Acute myeloid leukemia (AML) is an aggressive malignancy with median age at diagnosis of 67 years,1 stressing the importance of developing treatments that are both effective and tolerable in elderly patients. Hypomethylating agents (HMAs) have been extensively studied in AML, typically in older adults who are deemed to be unfit for standard induction chemotherapy, demonstrating improved complete response (CR) rate and trends toward improvement in overall survival (OS) compared with conventional care regimens in the phase III setting with less favorable responses in the relapsed/refractory setting.32 Genome-wide studies of AML patients have revealed that 44% of patients with de novo AML harbor mutations in genes affecting DNA-methylation, including DNMT3A (26%), IDH1/2 (20%), TET2 (8%), and WT1 (6%).4 The presence of these mutations has been suggested to have therapeutic implications in small, retrospective series.85 Using data from two large referral centers together with previously reported data, we sought to investigate the relationship between somatic gene mutations affecting DNA methylation and HMA response in an expanded AML patient cohort. We did not observe a relationship between response to HMAs and IDH1/2 and TET2 mutations. We identified DNMT3A mutations to predict response to HMAs in patients treated in the frontline setting [odds ratio (OR), 3.12; P=0.001], but not in the total cohort when including relapsed/refractory patients (OR 1.72; P=0.23).
This is a dual institution, retrospective study. Permission to review medical records was obtained by the Institutional Review Board of each participating institution. From March 2010 to December 2014, 242 patients were identified at Memorial Sloan Kettering Cancer Center (MSKCC) who had a diagnosis of AML by World Health Organization (WHO) criteria and next-generation sequencing (NGS) performed. Of these, 62 evaluable patients were treated with HMAs alone or in combination. From May 2013 to October 2014, 82 patients were identified in the Moffitt Cancer Center (MCC) cohort with a diagnosis of AML by WHO criteria and NGS. HMA therapy was administered alone or in combination in 21 evaluable patients (Table 1). NGS was performed as previously described.109
A systematic search was conducted to identify studies examining the response to HMAs in patients with AML in relation to presence/absence of mutations in DNMT3A, IDH1/2, and/or TET2. Studies containing both AML and MDS patients were included provided that the number of AML patients exceeded 20 and individual patient level data for AML patients could be obtained. Studies analyzing HMAs in combination with a non-intensive agent were included; however, studies analyzing HMAs in combination with intensive induction chemotherapy were excluded. Two reviewers (CCC and DAS) independently performed literature searches using Pubmed and Ovid in order to assess study eligibility. Search terms included “acute myeloid leukemia” AND “hypomethylating agent” AND “azacitidine” AND “Vidaza” AND “Dacogen” AND “decitabine” AND “DNMT3A” AND “TET2” AND “IDH1” AND “IDH2” AND “isocitrate dehydrogenase” as Medical Subject Heading and text terms. Boolean operators (NOT, AND, OR) were also used in succession to narrow and widen the search. For included publications,85 references and subsequent citing publications were also reviewed.
We extracted data from selected publications regarding clinical parameters and response to HMA therapy (Table 2).85 For one publication containing data on both MDS and oligoblastic AML,5 we obtained primary data from study authors to determine individual treatment responses for AML patients. Logistic regression was used to estimate the association between mutation status and CR attainment both overall and among the subset of patients receiving HMA as part of frontline therapy. To account for potential correlation among treatment centers or publications, robust estimates of the variance were obtained using the sandwich estimator when reporting confidence intervals and associated P-values. As a sensitivity analysis, each regression model was repeated including site/publication as a fixed effect in the model. All analyses were conducted in R (v.3.2.3, R Foundation for Statistical Computing).
A total of 83 patients from MSKCC and MCC met the study criteria (Table 1). NGS was performed prior to initiation of HMA in 70% of patients. Our cohort was enriched for higher risk disease as therapy-related AML occurred in 17% of patients (n=14), and 36% patients (n=30) had AML that evolved from an antecedent hematologic disorder (4 were also considered as having therapy-related disease). The remaining 52% of patients (n=43) had de novo AML.
