Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Azacitidine (AZA) is a nucleoside analog used for treatment of myelodysplasia and the prediction of AZA responsiveness is important for the therapy management.

METHODS:

Using microarrays and reverse-transcription quantitative-PCR, we analyzed microRNA (miRNA) expression in bone marrow CD34+ cells of 27 patients with higher-risk myelodysplastic syndromes or acute myeloid leukemia with myelodysplasia-related changes before and during AZA treatment.

RESULTS:

At baseline, we found that future overall response rate was significantly higher in patients with upregulated miR-17-3p and downregulated miR-100-5p and miR-133b. Importantly, the high level of miR-100-5p at baseline was associated with shorter overall survival (HR $=$ 4.066, $P=$ 0.008). After AZA treatment, we observed deregulation of 30 miRNAs in responders (including downregulation of miR-10b-5p, miR-15a-5p/b-5p, miR-24-3p, and miR-148b-3p), while their levels remained unchanged in non-responders.

CONCLUSIONS:

Our study demonstrates that responders and non-responders have distinct miRNA patterns and that the level of specific miRNAs before therapy may predict the efficacy of AZA treatment.

Keywords

Myelodysplastic syndromes microRNA azacitidine response prediction

1. Introduction

Myelodysplastic syndromes (MDS) are a heterogeneous group of hematopoietic stem cell disorders characterized by ineffective hematopoiesis leading to morphologic dysplasia and peripheral blood cytopenia in one or more lineages. Approximately one-third of all patients with MDS transform to acute myeloid leukemia with myelodysplasia-related changes (AML-MRC) [1]. Prognosis of MDS is assessed based on the Revised International Prognostic Scoring System (IPSS-R) and depends on the number and severity of cytopenia, percentage of bone marrow (BM) blasts, and cytogenetics [2].

Epigenetic changes play a significant role in the pathogenesis of MDS and AML [3, 4], and a therapy with hypomethylating agents (HMAs) such as azacitidine (5-azacytidine, AZA) or decitabine (5-aza-2’-deoxycytidine, DEC) is currently considered the standard therapy for higher-risk MDS and AML-MRC [5, 6]. AZA is a cytidine analog which, at low doses, functions as a DNA methyltransferase inhibitor causing DNA hypomethylation, thus appearing to work by preventing the inactivation of tumor suppressor genes. At high doses, AZA is directly cytotoxic to abnormal BM hematopoietic cells because of its incorporation into DNA and RNA, resulting in cell death [7]. AZA has been shown to prolong patient survival, improve clinical outcomes and quality of life, and delay progression to AML [5, 8]. However, the overall response rate (ORR) is only between 40–50% [9], and the outcome of patients after AZA treatment failure is very poor. Therefore, the identification of biomarkers predictive of response to HMAs is an area of intensive research.

MicroRNAs (miRNAs) are non-coding RNAs of $\sim$ 21 nucleotides in length that regulate gene expression at the post-transcriptional level. They play important roles in numerous cellular processes and diseases, including hematopoiesis and MDS [10, 11, 12]. To date, only a few studies have investigated miRNA expression with respect to HMA treatment in MDS and AML [13, 14, 15, 16]. For example, Blum et al. [13] proposed miR-29b as a predictive factor for the stratification of older AML patients treated with DEC; however, this was not confirmed by other studies [14]. Butrym et al. [15] showed that low miR-181 expression at diagnosis was a predictor for complete remission and prolonged survival in a subset of older AML patients treated with AZA. In another study, the level of miR-21 in the serum was associated with ORR and progression-free survival in the HMA-treated MDS patients [16].

In this report, we studied changes in miRNA expression profiles of BM CD34+ cells in a small cohort of higher-risk MDS and AML-MRC patients treated with AZA at the genome-wide level. We focused on differentially expressed miRNAs at baseline between responders and non-responders with the aim to find candidate molecular markers predicting the response to AZA.

