Abstract
BACKGROUND: Alloimmunization to minor blood group antigens is more common in sickle cell disease (SCD) than in the general population. Causes include the high frequency of transfusion as well as phenotypic differences between the largely African-American SCD population and mostly Caucasian blood donors. Alloimmunization, as well as the risk of alloimmunization, results in higher costs as well as reduced availability of and delays in obtaining antigen-matched blood to meet clinical needs. We used a candidate gene approach to determine if specific genetic signatures were associated with alloimmunization and thus could identify patients at high risk for alloimmunization and therefore requiring prospective antigen matching for transfusion.
METHODS: Patient data were previously collected through a multicenter study to identify genetic factors associated with SCD complications (Duke and UNC; DU) and through the Cooperative Study of SCD (CSSCD). All subjects were adults with HbSS genotype (DU mean yrs 34.9 ± 12.1, CSSCD mean yrs 29.1 ± 9.5, p < .0001). The DU data set was 43.7% male while the CSSCD data set was 37.1% male (p = .18). A genome-wide association study (GWAS) was conducted with the Illumina 610 BeadChips for 390 individuals (215 DU, 175 CSSCD). Quality control processing of these data was published previously (Solovieff et al, 2010). Additionally, imputation of the DU data set was conducted using IMPUTE2 (Howie et al, 2009) and a global reference panel from 1000 Genomes in order to have consistent genotypes across both studies. Based on previous literature, we studied 255 SNPs in the HLA locus as well as 684 SNPs in 53 genes previously associated with immune responses, totaling 939 SNPs. Additive logistic regression was performed separately for the DU and CSSCD data sets with PLINK (Purcell et al, 2008), controlling for age at enrollment. Results from the two data sets were combined by meta-analysis, using the invariance weight method in METAL (Willer et al, 2010). False discovery rate (FDR) q-values were generated using PROC MULTTEST in SAS v9.4 (SAS Systems). Contingency tables were produced using SAS and association between candidate SNPs and specific antibodies was tested using PLINK.
RESULTS: Alloimmunization rates were different in the DU and CSSCD cohorts (30.7% vs. 18.9%, respectively; p = .0075) but were not associated with age or number of transfusions > 0 in either cohort. The most significant genetic associations with alloimmunization (FDR q = .28) occurred for 46 SNPs in the HLA locus and 10 other genes (Table 1). Two of these SNPs were missense mutations in GZMB (granzyme B (granzyme 2, cytotoxic T-lymphocyte-associated serine esterase 1)), one of which (rs8192917) has previously been associated with acute kidney transplant rejection and acute GVHD after stem cell transplant. A missense mutation in TRIM31 (E3 ubiquitin-protein ligase) was also associated with alloimmunization (p = .01). TRIM31 regulates Src-induced anchorage-independent cell growth and has been implicated in antiretroviral defense, cancer and ADHD. Within the DU cohort, we conducted a sub-analysis among only antigen-negative individuals, in whom immunization rates were 15.8%, 14%, 4.5%, 10.2%, 11%, and 8.3% to E, C, K, S, Jkb and Fya, respectively. Regression analysis of all 939 SNPs identified nominally significant associations predicting specific antibody alloimmunization. A missense mutation in PRKCQ (rs2236379) was associated with both anti-E and –C alloimmunization. Additional missense mutations were identified as follows: in TGFBR3 and GZMB with anti-C; in CFH, andTLR6 with anti-K, and in TLR10 with anti-Jkb.
CONCLUSIONS: We have identified several immune response genes that are also associated with increased risk of alloimmunization in SCD (nominal p ≤ .006, FDR q = .28). Some associations may arise from the prevalence of alloimmunization to specific antigens. For example, PRKCQ was associated with development of alloantibodies to both E and C antigens, which reside on the same nonglycosylated RH protein. Expanded studies may identify those patients who would benefit from intensive or specific antigen-matching transfusion protocols.
Gene or Locus (ordered by descending p value) | SNPs with FDR q = 0.28 | SNPs with p ≤ .006 |
GZMB | 3 (2 missense) | 3 |
TGFBR2 | 4 | 2 |
TLR3 | 1 | 1 |
HLA locus | 26 | 2 |
TGFBR3 | 3 | 2 |
PRKCQ | 3 | |
CD80 | 1 | |
VAV2 | 1 | |
TRIM31 | 2 (1 missense) | |
RNF39 | 1 | |
ZNRD1 | 1 |
Gene or Locus (ordered by descending p value) | SNPs with FDR q = 0.28 | SNPs with p ≤ .006 |
GZMB | 3 (2 missense) | 3 |
TGFBR2 | 4 | 2 |
TLR3 | 1 | 1 |
HLA locus | 26 | 2 |
TGFBR3 | 3 | 2 |
PRKCQ | 3 | |
CD80 | 1 | |
VAV2 | 1 | |
TRIM31 | 2 (1 missense) | |
RNF39 | 1 | |
ZNRD1 | 1 |
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.
This feature is available to Subscribers Only
Sign In or Create an Account Close Modal