Factors influencing treatments outcomes in myeloid leukemia

Abstract

Acute myeloid leukemia (AML) is a rapidly proliferating clonal disorder of hematopoietic stem cells. Since AML is a clinical and biological heterogenous disease, AML patients are divided into three cytogenetically defined risk groups with significant differences in overall survival (OS). However, large inter‑individual differences in treatment response and development of resistance are still major drawbacks in AML. Cytarabine arabinoside (ara‑C) is the key agent for treating AML, but there is also considerable heterogeneity in the outcomes for individual patients in same risk group. In addition, up to 50% of AML patients show no abnormalities by conventional cytogenetics at diagnosis. These normal karyotype AML (NK‑AML) patients are prognostically heterogeneous although categorized in the intermediate‑risk group. There is also an inter‑ethnic difference in treatment outcomes among AML patients. Genetic or genomic alterations may affect the expression and/or function of specific drug protein targets and explain, at least in part, the inter‑individual variations in the response to specific treatments. The genetic variations, such as single nucleotide polymorphisms (SNPs), in the genes encoding the ara‑C transport and metabolizing pathways may play an important role in the clinical outcomes in AML patients. Since the drug response is the result of an interaction of numerous genetic combinations, the combined effects of SNPs via gene‑gene interactions as well as the effect of individual SNP may explain the different clinical outcomes between patients. Copy number variation (CNV) is a common type of genomic structure variation. CNVs also have recently attracted considerable interest as a source of genomic variation because they may play an important role in the etiology of complex diseases and in evolution. CNVs, depending on their size and location, are as important as SNPs for producing variations in treatment efficacy and/or adverse responses to chemotherapy. To identify the susceptible genetic or genomic alterations affecting the clinical outcomes of AML patients receiving ara‑C based chemotherapy, we genotyped 139 SNPs of 10 candidate genes within the ara‑C transport and metabolic pathway using the Illumina GoldenGate Genotyping Assay (Illumina Inc., San Diego, CA, USA) in 97 patients with previously non‑treated de novo AML other than M3. For 30 NK‑AML patients, we determined the frequency of genome‑wide cytogenetic CNV aberrations using HelixTreeⓇ software version 5.2.0 (Golden Helix Inc., Bozeman, MT, USA). Bone marrow aspirates and blood from AML patients were provided at the time of diagnosis for genotyping and copy number analysis. For SNP analysis, both effect of single SNP and SNP‑SNP interaction on treatment outcomes were tested. And we tested three different genetic models, including dominant, recessive, and additive model. In multivariate anlaysis, SNP rs4694362 (CC genotype) in DCK gene, individually, was a significant poor prognostic factor for OS (HR, 33.202 [95% CI, 4.937-223.273], P < 0.0001, PBonferroni = 0.017). In addition to the single SNP effect on treatment outcomes, multivariate analysis revealed that the presence of the SLC29A1 rs3734703 (AA or AC genotype) in combination with TYMS rs2612100 (AA genotype) was significantly associated with shorter relapse free survival (RFS) compared to the combination with wild type (HR, 17.630 [95% CI, 4.829-64.369], P < 0.0001, PBonferroni = 0.021). The effect of these SNP‑SNP interaction also decreased the survival time, although not statistically significant after the multiple test adjustment (HR, 23.523 [95% CI, 4.616-119.873], P = 0.0001). In addition, CDA rs10916827 (GG genotype) in combination with DCTD rs17331744 (TC or CC genotype) was associated with less survival time (HR, 31.680, [95% CI, 6.152-162.905], P < 0.0001, PBonferroni = 0.052). These results suggest that a single SNP and SNP‑SNP interactions may help to predict the drug response and provide a guide in developing individualized chemotherapy for AML patients receiving ara‑C based chemotherapy. For 30 NK‑AML patients, possible associations between cytogenetic aberrations and clinical parameters were analyzed. CNVs were identified in 23 (76.7%) of the 30 cases tested. Multivariate analyses controlled for other clinical covariates showed that patients having copy number loss had a decreased probability of complete remission (OR, 0.015 [95% CI, 0-0.737], P = 0.035). And patients who had a copy number gain of more than four regions tended to have shorter RFS (P = 0.083) with multivariate analysis showing that CNV increase is an independent predictive factor for increased risk of relapse (HR, 22.104 [95% CI, 1.644-297.157], P = 0.020). In addition, we identified genes in recurrent CNV regions utilizing data from the University of Canada database (http://projects.tcag.ca/variation/project.html) and the PharmGKB (http://www.pharmgkb.org/index.jsp). It involved nineteen previously reported AML-related genes, including HES5, PRDM16, TNFRSF25, MTX2, TERT, ABCB8, PTP4A3, PBX3, VENTX, AKT1, KIAA0284, ABCA3, CBFA2T3, FANCA, MLLT6, CD7, PRTN3, CEBPA, and TYMP. Among drug-related genes, NOS3, ERCC1, ERCC2, ATP5I, ATP5D, CYBA, NDUFS7, SLC19A1, and P2RX1 are known to be related to ara‑C or anthracycline response. These results suggest that CNVs may affect the success of ara‑C and anthracycline based chemotherapy in Korean NK‑AML patients. In addition, the population specificity in allele frequencies of the 139 SNPs through inter‑ethnic comparisons was assessed in this study. For this analysis, the International HapMap (http://hapmap.ncbi.nlm.nih.gov/) and the 1000 genomes database (http://www.1000genomes.org/) was used. FST statistic are calculated between Korean and other populations. Overall, there were large differences in allele frequencies between Korean and Caucasian or African, whereas Chinese and Japanese populations were extremely similar to Korean. The SNPs which showed a significant relationship with the response to ara‑C based chemotherapy in this study represented a large divergence for the comparisons with other populations

DCK rs4694362 (comparison with African, FST = 0.519), SLC29A1 rs3734703 (comparison of Caucasian, FST = 0.136), and TYMS rs2612100 (comparison of Caucasian, FST = 0.195). In conclusion, the results of this study have important implications in providing fundamental and useful information for predicting the treatment outcomes in Korean AML patients, and may help the development of more appropriate therapeutic modalities.