Identification of Chromosomal Abnormalities in Healthy Bone Marrow Using 250K SNP Arrays.

Myelodysplastic syndromes (MDS) are characterized by the presence of clonal chromosomal abnormalities detectable by traditional cytogenetics in around 50% of patients. We demonstrated that a higher percentage of unbalanced clonal chromosomal defects and uniparental disomy (UPD) can be identified in MDS using high-density SNP arrays (50 or 250K SNP-A; Gondek et al, 2005). The higher detection rate may have important clinical consequences. However, such findings must be considered in the context of normal karyotypic variation. Before the clinical relevance of new lesions identified by SNP-A can be presumed, several issues must be addressed.

• The increased precision of karyotypic analysis may lead to the detection of lesions in normal bone marrow. This is especially relevant to elderly patients with MDS for whom adequate age-matched comparisons should be performed.

• The normal distribution of chromosomal changes across the genome must be defined as it may overlap with that in disease.

• The minimal clonal size detectable by SNP-A analysis is of importance, as hematopoiesis may be oligo- rather than monoclonal in many conditions.

We stipulated that the clinical applicability of SNP-A-based karyotyping will depend upon the findings in healthy controls and have studied 36 normal bone marrows using SNP-A karyotype analysis. We utilized the Affymetrix 250K SNP chip and the CNAG software for copy number and LOH analysis. Using a stringent set of criteria, suspicious lesions were identified in 83% of samples. Loci altered in 2 or more samples, most likely reflecting copy number polymorphisms (CNP), were identified in 70% of individuals. With these CNP excluded, 69% of controls harbored putative lesions. 11 marrows contained 1 chromosomal change, while 2-4 were found in 14 marrows. Both loss and gain of sequences were detected. The size of the largest deletion and duplication was 1.8 Mb and 1.64 Mb, respectively. Karyotypic abnormalities identified in control samples appeared to be randomly distributed across the genome. No lesions were identified on chromosome 8 or chromosome 5q, although one deletion on chromosome 5p was found. A small deletion of 7q was detected (1.2 Mb); overall, the regions frequently affected in MDS were rarely altered in controls. Finally, we identified LOH due to UPD spanning 13q21.3 to 13q32.1 (4.95 Mb) in one healthy control. To assess the minimal detectable clonal size, we next examined the sensitivity of SNP-A analysis to the admixture of normal cells. When samples identified as abnormal (deletion of 7q, and trisomy 8) by traditional cytogenetics were serially diluted with normal genomic DNA, the previously identified lesions could still be detected when the sample contained 25% normal genomic DNA, but at 50% the copy number analysis appeared normal. Our result underscores that, only significantly expanded clones can be systematically detected by SNP-A analysis. Pathogeneic defects have to exclude regions of CNP and must be larger than changes in healthy controls. Additionally, they would preferentially occur in regions not frequently affected in normal bone marrow. Our studies reveal the physiologic level of chromosomal abnormalities present in healthy controls and allow us to design criteria for defining abnormal karyotypes as measured by SNP-A.