High-Density Genomic Scan with 50k SNP Arrays May Reveal Existence of Germline Allelic Polymorphisms That Determine Predisposition to Aplastic Anemia.

Although immune mechanisms are likely responsible for most cases of acquired idiopathic AA, the pathogenesis of this disease is not known and various causes, including viral infections and chemical exposures have been implicated. Irrespective of the cause it appears clear that, in analogy to other autoimmune conditions, only certain individuals are susceptible to the development of AA. Theoretically, inherited complex traits may, in combination with exogenous factors, determine the risk to AA. A spectrum of such possible genetic determinants ranges from immunogenetic factors, detoxifying enzymes to DNA repair genes. Overrepresentation of otherwise rare SNP may suggest existence of cryptic changes involving adjacent portions of the genome. The goals of our study were: i) to establish the feasibility of high density SNP scan in the study of genetic factors predisposing to AA and ii) to develop principles for efficient identification of putative disease-associated variants in a large number of AA patients. First, we have studied individual patients with typical AA to identify globally overrepresented alleles. Second, we have examined ability of detection of overrepresented SNPs in pooled DNA samples derived from the cohort of patients. We have applied 50K SNP Affymetrix arrays to the analysis of genome in AA. Lymphocytes from 8 patients were used as a source of germline DNA. DNA was digested with Xba III; hybridization to SNP arrays resulted in average 96.2% genotype call rate. Subsequent analysis concentrated on the detection of polymorphic genotypes rather than changes in gene copy number as seen in MDS. Globally, SNP array allowed for classification of 56725 SNPs in 8 patients (>0.453x10⁶ genotypes). Of them 28% showed heterozygozity, while 36% were homozygous for either genetic variant. Complex analysis of such a large number of genotypes may allow for determination of potential associations between the genetic makeup and disease risk. Based on the sampling size and comparison with control population, we have identified 370 and 88 polymorphisms (in hetero- or homozygous form) that were observed at higher than the expected control frequency (p<.05 and p<.001, respectively). For some variants, SNP were present in >80% of patients. Examples of 70 overrepresented SNPs (>62% in AA;<30% in controls) include IL20R, IL22R, CD109, CD51 and catenin. For some of the SNPs, a homozygous configuration was observed e.g., for DEK (bb in 4/8) and CBP2 (bb in 3/8); for both genes the homozygote frequency in general population was <.0001.

The principle of pooling DNA samples relies on the theory that randomly distributed SNP will be diluted out. To determine the possible sensitivity of a significant SNP overrepresentation, individually determined polymorphic frequencies were compared to the results obtained in a pooled sample (equimolar DNA concentrations). Individually detected allelic frequencies of ≥50 were correctly genotyped with the pooled array. Thus, our technique may be helpful in eliminating variants that show lower prevalence within the patient population. While investigations of higher numbers of patients with various clinical, including PNH are under way, our study reveals feasibility of application of the high density SNP arrays in the study genotypic profiles that may potentially contribute to inherited predisposition traits in AA.