Abstract
Polymorphisms in genes related to the metabolism of vitamin B12 haven’t been examined in a Brazilian population. To (a) determine the correlation between the local genetic ancestry components and vitamin B12 levels using ninety B12-related genes; (b) determine associations between these genes and their SNPs with vitamin B12 levels; (c) determine a polygenic risk score (PRS) using significant variants. This cross-sectional study included 168 children and adolescents, aged 9–13 years old. Total cobalamin was measured in plasma. Genotyping arrays and whole exome data were combined to yield ~ 7000 SNPs in 90 genes related to vitamin B12. The Efficient Local Ancestry Inference was used to estimate local ancestry for African (AFR), Native American, and European (EUR). The association between the genotypes and vitamin B12 levels were determined with generalized estimating equation. Vitamin B12 levels were driven by positive (EUR) and negative (AFR, AMR) correlations with genetic ancestry. A set of 36 variants were used to create a PRS that explained 42% of vitamin level variation. Vitamin B12 levels are influenced by genetic ancestry and a PRS explained almost 50% of the variation in plasma cobalamin in Brazilian children and adolescents.
Similar content being viewed by others
Background
Public health recommendations for the intake of micronutrients are designed to meet the requirements of the majority (97–98%) of healthy individuals within a population group1. However, no standard approach has been developed for deriving vitamin and mineral recommendations2,3, and large variations exist across countries causing confusion among consumers, food producers, and policy makers. Advising diets and food patterns for individuals of different ages, conditions (e.g., pregnancy, lactation, athletes), and diseases is even more challenging because physiologies are affected by interactions between genetic makeup and environmental conditions which also change over time.
A large number of studies have focused on how genetic variation might affect micronutrient metabolism, clinical and metabolomic measurements, and phenotypic expression4,5,6 with the goal of personalizing recommendations based on genetic variation7. Over 2300 publications on associations between single-nucleotide polymorphisms (SNPs) in candidate genes involved in nutrient (including vitamins and minerals) metabolism or response, as well as with disease8,9 have been published since 200110. The majority of reports found statistical correlations but effect sizes were uniformly very small (usually < 1% of total phenotype) and reproducibility between studies is low. These results can be explained by gene–gene interactions, especially between different ancestral populations, and variation in diet and environmental/lifestyle factors that underlay gene-nutrient interactions11,12,13. In addition, reductionistic approaches ignore the many metabolite–protein (and therefore gene) and protein–protein interactions (i.e., gene–gene interactions) that produce an observable and measurable phenotype14.
More extensive omics measurements4,15,16 and systems biology approaches11 may provide a pathway to map the interactions between genotype, environmental factors, nutritional intakes and more exhaustive physiological analyses in different ethnic populations4,12,17,18,19,20. We tested the influence of genetic ancestry on baseline vitamin levels and found negative association between Native American (AMR) ancestry and vitamin B12 levels17. Genetic ancestry is an important cofounding variable for each specific phenotype, as admixed individuals inherit different combinations of causative variants based on the ancestral population in which those variants were present21.
Vitamin B12 is an essential water-soluble B vitamin that plays an elemental role in DNA synthesis, methylation reactions, and genomic stability, as well as mitochondrial metabolism22,23,24,25. Vitamin B12 may impact cognitive function12,26, cardiovascular disease through its role in homocysteine metabolism, and the phosphatidylethanolamine to phosphatidylcholine pathway, thereby influencing the concentrations of polyunsaturated fatty acids in plasma and in red blood cells27,28.
Inadequate intake or bioavailability and malabsorption are causes of vitamin B12 deficiency. Sub-clinical deficiency may affect 2.5–26% of the population, depending on the definition, as the cut-offs are undefined29. Studies with children in Brazil have shown different vitamin B12 deficiency rates, such as 3.7%30 11.7%31 and 15%32. The mandatory flour fortification with folic acid in Brazil has proved to be effective in increasing serum folate concentrations in children and adolescents33 but some researchers speculate that excess folate may interfere with vitamin B12 metabolism and may worsen the functional consequences of impaired vitamin B12 deficiency34. Although some Brazilian authors suggest that vitamin B12 deficiency is practically nonexistent in children35, the screening for vitamin B12 deficiency may be particularly relevant in our population because the combination of high serum folate and normal vitamin B12 status has been associated with a lower frequency of anemia in these children30,32. Severe deficiency without treatment may lead to high homocysteine levels in Brazilian children35.
Enzymes and transporter proteins play important roles in metabolism of vitamin B12 and, therefore in its status. Polymorphisms in genes related to the metabolism of vitamin B12 have been examined in several studies, including genome-wide association ones36,37,38,39,40,41,42,43, but the results are still ambiguous or inclusive and have not been well defined in Brazilian population. Determining the genetic factors that may influence vitamin B12 requirements may help form the basis for personalizing B12 intake recommendations. Variants within B12-related genes that are significantly associated with vitamin B12 levels36,37 may be used, in a middle out approach44, to construct a polygenic risk score (PRS) after adjusting for cofounding variables, such as genetic ancestry. The PRS of an individual is defined as a quantitative measure of the total genetic risk burden of the phenotype over multiple susceptibility loci45. Genetic risk estimation is the most basic measurable contributor to common heritable disease risk. Recent studies suggest that, for a subset of diseases, polygenic risk profiling provides personal and clinical utility as well as for therapeutic intervention and/or life planning46.
Genetic makeup may play an important role in which populations or subgroups will be more sensitive to differences in B12 vitamin availability, and also explain the prevalence of hyperhomocysteinemia and methylmalonic acidemia since most Western diets provide adequate supplies of B12, B6 and folic acid12. The present study aimed to (i) determine the correlation between the local ancestry components (AFR, AMR, EUR) and baseline vitamin B12 levels using SNPs in ninety B12-related genes; (ii) determine associations of 90 vitamin B12 related genes and vitamin B12 levels using generalized estimation equation, adjusted for sex, BMI, age, healthy eating index (HEI) and mean ancestry; (iii) determine a PRS using only the variants within the 90 related vitamin B12 genes that were significantly associated with vitamin B12 levels.