2. Materials and methods

2.1 Samples

Bone marrow aspirates of MDS/AML-MRC patients treated at the General University Hospital and at the Institute of Hematology and Blood Transfusion in Prague were obtained before and during AZA therapy. The study cohort included 27 patients (median age 68 years, 15 males and 12 females) with MDS or AML-MRC. None of the patients received chemotherapy or hematopoietic stem cell transplantation prior to the study. Samples from 11 age-matched healthy donors (median age 67 years, 8 males and 3 females) with no history of hematological malignancies were used as controls. Written informed consent was obtained from all tested subjects in accordance with the institutional review boards. The patients’ diagnoses were assessed according to the World Health Organization (WHO) classification criteria [17]. Azacitidine was administered as 75 mg/m ${}^{2}$ /day for 7 consecutive days every 28 days. The hematological evaluation of response to the treatment was performed after cycle 4 according to the International Working Group (IWG) criteria for MDS [18] and AML [19]. Patients who achieved complete remission (CR), partial remission (PR), marrow complete remission (mCR), or hematological improvement (HI) were considered responders; patients with stable disease (SD) or progressed disease (PD) were considered non-responders.

2.2 Cell separation and RNA extraction

Mononuclear cells (MNCs) were separated from the BM aspirates by Ficoll-Paque density centrifugation (GE Healthcare, Munich, Germany). CD34+ cells were isolated from MNCs with the Direct CD34 Progenitor Cell Isolation MACS Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Total RNA was extracted using the acid guanidinium-thiocyanate-phenol-chloroform method [20]. The integrity of total RNA was evaluated with Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA), and RNA concentration was determined fluorometrically with Qubit 2.0 (Thermo Fisher Scientific, Waltham, MA, USA).

2.3 Microarray profiling

Human miRNA Microarrays, Release 19.0, 8 $\times$ 60 K (Agilent Technologies), were used to assess the expression levels of 2,006 human miRNAs. The microarray profiling was performed on a discovery cohort (28 samples from 12 patients prior to and during AZA therapy and four healthy donors). The input amount of total RNA was 200 ng per sample. All samples were processed according to the manufacturer’s protocol. Scanning was performed on an Agilent Microarray Scanner C. Data extraction and quality control were done by Agilent Feature Extraction Software (ver. 10.7.3.1). The data were normalized and processed in R statistical software (www.r-project.org) with the AgiMicroRNA [21] and its web interface MagiCMicroRna [22] packages. Differential gene expression analyses were performed in R using the gtools package ( $|$ FC $|$ and its modifications) and in MeV software (ver. 4.8.1) [23]. For the analysis of differential expression, only the miRNAs expressed at the detectable level were included. Analysis of variance (ANOVA) was performed to identify differentially expressed miRNAs between the different groups of samples. Welch’s approximate t-test was applied to determine the differentially expressed miRNAs between the two sample groups. Multiple testing correction was performed using the Holm-Bonferroni method. Hierarchical clustering of the miRNA expression data was done by using average linkage and Euclidean distance.

2.4 RT-qPCR

Reverse-transcription quantitative-PCR (RT-qPCR) was applied to measure expression levels of individual miRNAs (miR-17-3p, miR-27b-3p, miR-100-5p, miR-133b, and miR-142-3p) in an extended cohort (50 samples from 27 patients prior to and during AZA therapy and 11 controls). TaqMan microRNA cDNA synthesis kit and TaqMan microRNA expression assays (Thermo Fisher Scientific) were used following instructions of the manufacturer. The data were normalized to RNU48 and processed by the 2 ${}^{-\Delta\Delta\text{CT}}$ method [24].

2.5 Statistical analysis

Statistical analyses were performed with GraphPad Prism (GraphPad Software, La Jolla, CA) and MedCalc (MedCalc Software, Ostend, Belgium). Unpaired and paired t-tests were used to compare clinical parameters and miRNA levels between different groups of samples. The sensitivity and specificity of the optimum cut-off points of miRNA levels were defined as the values that maximized the area under the receiver operating characteristic (ROC) curve (AUC). The chi-squared test was applied for the comparison of categorical clinical variables. Overall survival (OS) curves were generated by the Kaplan-Meier method and the results were compared by the log-rank test. Cox-regression analysis was used for univariate analysis of response for OS. The OS was defined as the time from the beginning of treatment until death from any cause; the patients who were alive were censored at the time of last follow-up. Multivariate analysis was performed by Cox proportional hazard regression analysis. Differences were considered statistically significant if $P<$ 0.05.

2.6 miRNA target prediction

The KEGG pathways and genes targeted by selected miRNAs were predicted by DIANA-mirPath v.3 web server [25]. Only the pathways with enrichment $P$ -value $P<$ 0.01 were considered.

