Genomic and epigenomic predictors of response to guadecitabine in relapsed/refractory acute myelogenous leukemia

Abstract

Background: Guadecitabine is a novel DNA methyltransferase (DNMT) inhibitor with improved pharmacokinetics and clinical activity in a subset of patients with relapsed/refractory acute myeloid leukemia (r/r AML), but identification of this subset remains difficult.

Methods: To search for biomarkers of response, we measured genome-wide DNA methylation, mutations of 54 genes, and expression of a panel of 7 genes in pre-treatment samples from 128 patients treated at therapeutic doses in a phase I/II study.

Results: Response rate to guadecitabine was 17% (2 complete remission (CR), 3 CR with incomplete blood count recovery (CRi), or CR with incomplete platelets recovery (CRp)) in the phase I component and 23% (14 CR, 9 CRi/CRp) in phase II. There were no strong mutation or methylation predictors of response. Gene expression clustering defined a subset of patients (~ 20%) that had (i) high DNMT3B and low CDKN2B, CTCF, and CDA expression; (ii) enrichment for KRAS/NRAS mutations; (iii) frequent CpG island hypermethylation; (iv) low long interspersed nuclear element 1 (LINE-1) hypomethylation after treatment; and (v) resistance to guadecitabine in both phase I (response rate 0% vs. 33%, p = 0.07) and phase II components of the study (response rate 5% vs. 30%, p = 0.02). Multivariate analysis identified peripheral blood (PB) blasts and hemoglobin as predictors of response and cytogenetics, gene expression, RAS mutations, and hemoglobin as predictors of survival.

Conclusions: A subset of patients (~ 20%) with r/r AML is unlikely to benefit from guadecitabine as a single agent. In the remaining 80%, guadecitabine is a viable option with a median survival of 8 months and a 2-year survival rate of 21%.

Trial registration: NCT01261312 .

Keywords: AML; DNA methylation inhibitor; Drug resistance; Gene expression; Guadecitabine; Mutations.

Conflict of interest statement

XYS and MA are employees of Astex Pharmaceuticals. GJR and J-PJI received consulting fees from an advisory committee, and J-PJI received research funding from Astex Pharmaceuticals. WC, ADK, PK, HF, JJ, and HMK declare that they have no competing interests.

Figures

Fig. 1

DNA methylation and response to guadecitabine. a Unsupervised hierarchical clustering of 116 r/r AML patients based on 2774 hypervariable sites (standard deviation > 10%) divided the cases into three clusters which were normal-like, CIMP-like, and intermediate methylation. b Kaplan-Meier survival analysis of 116 cases based on the clusters derived in a. There was a trend for longer survival in the “normal-like” cluster, but this trend was not statistically significant (p = 0.21 by log-rank test)

Fig. 2

Mutation spectrum and response to guadecitabine. a Association of mutation spectrum, clinical characteristics, and response to guadecitabine in r/r AML (n = 122). Genomic mutation analysis was performed using the TruSight Myeloid Sequencing Panel (Illumina). The gene rows in the graph represent individual genomic lesions, the clinical characteristics row represent simplified clinical information, and the columns represent patients in the study. Black in the gene row indicates the presence of a specified mutation in a patient, and colors in the clinical characteristics row represent low (blue) to high (red). b Comparison of the mutation rate of CR vs. non-CR patients. None of the genes showed a significant correlation with CR, but there was a strong trend for RAS mutations (N or K) to be associated with resistance to guadecitabine (CR were seen in 0/22 patients with RAS mutations compared to 15/100 patients without RAS mutations, p = 0.07). c Kaplan-Meier survival analysis stratified by RAS mutation status. The presence of RAS mutations was associated with a significantly worse survival (p = 0.0004 by log-rank test)

Fig. 3

Selected gene expression profile and response to guadecitabine. a Unsupervised hierarchical clustering by baseline expression of the 7 gene panel grouped the phase I patients into two clusters (n = 27). Cluster R patients were clearly resistant to guadecitabine, with CRc seen in 0/9 patients compared to 5/18 patients for cluster S (p = 0.14). b Similar analyses of phase II patients (n = 95). These cluster R patients also showed resistance to guadecitabine (CRc seen in 1/21 patients compared to 22/74 patients for cluster S (p = 0.02), thus confirming the initial data in phase I patients. c A combined analysis of all 122 patients refined the clusters; 27 patients (17.2%) were in cluster R and had lower responses to guadecitabine (CRc seen in 0/27 patients compared to 28/95 patients for cluster S (p = 0.0005)). d Kaplan-Meier survival analysis of all patients by the clusters derived in c. Cluster R (n = 27) had a significantly worse survival after guadecitabine treatment (p = 0.003 by log-rank test)

Fig. 4

Univariate and multivariate COX regression to study survival after guadecitabine. a In univariate COX regression analyses, significant factors in univariate analyses were PB blasts (HR = 1.01, 95% CI 1.01–1.02, p < 0.0001), z4-score (HR = 0.85, 95% CI 0.78–0.91, p < 0.0001), BM blasts (HR = 1.02, 95% CI 1.01–1.02, p = 0.0002), RAS mutation (HR = 2.53, 95% CI 1.45–3.92, p = 0.0004), cytogenetic risk (HR = 2.18, 95% CI 1.41–3.17, p = 0.0004), cluster R (HR = 2.03, 95% CI 1.26–3.17, p = 0.003), hemoglobin value (HR = 0.86, 95% CI 0.75–0.99, p = 0.03), IDH2 mutation (HR = 0.47, 95% CI 0.23–0.98, p = 0.044), and mutated gene number (HR = 1.20, 95% CI 1.00–1.43, p = 0.048). b In a multivariate analysis by backward regression, cytogenetic risk (HR = 2.25, 95% CI 1.46–3.48, p = 0.0003), z4-score (HR = 0.89, 95% CI 0.81–0.97, p = 0.01), the presence of RAS mutation (HR = 2.12, 95% CI 1.19–3.76, p = 0.01), and hemoglobin value (HR = 0.86, 95% CI 0.74–0.99, p = 0.04) were significant predictors of survival