Somatic Mutations Could Predict Response to Hypomethylating Agents in MDS

A biomarker model that incorporates the mutational status of multiple somatic genes could be used to predict response to hypomethylating agents for patients with myelodysplastic syndromes.

Aziz Nazha, MD

A biomarker model that incorporates the mutational status of multiple somatic genes could be used to predict response to hypomethylating agents for patients with myelodysplastic syndromes (MDS), according to findings from a small pilot study presented at the 2015 International MDS Symposium.

The new predictive model builds upon previous research showing that mutations in TET2, DNMT3A, and ASXL1 could predict response to hypomethylating agents. Based on the presence or absence of these and other somatic mutations, the new model was able to identify a 95% probability of response to hypomethylating agents, such as decitabine and azacitidine.

"Approximately 40% to 50% of patients respond to treatment, and the majority of patients take 6 months or more to achieve this response," said lead author Aziz Nazha, MD, a hematologist in the Leukemia Program at the Cleveland Clinic, Taussig Cancer Institute. "Identifying biomarkers that could predict response is very important clinically because it would prevent prolonged exposure to ineffective therapy, avoid toxicities, and decrease unnecessary treatment costs."

The study included 122 patients with MDS (73%), MDS/myeloproliferative disease (17%), and secondary acute myelogenous leukemia (10%). Patients were treated with a median of 7 cycles of single-agent azacitidine (47%) or decitabine (35%). Overall, 18% of patients had received both treatments.

The majority of individuals studied were male (66%) with a median age of 69 years. Fifty-two percent of patients had high or intermediate-2 risk disease by IPSS. By IPSS-R, 27% were intermediate and 14% were high/very high risk. Next-generation and whole exome sequencing was utilized to examine a panel of 62 genes.

Responses to hypomethylating therapy consisted of complete responses (CR), including bone marrow CR, in 22% of patients. The partial response rate was 3%, and 23% of patients experienced a 5% hematologic improvement. Based on these data, patients were divided into responder (n = 60) and nonresponder (n = 62) groups.

In total, 74% of patients had at least 1 gene mutation. The most commonly mutated genes were: ASXL1 (15%), TET2 (12%), DNMT3A (12%), SF3B1 (12%), U2AF1 (12%), RUNX1 (11%), EZH2/EED/SUZ12 (11%), STAG2/RAD21 (8%), SETBP1 (6%), CBL (7%), and IDH1/IDH2 (5%). Overall, mutations in TP53, EZH2, and ASXL1 were associated with a worse prognosis and shorter overall survival, Nazha noted.

"For the comparison of responders and nonresponders, to overcome some of the small number of somatic mutations we combined them into their respective pathways. However, if you look at each separately, they are still significant," he said. "TET2/IDH1/IDH2 as well as mutations in STAG2/RAD21 predicted response to hypomethylating agents."

TET2/IDH1/IDH2 alterations were present in 25% of responders versus 10% of those who did not respond (P = .03). Additionally, mutations in the cohesin genes STAG2/RAD21 were found in 13% of responders versus 3% of nonresponders (P = .05). TP53 mutations were absent in nonresponders but were discovered in 7% of those who responded to therapy (P = .05).

"There is some evidence to suggest that response to hypomethylating agents is TP53 independent; however, this patient cohort is small and made it difficult to derive a clinical conclusion on this," Nazha explained.

Mutations in ASXL1 were found in 21% of nonresponders compared with 8% of responders (P = .07). Additionally, U2AF1 alterations were more common in non-responders compared with responders (8% vs 16%; P = .27).

A multivariate regression analysis was applied to the mutation data, adjusting for age and IPSS score. From this analysis it was established that TET2/IDH1/IDH2 (odds ratio [OR] = 3.07; P = .04), STAG2/RAD21 (OR = 7.48; P = .03), and ASXL1 (OR = 0.19; P = .02) were independent prognostic factors for response to hypomethylating agents.

Table. Probability of Response Based on Somatic Mutation Score


Probability of Response



















The fitted coefficients from each of these factors were obtained from linear predictive scoring to produce a formula that was based on the presence or absence of each somatic mutation. In the scoring formula, the presence of TET2/IDH1/IDH2 mutations equaled 1.0, STAG2/RAD21 mutations were 2.0, and ASXL1 alterations were defined as -1.5 (Table).

"If you have a higher score you have a higher probability of response," Nazha explained. "Patients with mutations within DNA methylation as well as mutations on the cohesins have around a 95% chance of responding."

Work is underway to enroll patients into a validation study, Nazha said. Additionally, further analysis is being conducted on the current datasets. If validated, the predictive model could be used to guide treatment selection for patients with MDS.

Nazha A, Radivoyevitch T, Thota S, et al. Somatic mutational model to predict response to hypomethylating agents in myelodysplastic syndromes. Leukemia Research. 2015;39:1s (suppl; abstr 20).


View more from the 2015 MDS Symposium