Investigation of Genetic Variants of Genes of the Hemochromatosis Pathway and Their Role in Breast Cancer

Iron is an important micronutrient required for ribonucleotide reductase, a key enzyme in DNA synthesis as well as respiratory and oxidative cell metabolism. In mammals iron homeostasis is regulated at the level of uptake because there is no excretory route. The precise mechanism and regulation of absorption is attributed to the transferrin receptor system, but another yet not completely understood system is likely to play a role (1). There is epidemiologic, clinical, and experimental evidence for iron to favor neoplastic growth (2). The free metal is carcinogenic due to its catalytic effect on the formation of mutagenic hydroxyl radicals, the suppression of host defense cells, and being a nutrient for unrestricted cancer cell multiplication (2). In animals and humans, primary neoplasms develop at extensive sites of iron deposits. In a carcinogen-induced experimental breast cancer model, rats developed sooner and twice as many mammary tumors on excess dietary iron administration compared with controls on a normal diet (3). In human breast cancer tissues iron overload has been noticed and may reflect a hunger for iron in those cancer cells. Particularly, transferrin receptor expression has been noticed as a feature of clinical breast cancer samples (4, 5) and specific transferrin receptor binding capacity was identified in microsomal preparations (6, 7). Also, elevated ferritin concentrations were identified in breast carcinomas (8), which were assigned to the malignant epithelium (9).

Insight into the molecular basis of iron storage disorders comes from hereditary hemochromatosis (OMIM 235200) that develops on germline mutations within the hemochromatosis pathway comprising genes involved in iron absorption and cellular uptake. These involve the hemochromatosis gene HFE, the ferroprotein 1 gene FPN1, and the transferrin receptor gene TFR2 (10-12). Variations of genes causing dysregulation of iron uptake may serve as corollaries for an increased breast cancer risk. More than 35 HFE allelic variants are known but most hereditary hemochromatosis cases develop on homozygosity of the HFE_845_G > A variant that substitutes tyrosine to cytosine at residue 282 (Cys282Tyr; ref. 13). Hemochromatosis is a MHC class I–like molecule that regulates duodenal iron absorption when stabilized by β2-miocroglobulin and bound to cell surface–associated transferrin receptor (14). Hemochromatosis Tyr282 homozygotes fail to properly assemble the HFE-β2m-TFR complex and are at increased risk of iron accumulation and its sequelae (15). This iron storage disorder eventually may lead to liver cirrhosis and hepatocellular carcinoma, diabetes mellitus, hypogonadism, cardiac manifestations including cardiomyopathy and arrhythmias, hyperpigmentation, arthritis, and premature death (16).

Hereditary hemochromatosis is the commonest autosomal recessive condition in individuals of Northern European descent and the prevalence of HFE Cys282Tyr heterozygotes among white Caucasians of Northern Europe, North America, South Africa, and New Zealand is about 10% (13). Increased cancer risks for extrahepatic neoplasms have been reported including that for hematologic, gastric, and colorectal tumors (17-19). Recently, an increased prevalence of the HFE Tyr282 allele has been reported for women with breast cancer (20), however, other previous breast cancer studies did not identify such an association (21, 22). With breast cancer being the most frequent female neoplasm in the western hemisphere and the HFE Cys282Tyr polymorphism being the commonest disease causing variation in Caucasian populations, it is important to clarify a possible relationship. Here we report the prevalence of 19 variations within HFE and TFR genes in a German population and discuss the role of HFE, FPN1, TFR1, and TFR2 in breast cancer susceptibility.

We analyzed incident breast cancer cases and population controls of the GENICA study population recruited between August 2000 and October 2002 from the Greater Bonn Region in Germany, an area of more than 1 million inhabitants. This is part of a wider effort of the Interdisciplinary Study Group on Gene Environment Interactions and Breast Cancer in Germany (GENICA) which focuses on the identification of breast cancer risks (23). Works are therefore referred to as the GENICA study. The study includes 688 breast cancer cases with a first time diagnosis of primary breast cancer that was histologically confirmed within 6 months of enrollment, and 724 population-based controls matched by age in 5-year classes. Cases and controls were included when they were below the age of 80 years, of Caucasian ethnicity, and residing in the study region. Risk factor information was collected via in-person interviews and included assessment of menopausal status, menstrual cycle, lifestyle factors, medical history, as well as personal and family history of cancer. Clinical and histopathologic variables were collected from medical records and pathology reports. The response rate was 88% for cases and 67% for controls. Study participants provided a blood sample drawn into heparin tubes (no. 367680 green, Becton Dickinson, Franklin Lake, NJ). The GENICA study was approved by the Ethics Committee of the University of Bonn. All study participants gave written informed consent.