Table 1
Clinical characteristics of patients

Patients, n	27
Age (years), median (range)	68	(63–82)
Sex, n (%)
Male	15	(56%)
Female	12	(44%)
Diagnosis, n (%)
MDS-RCMD	1	(4%)
RAEB1	2	(7%)
RAEB2	16	(59%)
AML-MRC	8	(30%)
Marrow blasts (%), median (range)	18	(4–27)
Karyotype, n (%)
Isolated del(5q)/del(5q) $+$ any other abnormality	5 (19%)/10 (37%)
Isolated trisomy 8/trisomy 8 $+$ any other abnormality	2 (7%)/5 (19%)
Normal	8	(30%)
Cytogenetics by IPSS-R ${}^{*}$ , n (%)
Good	17	(63%)
Intermediate	6	(22%)
Poor	1	(4%)
Very poor	2	(7%)
n.a.	1	(4%)
IPSS-R (MDS patients only), n (%)
Intermediate	5	(26%)
High	9	(47%)
Very high	5	(26%)
AZA cycles administered, n, median (range)	8	(3–34)
Follow-up, months, median (range)	21.1	(3.4–47.1)
Alive, n (%)	2	(7%)
Response, n (%)
Complete response	4	(15%)
Partial response	3	(11%)
Marrow complete response	2	(7%)
Hematological improvement	2	(7%)
Stable disease	8	(30%)
Progressive disease	8	(30%)
ORR, n (%)	11	(40%)

ORR – overall response rate, ${}^{*}$ The IPSS-R cytogenetic risk score has been established for patients with MDS and is not commonly used to stratify AML patients. However, we used this score for both MDS and AML patients, with the aim to compare frequencies of karyotypes across these patient groups.

Table 2

Univariate analysis for overall response rate and overall survival

Variable	n	ORR, n (%)	$P$ value	OS, median (mo.)	$P$ value
Age
$<$ 70 years	15	6 (40%)	0.930	19.6	0.385
$\geqslant$ 70 years	12	5 (42%)		21.2
Sex
Male	15	6 (40%)	0.930	19.6	0.837
Female	12	5 (42%)		24.9
Blasts
$<$ 10%	7	3 (43%)	0.895	23.9	0.515
$\geqslant$ 10%	20	8 (40%)		19.4
Karyotype
Normal	8	3 (38%)	0.330	21.2	0.853
Isolated del(5q)	5	3 (56%)		25.9
Other	13	3 (33%)		13.0
Cytogenetics by IPSS-R
Good	17	7 (41%)	0.334	25.9	0.033
Int/poor/very poor	9	2 (22%)		11.3
Diagnosis
MDS	19	8 (42%)	0.824	21.1	0.953
AML-MRC	8	3 (38%)		20.4
miR-17-3p
Low	15	3 (20%)	0.014	21.2	0.215
High	12	8 (67%)		19.1
miR-100-5p
Low	13	8 (62%)	0.034	25.9	0.047
High	14	3 (21%)		16.4
miR-133b
Low	9	7 (78%)	0.005	19.1	0.915
High	18	4 (22%)		22.4

Int – intermediate, ORR – overall response rate, OS – overall survival.

3. Results

3.1 Patient characteristics

The study included 50 samples from 27 patients with MDS/AML-MRC at baseline (i.e., patients before AZA administration) and during AZA therapy (collected at the time of the best response, between cycles 4–11). Serial samples (i.e., samples taken from one patient at baseline and during the therapy) were available for 23 patients; four patients were screened only before AZA therapy. The detailed clinical characteristics of patients are summarized in Table 1. The diagnoses included one case of refractory anemia with multilineage dysplasia (RCMD) (4%), two cases of refractory anemia with excess blasts-1 (RAEB-1) (7%), 16 cases of RAEB-2 (59%), and eight cases of AML-MRC (30%). The IPSS-R categories of MDS patients were intermediate for five cases (26%), high for nine cases (47%), and very high for 5 cases (26%).

The median number of administered AZA cycles was eight (range, 3–34 cycles), the median time to the best response in the responder cohort was five months (range, 3.2–11.0 months), and the median duration of response was 16.8 months (range, 8.0–29.1 months). ORR that included rates for all responders was 40% within the cohort (CR in four cases (15%), PR in three cases (11%), mCR in two cases (7%), and HI in two cases (7%)). The non-responders (60%) consisted of eight patients with SD (30%) and eight patients with PD (30%). The median patient follow-up from the initiation of AZA treatment was 21.1 months (range 3.4–47.1 months); 25 patients died and two patients were censored at the time of last follow-up. Based on Cox-regression analysis, the responders and non-responders showed significantly different OS (hazard ratio [HR] $=$ 0.195, 95% confidence interval [CI], 0.070 to 0.542, $P=$ 0.002) (SI Fig. 1). Further statistical analyses were performed to evaluate the correlation of clinical parameters at baseline with patient response to AZA treatment. However, the ORR was not associated with any of the tested clinical variables (Table 2).