Among patients treated with HMAs in the frontline setting (n=45), we noted a 60% CR rate [including CR and incomplete blood count recovery (CRi)] in DNMT3A mutants compared to 33% of those with wild-type DNMT3A, although numbers were insufficient to allow statistical comparisons. We noted a trend toward a higher CR rate in IDH1/2 mutants compared to wild-type IDH1/2 (60% vs. 29%, odds ratio (OR) 3.67; P=0.08). We did not observe an association between TET2 mutation status and response (18% CR rate in TET2 mutants vs. 41% in wild-type TET2, OR 0.33; P=0.195). Given the mutual exclusivity of IDH1/2 and TET2 mutations, we compared response rates of patients with presence of IDH1/2 and TET2 to patients with neither mutation, but could not demonstrate an association with presence of mutation and CR rate (38% CR in mutants compared to 33% in non-mutants; P=0.62). When considering all patients (n=83) and treatment setting (frontline and relapsed/refractory), there were no statistically significant associations (Table 1).
We next sought to combine the MSKCC/MCC cohort with previously published cohorts examining the role of mutations and response to HMAs (Table 3). This included 197 patients with DNMT3A mutation status, 239 with IDH1/2 mutation status, and 152 with TET2 mutation status. In the frontline setting only, there was a statistically significant association between presence of DNMT3A mutation and attainment of CR [57% vs. 29%, OR 3.12 (1.63–5.94) with P=0.001]. Presence of mutation in both DNMT3A and NPM1 demonstrated a CR rate of 73% compared to 21% in patients without co-mutation of these genes [OR 2.82 (1.33–6.00) with P=0.007]. There was no correlation between CR rate and mutation status for IDH1/2 and TET2.
When examining all patients treated in both the frontline and relapsed/refractory setting, we noted a significantly higher CR in DNMT3A/NPM1 co-mutants compared to others [62% vs. 28%, OR 2.57 (1.18–5.94) with P=0.017]. However, there were no statistically significant associations for DNMT3A, IDH1/2, or TET2 when considered alone.
The mechanistic link between mutations in DNMT3A and increased response to HMAs has not been elucidated. The majority of DNMT3A mutations seen in AML are a heterozygous substitution at arginine 882, which reduces methyltransferase activity leading to global hypomethylation, making it counterintuitive that further reduction in methylation could predict clinical response.11 However, the mechanism of action of HMAs is poorly understood, as there is not a clear relationship between the amount of demethylation following HMA therapy and clinical response.12 Cancer cells are often hypermethylated at promoter regions, leading to down-regulated expression of tumor suppressor genes but hypomethylated at repetitive DNA regions, leading to chromosomal instability and to malignant transformation. Hence, current methods measuring global methylation may not be the optimal approach, as these depict the sum of hyper- and hypomethylation at respective sites.12
Limitations of this study include the retrospective nature and cohort heterogeneity of both our series and previously published series, as we included patients from multiple institutions treated in the frontline and relapsed/refractory setting, with both HMAs alone and in combination with additional drugs. While the majority of patients had samples at time of newly diagnosed AML, a subset of patients in our cohort had NGS later in their clinical course with a small subset having NGS performed after initiation of HMA, in which clonal evolution could conceivably have led to acquisition or loss of mutations. Given that IDH1/2 and TET2 mutations lead to a hypermethylated state,13 one would expect an association with mutation status in these genes and HMA response. We may be underpowered to determine true associations with IDH1/2, TET2, and HMA response.
In conclusion, we have identified a significantly higher CR rate in AML patients harboring DNMT3A mutations who were treated in the frontline setting with HMAs. This increased response rate was not observed when our analysis included patients in the relapsed/refractory setting. We also identified a statistically significant improved CR rate in DNMT3A/NPM1 co-mutants who were treated in both frontline and relapsed/refractory settings. To our knowledge, this is the largest dataset assembled to address the impact of mutations in epigenetic modifiers in AML patients treated with HMAs. Notably, the response rate reported here compares favorably with those reported in elderly AML patients treated with induction chemotherapy.14 In the context of the relatively favorable side-effect profile of HMAs, this may have important implications for therapeutic decision making when considering older patients who are borderline candidates for induction chemotherapy, especially as DNMT3A-mutant AML has been associated with improved responses to daunorubicin dose intensification.15 This series has not examined whether the increase in CR translates to an OS benefit. Notably, in the HMA phase III trials for AML, comprehensive molecular profiling including DNMT3A, IDH1/2, and TET2 was not reported, so prospective data regarding molecular predictors of HMA response are limited.32 Therefore, a prospective therapeutic trial utilizing HMAs with comprehensive molecular profiling of all patients at baseline is warranted with both CR and OS as end points.
- Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin. 2016; 66(1):7-30. PubMedhttps://doi.org/10.3322/caac.21332Google Scholar
- Kantarjian HM, Thomas XG, Dmoszynska A. Multicenter, randomized, open-label, phase III trial of decitabine versus patient choice, with physician advice, of either supportive care or low-dose cytarabine for the treatment of older patients with newly diagnosed acute myeloid leukemia. J Clin Oncol. 2012; 30(21):2670-2677. PubMedhttps://doi.org/10.1200/JCO.2011.38.9429Google Scholar
- Dombret H, Seymour JF, Butrym A. International phase 3 study of azacitidine vs conventional care regimens in older patients with newly diagnosed AML with >30% blasts. Blood. 2015; 126(3):291-299. PubMedhttps://doi.org/10.1182/blood-2015-01-621664Google Scholar
- Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med. 2013; 368(22):2059-2074. PubMedhttps://doi.org/10.1056/NEJMoa1301689Google Scholar
- Itzykson R, Kosmider O, Cluzeau T. Impact of TET2 mutations on response rate to azacitidine in myelodysplastic syndromes and low blast count acute myeloid leukemias. Leukemia. 2011; 25(7):1147-1152. PubMedhttps://doi.org/10.1038/leu.2011.71Google Scholar
- Emadi A, Faramand R, Carter-Cooper B. Presence of isocitrate dehydrogenase mutations may predict clinical response to hypomethylating agents in patients with acute myeloid leukemia. Am J Hematol. 2015; 90(5):E77-79. PubMedhttps://doi.org/10.1002/ajh.23965Google Scholar
- DiNardo CD, Patel KP, Garcia-Manero G. Lack of association of IDH1, IDH2 and DNMT3A mutations with outcome in older patients with acute myeloid leukemia treated with hypomethylating agents. Leuk Lymphoma. 2014; 55(8):1925-1929. PubMedhttps://doi.org/10.3109/10428194.2013.855309Google Scholar
- Metzeler KH, Walker A, Geyer S. DNMT3A mutations and response to the hypomethylating agent decitabine in acute myeloid leukemia. Leukemia. 2012; 26(5):1106-1107. PubMedhttps://doi.org/10.1038/leu.2011.342Google Scholar
- Sallman DA, Komrokji R, Vaupel C. Impact of TP53 mutation variant allele frequency on phenotype and outcomes in myelodysplastic syndromes. Leukemia. 2016; 30(3):666-673. PubMedhttps://doi.org/10.1038/leu.2015.304Google Scholar
- Cheng DT, Cheng J, Mitchell TN. Detection of mutations in myeloid malignancies through paired-sample analysis of microdroplet-PCR deep sequencing data. J Mol Diagn. 2014; 16(5):504-518. PubMedhttps://doi.org/10.1016/j.jmoldx.2014.05.006Google Scholar
- Russler-Germain DA, Spencer DH, Young MA. The R882H DNMT3A mutation associated with AML dominantly inhibits wild-type DNMT3A by blocking its ability to form active tetramers. Cancer Cell. 2014; 25(4):442-454. PubMedhttps://doi.org/10.1016/j.ccr.2014.02.010Google Scholar
- Voso MT, Santini V, Fabiani E. Why methylation is not a marker predictive of response to hypomethylating agents. Haematologica. 2014; 99(4):613-619. PubMedhttps://doi.org/10.3324/haematol.2013.099549Google Scholar
- Figueroa ME, Abdel-Wahab O, Lu C. Leukemic IDH1 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair hematopoietic differentiation. Cancer Cell. 2010; 18(6):553-567. PubMedhttps://doi.org/10.1016/j.ccr.2010.11.015Google Scholar
- Lowenberg B, Ossenkoppele GJ, van Putten W. High-dose daunorubicin in older patients with acute myeloid leukemia. N Engl J Med. 2009; 361(13):1235-1248. PubMedhttps://doi.org/10.1056/NEJMoa0901409Google Scholar
- Patel JP, Gonen M, Figueroa ME. Prognostic relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med. 2012; 366(12):1079-1089. PubMedhttps://doi.org/10.1056/NEJMoa1112304Google Scholar