Genomic DNA was extracted from 20 mL heparin blood samples using the Puregene kit (Gentra Systems, Inc., Mineapolis, MN) according to the instructions of the manufacturer. DNA samples were available from 610 of 688 (89%) cases and from 651 of 724 (90%) controls (23).

For genotyping, we applied TaqMan allelic discrimination (Perkin-Elmer ABI Prism 7700 Sequence Detection System; Applied Biosystems, Foster City, CA) and Hemochromatosis Strip Assay analyses (ViennaLab, Vienna, Austria). All DNA samples were genotyped at HFE_187_C > G (His63Asp, rs1799945), HFE_193_A > T (Ser65Cys, rs1800730), HFE_845_G > A (Cys282Tyr, rs1800562) and TFR1_424_A > G (Ser142Gly, rs3817672) using the TaqMan methodology. PCRs were carried out in reaction volumes of 25 μL containing 50 ng DNA, 1× TaqMan Universal PCR Master Mix (Applied Biosystems), 280 nmol/L of each primer, 6-carboxyfluorescein–labeled minor groove binder probe, and VIC-labeled minor groove binder probe (200 nmol/L for HFE_193_A > T and TFR1_424_A > G, 400 nmol/L for HFE_187_C > G and HFE_845_G > A). Primers were synthesized by MWG-Biotech AG (Ebersberg, Germany) and minor groove binder probes were from Applied Biosystems. Primer sequences and minor groove binder probes are given in Table 1. PCR conditions included an initial incubation at 50°C for 2 minutes, 95°C for 10 minutes followed by 35 cycles of 15 seconds at 92°C and 1 minute at 60°C. Genotype analysis was done using the ABI Prism 7700 SDS software (Applied Biosystems).

For controls, homozygote DNA samples were included that had been verified before genotyping by sequence analysis using Big-Dye Terminator mix and ABI PRISM 310 Genetic Analyzer (Applied Biosystems).

DNA samples of 321 cases were genotyped at 18 polymorphic sites at HFE (Val53Met, Val59Met, His63Asp, His63His, Ser65Cys, Gln127His, Pro160delC, Glu168Gln, Glu168Stop, Trp169Stop, Cys282Tyr, and Gln283Pro), TFR2 (Glu60Stop, Met172Lys, Tyr250Stop, and AlaValAlaGln594-597del), and FPN1 (Asn144His and Val162del) using the Hemochromatosis Strip Assay (ViennaLab). HFE, TFR2, and FPN1 polymorphic sequences were amplified simultaneously in multiplex PCR using biotin-labeled oligonucleotides. Genotyping was done by reverse hybridization with immobilized corresponding oligonucleotides using streptavidin-alkaline phosphatase and color substrates according to a published protocol (12).

Statistical analyses of differences between breast cancer cases and controls with respect to epidemiologic characteristics were done using Fishers' exact test. Genotype frequencies were tested for deviations from Hardy-Weinberg equilibrium by Pearson χ² square test. Data were considered in Hardy-Weinberg equilibrium when P values were >0.05. Epidemiologic risk factors, genotype frequencies, and adjusted odds ratios were calculated using SAS/STAT software version 8.02. Risk estimates for the development of breast cancer were calculated as odds ratios with 95% confidence interval using logistic regression analysis, conditional on age (<45, 45-50, 50-55, 55-60, 60-65, 65<75, ≥70 years). Additionally, we adjusted the risks for potential breast cancer risk factors such as smoking (never, ever), history of cancer in mother or sisters, and hormone replacement treatment in 10-year groups (0, <10, ≥10 years) referred to as adjusted odds ratios. Statistical analyses of differences in genotype frequencies between cases and controls or between studies were done by χ² or Fishers' exact test where applicable. Fishers' exact test for count data was used for the classification of HFE and TFR1 genotypes by nodal status of breast cancer cases.

Baseline characteristics of the GENICA controls and breast cancer cases are given in Table 2. These include age, menopausal status, menstrual cycle, potential breast cancer risk factors, iron storage disorder–related conditions, personal and family history of cancer, as well as histopathologic variables. There were no significant differences between patients and controls with the exception of known risk factors such as breast cancer in mother and sisters (P = 0.01) as well as hormone replacement treatment use (P = 0.047).