Figure 1.

Hierarchical clustering of differentially expressed miRNAs in the baseline samples. ANOVA was performed on patient samples prior to AZA treatment that were divided according to their later response status. The heatmap shows the binary logarithm of fold change (logFC) compared to the median of control samples, which is expressed using a color gradient intensity scale (blue – downregulation, red – upregulation compared to the median of controls). CTR – healthy control, CR – patient with complete response, PR – partial response, SD – stable disease, and PD – progressed disease.

Figure 2.

Heatmap of miRNAs with differentially changed expression between responders and non-responders after AZA treatment. Statistical analysis was used to compare the changes in miRNA expression (ratios of signals between paired samples during treatment vs. at baseline) in the AZA-treated samples from responders vs. non-responders. The changes in the miRNA levels are expressed using a color gradient intensity scale (blue – downregulation, red – upregulation compared to the expression level of a given miRNA in a paired sample at baseline). CR – patient with complete response, PR – partial response, SD – stable disease, and PD – progressed disease.

Figure 3.

Expression of selected miRNAs measured by RT-qPCR. Relative miRNA levels are presented as 2 ${}^{-\Delta\Delta\text{CT}}$ . CTR – controls, RS – responders at baseline, RS AZA – AZA-treated responders, nRS – non-responders at baseline, nRS AZA – AZA-treated non-responders, ${}^{*}P<$ 0.05, ${}^{**}P<$ 0.01, ${}^{***}P<$ 0.001.

3.2 miRNA profiling at baseline

miRNA microarray profiling was done on a smaller set of samples (12 patients at baseline and four healthy donors) to preselect miRNAs with potential prognostic significance. The normalized microarray expression data are included in SI Table 2. ANOVA performed on the samples at baseline stratified by response status (three CR/two PR/four SD/three PD) identified significant differences in the expression of 64 miRNAs (Fig. 1). Hierarchical clustering of these miRNAs clearly defined four sample clusters with different miRNA profiles: i) controls, ii) patients with CR, iii) patients with PR, and iv) a mixed cluster of non-responders (SD and PD). The baseline samples obtained from patients who later achieved CR exhibited the most distinct profile with the downregulation of let-7c, miR-27b-3p, miR-100-5p, mir-140-3p, and miR-423-5p and upregulation of miR-21-5p, miR-211-3p, miR-1246, miR-5739, miR-6085, miR-6124, miR-6132, and miR-6165. In the non-responders, significant downregulation of miR-10a/b-5p and upregulation of miR-1 and miR-133b were observed compared to the patients with CR and PR. Regarding the fact that our study comprises only limited number of samples and because we did not find any significant difference in miRNA profiles between SD and PD samples, we divided the patients only into two groups (responders and non-responders) for all subsequent analyses.

3.3 miRNA profiling before and during AZA treatment

To identify the miRNAs affected by AZA, miRNA expression profiles of 12 patients at baseline and during AZA treatment were compared by paired t-tests. The miRNAs with median expression level change of $\geqslant$ 30% compared to their level at baseline included 20 downregulated (including let-7f-5p/7g-5p, miR-10a-5p, miR-142-3p, miR-146a-5p, miR-148a-3p, miR-181a-3p/b-5p/c-5p, miR-196b-5p, and miR-451a) and 14 upregulated miRNAs (including miR-1202, miR-1260a, and miR-3656) in all the tested patients irrespective to response status (SI Fig. 2).

To determine the miRNAs involved in response to AZA, we compared the miRNA expressions between responders and non-responders after AZA treatment. We found 30 miRNAs with significantly different expression exclusively in responders and unchanged levels in non-responders (Fig. 2). Among these, miR-10b-5p, miR-15a-5p/b-5p, miR-24-3p, miR-148b-3p, and miR-199a-3p were downregulated and miR-1202 or miR-1260a were upregulated in responders after AZA treatment compared to the baseline samples.

