The most common blood type among the population in the United States is O-positive. Around 53 percent of the Latino-American population in the U.S. has blood type O-positive, while only around 37 percent of the Caucasian population has this blood type. The second most common blood type in the United States is A-positive. Around 33 percent of the Caucasian population in the United States has A-positive blood type. Blood type O-negative Those with blood type O-negative are universal donors as this type of blood can be used in transfusions for any blood type. O-negative blood type is most common in the U.S. among Caucasian adults. Around eight percent of the Caucasian population has type O-negative blood, while only around one percent of the Asian population has this blood type. Only around seven percent of all adults in the United States have O-negative blood type. Blood Donations The American Red Cross estimates that someone in the United States needs blood every two seconds. However, only around three percent of age-eligible people donate blood yearly. The percentage of adults who donated blood in the United States has not fluctuated much for the past two decades. In 2021, around 15 percent of U.S. adults donated blood, the same share reported in the year 2003.

The eight main blood types are A+, A-, B+, B-, O+, O-, AB+, and AB-. The most common blood type in the United States is O-positive, with around 38 percent of the population having this type of blood. However, blood type O-positive is more common in Latino-Americans than other ethnicities, with around 53 percent of Latino-Americans with this blood type, compared to 47 percent of African Americans and 37 percent of Caucasians. Blood donation The American Red Cross estimates that every two seconds someone in the United States needs blood or platelets, highlighting the importance of blood donation. It was estimated that in 2021, around 6.5 million people in the U.S. donated blood, with around 1.7 million of these people donating for the first time. Those with blood type O-negative are universal blood donors, meaning their blood can be transfused for any blood type. Therefore, this blood type is the most requested by hospitals. However, only about seven percent of the U.S. population has this blood type. Blood transfusion Blood transfusion is a routine procedure that involves adding donated blood to a patient’s body. There are many reasons why a patient may need a blood transfusion, including surgery, cancer treatment, severe injury, or chronic illness. In 2021, there were around 10.76 million blood transfusions in the United States. Most blood transfusions in the United States occur in an inpatient medicine setting, while critical care accounts for the second highest number of transfusions.

The U antigen (MNS5) is one of 49 antigens belonging to the MNS blood group system (ISBT002) carried on glycophorins A (GPA) and B (GPB). U is present on the red blood cells in almost all Europeans and Asians but absent in approximately 1.0% of Black Africans. U negativity coincides with negativity for S (MNS3) and s (MNS4) on GPB, thus be called S–s–U–, and is thought to arise from homozygous deletion of GYPB. Little is known about the molecular background of these deletions. Bioinformatic analysis of the 1000 Genomes Project data revealed several candidate regions with apparent deletions in GYPB. Highly specific Gap-PCRs, only resulting in positive amplification from DNAs with deletions present, allowed for the exact genetic localization of 3 different breakpoints; 110.24- and 103.26-kb deletions were proven to be the most frequent in Black Americans and Africans. Among 157 CEPH DNAs, deletions in 6 out of 8 African ethnicities were present. Allele frequencies of the deletions within African ethnicities varied greatly and reached a cumulative 23.3% among the Mbuti Pygmy people from the Congo. Similar observations were made for U+var alleles, known to cause strongly reduced GPB expression. The 110- and 103-kb deletional GYPB haplotypes were found to represent the most prevalent hereditary factors causative of the MNS blood group phenotype S–s–U–. Respective GYPB deletions are now accessible by molecular detection of homo- and hemizygous transmission.

We examined the association between DNA methylation in circulating leukocytes and blood lipids in a multi-ethnic sample of 16,265 subjects. We identified 148, 35, and 4 novel associations among Europeans, African Americans, and Hispanics, respectively, and an additional 186 novel associations through a trans-ethnic meta-analysis. We observed a high concordance in the direction of effects across racial/ethnic groups, a high correlation of effect sizes between high-density lipoprotein and triglycerides, a modest overlap of associations with epigenome-wide association studies of other cardio-metabolic traits, and a low overlap with lipid loci identified to date through genome-wide association studies. Thirty CpGs reached significance in at least 2 racial/ethnic groups including 7 that showed association with the expression of an annotated gene. CpGs annotated to CPT1A showed evidence of being influenced by triglycerides levels. DNA methylation levels of circulating leukocytes show robust and consistent association with blood lipid levels across multiple racial/ethnic groups.

Methods A total of 15 cohorts (N=16,265) from the epigenetics working group in the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) consortium participated in this study. These included the Old Order Amish (OOA), Atherosclerosis Risk in Communities (ARIC), Bogalusa Heart Study (BHS), Cardiovascular Health Study (CHS), Framingham Heart Study (FHS), Genetic Epidemiology Network of Arteriopathy (GENOA), Genetics of Lipid Lowering Drugs and Diet Network (GOLDN), Hypertension Genetic Epidemiology Network (HyperGEN), Cooperative health research in the Region of Augsburg (KORA), Normative Aging Study (NAS), Prospective Investigation of Vascularity of Uppsala Elders Study (PIVUS), Rotterdam Study (RS), UK Adult Twin Registry (TwinsUK), Women’s Health Initiative Broad Agency Announcement 23 (WHI-BA23), and the Women’s Health Initiative Epigenetic Mechanisms of PM-Mediated CVD (WHI-EMPC) cohorts. Four cohorts, BHS, CHS, WHI-BA23, and WHI-EMPC, examined more than one racial/ethnic group. The total number of cohorts in the European (EA), African (AA), and Hispanic (HA) study populations is 12 (N=11,114), 9 (N=4,452), and 2 (N=699), respectively.