Methods
Study population and design
The data described in this cross-sectional study were from the crossover N-of-1 micronutrient intervention study previously reported17. Briefly, a six-week multivitamin/mineral intervention was conducted in 9–13-year-old children and adolescents. Participants were: (i) their own control (N-of-1); (ii) monitored for compliance; and (iii) measured for food intake, anthropometric and metabolites in plasma and RBCs, at baseline (Visit 1), post intervention (Visit 2), and following a 6-week washout (Visit 3) in two consecutive years, 2013 and 2014. Genetic profile was analyzed at baseline. To avoid the influence of these supplements on plasma B12 vitamin levels, only the baseline data (Visit 1) were used in the analyses described here.
Data collection was performed at the Ribeirão Preto Medical School Hospital (HCFMRP-USP), University of São Paulo, Brazil. The study was approved by the internal ethics committee (Process HCRP No. 14255/2010) and by the National Research Ethics Commission (No. 00969412.6 CAAE. 0000.5440). The trial was registered on ClinicalTrials.gov (NCT01823744–April 4th 2013). All the experiments were performed in accordance with relevant guidelines and regulations. The participants were informed about the purpose and procedures of the study and signed a statement of informed consent. Parents of each participant signed informed consent17,28.
Participants in this study were clinically stable children and adolescents, i.e., without injury, or infectious diseases, as specified in exclusion criteria. Children and adolescents aged 9–13 years, were recruited from three schools in the west side of Ribeirão Preto. This municipality is in the northeastern region of the state of São Paulo in Brazil17,28. Exclusion criteria were individuals: (i) with one or more episodes of axillary temperature higher than 37 °C in the 15 days preceding the blood collection; (ii) with three or more episodes of liquid stools in the 24 h before assessment; (iii) with intake of any kind of vitamin or mineral supplement; (iv) on a supervised diet for reducing weight or any other type of dietary restriction; (v) with a diagnosis of chronic disease that may interfere with data collection; and (vi) who participated in another clinical trial in the four weeks preceding the study.
The upper age cut-off was 13 years, 11 months and 29 days at registration visit (Visit 1). Individuals in all weight groups were included. A total of 280 participants met the inclusion criteria. After removing siblings and outliers of clinical and vitamins levels, 168 participants were considered for analysis in the present study17,28.
Anthropometry and socioeconomic rating
A dietitian measured height and weight of participants immediately after fasted blood collection (12 h) and BMI was calculated according to World Health Organization (WHO)47. The questionnaire for socioeconomic rating was from Associação Brasileira de Empresas de Pesquisa48.
Determination of metabolites levels
The analysis of metabolites investigated in this study has already been detailed described by Mathias et al.17 and by Ued et al.28. Total cobalamin and folate in plasma using AM-396 and MonoBind ELISA (Folate/Vitamin B12 Anemia Panel VAST test system, Monobind, Lake Forest, CA 92630, USA) was measured by Vitas (Norway). Vitamin B12 levels were defined as follow49: Low plasma levels (< 148 pmol/L), normal range (≥ 148 < 295 pmol/L); above normal range ≥ 295 pmol/L). Total homocysteine (tHcy) was measured in red blood cells (RBC) by liquid chromatography tandem mass spectrometry (LC–MS/MS) according to da Silva et al. 201650. Riboflavin and pyridoxal were analyzed in plasma by LC–MS/MS according to Meisser et al.51.
Genotype and exome sequencing data
Genotype data were generated using the HumanOmni5Exome 4v1_A (IlluminaTM) according to the Infinitum® LCG Quad Assay Protocol Guide52. Exome capture and sequencing used 3 ug of genomic DNA extracted from blood which was fragmented on a E220 Covaris to an average size of 150–200 bp. Following purification of the fragmented DNA using Ampure XP magnetic beads (Agencourt), libraries were prepared using the Agilent library kit (SureSelectXT Automated Target Enrichment for Illumina Paired-End Multiplexed Sequencing Automated using Agilent NGS Workstation Option B, Version F.3, October 2014). Agilent SureSelect V5 plus UTR kit, which targets 75 Mb of genomic regions, was used for exome capture following manufacturer’s instructions. Pools of 96 libraries were assembled and pooling homogeneity was evaluated by sequencing on MiSeq Nano Kit V2 with around 1.2 M reads pass filter. Deep sequencing was next performed on HiSeq2500 (Illumina) with PE 125 reads using V3 chemistry, with targeted mean exon coverage of 100× with a minimum threshold set at 70×. The SNPs were assigned to the coordinates of Genome Reference Consortium Human genome build 37 (GRCh37)53.
Local ancestry determination and their correlation with vitamin B12 levels
The Efficient Local Ancestry Inference (ELAI)54 was used to estimate local ancestry for each individual for three-way admixture for the following ancestral groups (components): African, Native American, and European. A total of 76,307 SNPs from samples of this study population mapped to both Human Genome Diversity Project (HGDP)55 and 1000 Genomes Project56 reference populations, according to Mathias et al.17. Averages for ancestry were calculated for each SNP from each individual, with 10 calculations at each of 20 steps along each chromosome by specifying the number of admixing generations equal to 15 spanning from a defined upper-layer cluster equal to 3 and a lower-layer cluster equal to 1554. In order to check for consistency, the ADMIXTURE tool57 was also used to determine the global ancestry to compare with mean ancestry values obtained from local ancestry. Spearman correlation was determined between the local ancestry components (AFR, AMR, EUR) and baseline vitamin B12 levels with the p value calculated after 10,000 permutations (pperm). The results were used to generate a correlation score defined as – log10(pperm) multiplied by the sign of the correlation. Significance was accepted at pperm < 0.1.
