Acute myeloid leukemia (AML) is a heterogeneous disease resulting in different outcomes for patients after chemotherapy treatment. The variation in prognosis of AML patients depends on multiple intrinsic factors, including karyotype and molecular alterations as well as the expression of stem cell genes.1, 2 MicroRNAs (miRNAs) have emerged as epigenetic regulators affecting both normal as well as pathological processes by regulating protein expression.3 During formation of the hematopoietic system specific miRNAs are expressed at particular developmental stages, where they can regulate and control differentiation of hematopoietic stem cells (HSC) into lineage determined cells.4 The aberrant expression of miRNAs can influence apoptosis, block hematopoietic differentiation and induce myeloproliferative disorders and acute leukemias.5, 6, 7, 8 In AML, aberrant expression of miRNAs can contribute to the character of the disease and consequently is associated with treatment outcome.9
Recently, we published miRNA expression profiles of residual HSC, leukemic stem cells and leukemic progenitor fractions which were all obtained from the same AML patients bone marrow.10 Within these profiles, we found MIR551B highly expressed in HSC but absent in the leukemic fractions of most AML cases, suggesting MIR551B to be HSC specific.
To gain insight in the expression of MIR551B in the various hematopoietic cell populations of healthy bone marrow we purified six well characterized stem and progenitor cell populations together with mature monocytes and lymphocytes (Supplementary Figures 1A and B) from four healthy individuals and determined MIR551B expression in these populations (Figure 1a). The highest expression of MIR551B was seen in the most primitive cell populations; the HSC and multipotent progenitors (MPP), while more differentiated progenitors showed decreased expression and monocytes and lymphocytes lacked MIR551B expression. This expression pattern might indicate a role for MIR551B in early hematopoiesis and stem cells.
As we showed that MIR551B is highly expressed in normal HSC and MPP, we hypothesized that its expression in AML might be indicative for an immature leukemia with stem cell features. As a consequence AML cases with high MIR551B expression might be more chemotherapy resistant, more potent in re-initiating leukemia and associated with a poorer outcome. We measured MIR551B expression in 154 diagnosis AML patients (⩽60 years) and observed that patients with high expression have similar expression levels as HSC/MPP in healthy individuals and those residing within AML bone marrows (Supplementary Figure 1C). In these patients (Supplementary Table 1, discovery cohort), the expression of MIR551B is highly predictive for poor overall survival (OS) (hazard ratio (HR)=1.89, P=0.005) (Figure 1b) and relapse free survival (RFS) (HR=2.30, P=0.003) (Figure 1c). Moreover, patients with high MIR551B expression showed a decreased complete remission rate after the first cycle of standard-dose remission induction therapy (45 vs 68%, P=0.015)(data not shown) compared with patients with low MIR551B. We confirmed this result in patients (⩽60 years) of a secondary independent AML cohort (The Cancer Genome Atlas, TCGA; https://tcga-dat.nci.nih.gov/tcga/).11 Although, patients from the TCGA cohort have a worse outcome than those from our discovery cohort, and optimal cutoff values for high and low MIR551B expression and effect on outcome are different in the two cohorts (Supplementary Figures 2A and B for the discovery and Supplementary Figures 2C and D for the TCGA cohort), high MIR551B expression is also associated with poor OS (HR=2.02, P=0.028) (Figure 1d) and RFS (HR=2.17, P=0.030) (Figure 1e) in the TCGA cohort. AML patients from the TCGA cohort with high MIR551B expression also have significant lower complete remission rates compared with patients with low expression (71 vs 91%, P=0.023)(data not shown).
Importantly, in the patient subgroup with cytogenetically normal AML and an intermediate prognosis, high MIR551B expression predicts for both OS (HR=2.52, P=0.037) (Figure 2a) and RFS (HR=3.61, P=0.010) (Figure 2b) in the discovery cohort and for RFS (HR=2.93, P=0.034) in the TCGA cohort (Figures 2c and d).