3.4 Quantification of individual miRNAs

To verify the microarray results, RT-qPCR was used to evaluate the levels of miR-17-3p, miR-27b-3p, miR-100-5p, miR-133b, and miR-142-3p in the extended cohort (Fig. 3). These miRNAs were selected based on their significant differential expression in the microarray experiments. In concordance with the microarray data, miR-17-3p, miR-100-5p and miR-133b were differentially expressed between the responders and non-responders; the level of miR-17-3p was significantly increased (2 ${}^{-\Delta\Delta\text{CT}}$ of 2.907 $\pm$ 1.372 vs. 1.674 $\pm$ 0.162, $P=$ 0.0214), and the levels of miR-100-5p and miR-133b were significantly reduced (miR-100-5p: 0.293 $\pm$ 0.130 vs. 1.161 $\pm$ 0.304, $P=$ 0.003; miR-133b: 0.830 $\pm$ 0.323 vs. 4.015 $\pm$ 0.972, $P=$ 0.007, respectively) in responders vs. non-responders. Comparison of the baseline samples with those collected during AZA therapy revealed reduced level of miR-27b-3p and increased level of miR-142-3p at baseline that get closer to the control levels in relation to AZA treatment. The level of miR-27b-3p was 1.494 $\pm$ 0.345 in controls, 0.583 $\pm$ 0.068 in patients at baseline and 1.154 $\pm$ 0.322 during therapy ( $P=$ 0.0692); the level of miR-142-3p was 0.271 $\pm$ 0.229 in controls, 1.621 $\pm$ 0.119 in patients at baseline, and 0.9463 $\pm$ 0.210 during therapy ( $P=$ 0.0016).

Figure 4.

Kaplan-Meier curves for the overall survival of patients stratified according to IPSS-R cytogenetics and miR-100-5p level.

3.5 Correlation of clinical parameters with miRNA levels

Because the examination of miRNA levels at baseline by RT-qPCR showed significant elevation of miR-17-3p and reduction of miR-100-5p and miR-133b in the responders vs. non-responders, we tested whether the levels of these miRNAs might be potential markers predicting the response to AZA. Based on the analysis of the ROC curve, the cut-off value for miR-17-3p was defined at 2.0 (with an AUC of 0.641, 95% CI, 0.427 to 0.855), the cut-off value for miR-100-5p at 0.4 (AUC of 0.713, 95% CI, 0.518 to 0.908), and the cut-off value for miR-133b at 0.6 (AUC of 0.715, 95% CI, 0.509 to 0.921). The patients were divided according to these cut-offs into “high-miR” and “low-miR” groups, and a series of statistical tests was performed to compare these groups. The ORR was significantly higher in the groups of patients with high-miR-17-3p (67% ORR in the high-miR-17-3p group vs. 20% ORR in the low-miR-17-3p group, $P=$ 0.014), low-miR-100-5p (21% ORR in the high-miR-100-5p group vs. 62% ORR in the low-miR-100-5p group, $P=$ 0.034), and low-miR-133b (22% ORR in the high-miR-133b group vs. 78% ORR in the low-miR-133b group, $P=$ 0.005) (Table 2). There were no differences in patient age, sex, percentage of BM blasts, cytogenetics, and diagnosis between the groups divided according to the levels of these miRNAs. Similarly, the number of AZA cycles, follow-up period, and the proportion of patients who died during this period were comparable between these groups (SI Table 1).

Further statistical testing examined an impact of clinical and molecular variables on OS after AZA initiation. Univariate analysis revealed that OS was not significantly associated with age, sex, percentage of BM blasts, WHO-based diagnosis, miR-17-3p or miR-133b level. The only variables significantly associated with OS in our patient cohort were the IPSS-R cytogenetics and the level of miR-100-5p (Table 2 and Fig. 4). The patients with good karyotype had significantly longer OS than those with unfavorable karyotypes (25.9 vs. 11.3.4 months from the beginning of treatment, $P=$ 0.033). Stratification by miR-100-5p expression showed prolonged OS in the low-miR-100-5p group compared to the high-miR-100-5p group (25.9 vs. 16.4 months from the beginning of treatment, $P=$ 0.047). Multivariate Cox regression analysis confirmed that both the IPSS-R-based cytogenetics (HR $=$ 10.040, 95% CI 2.649 to 38.058, $P=$ 0.001) and the level of miR-100-5p (HR $=$ 4.066, 95% CI 1.444 to 11.445, $P=$ 0.008) represented prognostic factors for the OS (Table 3).

