Array genotyping of transfusion-relevant blood cell antigens in 6946 ancestrally diverse study participants

Abstract

Blood transfusions save millions of lives worldwide each year, yet formation of antibodies against nonself antigens remains a significant problem, particularly in patients who receive frequent transfusions. We designed and tested the Universal Blood Donor Typing (UBDT_PC1) array for automated high-throughput simultaneous typing of human erythrocyte antigens (HEAs), platelet antigens (HPAs), leukocyte antigens (HLAs), and neutrophil antigens to support selection of blood products matched beyond ABO/Rh. Typing samples from 6946 study participants of European, African, Admixed American, South Asian, and East Asian ancestry at 2 different laboratories showed a genotype reproducibility of ≥99% for 17 244 variants, translating to 99.98%, 99.90%, and 99.93% concordance across 338 372 HEA, 53 270 HPA, and 107 094 HLA genotypes, respectively. Compared with previous clinical typing data, concordance was 99.9% and 99.6% for 245 874 HEA and 3726 HPA comparisons, respectively. HLA types were 99.1% concordant with clinical typing across 8130 comparisons, with imputation accuracy higher in Europeans vs non-Europeans. Seven variant RHD alleles, a GYPB deletion underlying the U- phenotype, and 14 high-frequency antigen-negative types were also detected. Beyond blood typing, hereditary hemochromatosis-associated HFE variants were identified in 276 participants. We found that the UBDT_PC1 array can reliably type a wide range of blood cell antigens across diverse ancestries. Reproducibility and accuracy were retained when transfusion-relevant targets from the UBDT_PC1 array were incorporated into the UKBB_v2.2 genome-wide typing array. The results represent the potential for significant advancement toward improved patient care by reducing harm in transfusion recipients through extended matching.

Graphical abstract

Figure 1.

Study design. (A) Samples: number of DNA samples provided by the 7 blood services (NHSBT, SANQUIN, NYBC, ARCLB, CBS, FRCBS, SANBS). (B) Array content: bar plot indicating the number of probes per category in the transfusion module. HLA, HEA, HPA, and HNA. (C) Genotyping: 6946 identical DNA samples were genotyped with the UBDT_PC1 Transfusion Array at Sanquin and NYBC, with 3938 of these samples also genotyped using the UKBB_v2.2 GWAS array by NHSBT. (D) QC: Heat map gives the reason for, and number of, samples failing QC for the 3 genotyping laboratories. Venn diagrams show overlap in samples that failed Axiom BP QC, gender-vs-sex discordance (sex discordant), and evidence of contamination (contamination). (E) Venn diagram showing the overlap in samples passing QC. (F) Ancestry: (left) bar plot showing genetically inferred ancestry of samples typed successfully by Sanquin and NYBC (6679 samples). EUR, AFR, AMR, SAS, EAS, OTH are shown. Right: heat map showing concordance between the ancestry inferred from the Sanquin and NYBC genotyping results, respectively. AFR, African; AMR, Admixed American; ARCLB, Australian Red Cross Lifeblood; Axiom BP, Axiom Best Practices; CBS, Canadian Blood Services; EAS, East Asian; EUR, European; FRCBS, Finnish Red Cross Blood Service; OTH, Other; SANBS, South African National Blood Service; SANQUIN, Sanquin Blood Supply Foundation; SAS, South Asian.

Figure 2.

