SNP-Array Based Karyotyping Complements Routine Cytogenetics in Diagnosis and Risk Stratification Schemes of MDS

While current pathomorphologic criteria distinguish MDS, MDS/MPD and MDSderived AML (sAML), these conditions share common unbalanced chromosomal abnormalities highly predictive of prognosis. Reflecting their clinical importance, IPSS is highly influenced by metaphase cytogenetics (MC), the standard for detection of chromosomal abnormalities. Due to its low resolution and need for cell divisions, MC can only detect abnormalities in 50% of patients with MDS, novel technologies have been developed that could improve the diagnostic yield. Among these methods, single nucleotide polymorphism arrays (SNP-A) have been adopted as a cytogenetic platform. SNP-A karyotyping can be performed on interphase cells and allows for detection of smaller lesions as well as for copy-neutral loss of heterozygosity (LOH). As a result, some reference laboratories offer this test for routine cytogenetic diagnostics. However, the clinical relevance of lesions detected by SNP-A remains unclear and, consequently, we designed this comprehensive study to clarify the clinical impact of cytogenetic results generated through a combined application of SNP-A and MC. A cohort of 352 patients (218 MDS, 59 MDS/MPD, 75 sAML) was studied using Affymetrix 250K (N=160), Affymetrix 6.0 (N=190) and both platforms (N=95). Controls included 362 and 118 samples analyzed by 250k and 6.0 arrays, respectively. CNAGv3 or GCv2.1 software was used for analysis. To avoid biological and technical artifacts, known copy number variants as well as areas of LOH<2.5Mba were excluded. All other lesions, if not present on MC, were confirmed in germ line DNA if possible. The median follow-up time was 29 months (95% CI; 21-38 mo) and for many patients serial samples were studied, showing sequential acquisition of chromosomal defects. Overall, the detection rate of chromosomal abnormalities was improved compared to MC (57% vs. 44%; p=.0096), MDS/MPD (66% vs. 39%, p=.0055) and sAML (63% vs. 45%, p=.048). Somatic uniparental disomy (UPD) accounted for a significant proportion of newly detected defects (26% of new lesions). Overall, UPD was detected in 100 patients (78 as the sole and 22 as an additional lesion). Chromosomal lesions were identified in 98/180 with normal MC. In 61% of patients with non-informative MC (N=18), SNP-A demonstrated an abnormal karyotype; those with newly detected lesions showed inferior outcomes in overall survival (OS) [16 vs. 36 mo, p=.03]. Analysis of OS showed that patients with new defects detected by SNP-A and previously normal MC have worse outcomes compared to those in whom normal chromosomes were confirmed (39 vs. 73 mo, p=.03). Moreover, additional lesions also conferred worse prognosis for patients with abnormal karyotype by MC (17 vs. 38 mo, p=.01). This trend was even more pronounced when specific subgroups were analyzed based on OS and event free survival (EFS). For example, detection of new lesions in MDS (OS p=.02), MDS/MPD (OS p=.009, EFS p=.01) or sAML (OS p=.02, EFS p=.008) results in significant worsening of OS and EFS as compared to patients with truly normal cytogenetics. Newly detected lesions allowed for more precise prognostication within established IPSS strata of MDS and MDS/MPD patients. The detection of SNP-A defects negatively impacted OS/EFS in patients with low IPSS (NR vs. 69 mo, p=<.0001; 42 vs. 112 mo, p=<.0001), decreased EFS in Int-1/2 (24 vs. 29 mo; p=<.0001) and decreased OS/ EFS in high IPSS categories (11 vs. 26 mo, p=<.0001; 7 vs. 14 mo, p=<.0001). In sAML, patients with SNP-A lesions, regardless of MC, have worse OS (6 vs. 21 mo, p=.009) and EFS (5 vs. 13 mo, p=.01). In summary, SNP-A karyotyping facilitates detection of lesions which can complement MC and contribute to a better molecular diagnosis. SNPdetected lesions have impact on prognostic parameters which will likely affect future risk stratification schemes and have prognostic significance regardless of MC in sAML. As a result, new molecular subtypes of MDS may be identified and better-defined recurrent lesions will contribute to identification of causative genes.