Table 3
Multivariate Cox analysis for overall survival

Variable	HR	95% CI	$P$ -value
Age ( $\geqslant$ 70 years)	2.540	0.795 to 8.123	0.116
Female sex	0.659	0.243 to 1.788	0.413
Blasts ( $\geqslant$ 10%)	4.245	0.930 to 19.372	0.062
Cytogenetics by IPSS-R	10.040	2.649 to 38.058	0.001
(int/poor/very poor)
AML-MRC diagnosis	1.289	0.474 to 3.508	0.620
High miR-100-5p	4.066	1.444 to 11.445	0.008

Int – intermediate, HR – hazard ratio, CI – confidence interval.

4. Discussion

Although AZA considerably improves the outcomes of MDS patients, about half of the patients fail to respond and their outcomes remain very poor. In this context, the identification of biomarkers predicting the response to AZA would facilitate an individualized risk-adapted therapy. In this study, we performed a comprehensive analysis of miRNA expressions at the whole-genome level in AZA-treated patients with MDS or AML-MRC. We aimed to determine predictive miRNA profiles associated with response status and to define miRNA markers that could predict response to the therapy.

Using a microarray platform, we demonstrated that the miRNA profile was distinct already at baseline, predicting the future response status. Patients who later achieved a response to AZA therapy had different miRNA patterns compared to non-responders. Interestingly, patients with SD and PD had similar patterns without significant differences and did not form two separate groups in clustering analysis. With respect to this finding, subsequent testing was done only between responders and non-responders, with the aim to increase the statistical rigor of the small study and to make relevant conclusions.

In the next step, we focused on several differentially expressed miRNAs in more detail and found that the patients with upregulated miR-17-3p and downregulated miR-100-5p and miR-133b achieved a response to AZA more likely and that low level of miR-100-5p was associated with prolonged OS of the AZA-treated patients. To illustrate biological roles of these three miRNAs, we summarized their target genes and indicated putative signaling pathways in SI Table 3.

Although reduction in the expression of miR-100 has been found in solid tumors [26, 27, 28, 29], the upregulation of miR-100 has been demonstrated in AML [30, 31], suggesting that it may function either as a tumor suppressor or an oncogene in different cancers. Zheng et al. [30] found a high expression of miR-100 in AML and demonstrated that its level was related to the stage of the maturation block underlying the myeloid leukemia subtypes. Bai et al. [31] revealed that increased expression of miR-100 was significantly associated with advanced clinical features of pediatric AML patients and that the high level of miR-100 was predictive of shorter relapse-free and overall survival. Therefore, we propose that the reduced expression of miR-100 detected in the MDS/AML-MRC patients who later responded to AZA may reflect lower aggressiveness of the disease that might have been modulated by AZA administration.

The comparison of baseline expression profiles between the responders and non-responders identified some other miRNAs that are implicated in cell proliferation and oncogenesis. miR-1 and miR-133b, whose levels were increased in the AZA non-responders, are encoded by homologous clusters of miR-1/133a and miR-206/133b. Although miR-1 is downregulated in many tumors [32], its overexpression was detected in AML where it promoted cell proliferation, suggesting that it may act as an oncogene in hematologic malignancies [33]. The HOX-related miRNAs, miR-10a and miR-10b, showed reduced levels in the non-responders and elevated levels in responders. It has been shown that these miRNAs are aberrantly expressed in myeloid malignancies [34, 35, 36] and high baseline expression of the miR-10 family in untreated AML patients is associated with complete remission in response to induction chemotherapy [36].

Further, we identified miRNAs modulated by the AZA treatment, providing more insights into the molecular mechanism of action of AZA. Numerous hematopoiesis/oncology-related miRNAs (such as let-7f, miR-10, miR-15, and miR-181 families) were downregulated after AZA treatment compared to their levels at baseline. For example, the expression of members of the miR-181 family was significantly reduced after treatment. The miR-181 family plays an important role in the regulation of hematopoiesis, including proliferation and differentiation of stem/progenitor cells and megakaryocytic lineage development [37]. Selective overexpression of miR-181 family members was previously detected in higher-risk MDS [38] and M1 and M2 subtypes of AML [39]. The reduction in the level of miR-181 may be therefore one of the downstream effects of AZA that modulates the aberrant proliferation and differentiation in myelodysplasia.