Reproducibility of typing results between Sanquin and NYBC for the 6679 DNA samples of the unified data. (A) Genotype reproducibility for 20 681 biallelic probe-variant pairs included in the UBDT_PC1 array design. Reproducibility expressed as percentage of concordant genotype comparisons, and gnomAD MAF for each variant are displayed on the x- and y-axes, respectively. Blue hexagons and red dots on the central scatterplot represent the density of probes with reproducibility of ≥99% and individual probes with <99% concordance, respectively. Marginal histograms show probe counts on a log scale. (B) Correlation of the MAF in EUR study participants vs (non-Finnish) EUR participants from the gnomAD database for each probe-variant pair. Probes with ≥99% and <99% genotype reproducibility are shown in blue and red, respectively. Contour lines represent boundaries of statistical significance with corresponding P values calculated using the χ² test. (C) Genotype reproducibility for critical blood antigen types and iron homeostasis probes. Box plots show the percentage reproducibility between genotypes, split across 2 y-axes ranges to highlight high-reproducibility results (99%-100%) and broader distribution patterns (40%-99%). Data are shown for HEAs, HPAs, HNAs, and iron homeostasis variants in blue, orange, green, and red, respectively. Box plots display the median (center line), interquartile range (IQR; box), whiskers (1.5 × IQR), and outliers (black circles). Outlier variants are annotated with relevant antigen types. (D) Reproducibility between HEA types generated by the Sanquin and NYBC laboratories. The reproducibility is given as a percentage between on the y-axis for the 51 HEA types on the x-axis. Results are stratified for the 5 ancestry groups. When the bars for different ancestries are at identical values, only 1 bar is shown in the order of the legend, that is blue for EUR participants in most cases. (E) The percentage of no-type results is given on the y-axis for the 51 HEA types on the x-axis. HEA types with identical percentage of no-type results are visualized according to the principles of panel D. AFR, African; AMR, Admixed American; EAS, East Asian; EUR, European; MAF, minor allele frequency; NFE, non-Finnish European; SAS, South Asian.

Figure 3.

Array-generated typing for HEAs. (A) Comparison of clinical and UBDT_PC1 array-based HEA-typing density. In the graph, the presence of color represents a typing result (positive or negative), and the absence of color indicates a lack of a typing result. The graph is stratified according to the ancestry of the study participants, with a top and bottom panel for each ancestry group representing the density of clinical and array typing results, respectively. EUR, AFR, AMR, SAS, EAS, and OTH ancestries are shown. The HEA systems and relevant antigen types are indicated on top of the graph, and the typing density as a percentage of the total possible types is given on the right of the graph. (B) Concordance between clinical and array-generated HEA types. Bar plots showing the number of comparisons (y-axis) per HEA type (x-axis) with concordant results obtained by both Sanquin and NYBC in green and discordant results by both Sanquin and NYBC in blue, Sanquin only in orange and NYBC only in red. Ascending and descending bars represent the number of comparisons to positive clinical or negative clinical antigen types, respectively. Bar plots show the number of comparisons on a log scale. (C) HEA typing discordances. Heat map showing the cause of the discordance (columns) between clinical and array-generated types by HEA system (rows). The number of unique discordances per system and the number per cause of discordance are given on the right and top marginal bar plots, respectively. AFR, African; AMR, Admixed American; EAS, East Asian; EUR, European; OTH, other; SAS, South Asian.

Figure 4.

Common and rare HEA types. (A) Ancestral differences in frequencies of some common HEA types, which frequently elicit alloantibody formation. Heat map with the ancestry stratified frequencies of the common MNS, Rh, FY, and JK types in the unified set of 6679 DNA samples. Heat map colors range from yellow (0%) to deep blue (100%), showing HEA-type frequencies within each ancestry group. (B) Number of HFA⁻ samples identified in the unified set of 6679 DNA samples with those identified by Sanquin and NYBC on the x- and y-axes, respectively. True negative, false negative in Sanquin, false positive and no-type in NYBC, no-type in Sanquin, no-type in NYBC, no-type in Sanquin and NYBC, and false negative in NYBC are showing in blue, red, green, orange, brown, gray, and magenta, respectively. (C) Number of patients typed negative for 16 HFA identified in the extended unified sample set. Bar plot shows phenotype and the count of negative typing results on the x- and y-axes, respectively. Typing results concordant with clinical type, array detected and confirmed, array detected and unconfirmed, false negative array types, and no-type results are shown in purple, blue, orange, yellow, and green, respectively. (D) Concordance between clinical and array-generated results for DNA samples harboring complex Rh genotypes. A graphical representation of 8 alleles of the RHD gene, in descending order: D⁺ (RHD∗01), weak D type 1 (RHD∗01W.1), weak D type 2 (RHD∗01W.2), weak D type 3 (RHD∗01W.3), D^– (RHD∗01N.01), r'^S type 1 (RHD∗03N.01), D pseudogene (RHD∗08N.01), and DEL1 (RHD∗01EL.01). Counts on the right show the number of alleles detected, confirmed by clinical type, and discordant in the extended unified sample set in black, green, and orange, respectively. AFR, African; AMR, Admixed American; EAS, East Asian; EUR, European; SAS, South Asian.