Statistical analysis on genes related to vitamin B12 levels
A list of genes related to vitamin B12 levels was obtained from a gene set called vitamin B12 deficiency from the DISEASES resource36, and includes 76 genes from text mining and manually curated disease-gene associations. This list was expanded with 14 other B12-related genes described in a recent review37 resulting in a total of 90 B12-related genes (Additional file 2: Table S1 and S2). The DISEASES resource36 is a comprehensive freely available database of disease-gene associations. The associations are extracted through automatic text mining with evidence from databases with permissive licenses, namely manually curated associations from Genetics Home Reference (GHR) and UniProt Knowledgebase (UniProtKB), GWAS results from DistiLD, and mutation data from Catalog of Somatic Mutations in Cancer (COSMIC). A recent review of 16 studies of vitamin B12-related genes provided other SNPs with statistically significant associations with blood vitamin B12 concentrations37. The populations included in the studies were African American, Brazilian, Canadian, Chinese, Danish, English, European, Icelandic, Indian, Italian, Latino, Northern Irish, Portuguese and residents in the United States.
The coordinates of the loci from transcription start to end sites were obtained from NCBI (37.3) gene definitions58 and were employed to mapping the variants observed in the samples. Variants were between 5 kb upstream from the start coordinate to 1.5 kb past the end coordinate were annotated to the respective genes. Variants extracted from exome data with VCF Tools59 were merged with genotype SNPs identified with PLINK60. Genetic ancestry per gene was determined as the mean local ancestry values calculated at SNP positions within each gene region. When no SNP determined local ancestry was within the region of the gene, the gene borders were interactively increased by 1 kb in order to search for average ancestry in its neighborhood (Additional file 2: Table S1). Spearman correlation was determined between the local ancestry components (AFR, AMR, EUR) per each 90 B12-related gene and baseline vitamin B12 levels with the pperm value calculated after 10,000 permutations. The genes were correlated with vitamin B12 levels within a significant level of pperm < 0.1.
The levels of vitamin B12 were associated with variants mapped to previously defined gene regions. The software PLINK 1.960 was used for quality control (QC) analysis excluding variants with Hardy–Weinberg equilibrium with exact test p-values < 10–3 and with minor allele frequency (MAF) < 0.01. The association between the genotypes and vitamin B12 levels were determined with generalized estimating equation (GEE) employing the function geese of geepach R library61 and using as covariates sex, BMI, age, total HEI and mean ancestry components (AFR, AMR, and EUR) per each individual. Vitamin B12 levels were log2 transformed in order to obtain a normal distribution. The SNPs in the 90 B12 related genes were then clumped into blocks of linkage disequilibrium (LD) based on an r2 threshold > 0.2 within a window of 250 kb. For each clumped block the more significant associated SNP was denoted as SNPref. The p-values for the resulting list of SNPref were corrected (padj) for multiple testing using the Benjamini and Hochberg method62 and a significant association was accepted at padj < 0.1.
String v.11 was used to interrogate known protein–protein network associations63 and p-value was adjusted for multiple comparisons (False Discovery Rate—FDR). A significant association was accepted at p < 0.05.
Polygenic risk score (PRS) construction
SNPs within B12-related genes that were significantly associated with vitaminB12 levels in GEE analysis were used to construct a PRS. The PRS was computed by summing the number of risk alleles (0, 1, or 2) weighted by the effect size from GEE results64.
Stratification of individuals into PRS terciles
The individuals were stratified into terciles based on the PRS in order to determine how variables related to individuals, such as age, sex, body mass index, total HEI, meat and milk intake (HEI components), socioeconomic status and levels of vitamin B12, tHcy, folate, pyridoxal and riboflavin, are distributed across the terciles. Kruskal–Wallis was employed to evaluate the significance of differences among terciles within a significance level of 0.05.
Ethics approval and consent to participate
The study was approved by the internal ethics committee (Process HCRP No. 14255/2010) and by the National Research Ethics Commission (No. 00969412.6CAAE.0000.5440). The trial was registered on ClinicalTrials.gov (NCT01823744). The participants were informed about the purpose and procedures of the study and signed a statement of informed consent. Parents of each participant signed informed consent.
Results
Ancestry along the genome and their per gene correlation with phenotype
After removing siblings and outliers of clinical and vitamins levels, 168 participants were considered for analysis in the present study. The investigation of local ancestry along the genome was motivated by previous results17 that showed a negative association between baseline levels of vitamin B12 with AMR. First, the mean values for the local ancestry components AFR, AMR and EUR calculated for each individual were compared with global ancestry from ADMIXTURE tool to check for consistency of the ancestry levels obtained in Mathias et al.17. This comparison shows that both tools generated the same patterns of ancestry contributions (Additional file 1: Figure S1).
The highest ancestral contribution throughout the autosomal chromosomes is from EUR with mean values ranging from 0.54 to 0.67, followed by AFR (0.18–0.31), and AMR (0.10–0.21) (Additional file 1: Figure S2), consistent with the overall genome-wide mean ancestry for each component.
The decomposition of ancestry components along the genome was used to identify regions associated with variations in the levels of vitamin B12. For this purpose, the local ancestry along the chromosomes for all children were correlated with vitamin B12 resulting in correlation scores as defined in methods (Fig. 1a). The results showed that baseline vitamin B12 levels were driven by positive correlations with EUR and negatively correlated with AFR and AMR ancestries. These associations corroborated the observed negative association with AMR in Mathias et al.17 study. These results also revealed that genetic local ancestry has an important role in vitamin B12 levels.
We also determined the mean ancestry for regions encoding each of the 90 genes involved in B12 metabolism and the genes that carried ancestral components most correlated with variations in vitamin B12 levels (Table 1). Significant correlations were observed for 51 (of 90 vitamin B12 related genes) gene regions with negative signals for both AFR and AMR and positive for EUR (Fig. 1b), consistent with local correlations previously observed. The number of gene regions correlated with AFR, AMR and EUR ancestries were 17, 28, and 39, respectively.
Association between variants in the set of 90 B12-related genes and B12 levels
Of a total of 3,406,465 SNP variants from genotype and exome analysis, 6699 were found to be present in the 90 B12-related genes. These SNPs were grouped in 1971 clumps where the most significant SNPs from GEE were considered as the representative SNP of the clump (SNPref). After multiple testing correction of p-values obtained from GEE for these 1971 SNPref, thirty-six variants in 26 genes were associated with the B12 baseline vitamin levels at padj < 0.1 (Table 2). The most significant SNPref (kgp6592612, padj = 4.22 × 10–12) was in FUT6 gene (Chr19) which showed a negative association with vitamin B12 levels. Fifteen SNPref showed a negative association with phenotype, while 21 showed a positive association. Among the SNPref, 16 had no SNP in LD, whereas 6 had one SNP in LD, 3 had two SNP in LD and 11 had 3 or more SNPs in LD.
