Cytarabine (1-β-D-arabinofuranosylcytosine; Ara-C) forms the backbone of chemotherapeutic regimes for acute myeloid leukemia (AML) and has a high success rate in inducing complete remission. However, the majority of AML patients will relapse within 2 years, often with an aggressive, chemoresistant form of the disease. Mechanisms associated with the development of relapsed disease remain to be fully elucidated, although recent whole-genome deep sequencing data derived from matched primary AML-relapse pairs demonstrated that relapse is associated with gain of novel mutations not seen at initial diagnosis [1] . Furthermore, the nature of relapse-specific mutations suggests these to be shaped, at least in part, by the chemotherapy used to induce and maintain remission.
In order to investigate this further we have assessed the mutagenicity of Ara-C. We show here that the frequency of induction of mutations at thymidine kinase (TK) and hypoxanthine phosphoribosyltransferase (HPRT) loci in vitro is increased following exposure to Ara-C, confirming this agent to be mutagenic in human cells. In addition, molecular characterization of Ara-C-induced mutations at the HPRT locus revealed that this commonly used nucleoside analogue induces a mutational fingerprint in vitro that can also be identified in vivo. Specifically, our data demonstrate that Ara-C is able to induce both deletion and base substitution mutations and has a strong propensity to induce mutations at 5′TGA3′/5′TCA3′ sequences in the genome, and particularly at the central G:C position consistent with its function as a cytosine analogue (p<0.001). Subsequent analysis of 4039 somatic AML mutations (3587 at disease presentation and 452 at relapse) derived from genome-wide deep sequencing data revealed that the frequency of putative Ara-C-associated G:C mutations in 5′TGA3′/5′TCA3′ sequences was significantly higher in AML relapse following Ara-C therapy compared to chemotherapy-naïve leukemia at initial presentation (OR 2.2, 95% CI 1.4-3.6; p=0.003).
Given that most cancer patients die from progressive disease, understanding the etiology of somatic mutations responsible for driving the evolution of relapse is a substantial challenge in the management of malignant disease. Although long suspected to exist, our study is the first to identify a chemotherapy-induced mutational fingerprint in a relapsing cancer and demonstrates that chemotherapy plays a role in driving evolution of the somatic cancer genome. The approaches we present could be used to discern the contribution of other treatments to evolution of relapsed disease, informing both personalized medicine and drug development strategies.
[1] Ding L et al. Nature 2012; 481(7382):506-10