative accuracy and with a sensitivity of approximately 1 µg/g in biological material, such as tissue sections and cells. The concentration of matrix constituents was obtained by RBS. The distribution images of Fe were superimposed on the density images. Consequently, Fe concentrations can be associated directly with morphological details of the skin. Concentration profiles were obtained by analysing a sequential number of points (corresponding approximately to the focused beam area) along a transept of the scanned skin section. Skin cryo-sections were analysed in vacuum, and iron concentrations in the skin were expressed in µmol/g on a dry weight basis.

System characteristics, data acquisition, and the method for calculation of concentration have been described previously (16).

MRI was performed at the Radiology Service of Hospital de Santa Maria, with a 0.5-T Gyroscan T5-NT system (Philips Medical System, Best, The Netherlands). The MRI results were analysed by an experienced radiologist. Briefly, the MRI technique consists of 3 gradient echo sequences (T1, proton density, and long-echo time sequence T2) following the protocol established by Gandon et al. (18). This protocol is based on a T2-weighted gradient-recalled-echo, GRE sequence, which was found to be the most sensitive, allowing the detection of clinically relevant liver iron overload greater than 60 µmol/g, with 89% sensitivity and 80% specificity, and was validated with biochemical measures of hepatic iron concentration (18). The repetition time (TR), echo time (TE) and pulse angle used in the gradient echo sequences to obtain the T1, proton density and T2 images were: T1: 120 ms, 14 ms, 90º; proton density: 120 ms, 14 ms, 20º; and T2: 120 ms, 28 ms, 20º, respectively. To assess possible hepatic lesions the whole liver was first inspected using T1-weighted spin-echo (T1-SE) sequence. Afterwards, 6 magnetic resonance sections were taken for each sequence and one of the sections chosen to evaluate the signal intensity (SI), in order to avoid global variations in SI between images. The SI in the hepatic parenchyma was measured in 3 regions of interest > 1 cm2 and in the paraspinal muscles in 2 regions of interest. After calculating the mean of the measurements, the liver-to-muscle SI ratio was calculated for each sequence and these values inputted to the program for calculation of LIC (available at the University of Rennes Website; www.radio.univ-rennes1.fr). This calculation was performed for all patients.

Statistical analysis

Statistical analysis of the data included a χ2 test to compare categorical variables, and non-parametric Mann–Whitney test to compare continuous variables. Differences between treatment phases were calculated with non-parametric Wilcoxon pair test. Correlations between iron concentration in the skin and blood test values were examined by Spearman’s correlation coefficient using rank and linear regression analyses. A linear mixed-effects regression model was used to fit the phase-dependent iron content variations in the liver and skin, taking sex, age, co-morbidities and genotype into consideration as confounders. The model took into account that, for each patient, the measurements at 3 different time-points are associated. To comply with normality requirements, iron concentration variables were log-transformed.

Values of p < 0.05 (two-tailed) were considered statistically significant. Calculations were performed using SPSS (version 17.0) and R (version 2.7.0) software packages.

Results

Iron distribution in the skin

Fig. 1 shows images of transversal cryo-sections of skin obtained by nuclear microscopy techniques with spatial resolution of approximately 2 µm. By measuring the energy loss of each ion impinging at each position in the scanned area, an image can be formed that represents the mass distribution within the tissue. Mass density images of selected skin regions evidenced morphological details (Fig. 1A) that can easily be associated with histological features. By measuring the emitted X-rays from the iron atoms present in the tissue, an image of the distribution of the element is also produced (Fig. 1B). Superposition of the distribution with the mass density map allows determination of the exact location of the element. Fig. 1C shows that the iron concentration increases from the stratum corneum (where iron levels were lowest) to the inner layers, peaking at the basal layer and decreasing in the papillary dermis. For comparison, data extracted from healthy skin is also shown in Fig. 1C. A similar iron distribution profile is observed, which agrees with that reported in previous studies (16, 19).

