Response to Hypomethylating Agent Therapy in Acute Myeloid Leukemia Based upon Mutations in the DNA Methylation Pathway

Background:

Acute myeloid leukemia (AML) is an aggressive malignancy with poor prognosis, especially in the elderly for whom hypomethylating agent therapy may be the only standard treatment option. Data from The Cancer Genome Atlas Research Network indicates 44% of patients with de novo AML carry mutations in genes that regulate DNA modifications, such as IDH1/2, DNMT3A, and TET2. Small retrospective series have reported differential response to hypomethylating agents (HMAs) in patients with mutations in the DNA modification pathway. For example, Itzykson et al. reported an increased response rate with azacitidine in MDS and AML patients with TET2 mutations compared with those without (82% vs 45%, p= 0.007). Next, Metzeler et al. reported an increase in CR rate with decitabine therapy in patients with DNMT3A mutations as compared with those without (75% vs 34% CR, p= 0.05). Emadi et al. noted a higher response rate in AML patients with IDH1/2 mutations compared to those without (71% vs 23%, p= 0.01) in patients receiving either decitabine or azacitidine. In contrast, Dinardo et al. did not find an association between responses to HMAs in patients with IDH1/2 and DNMT3A mutations in a retrospective analysis of 68 elderly AML patients. Recent evidence demonstrates that WT1 mutations lead to loss of TET2 function, suggesting that AML pts with WT1 mutations may also be sensitive to HMAs. These findings suggest that mutations in the TET/IDH/WT1/DNMT3A pathway might confer increased sensitivity to HMAs. We aimed to determine if there was an association between response and mutations in the DNA methylation pathway in AML patients treated with HMAs at our institution.

Methods:

This is a single institution, retrospective study. Of a cohort of 288 molecularly characterized AML patients, 76 patients were treated with HMAs; either alone (n= 59) or in combination with another agent (n= 17). In our cohort, 39 patients were treated with HMA in the frontline setting and 37 were treated in the relapsed/refractory setting.

DNA was obtained from peripheral blood or bone marrow aspiration samples.  Testing was performed using a targeted enrichment NGS assay designed with the RainDance DeepSeq system, covering clinically relevant regions in 30 genes known to be mutated in myeloid malignancies.  Samples were subjected to microdroplet emulsion PCR and sequenced on an Illumina MiSeq to an average depth of at least 500X. 51.3% of patients had mutational profiling at time of diagnosis, and the remainder had mutational profiling at time of relapse/refractory disease. Genes affecting DNA modification pathway (DNMT3A, IDH1/2, TET2, WT1) were analyzed as a composite (present/absent).

Results:

Among patients receiving HMA for AML, the mean age was 67.1 years and 68% of patients were male. Using the NCCN cytogenetic/molecular risk categories, 7/76 had “better” risk, 35/76 had “intermediate” risk, 28/76 had “poor” risk and 6/76 patients’ risk status could not be determined.  Four of the 76 patients had incomplete molecular data available and were excluded from analysis; therefore, the final cohort under study was 72 patients. 34/39 of the frontline HMA patients were evaluable for response and 32/37 of the relapsed/refractory patients were evaluable for response.

Among the 34 patients treated with HMAs in the frontline setting, 16 had mutations affecting DNA methylation and 18 patients did not. The CR rate (CR + CRi) in patients with DNA pathway mutations was 8/18 (44%) vs. 3/16 (19%) in patients without mutations (p= 0.15). Among the patients treated with HMAs in the relapsed/refractory setting, responses were rare with only 3 patients (9.4%) achieving CR. Responses were too few to demonstrate statistical relationships.

Conclusions:

We noted a trend toward an improved CR rate in patients with DNA modification mutations. However, the difference was not statistically significant in this small study cohort. Given the rarity of mutations and size of published cohorts, including ours, it has been difficult to demonstrate statistically significant associations between mutations and response to HMAs. A meta-analysis of existing data may overcome barriers of small numbers in individual studies, though retrospective nature of studies and publication bias may influence these results. A prospective trial with comprehensive molecular profiling of all patients is warranted to determine true associations.