Determinants Of Demethylation and Clinical Response In AML Patients Treated With SGI-110, a Novel Subcutaneous (SQ) Hypomethylating Agent (HMA) In a Phase 1 Study

SGI-110 is a dinucleotide of decitabine and deoxyguanosine and is a potent inhibitor of DNA methylation in-vitro and in-vivo. A Phase 1 study established the biologically effective dose of SGI-110 in MDS/AML patients as 60 mg/m² SQ daily x 5 and demonstrated clinical responses that correlated with hypomethylation induction. Here, we analyze AML patients treated at pharmacologically effective doses of SGI-110 to explore determinants of hypomethylation and response.

We studied patients with relapsed/refractory AML treated at therapeutic dose ranges of SGI-110 (36 mg/m²-125 mg/m²). DNA methylation pre/post treatment (for pharmacodynamics [PD]) was measured by bisulfite-pyrosequencing for the LINE-1 repetitive element and the INS6 CpG island gene promoter which is highly methylated in all somatic tissues. Gene expression at baseline and after treatment was measured by qPCR.

We analyzed samples from 27 patients with AML. Median age was 64, (range, 29–86), 18 were Males (67%), 13 (48%) had poor risk cytogenetics at study entry and 59% had prior exposure to a hypomethylating agent. Overall, peak LINE-1 demethylation generally occurred on Day 8 and correlated strongly with INSL6 demethylation (R=0.95, p<0.001). In individual patients, peak LINE-1 demethylation ranged from +1% to -39%. We next examined baseline expression of a panel of genes (CDA, P15, P21, DNMT3B, DNMT3A, DNMT1, CTCF) as possible predictors of SGI-110 response. High expression of DNMT3b (but not DNMT1) was associated with a trend for reduced demethylation (R=-0.20, p>0.05). Unsupervised classification grouped the patients into four clusters: A (N=2), B (N=6), C (N=10), and D (N=9). Cluster D is characterized by high DNMT3b expression, low P15 expression, low CDA expression and reduced demethylation (demethylation average -10.9% in cluster D compared to -22.7% in clusters B and C, p=0.06). Next, we examined SGI-110 mediated induction of gene expression for the P15, P21, DNMT3B and CTCF genes. P15 was significantly induced in patients who were treated on the daily x 5 regimen (p=0.03) but not in patients receiving the weekly x 3 regimen. In this group of 14 patients with induced P15, P15 induction peaked on Day 8 and averaged a 2.4-fold increase. P15 induction was associated with a trend for increased demethylation on Day 8 (R=0.28) and on Day 29 (R=0.37), p>0.05 for both. Of the 27 patients, 5 showed major clinical responses (2 CR, 3 CRi/CRp). LINE-1 and INSL6 demethylation averaged -21.1% and -16.4% in responders compared to -13.1% and -11.3% in non-responders. A three gene classifier score (low CDA, low P15 and high DNMT3B) was associated with low LINE-1 demethylation (R=0.43, p=0.025) as well as resistance to SGI-110 (mean score 6.2 in non-responders compared to 0.5 in responders, p=0.047). Finally, peak induction of P15 was similar in responders and non-responders, but sustained induction (at Day 29) was higher in responders (3.1 fold) than in non-responders (1.0 fold). While a genetic signature of response could not be identified among those 8 genes that were examined by exome sequencing, mutations in IDH1 or IDH2 were identified in 5 patients. One of these patients was positive for substitution R132H of IDH1 (described to induce epigenetic alterations and may predict poor clinical outcome in AML) and achieved a CR in response to SGI-110 treatment. The TP53 polymorphism NM_000546:c.C215G:p.P72R, (associated with differential response to chemotherapy in AML), was identified in 9 subjects, 3 of whom responded to SGI-110.

At therapeutic doses of SGI-110, we identified a gene expression signature (high DNMT3B, low P15, and low CDA) that differentiates responders from non-responders to SGI-110 and we find trends for associations between demethylation and response, as well as sustained P15 induction and response. These associations will be further investigated in the ongoing Phase 2 study of SGI-110 in AML.

Chung:Astex Pharmaceuticals Inc.: Research Funding. Off Label Use: SGI-110. Taverna:Astex Pharmaceuticals Inc.: Employment. Lyons:Astex Pharmaceuticals Inc.: Employment. Hao:Astex Pharmaceuticals Inc.: Employment. Azab:Astex Pharmaceuticals Inc.: Employment. Kantarjian:Astex Pharmaceuticals Inc.: Research Funding. Kropf:Astex Pharmaceuticals Inc.: Research Funding. Issa:Astex Pharmaceuticals Inc.: Honoraria, Research Funding.