Study participants were genotyped for 19 polymorphic loci at HFE (Val53Met, Val59Met, His63Asp, His63His, Ser65Cys, Gln127His, Pro160delC, Glu168Gln, Glu168Stop, Trp169Stop, Cys282Tyr, and Gln283Pro), TFR1 (Ser142Gly), TFR2 (Glu60Stop, Met172Lys, Tyr250Stop, and AlaValAlaGln594-597del), and FPN1 (Asn144His and Val162del). Overall success rate with respect to genotype calling was 97.3%. Frequent polymorphisms HFE_187_C > G (His63Asp), HFE_193_A > T (Ser65Cys), and HFE_845_G > A (Cys282Tyr) were analyzed in duplicate by TaqMan and Strip Assay analyses for 321 DNA patient samples. Both assays showed concordance of results in 100% of samples. Because TFR1_424_A > G (Ser142Gly) was not included in the Strip Assay, we obtained genotypes by TaqMan analysis.

HFE_187_C > G (His63Asp). In GENICA population controls, the frequency of the C allele was 84.1% and the frequency of the G allele was 15.9%. Observed genotype frequencies were in Hardy-Weinberg equilibrium with 70.7% CC, 26.8% CG, and 2.5% GG. In cases, allele and genotype frequencies were comparable with those of controls (Table 3).

HFE_193_A >T (Ser65Cys). In GENICA population controls, the frequency of the A allele was 98.1% and the frequency of the T allele was 1.9%. Observed genotype frequencies were in Hardy-Weinberg equilibrium with 96.3% AA, 3.7% AT, and 0% TT. In cases, allele and genotype frequencies were comparable with those of controls (Table 3).

HFE_845_G >A (Cys282Tyr). In GENICA population controls, the frequency of the G allele was 94.4% and the frequency of the A allele was 5.6%. Observed genotype frequencies were in Hardy-Weinberg equilibrium with 88.9% GG, 10.9% GA, and 0.2% AA. Allele and genotype frequencies were comparable between cases and controls (Table 3). When we compared both genotype and allele frequencies of the GENICA breast cancer patients with those published by others (Table 4), our data were not significantly different (P = 0.15) from a Swedish breast cancer cohort but differed significantly (P < 0.0001) from an American breast cancer cohort. The latter showed an increased frequency of the minor A allele, which was responsible for an overall percentage of 24.4% for the heterozygote GA and 12.2% for the homozygote AA genotypes.

TFR1_424_A > G (Ser142Gly). In GENICA population controls, the frequency of the A allele was 53.9% and the frequency of the G allele was 46.1%. Observed genotype frequencies were in Hardy-Weinberg equilibrium with 29.7% AA, 48.5% AG, and 21.8% GG. Allele and genotype frequencies as well as Hardy-Weinberg equilibrium in the group of breast cancer patients were comparable with controls (Table 3). When we compared both allele and genotype frequencies of the GENICA breast cancer patients with those published by others (Table 4), our data were not significantly different (P = 0.22) from a Swedish breast cancer cohort.

Rare Variants at HFE, TFR2, and FPN1. Among GENICA breast cancer patients none of the known rare variants at HFE, TFR2, and FPN1 were identified (Table 5) in patients tested. All 321 patients were carriers of frequent homozygous genotypes at all loci.

Because no significant differences in allele and genotype frequencies were identified between controls and breast cancer patients of the GENICA study population, we extended the search for a putative genetic breast cancer susceptibility to the compound HFE-TFR1 genotypes. The combination of HFE_845_G > A (Cys282Tyr) and TFR1_424_A > G (Ser142Gly) genotypes did not show significant differences between patients and controls (not shown). In a next step, we established compound genotypes of two HFE loci HFE_187_C > G (His63Asp) and HFE_845_G > A (Cys282Tyr) together with the corresponding genotype at TFR1_424_A > G (Ser142Gly) locus. We observed a minor increase of 2.9% with respect to the frequency of the compound genotype HFE_187_CC/HFE_845_GG/TFR1_424_AG in cases (n = 181, 32.9%) when compared with controls (n = 191, 30.0%; not shown). This difference was not significant but remained stable (2.8%) when we further extended this compound genotype by a third HFE polymorphism, HFE_193_A > T (Ser65Cys). Accordingly, the compound quadruple genotype HFE_187_CC/HFE_193_AA/HFE_845_GG/TFR1_424_AG was identified in 172 (31.3%) patients and 181 (28.5%) controls (Table 6). All other differences in compound genotype frequencies were lesser and ranged between 0% and 1.6% (Table 6).

HFE_187_C > G (His63Asp), HFE_193_A > T (Ser65Cys), HFE_845_G > A (Cys282Tyr), and TFR1_424_A > G (Ser142Gly) genotypes were stratified by menopausal status, chronic anemia, and diabetes. No significant increases of breast cancer risks were observed (not shown).