Polygenic risk score construction
Thirty-six SNPref from the additive model (Fig. 2a) were used to calculate a polygenic risk score. PRS explained 42% of phenotype (Fig. 2b). Figure 2c shows individuals in ascending order of PRS.
Although all the genes selected for this study were all associated with vitamin B12 metabolism, transport, or biochemistry, we used String version 11.063 to visualize protein interactions among the 26 genes used to constitute the PRS. A protein–protein enrichment p-value lower than 1.0 × 10–16 was found for the whole network (Fig. 3; Additional file 3: Excel files Supplement Material). Interacting proteins in the network were associated with many biological processes, including direct cobalamin metabolic processes (with FDR p-value equal to 6.74 × 10–10 for ABCD4, CUBN, LMBRD1, MTR, TCN1, TCN2) and cobalamin transport (p-value 1.73 × 10–7 for CUBN, LMBRD1, TCN1, TCN2), as expected. However, less significant but nonetheless interesting interactions suggest that vitamin B12 levels may impact lipid metabolism (p-value 0.0053 for CUBN, MVK, NDUFAB1, PEMT, SLC27A4) and coated pit processes (p-value 0.024 for CUBN, LRP2). During the first steps of the vesicle-mediated membrane transport, coated pits are internalized to form coated vesicles which transport proteins between distinct membrane-bound organelles. Vitamin B12 levels may also have a role in vitamin D metabolic process primarily through the low-density lipoprotein receptor-related protein 2 (p-value 0.009 for CUBN, LRP2) and in alpha-amino acid metabolic process (p-value 0.0005 for CBS, FPGS, MTR, PEMT, SARS).
We stratified PRS of individuals into terciles to assess its association with age, sex, BMI, total HEI, meat and milk intake (HEI components), socioeconomic status and levels of vitamin B12, tHcy, folate, pyridoxal and riboflavin, in order to build a hypothetical nutritional counseling diagram (Table 3; Fig. 4). In the total studied individuals, only 7% (n = 12) had low vitamin B12 levels, 45% (75) had normal values and 48% (81) had high levels.
Discussion
The development of broad public health strategies for disease prevention requires the identification of risk factors. Vitamin B12 insufficiency may be a risk factor that contributes to the substantial burden of diseases in the general population. Literature and database searches identified 90 genes related with vitamin B12 basal levels. We determined the genetic ancestry of each of these genes in individuals in the study population: thirty-nine genes were positively correlated with EUR ancestry, twenty-eight genes with Amerindian ancestry, and seventeen genes in regions of AFR ancestry. Brazilians share EUR, AMR and AFR ancestries65 along with minor contributions from many other geographic areas. Our results emphasize the contribution of individual genetic ancestry as a predictor for a clinically relevant phenotype and perhaps disease onset21.
Individual SNPs may contribute only modestly to a phenotype but in combination could explain a significant portion of the variation in phenotype incidence in the general population45. From the set of 36 SNPs associated with vitamin B12, FUT6, the gene encoding for a fucosyltransferase (an enzyme that allows for the addition of fucose to oligosaccharides) was the most significant. Polymorphisms in FUT genes have been shown to influence the quantity and quality of Human Milk Oligosaccharides (HMOs) in breast milk, and HMOs are exerting a prebiotic as well as immunomodulatory effects on the infant gut66. A polymorphism in an enzyme of the same family, FUT2, has been associated to susceptibility to Heliobacter pylori infection in Amazonian children67. Thus, bacterial overgrowth may be one of causes leading to vitamin B12 deficiency.
Taken together, the combined set of 36 SNPs statistically associated with vitamin B12 levels in this study were used to create a PRS to predict vitamin B12 levels. The PRS explained 42% of vitamin basal level variation, extending the results reported in Mathias et al.17. The heritability of B12 levels was estimated to be 59% in a study using monozygotic and dizygotic twins, indicating that the magnitude of genetic influence on vitamin B12 levels may be considerable68. Other genetic studies also indicate that vitamin B12 status is a multifactorial trait, where several single-nucleotide polymorphisms (SNPs) in multiple genes interact with the environment to cause the altered B12 status37,69. Genetic variants may alter vitamin B12 tissue status by affecting the proteins involved in vitamin B12 absorption, cellular uptake and intracellular metabolism37,70. The underlying genetic architecture for levels of other vitamins have not been characterized. The use of PRS are becoming clinically highly useful71,72 and the SNPs identified in this study may contribute to variations in B12 levels.
We used a middle out approach for choosing genes involved in vitamin B12 metabolism. Middle out is an emerging approach that uses a predetermined subset of high-dimensional data that are limited to a system of interest44.