Since AML patients above 60 years have an extremely poor prognosis identification of biomarkers for this group is needed. Therefore, we studied the association of MIR551B with prognosis in the group of patients above 60 years of age (TCGA cohort); however, did not observe a predictive value of MIR551B expression in this patient population (Supplementary Figures 3A and B).
To establish whether MIR551B expression can independently predict for survival and relapse in AML patients we performed univariate and multivariate cox regression analysis. In univariate analysis, MIR551B expression has prognostic value for prediction of OS and RFS in the discovery cohort (Supplementary Table 2A) and the TCGA cohort (Supplementary Table 2C). The HR values are comparable to the predictive value of cytogenetics, EVI1 and presence of a FLT-ITD. Multivariate cox regression analysis showed that MIR551B is not only an independent prognostic factor in predicting OS (HR=1.97, P=0.036) and RFS (HR=3.11, P=0.009) for the discovery cohort (Supplementary Table 2B) but also for OS (HR=2.02, P=0.054) and RFS (HR=2.11, P=0.049) for the TCGA cohort (Supplementary Table 2D). In univariate analysis, MIR551B expression has prognostic value for prediction of OS and RFS in cytogenetically normal AML from the discovery cohort (Supplementary Table 2E) but most importantly, in this patient group, MIR551B expression independently predicts for OS (HR3.24, P=0.026) (Supplementary Table 2F).
We divided the patients from both cohorts (discovery cohort; ⩽60 years and TCGA cohort; all ages) into high and low MIR551B expressing groups and investigated the clinical and molecular characteristics associated with high MIR551B expression. Lower white blood cell numbers, less NPM1 mutations and less FLT3-ITDs are observed in patients with high MIR551B (Supplementary Tables 1 and 3). Except for one patient, MIR551B was not expressed in core binding factor AML (favorable cytogenetics) or in AML with PML-RARA translocations. Moreover, patients with high MIR551B expression have more often an undifferentiated AML (FAB M0) and are more frequently positive for EVI1 (Supplementary Tables 1 and 3).
AML patients with low minimal residual disease (MRD) levels have lower relapse rates and a better OS than patients with high MRD.12, 13 Still, a substantial part of AML patients with low MRD levels relapses,13 indicating the need for additional factors predicting relapse in this group. Although we have only MRD data from a small subgroup of our patient cohort (n=52), we observed that high expression of MIR551B predicts for a poor OS (not significant, Figure 2e) and relapse (HR=6.28, P=0.01, Figure 2f) in patients without MRD after chemotherapy. Patients negative for MRD but with high expression of MIR551B have a median RFS time of 11.5 as compared with 50.0 months for patients without MRD and low MIR551B expression.
As MIR551B is highly expressed in HSC/MPP as well as in minimally differentiated and EVI1-positive AML its expression in AML might coincide with the expression of stem cell genes. From 179 AML cases within the TCGA AML patients cohort mRNA and miRNA sequencing data is available. In this cohort, MIR551B expression is significantly correlated with the expression of 1766 genes of which 1383 were positively and 383 were negatively associated (data not shown, Supplementary Figure 4A). The pseudogene EGFEM1P is the strongest positively co-expressed gene, suggesting simultaneous transcriptional regulation since MIR551B is located within the EGFEM1P gene. Comparison of these co-expressed genes with previously published HSC2 and minimally differentiated AML (FAB M0)14 gene expression profiles showed significant overlap (Supplementary Figures 4B and C). To compare the predictive value of MIR551B with that of a ‘core enriched’ HSC–leukemic stem cells signature,2 we determined the median expression of the core enriched genes in patients from the TCGA cohort and split the cohort in a high and low core enriched-signature expressing group (Supplementary Figure 5A). Univariate analysis (in patients ⩽60 years) showed that the HSC/leukemic stem cells signature has higher predictive value than expression of MIR551B for both OS and RFS (Supplementary Figure 5B). Interestingly, there is a significant positive correlation of MIR551B expression with expression of the core enriched genes (Supplementary Figures 5C and D). We also searched for miRNAs associated with MIR551B expression in AML and showed that the expression of 49 miRNAs is significantly positively correlated with MIR551B (Supplementary Table 4). From these 49 we observed that the top ones, MIR151, MIR126, Let-7c, MIR130A and MIR125B, are specifically expressed in residual HSC and/or leukemic stem cells within AML.10
