Total Genomic Loss Detected by High-Density Single Nucleotide Polymorphism Array Is Predictive of Azacitidine Response in Very Poor IPSS-Revised MDS or AML Patients

Azacitidine (AZA) has changed the outcome of patients with myelodysplastic syndromes (MDS) or acute myeloid leukemia with multi-lineage dysplasia (AML-MLD) unfit for intensive chemotherapy. AZA is a hypomethylating agent providing about 50% of responses in MDS and AML with low blast count (Fenaux et al., Lancet Oncol 2009, JCO 2010). IPSS-R scoring was evaluated in MDS and AML with 20–29% of blasts untreated patients. Conventional cytogenetic have major prognostic interest in this score (Schanz et al., JCO 2012). To date, SNP array analysis are not included in scoring system and no markers of AZA response have been identified with this analysis.

Methods: This analysis included MDS and WHO-AML-MLD (>30% blasts) patients treated by AZA in 6 centers. Genomewide single nucleotide polymorphism (SNP) analysis using SNP 6. 0 arrays (Affymetrix, High Wycombe, U. K) was performed on bone marrow (BM) samples. Patients having received ≥ 1 cycle of AZA and who had bone marrow evaluation after ≥ 4 cycles, or who died or progressed before completion of 4 cycles were considered evaluable (the last 2 groups were considered as treatment failures). Responses were scored according to IWG 2006 criteria for MDS and to Cheson et al. (JCO 2003) for AML. DNAs were extracted for hybridization according to the manufacturers' recommendations (Qiagen). Affymetrix CEL files for each sample were analyzed using the Genotyping Console software (v3. 0. 2). Genotyping was performed using Birdseed V2 algorithm. Unpaired Copy Number and LOH analysis was performed with Regional GC correction. For each patient, size of all deletions (loss) find with GC was added to have a total deletion size. The same was performed for total gain size.

Results: The study population included 51 patients: F/M: 23/28; median age 73 (range 42–88). Diagnosis at AZA onset was RAEB-2 in 19 patients (IPSS int-1 in 4, int-2 in 10, high in 5) and WHO-AML-MLD in 32 patients. Cytogenetic according to IPSS-R was good in 23, intermediate in 9, poor in 7 and very poor in 12. IPSS-R score was good in 1 patient, intermediate in 4 patients, poor in 25 patients and very poor in 21 patients. Median number of cycles was 7 (range 1–42). All patients received the approved (75mg/m² for 7 days every 4 weeks) or a reduced AZA schedule (75 mg/m² for 5 days every 4 weeks). Median overall survival (OS) of our cohort was 11 months. DNA samples from all patients were available for SNP analysis. In using smoothing spline Cox model analysis, we observed that relative risk (RR) of worse OS increased above 4 megabases (Mb) of total chromosomal loss size detected with SNP array. Increase of RR was found in IPSS-R intermediate, poor and very poor. OS of patients treated by AZA with more than 4 Mb of genomic loss detected by SNP array was significantly different of OS of patients with less than 4 Mb (6 months vs 14 months respectively, p=0. 0012). Negative effect of genomic loss above 4 Mb was mainly observed in IPSS-R very poor group. In IPSS-R very poor subgroup, we observed significant difference in OS between patients with more than 4 Mb of genomic loss versus patients with less than 4Mb (3 months vs 10 months respectively, p=0. 01). In our cohort, total genomic gain has no impact on OS.

Conclusion: New tools such as SNP array are not used in current practice. Therefore, prognostic impact of molecular karyotype is unclear in MDS/AML treated by AZA; our data suggest that total genomic loss superior to 4 Mb could be a worse prognosis, mainly in IPSS-R very poor patients. Impact of SNP array should be studied in larger cohort of patients treated by AZA in order to validate these results. Starczynowski et al. (Blood 2008) showed that total genomic alterations superior to 3Mb were associated with a poor OS in low-risk MDS. Taken together, these data are in favor of the use of SNP array's parameters in future prognosis scoring system.

No relevant conflicts of interest to declare.