Negative Prognostic Relevance of a Specific 3-Gene Cluster in Myelodysplastic Syndromes during Azacitidine and Lenalidomide Therapy

Background and Rationale. Azacitidine (AZA) is a standard first-line therapy in high-risk MDS. Also its combination with Lenalidomide (LEN) has been tested, but its molecular effect is still under investigation. Here we analyzed the effect of AZA+LEN therapy on gene mutations and microRNA expression in MDS patients. Patients and Methods. This study included 44 high-risk MDS patients treated with AZA (75 mg/m2/day, days 1-5, sc) and LEN (10 mg/day, days 1-21, orally) every 4 weeks. Patients showing complete remission (CR), partial remission (PR) or any hematologic improvement were considered as responders, while patients showing stable disease or disease progression were considered as non responders. Molecular analyses were performed at baseline and during the therapy. Gene mutations were studied by an Illumina Cancer Myeloid Panel and an Ion Torrent specific panel, whereas microRNAs expression was assessed using an Affymetrix miRNA 4.0 array. Results. 34/44 patients were considered evaluable for response, with an overall response rate of 76.25% (26/34 cases). 13 patients showed a positive response within the 4th cycle (T4) and maintained it at T8; 9 patients showed a positive response within T4 and lost response at T8; 4 patients responded after T4 and maintained the response at T8; 8 patients never responded. Molecular analyses were performed on serial samples (baseline, T4 and T8) available for 30 patients. Results from the Illumina analysis on cancer myeloid genes showed that 3/30 cases had no mutations at all, all other cases showed mutations both at baseline and during the therapy. The most frequently mutated genes were ASXL1 (14 cases = 47%), TET2 (11 cases = 37%), RUNX1 (8 cases = 27%) and SRSF2 (5 cases = 17%). All samples with a decreasing variant allele frequency (VAF) had a favourable response at T8 (CR, marrow CR or PR), while none of the non responders showed a decreasing VAF. Ion Torrent analysis of 24 inositide-specific genes showed that all patients had mutations both at baseline and during the therapy. Interestingly, all patients responding at T4 and losing response at T8, as well as cases that did not respond, acquired the same 3 point mutations at T8, affecting respectively PIK3CD (D133E), AKT3 (D280G) and PLCG2 (Q548R) genes. Patients responding at T4 and losing response at T8 showed these mutations even at T4. Kaplan-Meier analyses revealed that the presence of these mutations was significantly associated with a decreased duration of therapy (39.5 vs 8.5 months; p<0.05) and duration of response (36 vs 6 months; p<0.05). As for microRNA profiling, paired analysis between responders and non responders showed specific clusters of up- or down-regulated microRNAs. Interestingly, unpaired analysis on patients responding at T4 and losing response at T8 showed 18 up- and 11 down-regulated microRNAs, like miR-3613-3p and miR-6757-5p, whose predicted targets are our 3 genes among the others. Also in patients never responding to the therapy there was a specific cluster of 3 up- and 12 down-regulated microRNAs and, interestingly, 7 of these microRNAs, like miR-4786-5p or miR-6853-3p, targeting our 3-gene cluster among the others, were altered also in patients losing response. Conclusions. Our results show that the presence of a common cluster of point mutations affecting 3 inositide-specific genes (PI3KCD, AKT3, PLCG2, all regulating cell proliferation), is significantly associated with loss of response to AZA+LEN therapy. Moreover, also a cluster of 7 microRNAs, targeting our 3 genes among the others, is associated with unfavourable outcome. Further studies are warranted to confirm these data, to further analyze the role of this 3-gene cluster and to identify the specific targets for the dysregulated microRNAs identified.