High-density-lipoprotein (HDL, mg per dl) and triglycerides (TG, mg per dl) were directly measured in blood samples taken from participants after at least an 8 hour fast. Low-density-lipoprotein (LDL, mg per dl) was inferred using the Friedewald’s formula in all cohorts except for GOLDN, HyperGEN, and KORA where LDL was measured directly. We did not infer LDL in subjects with triglycerides > 400 mg per dL and we excluded lipid measure from subjects who did not fast for at least 8 hours. We also excluded outliers as defined by >5 standard deviations from the mean of a blood lipid in each cohort. To reduce skewness, HDL and triglycerides were natural log-transformed.

DNA methylation was produced by investigators from each cohort independently. Levels were measured from peripheral blood leukocytes isolated from whole blood in all studies except GOLDN where only CD4 + T cells were examined. The EZ DNA Methylation Gold Kit (Zymo Research, Orange CA) was used for bisulfite conversion. The Illumina® Infinium HumanMethylation450 BeadChip and the Illumina BeadXpress reader were used to perform the methylation assays. Either the SWAN method in the minfi R package, the Beta Mixture Quantile method (BMIQ), the DASEN method in the wateRmelon R package, or the GenomeStudio® Methylation Module was used for pre-processing and normalization of the data in each cohort. For each CpG site, a beta-value was calculated representing the percent methylation at that CpG site. Any single value with a detection p-value > 0.01 was set to missing. In each cohort, we excluded probes with a detection p-value > 0.01 in greater than 5% of samples. In addition, we excluded samples with a detection p-value > 0.01 in greater than 5% of probes. To avoid spurious signals in DNA methylation data, we excluded 29,233 CpGs that co-hybridize to alternate genomic sequences (highly homologous to the intended targets)

Epigenome-wide association analyses (EWAS) were performed in each cohort stratified by racial/ethnic group (European, African, and Hispanic). For Model 1, a linear mixed effects model was used to study the association between the DNA methylation level of a CpG (dependent variable) and each of the lipid measures (independent variable; HDL, LDL, or TG), adjusting for age, sex (reference = male), smoking variable (never/previous/current, reference = never), lipid medication (Yes or No, reference = No), the top four principal components from genotypes (SNPs), and the proportion of 5 types of cells estimated with the Houseman method (CD8 T lymphocytes, CD4 T lymphocytes, natural killer cells, B cells, and monocytes). We added random effects for plate, row, and column. We also included family structure as a random effect among family-based studies. For Model 2, we further adjusted for BMI in addition to Model 1 covariates. We also ran Model 3 and Model 4 which were analogous to Model 1 and 2, respectively, in the subset of individuals not taking lipid lowering medication.

We performed meta-analyses of all the participating cohorts (N=16,265) and also stratified by racial/ethnic group: European Americans (12 cohorts, N=11,114), African Americans (7 cohorts, N=4,452), and Hispanics (2 cohorts, N=699). These meta-analyses were performed for each of the 4 models, respectively. We used a random effects meta-analysis implemented in METASOFT 33 to take into account the heterogeneity of the effect sizes of different cohorts while achieving a higher or comparable statistical power compared to fixed effects meta-analysis. To avoid spurious findings from population substructure, we applied genomic control.

We investigated the association between either imputed or genotyped SNPs located within 25kb upstream or downstream of each CpG and DNA methylation levels to identify cis-acting methylation quantitative trait loci (mQTL). For imputed SNP data, we restricted the mQTL analysis to SNPs with a good quality imputation (IMPUTE info>=0.4 or MACH r^2>=0.3). Subjects taking lipid lowering medications were excluded from this analysis. Beta-values of DNA methylation levels were inverse-normal transformed and regressed on age, sex, smoking (current/former/never), BMI, at least 4 SNP PCs, cell proportions (WBC count and/or estimated WBC proportions (granulocytes as a reference)), and technical covariates (plate, row, and column as random effects) (two-sided test). Family information was also included as a random effect if a cohort was a family-based study. We then regressed the residuals on each SNP of interests stratified by racial/ethnic group. Five out of the 15 cohorts provided genetic data for this analysis including ARIC (NAA=1,717), GOLDN (NEA=713), KORA (NEA=1,379), WHI-BA23 (NEA=790, NAA=540, and NHISP=324), and WHI-EMPC (NEA=494, NAA=424, and NHISP=221). We restricted our analysis to SNPs located within 25 kilobases up- or downstream of the CpGs with a minor allele frequency (MAF) > 0.01 in each cohort and implemented a fixed effects meta-analysis within each of the three racial/ethnic groups.