For patients, genotypes at HFE_187_C > G (His63Asp), HFE_193_A > T (Ser65Cys), HFE_845_G > A (Cys282Tyr), and TFR1_424_A > G (Ser142Gly) were classified by tumor type, size, and grade as well as nodal status. With the exception of nodal status, histopathologic variables did not influence the homogeneity of distribution of genotypes among patient groups. However, when we tested for the distribution of genotypes among patients with different nodal status (N₀ + N₁ versus N₂ + N₃) there was a significant association of the HFE_845_G > A (Cys282Tyr) polymorphism with high nodal status (P = 0.032, data not shown). Similarly, when we compared the frequencies of the combined heterozygous (HFE_845_GA) and homozygous (HFE_845_AA) variant genotypes with the more prevalent HFE_845_GG genotype, we observed an increased risk for breast cancer of high nodal status with an odds ratio of 2.6 (95% confidence interval, 0.6-9.0). This difference was not observed with the other genotypes.

One of the cellular features of breast cancer is iron overload, a reason why the hemochromatosis and transferrin receptor complex attracted interest as potential contributors to breast cancer susceptibility. If dysregulation of iron uptake would be involved in a breast cancer risk, carriers of variations within the hemochromatosis and transferrin receptor systems should be more susceptible to breast cancer than noncarriers. Such a risk should be particularly obvious in populations of Northern European descent with a reputedly high prevalence of HFE mutations. Our contribution to this issue is the investigation of the prevalence of 19 known HFE, TFR1, TFR2, and FPN1 variations in a female German population and their possible effects on the incidence of breast cancer. Whereas investigations of the frequent HFE Cys282Tyr and TFR1 Ser142Gly polymorphisms in breast cancer populations have been previously reported, nothing is known on the putative health consequences of other known hemochromatosis and transferrin receptor pathway–related mutations. This is the first report of a comprehensive analysis of these genes and their possible role in breast cancer susceptibility.

Our population from Germany showed HFE and TFR1 variants at four loci including the three most frequent HFE polymorphisms, Cys282Tyr, His63Asp, and Ser65Cys, as well as the TFR1 Ser142Gly polymorphism. No other known variants at HFE, TFR1, TFR2, and FPN1 were identified. Frequencies of the minor alleles at HFE and TFR1 loci were similar in controls and cases. Particularly, the frequency of the major HFE Tyr282 allele was 5.6% in both controls and cases. This frequency obtained by us from a population in North Rhine-Westphalia (GENICA study population) matches that previously reported for a Bavarian population (24), but exceeds frequencies reported in other German studies ranging from 1.9% to 4.8% (13). Owing to sufficient size and population-based design as well as unlikely gender-specific contributions, we expect the GENICA study population to provide reliable data with respect to the prevalence of population-specific HFE and TFR variations. Our findings are in agreement with a Viking origin (25) of the HFE Cys282Tyr polymorphism and are also reminiscent to observed frequencies in Celtics (6.9%), Nordics (6.5%), and Anglo Saxons (6.0%), but differ with respect to observed frequencies in Southern Europeans (2.5%) and Russians (1.8%; ref. 26). As expected, frequencies of the HFE Asp63 allele were higher and those of the HFE Cys65 allele were lower. Case-control comparisons with respect to individual HFE loci and compound HFE genotypes did not indicate an association with breast cancer. To further explore possible relationships, we included the TFR1 Ser142Gly polymorphism in our analysis, which was previously suggested to cooperate with the HFE Tyr282 variant to increase breast cancer susceptibility (21). Although the prevalence of the TFR1 Gly142 variant was high, there was no difference between breast cancer cases and controls with respect to allele and genotype distributions.

We compared our results with those of the two other studies that previously investigated the possible role of HFE and TFR1 variants in breast cancer risk. A study from North Sweden did not find an obvious association between the HFE Tyr282 allele and breast cancer risk (21). However, when carriers of at least one HFE 282Tyr allele were also carriers of Ser142 at TFR1, the risk to develop breast cancer was more than 2-fold increased. This risk was even higher in compound HFE Cys282Tyr and His63Asp heterozygotes, but not significant. When we tested for similar relationships in our larger population-based breast cancer study, we observed a nonsignificant increase of compound heterozygotes in breast cancer patients for the triple compound HFE_His63_Cys282-TFR1_Ser142Gly and the quadruple compound HFE_His63_Ser65_Cys282-TFR1_Ser142Gly genotypes. Reasons for the differences in observations may be attributed to the large study size and population-based design of our study as well as possible variations of genotype frequencies between the Swedish and German populations.