Skin iron changes during treatment and follow-up

The iron concentrations during therapy were examined in the major strata, i.e. the stratum corneum (SC), stratum spinosum (SP), basal layer (BL), and papillary dermis (PD) . Iron concentrations were also measured in total epidermis (EPI) by combining the data for all epidermal strata, and in superficial skin (SSK) by combining the data for the epidermis and papillary dermis. Results obtained for patients with HH are summarized in Table SII (available from: http://www.medicaljournals.se/acta/content/?doi=10.2340/00015555-1601), which also reports data from healthy skin.

At Phase 1, when compared to control values, iron levels were significantly augmented in all skin regions with the exception of SC (Table I). After stopping the phlebotomy programme (Phase 2) skin iron content dropped drastically and continued to decrease until Phase 3. However, control levels were reached only at Phase 2 in SP and at Phase 3 in BL. At Phase 3, high iron levels were still measured in PD, EPI and SSK.

During therapy and 6 months after the end of the phlebotomies, the iron content measured in SC, SP and BL strata decreased to control values, but the iron levels in SSK and its major components, EPI and PD, remained significantly augmented. By contrast, no change in SC iron levels is observed during the 3 evaluation phases when compared with controls.

Iron distribution at the cellular level

Using the high-resolution mass density images, and taking advantage of the small beam size, epidermal cells can be visualized and the iron profiles of cells obtained. Extracellular and intracellular iron content can therefore be measured. Due to the complexity of the analysis, a sub-set of 6 patients with HH was randomly selected. The analysis was performed only in cryosections where basal keratinocytes could be individualized. In addition, cryosections were inspected under the light microscope (bright-field and/or dark-field reflection microscopy) prior to analysis, in order to assess cell overlap and cell sizes. In frozen-hydrated tissues, cells above the basal membrane ranged from 10 µm to 15 µm in diameter, corresponding approximately to the section thickness (14 µm). The tissue contracted slightly after the freeze-drying procedure, as observed in Fig. 2. Fig. 2 shows a mass density image of keratinocytes lying above the basal cells and the corresponding iron map. Iron deposits within the extracellular and cell membrane compartments can be observed. In phlebotomy naïve patients, extracellular iron concentration significantly exceeded the intracellular content (10.3 ± 1.3 µmol/g vs. 4.8 ± 0.9 µmol/g; p < 0.001). At Phases 2 and 3 the iron concentrations inside and outside the cells diminished remarkably. However, the extracellular rate of iron clearance was lower than that of intracellular clearance. At Phase 2, the extracellular iron concentration was 2.2-fold higher than that inside the cells (2.8 ± 2.2 µmol/g vs. 1.3 ± 1.0 µmol/g; p = 0.02) and, at Phase 3, the ratio increased to 3.2 (2.7 ± 2.0 µmol/g vs. 0.9 ± 0.5 µmol/g; p < 0.001).

Modifiers of iron concentrations in the skin

Iron concentrations in HH skin (whether considering major components or individual strata) were not dependent on sex or gender. Also, co-morbidities did not influence iron concentration in the skin. The genotype was significantly associated with iron concentration in the skin only at Phase 11.

Relationship between concentration of iron in the skin, blood indicators, and concentration of iron in the liver

At Phase 1, the concentration of iron in all regions of HH skin was correlated with blood TS, Ferr and P-Fe (p < 0.05). S-Fe did not correlate with EPI and SSK iron content. A strong correlation was found between LIC and SP, BL, EPI and SSK (p < 0.001). The decrease in LIC was comparable with that in skin, showing a progressive and significant decline up to 6 months after stopping the therapy programme. By contrast, data might suggest a trend towards an increase in Ferr, TS, SFe and PFe iron levels at Phase 3 (Table SII).

Despite different LIC and SSK values, a 2.2-fold iron content decrease was observed in going from Phase 1 to Phase 2, but this decrease was limited to 1.2-fold between Phase 2 and Phase 3 in both organs (values are shown in Tables I and SII)2.

Overall, it can be deduced that the iron concentration in BL, EPI and SSK was correlated with LIC throughout the treatment and follow-up. Iron content in the skin was correlated with Ferr and TS only in non-treated patients.