For the identification of miRNAs involved in signaling pathways modulated by AZA, we analyzed miRNA profiles at the time of the best response. However, it is important to take into account the fact that the expression of miRNAs as well as the composition of BM (rate of normal vs. myelodysplastic CD34+ cells) is changing in the course of AZA therapy. Because the collection of BM samples at different time points may affect the resulting data, time dependency of miRNA expression in the course of the treatment should have been evaluated in more detail in future.

Although limited in patient number, this is the first report providing a comprehensive analysis of the miRNAome in myelodysplastic patients treated with AZA with regard to their response status. Our data indicate that AZA responders and non-responders show distinct miRNA profiles and that the level of specific miRNAs before therapy initiation may predict the efficacy of AZA treatment. Validation of these results on a larger international cohort of patients is required to make clinically-relevant conclusions; however, the findings highlight the potential of miRNAs to be useful therapy biomarkers in MDS.

Footnotes

Acknowledgments

The authors thank Viktor Stranecky (Institute of Inherited Metabolic Disorders, Prague) for the microarray data analysis and Kyra Michalova and Zuzana Zemanova (Center of Oncocytogenetics, Faculty Hospital and First Faculty of Medicine, Charles University, Prague) for the cytogenetic data. This work was supported by grants No. 16-33617A and 17-31398A from the Ministry of Health of the Czech Republic, P205/12/G118 from the Grant Agency of the Czech Republic, LM201509 NCMG from the Ministry of Education, Youth and Sports of the Czech Republic, project ERDF OPPK CZ.2.16/3.1.00/28007 and by the project for conceptual development of research organization No. 00023736 from the Ministry of Health of the Czech Republic.

Conflict of interest

The authors declare that they have no competing interest.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-171029.

References

Nimer

S.D.

, Myelodysplastic syndromes, Blood 111(10) (2008), 4841–4851.

Greenberg

P.L.

et al., Revised international prognostic scoring system for myelodysplastic syndromes, Blood 120(12) (2012), 2454–2465.

Vasilatou

et al., Epigenetic alterations and microRNAs: new players in the pathogenesis of myelodysplastic syndromes, Epigenetics 8(6) (2013), 561–570.

Wouters

B.J.

and Delwel

, Epigenetics and approaches to targeted epigenetic therapy in acute myeloid leukemia, Blood 127(1) (2016), 42–52.

Fenaux

et al., Efficacy of azacitidine compared with that of conventional care regimens in the treatment of higher-risk myelodysplastic syndromes: a randomised, open-label, phase III study, Lancet Oncol 10(3) (2009), 223–232.

Vardiman

and Reichard

, Acute myeloid leukemia with myelodysplasia-related changes, Am J Clin Pathol 144(1) (2015), 29–43.

Loiseau

Ali

and Itzykson

, New therapeutic approaches in myelodysplastic syndromes: Hypomethylating agents and lenalidomide, Exp Hematol 43(8) (2015), 661–672.

Kornblith

A.B.

et al., Impact of azacytidine on the quality of life of patients with myelodysplastic syndrome treated in a randomized phase III trial: a Cancer and Leukemia Group B study, J Clin Oncol 20(10) (2002), 2441–2452.

García-Delgado

et al., Effectiveness and safety of different azacitidine dosage regimens in patients with myelodysplastic syndromes or acute myeloid leukemia, Leuk Res 38(7) (2014), 744–750.

10.

Milunović

et al., The role of microRNA in myelodysplastic syndromes: beyond DNA methylation and histone modification, Eur J Haematol 96(6) (2016), 553–563.

11.

Dostalova Merkerova

et al., Distinctive microRNA expression profiles in CD34+ bone marrow cells from patients with myelodysplastic syndrome, Eur J Hum Genet 19(3) (2011), 313–319.

12.

Krejčík

et al., Aberrant expression of the microRNA cluster in 14q32 is associated with del(5q) myelodysplastic syndrome and lenalidomide treatment, Cancer Genet 208(4) (2015), 156–161.

13.

Blum

et al., Clinical response and miR-29b predictive significance in older AML patients treated with a 10-day schedule of decitabine, Proc Natl Acad Sci U S A 107(16) (2010), 7473–7478.

14.

Yang

et al., Levels of miR-29b do not predict for response in patients with acute myelogenous leukemia treated with the combination of 5-azacytidine, valproic acid, and ATRA, Am J Hematol 86(2) (2011), 237–238.

15.

Butrym

et al., Expression of microRNA-181 determines response to treatment with azacitidine and predicts survival in elderly patients with acute myeloid leukaemia, Oncol Lett 12(4) (2016), 2296–2300.

