Dynamics of Methylation Profiling in Response to Tyrosine Kinase Inhibitor Therapy in Chronic Myeloid Leukemia

Background

Chronic myeloid leukemia (CML) is characterized by gene rearrangement event between the ABL1 gene and the BCR gene, yielding a bcr-abl fusion oncoprotein. Increasing evidence suggests that major DNA methylation change is associated with disease progression in CML. However, the impact of tyrosine kinase inhibitor (TKI) therapy on methylation in CML has never been analyzed on a genome-wide level. Accordingly, this study attempted to analyze methylation profiles in 78 CML patients prior to/after TKI therapy and its clinical relevance with respect to molecular response to TKI therapy.

Methods

Genome-wide methylation microarray was performed through the latest Illumina methylation profiling microarray with 850k probes, MethylationEPIC BeadChip in a series of 78 sample pairs from 78 CML patients, collected prior to and after TKI therapy. Response to TKI therapy was determined according to the ELN recommendation as optimal responder (n=58), failure but remained in chronic phase (n=16; CP) and progression to advanced disease (n=4). Healthy control samples were obtained through NCBI (series number: GSE54939). Methylation statuses prior to and after TKI were profiled and compared according to disease phase, response to TKI therapy, and somatic mutation profiles by targeted sequencing.

Results

Overall, the methylome 867,926 CpG sites were profiled, of which 606,030 (69.8%) passed quality control, followed by normalization procedure applied.

In comparison to control (23.1%), CML patients show higher proportion of methylation on CpG sites (34.2%, bonferroni corrected p-value <1e-22) prior to TKI therapy. Decrease in the percentage of methylated CpG sites was observed in CML sample following TKI therapy (30.7%, p-value < 1e-28), but not to the level in control (Figure 1). However, post TKI methylation status with advanced phase showed less decrease to 31.6%. In detail, 250,541 (41% of remained) CpG sites were differentially methylated (DM) between pre and post TKI therapy samples. Among all DM CpG sites, 113,692 CpG sites were hyper-methylated in post TKI samples, while 136,849 were hypo-methylated.

Next, comparison between the subgroups categorized according to the response to TKI therapy were conducted separately in pre and post TKI therapy samples, which were classified as responsive (R), resistant (T) and progressive (P). A total of 18 and 525 CpG sites was found to be significantly different in pre and post TKI samples, respectively. It implies that TKI therapy potentiates the difference of methylation on CpG sites, and methylation changes in CpG site correlates with the response to TKI therapy. In post TKI samples, 317,579 and 0 CpG sites were identified to be DM between R:T, R:P and T:P comparisons. On each DM site, the corresponding gene was annotated. In post-TKI R:P comparison, four genes were highlighted, i.e. CTC-479C5.6, EML1, IRF4 and RPS12P15. In addition, locus 16p68.01 was found as significant locus in epigenetic change.

Then, cumulative incidence analysis was performed to compare methylation status with respect to complete cytogenetic, major molecular response, MR4.5, MR5, failure, progression and overall survival. Nineteen recurrently CpG sites were reported to be associated with above 7 endpoints, recurrently. The list of associated genes are as follows: AIG1, ATAD2, ATP5F1, C11orf24, ETAA1, FAM63A, OVOL1, PRUNE, ST20, ST20-MTHFS, WDR77, WDYHV1, WNT7A, ZDHHC4, ZNF10, ZNF252P, ZNF252P-AS1, ZNF268, and ZNF891. There are five zinc finger protein genes.

Finally, integrative analysis of methylation profile with somatic mutation profile was conducted. As our group had reported previously (Kim, Blood 2017), we have divided them into 5 mutation patterns, then analyzed methylation profiles accordingly. Methylation profile was significantly different both in pre- and post-TKI samples in patient with pattern 2 compared to others. Pattern 2 was defined as acquisition of new mutation or increasing variant allele frequency over TKI therapy which correlates with TKI therapy failure or progression over treatment.

Conclusion

The present study suggested that methylation profile correlates with response to TKI therapy as well as progression to advanced disease over TKI therapy in CML patients. Also, correlation between somatic mutation profiles and methylation profile is also suggested in the context of TKI therapy.

Figure

View largeDownload slide

Figure

View largeDownload slide

Close modal