The LincRNA HOTAIRM1, Located in the HOXA genomic Region, impacts Prognosis in Acute Myeloid Leukemia and Is Associated with a Distinctive microRNA Signature

Background: Non-coding RNAs (ncRNAs) have recently emerged as key regulators of diverse cellular processes, including leukemia. ncRNAs are classified according to their size as short (eg, microRNAs) or long ncRNAs. lincRNAs are long ncRNAs located in intergenic regions and have multiple regulatory functions, including gene expression regulation. Interestingly, active crosstalk between microRNAs and lincRNAs has been observed. lincRNAs are known to be deregulated in some cancers but their importance in acute myeloid leukemia (AML) is so far unknown. HOX genes play an important role in hematopoiesis and are deregulated in AML. lincRNAs are especially abundant in the clusters of HOX genes. HOTAIRM1, a myeloid lineage-specific lincRNA, is located at the 3’end of the HOXA cluster and seems to play a regulatory role in myelopoiesis. However, to date the potential prognostic role of HOTAIRM1 expression in AML has not been examined.

Aims: To investigate first whether the expression of the lincRNA HOTAIRM1 is associated with the clinical, cytogenetic and molecular characteristics and microRNA expression in AML patients. Secondly, since intermediate risk (IR) AML patients have a highly diverse prognosis, we analyzed the potential prognostic value of HOTAIRM1 expression in IR-AML patients.

Methods: To explore the expression level of HOTAIRM1 among different AML subtypes, we analyzed samples from 244 AML patients including CBF-rearranged AML (n=5), APL (n=4), MLL-rearranged AML (n=3), EVI1-rearranged AML (n=3), t(6;9) AML (n=9), AML with monosomal karyotype (n=3), and a large cohort of IR-AML (described below). For the analysis of prognostic value of HOTAIRM1, we analyzed specifically the outcome of 217 IR-AML patients (median age, 52; 51% males) sequentially included in CETLAM trials during the period 1995-2009. Molecular genotyping of this group identified NPM1 mutation (NPM1mut), FLT3-ITD, and biallelic CEBPA mutation (CEBPA mut) in 99 (45%), 79 (36%) and 17 (11%), respectively. The expression of HOTAIRM1 was analyzed using TaqMan® Gene Expression Assays (Applied Biosystems). microRNA and mRNA expression data were obtained in previous studies (Díaz-Beyá, Leukemia 2013). Statistical analyses were performed with BRB Array Tools, SPSS v20 and R v3.0. MaxStat package from R software was used to determine the optimal cutoff point of HOTAIRM1 expression.

Results: Among all 244 patients, HOTAIRM1 expression was significantly different among the 7 included genetic subgroups (ANOVA p=0.0024), with the lowest levels observed in APL-AML patients and the highest in the t(6;9)AML patients. Within the IR-AML group, HOTAIRM1 overexpression was observed in NPM1mut patients (p<0.001).

The prognostic study showed that high HOTAIRM1 expression was associated with shorter 5-year overall survival (OS) (27+11% vs.47+8%; p=0.009) shorter 5-year disease-free survival (LFS) (22+12% vs. 53+9%; p<0.001), and a higher cumulative incidence of relapse (CIR) at 5 years (55+15% vs. 34+8%; p=0.004). The effect on outcome was maintained within the subgroup with favorable molecular features (i.e., NPM1mut and CEBPAmut without FLT3-ITD) (OS: 75+11% vs. 39+29%; p=0.026). In the multivariate analysis including age, sex, WBC, NPM1mut, FLT3-ITD and number of treatment cycles for CR achievement as covariates, HOTAIRM1 expression emerged as an independent prognostic marker in OS (HR=2.44; 95% CI: 1.51-3.93; p<0.0001), LFS (HR=2.07; 95% CI: 1.31-3.24; p=0.002) and CIR (HR=2.05; 95% CI: 1.18-3.55; p=0.01).

Supervised analysis by means of t-test based on multiple permutations revealed a distinctive 33-microRNA signature which correlated with HOTAIRM1 expression including miR-196b (p<0.001) located in the HOXA genomic region. Moreover, we correlated the expression of HOX genes and HOTAIRM1 and observed a positive correlation with HOXA4 gene expression (R²= 0.6; p=0.001).

Conclusion: The expression level of the lincRNA HOTAIRM1 varied among different molecularly-defined AML. Interestingly, HOTAIRM1 expression level showed independent prognostic value within the IR-AML group. Moreover, HOTAIRM1 expression strongly correlates with its neighboring HOXA4 gene and harbors a distinctive microRNA signature. Our findings can pave the way for further studies of HOX-related lincRNAs and microRNAs regulatory networks and their influence on clinical outcome.

Acknowledgments: ISCIII RH13/00205, SEHH, FIS13/00999