The present study shows that four SNPs, two in the TF gene and two others in the HFE gene, explain a large proportion of the genetic variation of serum transferrin in a group of menstruating women. In the best model calculated by BMA, the total phenotypic variation of serum transferrin explained by these SNPs is 16.9%. The estimated heritability of serum transferrin for Caucasian woman, which is the proportion of phenotypic variation in a population due to genetic variation, is 0.49 . Therefore, these four SNPs can explain approximately 35% of the total genetic variation or heritability for serum transferrin in this population group (calculated as 16.9/0.49).
The dietary study showed that mean dietary iron intake was below the Recommended Dietary Allowance for this population group (18 mg/day) , which, as previously reported by our research group [27, 28], is not easily reached in menstruating women. Although the influence of dietary components on iron absorption is well known , in the present study no association between nutrient intake and iron biomarkers was found. This is outstanding taking into account the wide range in iron biomarker levels that these volunteers presented. Moreover, these results are in agreement with other findings of our research group obtained in a study that compared iron status after consuming red meat or fish-based diets .
Serum transferrin and transferrin saturation are valid parameters that reflect iron status in this group of menstruating women as serum transferrin was inversely correlated with serum ferritin, haemoglobin, MCV, haematocrit and transferrin saturation .
Regarding the SNPs in TF, the two statistical analyses used led to coincident results, indicating that the presence of the minor allele A in rs3811647 is related to higher circulating transferrin levels whereas the minor allele T in rs1799852 relates to lower levels. These results from a sample of menstruating Caucasian women are consistent with and replicate data obtained by GWAS in other population groups [17, 18, 30].
SNP rs3811647 is situated in intron 11 and rs1799852 is a non-synonymous coding SNP in exon 17 of TF. As these TF variants are associated with serum transferrin concentration, they would appear to be involved at a transcriptional or postranscriptional level . In this sense, further studies should be done in order to study the influence that these two SNPs could have on the action of these transcription factors and on transferrin expression.
Present results also show that rs3811647 is associated with transferrin saturation, in agreement with the association observed with serum transferrin and with findings obtained in other populations . Considering that transferrin saturation is a marker of the iron supply to tissues, A carriers present a reduction in iron transport to tissues, as occurs in iron deficiency anaemia. Another study found that SNP rs3811647 was also associated with ferritin, although to a lesser degree than with transferrin and transferrin saturation . In the present study, however, an association between this SNP and serum ferritin was not observed, which can be explained as ferritin is the final marker of iron stores, indirectly related to iron transport, and also due to the sample size.
The other two SNPs included in the best model for the BMA test, rs1800562 and rs1799945 (C282Y and H63D mutation respectively), are in HFE. The minor allele of C282Y was associated with lower levels of serum transferrin, which is consistent with other studies [17, 18, 30]. Moreover, the present study shows that H63D was also associated with serum transferrin. Both SNPs are well known to be related to type I haemochromatosis , however the information concerning their roles in iron deficiency is very scarce. High iron status was observed in heterozygotes women, for either C282Y or H63D, compared with women lacking these mutations [32–34]. The present results confirm previous observations of our group  and support the hypothesis that C282Y and H63D mutations could have a protective effect against the development of anaemia in menstruating women. In this line, studies suggest that the haemochromatosis gene may have spread because of a selective advantage through protection of heterozygotes against iron deficiency [33, 35].
The study of interactions of these variants shows that the presence of G in rs3811647 and T in rs1799852 has a cumulative effect, with carriers of these variants presenting lower serum transferrin levels (Figure 1). In the presence of the minor allele A in rs3811647, that has been clearly associated with higher transferrin levels, the simultaneous presence of A in C282Y has a compensating favourable effect.
These four SNPs in TF and HFE may play a role in iron regulation because, as suggested in a recent study, the Tf/TfR2/HFE complex is critical for hepcidin regulation, and the concentration of diferric serum transferrin may act as a positive regulator, increasing hepcidin expression . Mutations in either TfR2 or HFE can result in iron overload, which is characterised by low hepcidin expression . However, there is no information concerning the possible compensatory effect on iron overload of the minor allele A of rs3811647. As these SNPs are found in genes for interrelated proteins that participate in one of the iron-metabolism pathways, it is suggested that the determination of these SNPs may be useful and may have clinical repercussions either in cases of iron deficiency or iron overload.
This work was focused on menstruating women because they present a high iron deficiency anaemia risk. According to WHO criteria , in this group of menstruating women, 10.6% were anaemic, 45.8% iron deficient and 43.7% iron sufficient women. This study replicates the associations between SNPs and serum transferrin and transferrin saturation observed previously in general population [17, 18, 30] and in HFE mutation carriers , which is remarkable, considering the size of this selected sample of menstruating women. In addition, a new finding concerning the association between H63D and serum transferrin, and the effects of the interactions between SNPs are presented.
It is well known that environmental factors, such as menstrual losses and nutrition, play an important role in the development of iron deficiency anaemia. Regarding menstruation, McLaren et al recently published a genetic association study excluding menstruating women in order to eliminate this possible confounding factor . However, our results are in agreement with the association between rs3811647 and TIBC found by those authors, and in addition, an association between this SNP and transferrin saturation was found.
These data suggest that, even though there are environmental and physiological factors involved in the development of iron deficiency anaemia, there are other factors, such as genetics, that could predispose to the disease. This may influence the choice of strategies for the prevention of iron deficiency anaemia in the population. In this regard, our group carried out a placebo-controlled nutritional intervention with an iron fortified food in iron deficient women. The iron-fortified food markedly increased iron status, except in women that presented the minor allele of SNP rs3811647 . These results may add new information to the area of nutrigenetics and nutrigenomics, since research on iron metabolism and gene-diet interaction is scarce.
To sum up, a large percentage of genetic variation of serum transferrin was explained by two SNPs located in the TF gene (rs3811647, rs1799852) and two in the HFE gene (C282Y, H63D) in menstruating women. In contrast to dietary intake, these SNPs are strongly associated with serum transferrin, and A carriers of rs3811647 present a reduction in iron transport to tissues, which might indicate higher iron deficiency anaemia risk. Moreover, the study of interactions of these variants shows that presence of the different alleles could have a cumulative or compensable effect in serum transferrin levels. Therefore, it is suggested that these genetic variants might potentially be used as markers of iron deficiency anaemia risk.
Further investigations in anaemic and non-anaemic subjects should be designed to increase the existing knowledge of the relationship between genetic variants and iron deficiency anaemia, and the possible modulating effect of diet and menstruation. Present results may be also useful to increase knowledge on iron overload disorders. It is also important to study the possible gene-diet interaction effects on the recovery of iron status. Finally, other studies should be carried out to explore the mechanisms by which these variants, or combination of variants as in haplotypes, affect iron metabolism at the transcriptional and postranscriptional levels and also at the functional level.