Multiple variants in a subset of these genes collectively explained more of the variation of vitamin level than could be explained by a single variant. Many recent Mendelian randomization investigations of complex traits (such as blood pressure, BMI, or blood lipids) have used multiple variants73,74 and the same can apply to vitamin levels. This middle out approach and subsequent analysis identified 26 genes encoding 36 SNPs associated with B12 levels. Functional interactions among these genes were analyzed using String63. Interactions among LMBRD1 (lysosomal cobalamin transport escort protein), CUBN (cubilin), TCN1 (transcobalamin 1), TCN2 (transcobalamin 2), and ABCD4 (lysosomal cobalamin transporter ABCD4) were expected since they are involved in the transport or in the lysosomal release of vitamin B12 into the cytoplasm. Disruptions in the transport machinery of vitamin B12 have then consequences on its bioavailability. While mutations in TNC2 may lead to anemia and profound neurological disorders, TNC2 polymorphisms may lead to differential blood concentration of transcobalamin29. Given the link between vitamin B12 and the methionine cycle, the genes CUBN, ABCD4 with MTR (exosome RNA helicase) were identified, as these participate in reactions that catalyze the transfer of a methyl group from methyl-cobalamin to homocysteine63. Interactions in a smaller subset of the identified genes are involved in lipid metabolism, which links to another metabolic role of vitamin B12, the involvement in mitochondrial metabolism: (i) NDUFAB1 (mitochondrial acyl carrier) is an acyl carrier protein of the growing fatty acid chain in fatty acid biosynthesis, (ii) MVK (mevalonate kinase) is a regulatory site in cholesterol biosynthetic pathway, (iii) PEMT (phosphatidylethanolamine N-methyltransferase) catalyzes the three sequential steps of the methylation pathway involving phosphatidylethanoamine (PE), phosphatidylmonomethylethanolaimne (PMME), phosphatidyldimethylethanolamine (PDME), phosphatidylcholine (PC), (iv) SLC27A4 (long-chain fatty acid transport protein 4) is a long-chain fatty acid transport protein, and (v) CUBN is a cotransporter lipoprotein metabolism63. In addition, megalin (LRP2, low-density lipoprotein receptor-related protein 2) acts together with CUBN to mediate endocytosis of high-density lipoproteins63. Intestinal vitamin B12 malabsorption is found in the rare disease Imerslund–Gräsbeck syndrome, that entails mutations in cubilin and amnionless (AMN)29. These potential interactions are consistent with a system view of metabolic processes in that genes involved in B12 metabolism also participate in other physiological processes.
Torkamani et al.46 proposed that PRS can be used for three classes of interventions: PRS-informed therapeutic intervention (the part that PRS can play in the selection of interventions to treat or prevent disease); PRS-informed disease screening (the role that PRS can have in the decision to initiate and the interpretation of disease screens); and PRS-informed life planning (the personal utility that PRSs can provide, even in the absence of preventive actions). Based on this categorization, we propose a hypothetical nutritional counseling based on PRS terciles, nutritional, and demographic data assuming that PRS can play a role in helping to screen, to prevent insufficiencies/deficiencies or to select interventions46. In addition, nutritional counseling based on PRS and considering cofounding variables may individualize management of vitamin B12 recommendation and personalize health care.
Strength of the study
From 6699 SNPs analyzed by genotyping and exome analysis within the 90 B12 related genes, 36 were associated with vitamin B12 levels after correcting for multiple testing and adjusting for covariates such as age, sex, BMI, quality of diet and genetic ancestry, as suggested by Goetz et al.21. Including the covariates may have increased statistical power. Some studies also adjust for other characteristics, including directly measured traits (such as age and sex)74. Association between a genetic variant and outcome may also be confounded by hidden population structure. This can be addressed by adjusting for genetic ancestry74 as done in this study. The 36 SNPs used to create the PRS may improve the accuracy and reliability of it.
Moreover, the resulting significant SNPs obtained from GEE were employed to clump (i.e., to create haplotypes) of variants within a window of 250 kb that are in LD in order to find the more significant SNPs for each clump. Associations accounting for LD have been shown to improve the performance of PRS models in some settings75.
Others have found that any significant association identified between genetic ancestry and disease or clinical condition is greatly attenuated after controlling for socioeconomic status and others environmental factors21,76,77. In the current study, socioeconomic status was similar among the PRS terciles.
Our results are consistent with the concept21 that genetic background impacts clinically relevant intermediate phenotypes, one of the first examples of which was the effect of ethnicity on a haplotype containing LTA4H on myocardial risk78. These types of results should motivate further research in the field of nutrition.
Limitations of the study
One-carbon metabolism is influenced by a variety of nutrients that interact and can compensate for one another when there is a deficiency of a single nutrient. The potential role of other micronutrients and metabolites in one-carbon metabolism, such as homocysteine, folate, riboflavin, and pyridoxine need to be examined, as they may influence both vitamin B12 metabolism26,79 and, consequently, the power of PRS. Although we did not adjust the PRS for these metabolites, the categorization of individuals into terciles showed no statistical difference for those B-vitamins. tHcy levels in RBC were higher in the first tercile but still within normal ranges.
We evaluated plasma B12 levels. Serum or plasma B12 measurement has advantages of (i) being widely available and inexpensive, (ii) not greatly affected by recent intake, (iii) concentrations within an individual are relatively constant, (iv) there is no need for fasting before sample collection and (v) is not influenced by age or infection, (vi) but are not overly-sensitive or -specific which would produce false positive and negative diagnoses80. A second indicator, such as serum methylmalonic acid (MMA) and/or holo-transcobalamin, should be assessed to improve problems with sensitivity and specificity81, which was not evaluated in the present study. The cut-off for vitamin B12 deficiency chosen was < 148 pmol/L, which corroborates other studies32,80,81,82,83. However, no consensus exists for a definition of a cut-off point low vitamin B12 levels in children and adolescents30,35,80.
The identification of genetic variants that contribute to more complex multifactorial conditions with a polygenic risk score may not be useful for predicting phenotype incidence rates in another population because the SNPs identified in one population may have different frequencies in a different population21. Therefore, PRS must be tested in populations with different genetic ancestries. In addition, greater uncertainty exists when using PRS because some of these SNPs may only be correlated with the causal genetic factor or factors. This may reduce the generalizability of PRS risk estimates in populations beyond the population studied. Missing heritability, the unknown component of genetic risk, is another source of uncertainty specific to polygenic risk estimates46.
The present study was based on small sample size for genetic analysis and considered only 3 major globally admixed populations. While the p-values for these SNPs were above the typical GWAS threshold of 5 × 10–8, the strategy was to identify the combinations of low effect SNPs that could be associated with B12 levels. Despite the inherent limitations of GWAS-PRS that are associated with sample size, the middle out approach used here investigated a possible PRS constructed on a given set of known phenotype related genes/SNPs data45. The ability to adjust statistical correlations for covariates, test other phenotypic variables (e.g., tHcy in RBC), and develop nutritional counseling strategies was a direct result of the depth of phenotypic analysis17 that was conducted in this study, a trade-off for larger study populations usually used for genetic studies.
The results presented should be validated in other studies and might serve as a guide for future studies of larger numbers of participants. We propose an approach to motivate further research and help move the biomedical research community towards greater sensitivity to global issues in population health.