Altogether, we identified that expression of MIR551B in AML is associated with stem cell features, an immature phenotype and predicts for relapse and a poor prognosis. Expression of MIR551B in AML might oppose stem cell features on leukemic cells however it might well be that in AML cases with high MIR551B expression AML-initiating mutations have their origin in HSC.15 We determined MIR551B expression in nine AML cell lines and found that expression was highest in the two most immature cell lines, KG1 and KG1a (Supplementary Figure 6A). To determine whether MIR551B can induce therapy resistance and/or a block in differentiation we overexpressed MIR551B in AML cell lines and measured cell survival and differentiation after incubation with chemotherapeutics or differentiation inducers. An increase in MIR551B expression (Supplementary Figures 6B and C) did not result in a change in sensitivity toward cytarabine, doxorubicin or etoposide (Supplementary Figures 6D and F). Overexpression of MIR551B in NB4, THP1 or HEL did not result in a change in all-trans retinoic acid or 12-O-tetradecanoylphorbol-13-acetate induced differentiation (Supplementary Figures 7A and D). However, differentiation induced by 12-O-tetradecanoylphorbol-13-acetate in HEL cells leads to downregulation of MIR551B expression (Supplementary Figure 7E), suggesting that MIR551B is a marker indicating immaturity of AML cells.
Recently, Xu et al.16 showed that MIR551B regulates expression of catalase and subsequently potentiated mucin1 expression. This modulation of the catalase/mucin1 pathway affected chemoresistance in lung cancer cells. We determined mRNA and protein expression of CAT and MUC1 in our AML cell lines overexpressing MIR551B (Supplementary Figure 6C) and found a small but significant downregulation of CAT mRNA and upregulation of MUC1 mRNA in two out of five cell lines (data not shown). Only KG1 showed both downregulation of CAT and enhanced expression of MUC1. Importantly, using a proteomic approach we did not observe modulation of catalase and mucin1 protein after MIR551B overexpression (data not shown) nor did we observe that expression of CAT and/or MUC1 is associated with MIR551B expression in AML patients (TCGA cohort). We hypothesize that the MIR551B/catalase/mucin1 signaling route might be present in AML however, may not be solely responsible for chemotherapy resistance.
HSC are intrinsically more resistant to chemotherapy compared with lineage restricted progenitors, which is partly due to their quiescent state, enhanced expression of efflux pumps and strong adherence to the chemotherapy protective microenvironment of the bone marrow niche in which these cells reside. We hypothesize that AML that arises from a mutation in the most immature cells of the hematopoietic system will inherit the expression of MIR551B as well as the intrinsic chemotherapy resistance and self-renewal potency of those cells and consequently AML cases with high MIR551B expression have a less favorable outcome.
Although, the size of the analyzed groups of patients with cytogenetically normal AML or known MRD status is small, our data suggests that MIR551B expression analysis can select for patients with a poor prognosis within these subgroups and MIR551B expression analysis might therefore add to risk stratification and therapy decision making for these subgroups of AML patients.
Löwenberg B . Acute myeloid leukemia: the challenge of capturing disease variety. Hematology Am Soc Hematol Educ Program 2008; 2008: 1–11.
Eppert K, Takenaka K, Lechman ER, Waldron L, Nilsson B, van Galen P et al. Stem cell gene expression programs influence clinical outcome in human leukemia. Nat Med 2011; 17: 1086–1093.
Bartel DP . Micrornas: genomics, biogenesis, mechanism, and function. Cell 2004; 116: 281–297.
