In a study in the North of England, we first confirmed the birth weight association in childhood ALL and hypothesized that iron-related gene polymorphisms may explain this association by increasing both birth weight and childhood ALL risk. We tested this hypothesis, on 995 infants and their mothers from The North Cumbria Community Genetics Project (NCCGP) and 168 cases with childhood ALL from the Newcastle Haematology Biobank in a comprehensive study of the hereditary hemochromatosis gene HFE and other selected iron-related genes including the transferrin receptor gene TFRC (Dorak et al, MS submitted). We found that certain materno-fetal genotype combinations involving HFE and TFRC that increase fetal iron exposure resulted in higher birth weight in boys only and elevated ALL risk mainly in girls. Maternal effect was evident in mother-child pairs when mothers were positive for HFE variants, presumably increasing the amount of iron available for materno-fetal transfer. Our interpretation was that iron-related gene polymorphisms affected placental iron transport and males were able to use excessive iron by increasing fetal growth but girls were not. Greater cell proliferation rate in male fetuses would cause this dichotomy. Ultimately, boys would suffer less from the genotoxic effects of (free) iron but girls would have high amounts of iron increasing their risk for leukemia. Limited data on cord blood iron levels suggested that that when (1) the mother was positive for any of the three HFE variants that increased birth weight, (2) the offspring was male and (3) the male infant was positive for HFE and TFRC variants, birth weight would reach extreme values and cord blood iron would still be high. Thus, only the materno-fetal genotype combinations that increase iron levels most extremely would increase both birth weight and ALL risk in boys. This study suffered from lack of statistical power, especially for the assessment of interactions, and lack of plasma samples in sufficient quantity from mother-child pairs. There was only limited opportunity to increase the sample size due to availability of DNA samples from further mother-child pairs.

We are applying to ALSPAC to replicate this study with greater statistical power in a larger cohort of mother-child pairs and to take advantage of (1) pre-existing data on iron levels at birth, especially in cord slices, and at intervals after birth, (2) pre-existing genetic data on other genetic polymorphisms influencing birth weight, (3) the availability of data on post-natal diet and growth as well as any health outcome. The ALSPAC cohort would also provide the study with data on gestational iron supplementation, birth order, sex of previous birth(s), maternal prepregnancy weight and maternal smoking, that have been reported to influence birth weight or may be relevant for the sex-specificity of the combined HFE and TFRC association with birth weight. For our purposes, the Children in Focus (CiF) subset of ALSPAC will be most useful. We propose to carry out initial studies on CiF samples and extend positive findings to the complete ALSPAC cohorts. Thus, initial genotyping studies will take place at HUMIGEN, Genomic Immunoepidemiology Laboratory (year 1) and any association found or suggested will be replicated in the whole cohort at K-Biosciences (year 2).

The Newcastle study and our ongoing studies in the Genomic Immunoepidemiology Laboratory in HUMIGEN have identified the low penetrance HFE mutations involved in iron overload, C282Y and H63D, to be relevant in birth weight and leukemia associations but have also identified an HFE intron 1 SNP (rs9366637) as a strong predictor of birth weight. In addition, a detailed analysis of the HFE region suggested new susceptibility markers for childhood leukemia. The SNP rs10425 outside HFE appears to be a marker for the wildtype HFE haplotype along with rs807212 and rs12346. In addition to the HFE SNPs, we plan to include the TFRC SNP rs3817672 (S142G), transferrin (TF) SNPs rs1049296 and rs4481157, hepcidin (HAMP) SNP rs7251432, hephaestin (HEPH) SNPs rs5919015, rs809363, rs4827365, ceruloplasmin (CP) SNPs rs710573, rs7652826, bone morphogenetic protein 2 (BMP2) rs235756, hemojuvelin (HFE2 / HJV) SNP rs4970862, ferroportin (HFE4 / SLC40A1) SNP rs2304704, solute carrier family 11 (proton-coupled divalent metal ion transporters), member 2 (SLC11A2 / NRAMP2) SNP rs422982 and transmembrane protease, serine 6 (TMPRSS6) SNP rs733655. We will include these SNPs as a minimal set of iron-related gene polymorphisms and add any other SNPs that our ongoing studies may identify in these genes and other iron pathway genes as new candidates. We will include two HLA-G polymorphisms (rs16375 and rs1233334) also in the extended HLA class I region (between HFE and the HLA complex) because of their potential influence on fetal growth rates. HLA-G has been implicated as the human equivalent of the mouse Ped gene which regulates fetal growth rate. We will examine the HLA-G polymorphisms for their role in birth weight determination and to exclude as a confounder of the HFE association. To achieve these genotypings, we will need 200-250 nanograms of genomic DNA from each mother-child pair in the CiF group. Depending on the amount of already available data in cord blood iron parameters, we may need cord blood serum for iron and ferritin measurements.

If we get the expected results, genotypes or their combinations increasing materno-fetal iron transfer will correlate with birth weight increase in boys but not in girls except the HFE intron 1 SNP rs9366637, which appeared to associate with birth weight in boys and girls in the NCCGP cohort. We will seek correlations between genetic markers and maternal and cord (tissue/blood) iron levels to elaborate our preliminary observations on the NCCGP samples. We expect to find a weak inverse correlation between birth weight and cord iron parameters in boys because of iron depletion by increased growth rate. Seemingly paradoxically, however, we expect to find increased cord iron levels in boys in the presence of certain materno-fetal combinations of iron-related gene polymorphisms. It is this subset that we propose as the basis of birth weight association in leukemia. It will be of great interest to analyze pre-existing data on blood iron level changes during early childhood in these boys to find out whether any genotype maintains high blood iron levels throughout childhood. In a future study, these data can also be correlated with (iron-induced) oxidative stress and genotoxicity assay results.

Iron measurement in cord blood or tissue is an end-point analysis. Our preliminary data suggest that cord iron levels are a cumulative result of materno-fetal iron transfer, fetal cell proliferation rate and duration of gestation. The overall dynamic process of fetal iron levels cannot be reliably quantitated by a single end-of-pregnancy measurement. According to our model, the fetus may be exposed to high levels of (free) iron that causes the genotoxic damage but the cord blood iron level may be low due to accelerated growth towards the end of pregnancy. Since real-time monitoring of fetal iron levels is not possible in humans, ultimately animal experiments will be required to examine this hypothesis. At this stage, we are not planning to propose any animal studies till we obtain more robust data supporting our preliminary findings.