Significance of time-dependent iron content changes studied by regression analysis

The overall iron contents of the skin and the liver were significantly correlated (Fig. 3). The changes in iron concentration as a function of time for all patients in SP, BL, EPI and SSK, from Phase 1 to Phase 2 and to Phase 3 were highly significant (p < 0.001)3. While the amount of iron removed is marginally correlated with the decrease in concentration of iron in the skin in going from Phase 1 to Phase 2 (p = 0.07), a high statistical significance is obtained with Phase 3, which reflects the therapeutic effect4.

Discussion

The present work demonstrates that, in patients with primary HH, the iron content of epidermis, reflects iron overload in the liver. Although LIC is now rarely determined through invasive liver biopsies, but rather by MRI measurements, it is worth noting that iron assessment by MRI may be influenced by tissue condition and relaxation properties of the tissue iron (20). Although iron deposition occurs in dermis around adnexal structures, the nuclear microscopy imaging demonstrates that the BL is the preferential site of uniform iron deposition in the skin, and is therefore adequate for quantitation purposes. This method has the unique capability to provide real-time morphological images, maps of multiple elements and, simultaneously, to quantify them down to the level of parts per million. Iron determination in skin sections can be achieved rapidly, in approximately 2 h; thus this method can be used routinely. It is therefore a powerful research tool for dermatology (16, 19), allowing precise follow-up of the clearance of iron from tissues during phlebotomy therapy.

The BL is the most metabolically active region of the skin, where cell proliferation and differentiation occur, iron being indispensable for these processes. Even in control skin, the iron content in the BL is higher than that in other parts of the skin. Interestingly, nuclear imaging demonstrates that iron is found preferentially in the extracellular space or bound to cell membrane. In addition, data obtained with this method suggest that the iron concentration in BL, EPI and SSK is correlated with the circulating ferritin content.

The apparent discrepancy between our results in HH epidermis and those described in the experimental mouse model by Adams et al. (15) is probably related to the skin isolation procedure, which may wash away most of the extracellular iron.

The consequence of the excess iron distribution in the skin of patients with HH must be discussed in the light of cell iron metabolism. In all cells, iron uptake and storage are tightly regulated to guarantee sufficient iron for essential cellular processes and to prevent the production of damaging free radicals. In hepatocyte membrane, it is hypothesized that a homozygous C282Y mutation of the HFE gene prohibits the assembly of the transferrin-receptor 1 (TFR1) with the HFE protein, thereby compromising the cellular influx of transferrin-bound iron from the endosomal compartment to the cytosol (21, 22).

The iron-binding capacity of transferrin in the plasma of patients with HH is often exceeded. In this study, 14 out of 28 patients had 70% or more transferrin saturation. Non-transferrin-bound iron is present as low-molecular-mass iron complexes of citrate and acetate (23, 24). These low-molecular-mass complexes can initiate free radical damage as they supply iron ions in a form capable of accelerating free radical reactions and lipid peroxidation (25). In this regard, iron sensitizes cultured human skin fibroblasts to the ultraviolet A (UVA) radiation when cells are treated with low molecular mass complexes or exposed to high ferritin and holotransferrin concentrations with evidence of increased membrane lipid peroxidation and cytotoxicity (26) and activation of MAP kinases (27). The present study, showing that the extracellular space is the major site of excess iron deposition in the skin of patients with HH (Figs 1B and 2), might provide evidence about the lack of notable UVA-induced skin sensitization.

Genotype may contribute to iron overload with variable penetrance and phenotypic expression may be influenced by cofactors (8, 28–30) and other additive individual factors (1, 5, 28). It is therefore reasonable to observe a lack of significance in the correlation between iron blood indicators and body stores, such as skin and liver over a long time-period. This raises the possibility that, in some cases, there is a differential cellular iron handling, which results in variable accumulation of iron (5, 6, 8, 15, 28–30).

Acknowledgements

The work was carried out under Portuguese Society of Dermatology and Venereology (SPDV) and IAEA RC 302-F1.20.19-POR-13262 research contracts. The reference group for skin was established under the research contract EC/QLK4-CT-2002-02678.

REFERENCES