Conclusion and future considerations
The basic components of disease risk are genetic susceptibility, environmental exposures, and lifestyle factors and these factors are often considered separately. Our study is consistent with the concept that genetic background impacts vitamin levels. The middle out approach used here investigated a possible PRS to predict vitamin B12 levels that explained 42% of vitamin basal level variation. Its performance should be evaluated in an independent sample, or using cross validation techniques. In the near future, large scale sequencing and imputation-based association studies will provide additional data and a more comprehensive assessment of the role of rare and low frequency variants that may confer more moderate risk to diseases45. Improvements in models through the incorporation of polygenic risk and possibly other predictive factors, may identify individuals at different levels of risk for developing diseases. Such data could be translated into improvements in primary and secondary prevention by tailoring interventions according to risk45 such as those within the tercile of genetic risk of low levels of vitamin B12 that are at increased risk. These types of considerations would strengthen the field of nutrition by creating more effective applications of nutrition to help public health.
Data availability
The datasets supporting the conclusions of this article are included within the article and its additional files. The raw data are available from the corresponding author on reasonable request.
Abbreviations
- AFR:
-
African
- AMR:
-
Native American
- BMI:
-
Body mass index
- Chr:
-
Chromosome
- ELAI:
-
Efficient Local Ancestry Inference
- EUR:
-
European
- FDR:
-
False Discovery Rate
- GEE:
-
Generalized estimating equation
- GRCh37:
-
Genome Reference Consortium Human genome build 37
- GWAS:
-
Genome-wide association study
- HCFMRP-USP:
-
Ribeirão Preto Medical School Hospital
- HEI:
-
Healthy eating index
- HGDP:
-
Human Genome Diversity Project
- LC–MS/MS:
-
Liquid chromatography tandem mass spectrometry
- LD:
-
Linkage disequilibrium
- LTA4H:
-
Leukotriene A4 hydrolase
- MAF:
-
Minor allele frequency
- PRS:
-
Polygenic risk score
- PRS-T:
-
Polygenic risk score tercile
- QC:
-
Quality control
- SD:
-
Standard deviation
- SNP:
-
Single nucleotide polymorphism
- RBC:
-
Red blood cell
- WHO:
-
World Health Organization
References
Pavlovic, M., Prentice, A., Thorsdottir, I., Wolfram, G. & Branca, F. Challenges in harmonizing energy and nutrient recommendations in Europe. Ann. Nutr. Metab. 51, 108–114 (2007).
Institute of Medicine. The Development of DRIs 1994–2004: Lessons Learned and New Challenges: Workshop Summary (National Academies Press, 2008).
Committee, D. G. A. Scientific Report of the 2020 Dietary Guidelines Advisory Committee Advisory Report to the Secretary of Agriculture and Secretary of Health and Human Services (U.S. Department of Agriculture, 2020).
Monteiro, J. P., Kussmann, M. & Kaput, J. The genomics of micronutrient requirements. Genes Nutr. 10, 466 (2015).
Van Ommen, B. et al. The Micronutrient Genomics Project: A community-driven knowledge base for micronutrient research. Genes Nutr. 5, 285–296 (2010).
Bouwman, J. et al. Visualization and identification of health space, based on personalized molecular phenotype and treatment response to relevant underlying biological processes. BMC Med. Genomics 5, 1 (2012).
Kaput, J. & Rodriguez, R. L. Nutritional genomics: the next frontier in the postgenomic era. Physiol. Genomics 16, 166–177 (2004).
Welter, D. et al. The NHGRI GWAS Catalog, a curated resource of SNP-trait associations. Nucleic Acids Res. 42, D1001 (2014).
MacArthur, J. et al. The new NHGRI-EBI Catalog of published genome-wide association studies (GWAS Catalog). Nucleic Acids Res. 45, D896–D901 (2017).
Nilsson, P. D., Newsome, J. M., Santos, H. M. & Schiller, M. R. Prioritization of variants for investigation of genotype-directed nutrition in human superpopulations. Int. J. Mol. Sci. 20, 3516 (2019).
Kaput, J., Perozzi, G., Radonjic, M. & Virgili, F. Propelling the paradigm shift from reductionism to systems nutrition. Genes Nutr. 12, 3 (2017).
Mitchell, E. S., Conus, N. & Kaput, J. B vitamin polymorphisms and behavior: Evidence of associations with neurodevelopment, depression, schizophrenia, bipolar disorder and cognitive decline. Neurosci. Biobehav. Rev. 47, 307–320 (2014).
Kaput, J. Lessons from application of data science strategies in nutritional research. Am. J. Clin. Nutr. 111, 4–5 (2020).
Kaput, J. et al. Consensus statement—understanding health and malnutrition through a systems approach: The ENOUGH program for early life. Genes Nutr. 9, 378 (2014).
Coelho-Landell, C. A. et al. Metabo groups in response to micronutrient intervention: Pilot study. Food Sci. Nutr. 8, 683–693 (2020).
Chakrabarti, A. et al. Impact of multi-micronutrient supplementation on lipidemia of children and adolescents. Clin. Nutr. 39, 2211–2219 (2020).
Mathias, M. G. et al. Clinical and vitamin response to a short-term multi-micronutrient intervention in Brazilian children and teens: from population data to interindividual responses. Mol. Nutr. Food Res. 62, 1700613 (2018).
Hillesheim, E. et al. Biomarker-based validity of a food frequency questionnaire estimating intake in Brazilian children and adolescents. Int. J. Food Sci. Nutr. 72, 1–12 (2020).
Barros, T. T. et al. DNA damage is inversely associated to blood levels of DHA and EPA fatty acids in Brazilian children and adolescents. Food Funct. 11, 5115–5121 (2020).
Toffano, R. et al. Validation of the Brazilian healthy eating index-revised using biomarkers in children and adolescents. Nutrients 10, 154 (2018).
Goetz, L. H., Uribe-Bruce, L., Quarless, D., Libiger, O. & Schork, N. J. Admixture and clinical phenotypic variation. Hum. Hered. 77, 73–86 (2014).
