Non Synonymous SNP-Array-Based Disease Association Analysis in Aplastic Anemia and Paroxysmal Nocturnal Hemoglobinuria.

The pathogenesis of aplastic anemia (AA) includes external triggers that induce an immune-mediated attack in susceptible individuals. A complex genetic background is likely the basis for AA predisposition; various genetic factors, including HLA, KIR and SNPs in immune response genes, were studied in AA. However, such an empiric approach, despite the rational target selection, is inefficient. SNP array (SNP-A) genotyping technology allows for investigation of complex genetic traits and is a suitable hypothesis-generating technology. We stipulated that application of SNP-A in AA will allow for identification of SNPs in pathogenic loci. We applied the Illumina 12K non-synonymous SNP-A to study 77 AA and ethnically matched controls (ctr; N=60; +170 historical ctr). The power of this technique is demonstrated by our ability to obtain >2.6 million genotypes with a fidelity (against PCR) of 98%. The training set included 64 AA patients and 56 controls. Initially, Exemplar automated analysis was used; due the Bonferroni correction this study underpowered. However, our strategy included ranking SNPs based on their P value, low frequencies of pathogenic genotypes in ctr, and high case/ctr ratio narrowing the selection of potentially informative SNPs to be tested in a validation set. We also applied Random Forests, a nonparametric tree method, whereby all SNP were used multivariately to predict disease association; this method does not relay on Bonferroni correction and more closely reflects complex polygenic traits. Results pointed towards many SNPs recognized by both approaches and a number of SNP was chosen for further analysis. E.g., rs8022805 (GA) in telomerase-associated protein-1 (TEP1) was found in 19% of AA patients but only in 2.7% of controls (N=220, p<.0001). TAP1 is a part of telomerase complex, and various mutations in telomerase machinery were linked to hereditary AA. Similarly, a CG SNP (rs1134648) in chemokine SCYE1 inducible by apoptosis was found in 8% of AA cases vs. 0.003% of ctrs. Given the high frequency of thromboembolic events in patients with PNH clones, an interesting association was that of factor 5 rs6019 SNP for which CG and GG frequencies were 23% and 2% in AA vs. 8% and 0% in ctr, respectively (p=.0009). When we analyzed the AA sub-cohorts based on presence of a PNH clone, we found that rs1079109 and rs753856 in HSP70 gene were significant. HSP70 is up-regulated and inhibits apoptosis in cells stress conditions. The genotype A/G of rs1079109 showed an increased frequency in patients with PNH (46%) and AA/PNH patients (33%) in comparison to both AA (without PNH clone) and normal ctr (6% and 7% respectively p=.001). Similar results were found for the genotype C/G of rs753856 (PNH 46%, AA/PNH 37%, AA 6%, NC 6% p<.0001;N=265). The role of HSP was investigated in AA and several autoimmune diseases. There is evidence that genetic polymorphisms in the HSP70 gene cluster are associated with carbamazepine-induced hypersensitivity and that upregulation of HSP is a marker of immune-mediated AA responsive to CsA. In conclusion our SNP approach allows for identification of immunogenetic polymorphisms possibly involved in the immune pathophysiology of AA and illustrates application of SNP-A to study AA.