Identification of Novel Methylation Biomarkers to Predict Clinical Response to SGI-110, a Second Generation Hypomethylating Agent (HMA), in Patients with Acute Myeloid Leukaemia (AML)

Introduction: SGI-110 is second generation HMA formulated as a dinucleotide of decitabine (DAC) and deoxyguanosine delivered as a small volume, pharmaceutically stable SQ injection allowing longer half-life and more extended decitabine exposure than DAC IV infusion. SGI-110 differentiated pharmacokinetic profile resulted in potent hypomethylation and clinical responses in previously treated MDS and AML patients in the phase 1 trial (Kantarjian et al. 2012).

Here, we have identified novel DNA-methylation biomarker candidates that may be predictive of response to SGI-110 using Differential Methylation Hybridisation (DMH) profiling of the NCI-60 cell line panel (Fassbender A et al, Methods Mol Biol 2010). Cell lines were stratified based on SGI-110 EC₅₀ values from Colony Forming Assay and degree of LINE-1 (Long Interspersed Nucleotide Elements) demethylation post-SGI-110 treatment. The stratification was used to classify cell lines into SGI-110 sensitive and resistant and to generate 249 genomic methylation sites as candidate biomarkers of response to SGI-110. Candidate markers from this analysis will now be compared in DNA samples derived from whole blood from treatment naïve and relapsed/refractory AML patients that were classified into responders and non-responders after treatment with SGI-110 in our phase 2 clinical study.

Methods: DMH was used for unbiased profiling of methylated loci throughout the genome. Briefly, the genome is cleaved into a defined fragment library using methylation unspecific restriction enzymes, followed by PCR-adapter ligation. In a second step, the fragments are digested by methylation-sensitive restriction enzymes cutting only at unmethylated restriction sites, while methylated fragments and fragments that do not have any restriction site were protected from the restriction digestion. The methylated fragments and the control fragments are then selectively amplified by PCR and the PCR product is labeled and hybridized to Epigenomics' proprietary DMH microarray. Direct Bisulfite Sequencing (DBS) is used to assess the DNA-methylation pattern of marker candidates at CpG level. Regions with CpGs of interest are amplified to generate PCR products that are directly sequenced using a 4 dye labelled ddNTP Sanger sequencing. Quantitative DNA methylation of CpG region with a lengths of up to 550 bp are derived from the resulting trace files using the ESME software (Lewin J et al, Bioinformatics. 2004 Nov 22;20(17):3005-12).

Results: Using this method, we have seen a clear correlation of high sensitivity to SGI-110 with high LINE-1 methylation level. Five (leukemia) cell lines classified as sensitive to SGI-110 by EC₅₀ values were among the 6 cell lines with the highest LINE-1 methylation (> 73% methylation) while resistant cell lines showed a LINE-1 methylation < 70%.

A high degree of correlation was observed between LINE-1 methylation and genome-wide DNA-Methylation measured by DMH in over 50k genomic sites and supported the role of LINE-1 as a useful indicator for sensitivity to SGI-110 and global DNA methylation. Fifty genomic fragments that better characterized sensitive and resistant cell lines were selected for further validation.

The differential DNA-methylation discriminating between SGI-110 sensitive and resistant cancer cell lines (n=18) was validated by DBS analysis at CpG level in genomic fragments. We will present data obtained from the DBS evaluation of methylation in these validated methylation marker candidates in up to 45 AML blood samples that were classified into responders and non-responders after treatment with SGI-110 in a phase 2 clinical trial.

Conclusions: We identified DNA methylation patterns associated to sensitivity and resistance to the novel HMA SGI-110. The validated methylation biomarker discovery process based on DMA and DBS approaches may help to identify and characterise specific subgroups of AML patients that could preferentially respond to SGI-110.