A recent report of an American breast cancer study from Tennessee showed a high increase in the prevalence of the HFE Tyr282 allele in a highly selected series of breast cancer patients when compared with female populations from Tennessee or the United States (20). Although this study included a total of 182 breast cancer patients, this observation was solely attributed to a subgroup of 41 recipients of escalated-dose chemotherapy and blood cell transplantation. Accordingly, patients were selected on the basis of accepted indications for autologous blood cell transfusion reflecting breast cancers of poor prognosis. Moreover, the mean age at diagnosis of patients was below 50 years, indicating a large proportion of early-onset types of breast cancer in that group of patients. The strong association observed in this study may be influenced by the preselection and small numbers of breast cancer cases, issues that are part of an ongoing debate in molecular epidemiologic studies (27). Yet, there is a chance that socioeconomic effects and differences in ethnicity between the North American and European populations might have influenced the outcome of that study. In contrast, the incidental breast cancer cases of the GENICA study were not preselected and therefore represent a normal distribution of breast cancer patients. Kallianpur et al. (20) suggested that their transplant cohort may have facilitated the detection of an association of the HFE Tyr282 allele with aggressiveness of the disease. Interestingly, we also observed a 2.6-fold increased prevalence of at least one HFE Tyr282 allele in patients with four or more affected lymph nodes. A supportive biological rationale for this observation may be inferred from the known increased iron uptake associated with the HFE Tyr282 allele (15) and a high iron content in breast cancer cells (4, 5). However, this conclusion is limited by the low numbers of patients with high nodal status in our study as shown by the wide confidence interval.

Based on our data obtained from a population-based study with high response rates for cases and controls as well as accurate genotyping by diagnostic and state-of-the-art procedures, we consider it rather unlikely that variants of the hemochromatosis-transferrin receptor system have an immediate effect on the incidence of breast cancer in Germany. The high breast cancer incidence in this population may therefore not be attributable to these genetic variants. Beyond that, putative effects on breast cancer progression and prognosis may be possible but remain elusive.

We thank all the women who participated in this study. We also acknowledge the support of the numerous participants of the GENICA Network—Scientific Advisory Board: Hermann M. Bolt, M.D., Ph.D., Institut für Arbeitsphysiologie an der Universität Dortmund, Michel Eichelbaum, M.D., Dr. Margarete Fischer-Bosch-Institut for Clinical Pharmacology, Stuttgart; Hans Vetter, M.D., Medizinische Universitäts-Poliklinik Bonn. Clinicians and hospitals: Walther Kuhn, M.D., Universitätsklinikum Bonn; Uwe-Jochen Goehring, M.D., and Horst Wilms, M.D., Johanniter-Krankenhaus Bonn; Heinrich Mause, M.D., Gemeinschaftskrankenhaus St. Elisabeth-St. Petrus Bonn; Volker Pelzer, M.D., Marien-Hospital Bonn; Michael Kaiser, M.D., Malteser-Krankenhaus Bonn-Hardtberg; Rolf-Eberhard Herzog, M.D., Evangelisches Waldkrankenhaus Bad-Godesberg; Sigrid Milz, M.D., and Wolfgang Nohl, M.D., St. Johannes Krankenhaus Bad-Honnef; Karl-Heinz Schlensker, M.D., Klinikum Rhein-Sieg; Robert Kampmann, M.D., St. Josef-Hospital Troisdorf; Bernd Werling, M.D., Krankenhaus Troisdorf-Sieglar; Norbert Golz, M.D., Marien-Hospital Euskirchen; Matthias Winkler, M.D., and Peter-Karl Abramowski, M.D., Kreiskrankenhaus Mechernich. Pathologists: Reinhard Büttner, M.D., Universitätsklinikum Bonn; Magdolna Bollmann M.D., Reinhard Bollmann, M.D., Dietmar Kindermann, M.D., and Rainer Nikorowitsch, M.D., Bonn; Jürgen Vogel, M.D., Mathias Feldmann, M.D., and Jürgen Gerlach, M.D., Siegburg. Epidemiologists: Ulrich Ranft, Institut für Umweltmedizinische Forschung (IUF), Düsseldorf; Heinz-Erich Wichmann, M.D., Ph.D., Rolf Holle, Ph.D., and Hannelore Nagl, GSF-National Research Center for Environment and Health, Institute of Epidemiology, Neuherberg. Interviewer: Mette Besuden, Martin Black, Sandra Gschwendtner, Miriam Knie, Annette Kramer, Robert Priemke, Astrid Prömse, Miriam Rossa, Heiko Schneider, Johanna Scholl, Katja Tamme, Matthias Wenghöfer, Bernd Wiesenhütter, Melanie Wollenschein. Laboratory assistant: Silke Schöneborn.