Scott, J. M. & Molloy, A. M. The discovery of vitamin B12. Ann. Nutr. Metab. 61, 239–245 (2012).
Rush, E. C., Katre, P. & Yajnik, C. S. Vitamin B12: One carbon metabolism, fetal growth and programming for chronic disease. Eur. J. Clin. Nutr. 68, 2–7 (2014).
Morris, M. S. The role of B vitamins in preventing and treating cognitive impairment and decline. Adv. Nutr. 3, 801–812 (2012).
Briani, C. et al. Cobalamin deficiency: Clinical picture and radiological findings. Nutrients 5, 4521–4539 (2013).
Venkatramanan, S., Armata, I. E., Strupp, B. J. & Finkelstein, J. L. Vitamin B-12 and cognition in children. Adv. Nutr. 7, 879–888 (2016).
Iglesia, I. et al. Folate and vitamin B12 concentrations are associated with plasma DHA and EPA fatty acids in European adolescents: the Healthy Lifestyle in Europe by Nutrition in Adolescence (HELENA) study. Br. J. Nutr. 117, 124–133 (2017).
Ued, F. V. et al. Vitamin B2 and folate concentrations are associated with ARA, EPA and DHA fatty acids in red blood cells of Brazilian children and adolescents. Nutrients 11, 2918 (2019).
Green, R. et al. Vitamin B12 deficiency. Nat. Rev. Dis. Prim. 3, 17040 (2017).
Cardoso, M. A., Scopel, K. K. G., Muniz, P. T., Villamor, E. & Ferreira, M. U. Underlying factors associated with anemia in amazonian children: A population-based, cross-sectional study. PLoS One 7, e36341 (2012).
Garcia, M. T., Granado, F. S. & Cardoso, M. A. Complementary feeding and nutritional status of 6–24-month-old children in Acrel̈ndia, Acre State, Western Brazilian Amazon. Cad. Saude Publica 27, 305–316 (2011).
Silva, L. L. S., Fawzi, W. W. & Cardoso, M. A. Serum folate and Vitamin B 12 status in young Brazilian children. Public Health Nutr. 22, 1223–1231 (2019).
Britto, J. C., Cançado, R. & Guerra-Shinohara, E. M. Concentrations of blood folate in Brazilian studies prior to and after fortification of wheat and cornmeal (maize flour) with folic acid: A review. Rev. Bras. Hematol. Hemoter. 36, 275–286 (2014).
Paul, L. & Selhub, J. Interaction between excess folate and low vitamin B12 status. Mol. Aspects Med. 53, 43–47 (2017).
Barnabé, A. et al. Folate, Vitamin B12 and Homocysteine status in the post-folic acid fortification era in different subgroups of the Brazilian population attended to at a public health care center. Nutr. J. 14, 19 (2015).
Pletscher-Frankild, S., Pallejà, A., Tsafou, K., Binder, J. X. & Jensen, L. J. DISEASES: Text mining and data integration of disease-gene associations. Methods 74, 83–89 (2015).
Surendran, S. et al. An update on vitamin B12-related gene polymorphisms and B12 status. Genes Nutr. 13, 2 (2018).
Grarup, N. et al. Genetic architecture of vitamin B12 and folate levels uncovered applying deeply sequenced large datasets. PLoS Genet. 9, e1003530 (2013).
Zinck, J. W. R., De Groh, M. & MacFarlane, A. J. Genetic modifiers of folate, vitamin B-12, and homocysteine status in a cross-sectional study of the Canadian population. Am. J. Clin. Nutr. 101, 1295–1304 (2015).
Lin, X. et al. Genome-wide association study identifies novel loci associated with serum level of vitamin B12 in Chinese men. Hum. Mol. Genet. 21, 2610–2617 (2012).
Hazra, A. et al. Genome-wide significant predictors of metabolites in the one-carbon metabolism pathway. Hum. Mol. Genet. 18, 4677–4687 (2009).
Nongmaithem, S. S. et al. GWAS identifies population-specific new regulatory variants in FUT6 associated with plasma B12 concentrations in Indians. Hum. Mol. Genet. 26, 2551–2564 (2017).
Hazra, A. et al. Common variants of FUT2 are associated with plasma vitamin B12 levels. Nat. Genet. 40, 1160–1162 (2008).
Monteiro, J. P. et al. Methylation potential associated with diet, genotype, protein, and metabolite levels in the Delta Obesity Vitamin Study. Genes Nutr. 9, 403 (2014).
Chatterjee, N., Shi, J. & García-Closas, M. Developing and evaluating polygenic risk prediction models for stratified disease prevention. Nat. Rev. Genet. 17, 392–406 (2016).
Torkamani, A., Wineinger, N. E. & Topol, E. J. The personal and clinical utility of polygenic risk scores. Nat. Rev. Genet. 19, 581–590 (2018).
de Onis, M. et al. Development of a WHO growth reference for school-aged children and adolescents. Bull. World Health Organ. 85, 660–667 (2007).
Associação Brasileira de Empresas de Pesquisa. Critério de Classificação Econômica Brasil. (Associação Brasileira de Empresas de Pesquisa, 2011).
Centers for Disease Control and Prevention & National Center for Environmental Health. Second National Report on Biochemical Indicators of Diet and Nutrition in the U.S. Population, 2012. (Createspace Independent, 2014).
Da Silva, L. et al. High-throughput method for the quantitation of metabolites and co-factors from homocysteine–methionine cycle for nutritional status assessment. Bioanalysis 8, 1937–1949 (2016).
Meisser, K., Longet, K., Bénet, S., Munari, C. & Campos-giménez, E. Simultaneous quantification of 21 water soluble vitamin circulating forms in human plasma by liquid chromatography-mass spectrometry. J. Chromatogr. A 1422, 89–98 (2015).
Illumina. Infinium LCG Assay Protocol Guide. https://support.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/infinium_assays/infinium_lcg_quad_assay/infinium-lcg-quad-assay-guide-15025908-d.pdf (2015).
Church, D. M. et al. Modernizing reference genome assemblies. PLoS Biol. 9, e1001091 (2011).
