Aberrant Expression of MIR9 Represents a Distinct Entity with Poor Prognosis in Acute Myeloid Leukemia

Abstract 2532

Inactivation of RUNX1 is critical in the development of AML. To test a hypothesis that dysregulation of RUNX1-inhibiting miRNAs could be another molecular basis, we screened expression levels of several candidate miRNAs whose seed sequences reside within the 3' UTR of human RUNX1 mRNA in bone marrow samples from 24 AML, 7 ALL and 2 mixed-lineage leukemia patients. Among them, two from AML patients and one from ALL patient exhibited extremely high levels of MIR9, while other leukemia samples showed hardly detectable levels. Based on the results, we further analyzed amounts of MIR9 in the other series of samples from 101 AML patients and 23 controls. All these AML patients were sequentially diagnosed in Dokkyo Medical University Hospital from 2000 to 2011 and selection of samples used in this study was strictly based on their availability.

Nineteen (19%) of the patients highly expressed MIR9 that was barely detected in other AML samples and controls. Therefore we divided the AML patients into two groups based on their expression levels of MIR9; MIR9 (+) and MIR9 (−) groups. Demographic, laboratory and clinical characteristics were compared between these two groups. Any parameter including sex, age, WBC count, percentage of blasts in the peripheral blood or bone marrow, LDH value or ratio of primary vs. secondary disease was not significantly different between these two groups. On the other hand, distribution of FAB subtypes (p=0.008) and chromosomal findings (p=0.017) statistically differed between them. It is worth noticing that no patients in the MIR9 (+) group were diagnosed as acute promyelocytic leukemia (APL) or core binding factor (CBF) leukemia carrying t(8;21) or inv(16)/t(16;16), while a substantial number of MIR9 (−) patients was diagnosed as the former (21%) and latter (15%), respectively.

The median observation period was 513 days. The median survival days were 349 and 574 in the MIR9 (+) and MIR9 (−) groups, respectively. Although complete remission (CR) rates after remission induction therapy were similar between the MIR9 (+) and MIR9 (−) groups, the median duration of CR was 267 days in the MIR9 (+) group, while 702 days in the MIR9 (−) group. In Kaplan-Meier analysis, MIR9 (+) group exhibited significantly inferior overall survival (OS) (Fig1; p=0.005) and relapse-free survival (RFS) (Fig2; p<0.001) compared with MIR9 (−) group. These data suggest a possibility that the shorter duration of CR caused the shorter OS in the MIR9 (+) group. In univariate Cox regression analyses, the presence of aberrant MIR9 expression was a strong predictor of decreased OS with a statistical significance (HR=2.602, p=0.006) along with age over 65 (HR=1.992, p=0.035), WBC count greater than 50 × 10⁹/L (HR=2.434, p=0.011) and LDH greater than 600 IU/L (HR=2.606, p=0.004). On the other hand, only the presence of aberrant MIR9 expression was a predictor of increased risk of relapse (HR=3.560, p=0.001). In multivariate Cox regression analyses, the presence of aberrant MIR9 expression kept independent prognostic significance for OS (HR=2.337, p=0.016) and RFS (HR=3.560, p=0.001).

It is important to note that there were no APL patients who aberrantly expressed MIR9. Subset analysis was performed for patients excluding APL patients, because this type of AML has been known to show better survival since the introduction of ATRA therapy. Also in this cohort of patients, the presence of MIR9 expression was demonstrated to be associated with unfavorable OS and RFS with statistical significances (Fig3; p=0.028 for OS, Fig4; p=0.009 for RFS) in Kaplan-Meier analysis. In Cox regression analyses, the presence of aberrant MIR9 expression kept independent prognostic significance for OS (HR=2.153, p=0.032 in univariate analysis and HR=2.088, p=0.040 in multivariate analysis) and RFS (HR=2.627, p=0.012 in univariate analysis and HR=2.627, p=0.012 in multivariate analysis). All these data indicate that aberrant expression of MIR9 represents a distinct entity with poor prognosis in AML.

Although we showed that MIR9 inhibited protein translation through the 3' UTR of RUNX1 mRNA in luciferase assays, we could not detect any significant repression of endogenous RUNX1 protein expression, when MIR9 was overexpressed in K562 and UT7/GM cells. These data suggest a possibility that RUNX1 mRNA could not be a major target of MIR9 in primary leukemia cells and that MIR9 might contribute to leukemogenesis through suppressing other target mRNAs.