Chen C-Z, Li L, Lodish HF, Bartel DP . Micrornas modulate hematopoietic lineage differentiation. Science 2004; 303: 83–86.
Gocek E, Wang X, Liu X, Liu C-G, Studzinski GP . Microrna-32 upregulation by 1,25-dihydroxyvitamin d3 in human myeloid leukemia cells leads to bim targeting and inhibition of arac-induced apoptosis. Cancer Res 2011; 71: 6230–6239.
Han YC, Park CY, Bhagat G, Zhang J, Wang Y, Fan JB et al. Microrna-29a induces aberrant self-renewal capacity in hematopoietic progenitors, biased myeloid development, and acute myeloid leukemia. J Exp Med 2010; 207: 475–489.
Bousquet M, Harris MH, Zhou B, Lodish HF . Microrna mir-125b causes leukemia. Proc Natl Acad Sci USA 2010; 107: 21558–21563.
O'Connell RM, Rao DS, Chaudhuri AA, Boldin MP, Taganov KD, Nicoll J et al. Sustained expression of microrna-155 in hematopoietic stem cells causes a myeloproliferative disorder. J Exp Med 2008; 205: 585–594.
Marcucci G, Maharry KS, Metzeler KH, Volinia S, Wu YZ, Mrózek K et al. Clinical role of micrornas in cytogenetically normal acute myeloid leukemia: mir-155 upregulation independently identifies high-risk patients. J Clin Oncol 2013; 31: 2086–2093.
de Leeuw DC, Denkers F, Olthof MC, Rutten AP, Pouwels W, Schuurhuis GJ et al. Attenuation of microrna-126 expression that drives CD34+38- stem/progenitor cells in acute myeloid leukemia leads to tumor eradication. Cancer Res 2014; 74: 2094–2105.
Cancer Genome Atlas Research Network. Genomic and epigenomic landscapes of adult de novo acute myeloid leukemia. N Engl J Med 2013; 368: 2059–2074.
Feller N, van der Pol MA, van Stijn A, Weijers GW, Westra AH, Evertse BW et al. MRD parameters using immunophenotypic detection methods are highly reliable in predicting survival in acute myeloid leukaemia. Leukemia 2004; 18: 1380–1390.
Terwijn M, van Putten WL, Kelder A, van der Velden VH, Brooimans RA, Pabst T et al. High prognostic impact of flow cytometric minimal residual disease detection in acute myeloid leukemia: data from the hovon/sakk aml 42a study. J Clin Oncol 2013; 31: 3889–3897.
Silva FP, Swagemakers SM, Erpelinck-Verschueren C, Wouters BJ, Delwel R, Vrieling H et al. Gene expression profiling of minimally differentiated acute myeloid leukemia: M0 is a distinct entity subdivided by runx1 mutation status. Blood 2009; 114: 3001–3007.
Shlush LI, Zandi S, Mitchell A, Chen WC, Brandwein JM, Gupta V et al. Identification of pre-leukaemic haematopoietic stem cells in acute leukaemia. Nature 2014; 506: 328–333.
Xu X, Wells A, Padilla MT, Kato K, Kim KC, Lin Y . A signaling pathway consisting of miR-551b, catalase and MUC1 contributes to acquired apoptosis resistance and chemoresistance. Carcinogenesis 2014; 35: 2457–2466.
Part of this research was financed by a grant (08-0075) from World Wide Cancer Research.
The authors declare no conflict of interest.
Supplementary Information accompanies this paper on the Leukemia website
About this article
Cite this article
de Leeuw, D., Verhagen, H., Denkers, F. et al. MicroRNA-551b is highly expressed in hematopoietic stem cells and a biomarker for relapse and poor prognosis in acute myeloid leukemia. Leukemia 30, 742–746 (2016). https://doi.org/10.1038/leu.2015.160
BMC Cancer (2018)
Clinical and Translational Oncology (2017)
Scientific Reports (2016)