Figure 5.

HLA typing. (A) Schematic depicting density and location of probes across the HLA region, with upper panel: location of the extended HLA locus on chromosome 6p indicated by a red box window; middle panel: zoomed in visual of the 6 Mb extended HLA locus spanning from genome coordinates 28 to 34 Mb and depicting 2 sets of 3, class I and class II, genes, respectively; lower panel: graph visualizing the number of probes per 50 000 base pair windows across the 6 Mb, with number of probes/window on the y-axis and genomic coordinates in Mb on the x-axis. (B) Genotype reproducibility for 7896 biallelic probe-variant pairs included in the UBDT_PC1 array design. Reproducibility expressed as percentage of concordant genotype comparisons, and gnomAD MAF for each variant are displayed on the x- and y-axes, respectively. Blue hexagons and red dots on the central scatterplot represent the density of probes with reproducibility ≥99% and individual probes with <99% concordance, respectively. Marginal histograms show probe counts on a log scale. (C) Allele diversity: upper panel: quantile-quantile plots illustrating the distribution of probability scores in calling alleles for European and non-European ancestries across the HLA-B (left) and HLA-DRB1 (right) genes. Quantiles of probabilities for EUR and non-European samples (AFR, AMR, SAS, and EAS) are shown on the x- and y-axes, respectively. Lower panel: stacked bar charts showing the frequency distribution of the top 4 alleles for the different ancestry groups for the HLA-B (left) and HLA-DRB1 (right) genes, respectively. For both genes, the frequencies of the top alleles are normalized to percentage values (y-axis) and ancestry groups are given on the x-axis. Shaded lines are drawn between bar segments representing the same allele. (D) Concordance between clinical and HLA∗IMP:02 imputed types are presented in a heat map, expressed as an agreeing percentage of total comparisons ranging from dark blue (85.7%) to bright yellow (100%). The vertical and horizontal marginal bar plots give on the y-axis the number of samples used for the concordance analysis stratified per ancestry group on a log scale and the number of comparisons made for each of the 6 HLA genes. AFR, African; AMR, Admixed American; EAS, East Asian; EUR, European; MAF, minor allele frequency; Mb, megabase; SAS, South Asian.

Figure 6.

Performance of transfusion module on UKBB_v2.2 array. (A) Genotype reproducibility for 17 070 biallelic probe-variant pairs included in the UKBB_v2.2 array design. Reproducibility expressed as percentage of concordant genotype comparisons, and gnomAD MAF for each variant are displayed on the x- and y-axes, respectively. Blue hexagons and red dots on the central scatterplot represent the density of probes with reproducibility ≥99% and individual probes with <99% concordance, respectively. Marginal histograms show probe counts on a log scale. (B) Correlation of the MAF in EUR study participants vs (non-Finnish) EUR participants from the gnomAD database for each probe-variant pair. Probes with ≥99% and <99% genotype reproducibility are shown in blue and red, respectively. Contour lines represent boundaries of statistical significance with corresponding P values calculated using the χ² test. (C) Concordance between clinical and array-generated HEA types for the unified samples genotyped in triplicate (n = 3791). Bar plots showing the number of comparisons (y-axis) per HEA type (x-axis) with concordant results obtained by all 3 test sites in green and discordant results by all test sites in blue, Sanquin only in orange, NYBC only in red, and NHSBT only in purple. Ascending and descending bars represent the number of comparisons to positive clinical or negative clinical antigen types, respectively. Bar plots show the number of comparisons on a log scale. EUR, European; MAF, minor allele frequency; NFE, non-Finnish European.