Guan, Y. Detecting structure of haplotypes and local ancestry. Genetics 196, 625–642 (2014).
Li, J. Z. et al. Worldwide human relationships inferred from genome-wide patterns of variation. Science 319, 1100–1104 (2008).
Altshuler, D. M. et al. An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56–65 (2012).
Alexander, D. H., Novembre, J. & Lange, K. Fast model-based estimation of ancestry in unrelated individuals. Genome Res. 19, 1655–1664 (2009).
Agarwala, R. et al. Database resources of the National Center for Biotechnology Information. Nucleic Acids Res. 44, D7–D19 (2016).
Danecek, P. et al. The variant call format and VCFtools. Bioinformatics 27, 2156–2158 (2011).
Chang, C. C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 4, 7 (2015).
Halekoh, U., Højsgaard, S. & Yan, J. The R package geepack for generalized estimating equations. J. Stat. Softw. 15, 1–11 (2006).
Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B 57, 289–300 (1995).
Szklarczyk, D. et al. STRING v11: Protein–protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 47, D607–D613 (2019).
Lewis, C. M. & Vassos, E. Polygenic risk scores: From research tools to clinical instruments. Genome Med. 12, 44 (2020).
Lins, T. C. et al. Association of serum lipid components and obesity with genetic ancestry in an admixed population of elderly women. Genet. Mol. Biol. 35, 575–582 (2012).
Ray, C. et al. Human milk oligosaccharides: The journey ahead. Int. J. Pediatr. 2019, 1–8 (2019).
Cobayashi, F., Tomita, L. Y., Augusto, R. A., D’Almeida, V. & Cardoso, M. A. Genetic and environmental factors associated with vitamin B12status in Amazonian children. Public Health Nutr. 18, 2202–2210 (2015).
Nilsson, S. E., Read, S., Berg, S. & Johansson, B. Heritabilities for fifteen routine biochemical values: Findings in 215 Swedish twin pairs 82 years of age or older. Scand. J. Clin. Lab. Invest. 69, 562–569 (2009).
Haggarty, P. B-vitamins, genotype and disease causality. Proc. Nutr. Soc. 66, 539–547 (2007).
Quadros, E. V. Advances in the understanding of cobalamin assimilation and metabolism. Br. J. Haematol. 148, 195–204 (2010).
Dudbridge, F. Power and predictive accuracy of polygenic risk scores. PLoS Genet. 9, e1003348 (2013).
Chatterjee, N. et al. Projecting the performance of risk prediction based on polygenic analyses of genome-wide association studies. Nat. Genet. 45, 400–405 (2013).
Benn, M. & Nordestgaard, B. G. From genome-wide association studies to Mendelian randomization: Novel opportunities for understanding cardiovascular disease causality, pathogenesis, prevention, and treatment. Cardiovasc. Res. 114, 1192–1208 (2018).
Davies, N. M., Holmes, M. V. & Davey Smith, G. Reading Mendelian randomisation studies: A guide, glossary, and checklist for clinicians. BMJ 362, k601 (2018).
Vilhjálmsson, B. J. et al. Modeling linkage disequilibrium increases accuracy of polygenic risk scores. Am. J. Hum. Genet. 97, 576–592 (2015).
Florez, J. C. et al. Strong association of socioeconomic status with genetic ancestry in Latinos: Implications for admixture studies of type 2 diabetes. Diabetologia 52, 1528–1536 (2009).
Campbell, D. D. et al. Amerind ancestry, socioeconomic status and the genetics of type 2 diabetes in a Colombian population. PLoS One 7, e33570 (2012).
Helgadottir, A. et al. A variant of the gene encoding leukotriene A4 hydrolase confers ethnicity-specific risk of myocardial infarction. Nat. Genet. 38, 68–74 (2006).
Swaminathan, S., Thomas, T. & Kurpad, A. V. B-vitamin interventions for women and children in low-income populations. Curr. Opin. Clin. Nutr. Metab. Care 18, 295–306 (2015).
Allen, L. H. et al. Biomarkers of nutrition for development (BOND): vitamin B-12 review. J. Nutr. 148, 1995S-2027S (2018).
Hannibal, L. et al. Biomarkers and algorithms for the diagnosis of vitamin B12 deficiency. Front. Mol. Biosci. 3, 27 (2016).
Carmel, R. Biomarkers of cobalamin (vitamin B-12) status in the epidemiologic setting: A critical overview of context, applications, and performance characteristics of cobalamin, methylmalonic acid, and holotranscobalamin II. Am. J. Clin. Nutr. 94, 348S (2011).
Finkelstein, J. L. et al. Vitamin B 12 status in pregnant adolescents and their infants. Nutrients 11, 397 (2019).
Acknowledgements
The authors thank the children, teens, and parents who participated in this study as well as the principals, teachers, and school district officials who made the schools available for activities related to the project.
Funding
Funding was provided by the Nestle Research (Lausanne, Switzerland) (contract reference RDHS 000054) and by Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (Process 2012/20421-8). This manuscript is the 10th publication from this study.
Author information
Authors and Affiliations
Contributions
J.P.M. and J.K. formulated the research question and the study design; C.A.F., F.V.U., S.M., O.C., S.M., J.C., S.P., A.C., P.D., J.K. and J.P.M. performed data collection. C.A.F., O.C., R.T., W.A.S.J., J.C., P.D., J.K. and J.P.M. analyzed and interpreted the data. C.A.F., F.V.U., J.K. and J.P.M. wrote the original draft, edited and revised the manuscript. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
J.K. was employee at Nestlé Research and participated in the experimental design of the study and in writing the final manuscript. He is currently employed by Vydiant, Folson, CA, USA. S.M., J.C., O.C., S.M., S.P., A.C. and P.D. are employees of Nestlé Research and did not participate in the experimental design. All the others authors declare no conflict of interest.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Fuzo, C.A., da Veiga Ued, F., Moco, S. et al. Contribution of genetic ancestry and polygenic risk score in meeting vitamin B12 needs in healthy Brazilian children and adolescents. Sci Rep 11, 11992 (2021). https://doi.org/10.1038/s41598-021-91530-7
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-021-91530-7
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.