16.

Kim

et al., Serum microRNA-21 as a potential biomarker for response to hypomethylating agents in myelodysplastic syndromes, PLoS One 9 (2014), e86933.

17.

Swerdlow

S.H.

, WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues, IARC Press, 2008.

18.

Cheson

B.D.

et al., Clinical application and proposal for modification of the International Working Group (IWG) response criteria in myelodysplasia, Blood 108(2) (2006), 419–425.

19.

Cheson

B.D.

et al., Revised recommendations of the International Working Group for Diagnosis, Standardization of Response Criteria, Treatment Outcomes, and Reporting Standards for Therapeutic Trials in Acute Myeloid Leukemia, J Clin Oncol 21(24) (2003), 4642–4649.

20.

Chomczynski

and Sacchi

, Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction, Anal Biochem 162 (1987), 156–159.

21.

López-Romero

, Pre-processing and differential expression analysis of Agilent microRNA arrays using the AgiMicroRna Bioconductor library, BMC Genomics 12 (2011), 64.

22.

Coonen

M.L.

et al., MagiCMicroRna: a web implementation of AgiMicroRna using shiny, Source Code Biol Med 10 (2015), 4.

23.

Saeed

A.I.

et al., TM4: a free, open-source system for microarray data management and analysis, Biotechniques 34(2) (2003), 374–378.

24.

Livak

K.J.

and Schmittgen

T.D.

, Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Δ⁢Δ⁢CT) Method, Methods 25 (2001), 402–408.

25.

Vlachos

I.S.

et al., DIANA-miRPath v3.0: deciphering microRNA function with experimental support, Nucleic Acids Research 43 (2015), 460–466.

26.

Henson

B.J.

et al., Decreased expression of miR-125b and miR-100 in oral cancer cells contributes to malignancy, Genes Chromosomes Cancer 48 (2009), 569–582.

27.

Doghman

et al., Regulation of insulin-like growth factor-mammalian target of rapamycin signaling by microRNA in childhood adrenocortical tumors, Cancer Res 70 (2010), 4666–4675.

28.

Cairo

et al., Stem cell-like micro-RNA signature driven by Myc in aggressive liver cancer, Proc Natl Acad Sci U S A 107 (2010), 20471–20476.

29.

Oliveira

J.C.

et al., MicroRNA-100 acts as a tumor suppressor in human bladder carcinoma 5637 cells, Asian Pac J Cancer Prev 12 (2011), 3001–3004.

30.

Zheng

Y.S.

et al., MiR-100 regulates cell differentiation and survival by targeting RBSP3, a phosphatase-like tumor suppressor in acute myeloid leukemia, Oncogene 31(1) (2012), 80–92.

31.

Bai

et al., Upregulation of microRNA-100 predicts poor prognosis in patients with pediatric acute myeloid leukemia, Onco Targets Ther 5 (2012), 213–219.

32.

Nohata

et al., microRNA-1/133a and microRNA-206/133b clusters: dysregulation and functional roles in human cancers, Oncotarget 3(1) (2012), 9–21.

33.

Gomez-Benito

et al., EVI1 controls proliferation in acute myeloid leukaemia through modulation of miR-1-2, Br J Cancer 103 (2010), 1292–1296.

34.

Dostalova Merkerova

et al., Distinctive microRNA expression profiles in CD34+ bone marrow cells from patients with myelodysplastic syndrome, Eur J Hum Genet 19(3) (2011), 313–319.

35.

Garzon

et al., MicroRNA signatures associated with cytogenetics and prognosis in acute myeloid leukemia, Blood 111 (2008), 3183–3189.

36.

Havelange

et al., Implications of the miR-10 family in chemotherapy response of NPM1-mutated AML, Blood 123 (2014), 2412–2415.

37.

Georgantas

R.W.

et al., CD34+ hematopoietic stem-progenitor cell microRNA expression and function: a circuit diagram of differentiation control, Proc Natl Acad Sci U S A 104 (2007), 2750–2755.

38.

Sokol

et al., Identification of a risk dependent microRNA expression signature in myelodysplastic syndromes, Br J Haematol 153(1) (2011), 24–32.

39.

Debernardi

et al., MicroRNA miR-181a correlates with morphological sub-class of acute myeloid leukaemia and the expression of its target genes in global genome-wide analysis, Leukemia 21(5) (2007), 912–916.