The Stat3/5 Signaling Biosignature in Hematopoietic Stem/Progenitor

Published OnlineFirst December 23, 2015; DOI: 10.1158/1078-0432.CCR-15-1288

Personalized Medicine and Imaging Clinical Cancer Research The Stat3/5 Signaling Biosignature in Hematopoietic Stem/Progenitor Cells Predicts Response and Outcome in Myelodysplastic Syndrome Patients Treated with Azacitidine Paraskevi Miltiades1, Eleftheria Lamprianidou1, Theodoros P. Vassilakopoulos2, Sotirios G. Papageorgiou3, Athanasios G. Galanopoulos4, Christos K. Kontos5, Panagiotis G. Adamopoulos5, Evangelia Nakou1,Soﬁa Vakalopoulou6,Vassilia Garypidou6, Maria Papaioannou7, Evdoxia Hatjiharissi8, Helen A. Papadaki9, Emmanuil Spanoudakis1, Vassiliki Pappa3, Andreas Scorilas5, Constantinos Tsatalas1, and Ioannis Kotsianidis1 on behalf of the Hellenic MDS Study Group

Abstract

Purpose: Azacitidine is the mainstay of high-risk myelodys- tive subpopulation were performed by flow cytometry and quan- plastic syndromes (MDS) therapy, but molecular predictors of titative real-time PCR in isolated MDS progenitors. response and the mechanisms of resistance to azacitidine remain Results: The pretreatment Stat3/5 signaling profiles in þ largely unidentified. Deregulation of signaling via Stat3 and Stat5 CD34 cells correlated strongly with response and cytogenetics in acute myeloid leukemia (AML) is associated with aggressive and independently predicted event-free survival. We further þ disease. Numerous genes involved in cell signaling are aberrantly identified a CD34 G-CSF–inducible Stat3/5 double-positive methylated in MDS, yet the alterations and the effect of azacitidine subpopulation (DP subset) whose pretreatment levels were treatment on Stat3/5 signaling in high-risk MDS have not been inversely associated with treatment response and cytogenetics. explored. The kinetics of the DP subset followed the response to azaci- Experimental Design: We assessed longitudinally constitutive tidine and the disease course, whereas its molecular character- and ligand-induced phospho-Stat3/5 signaling responses by mul- istics and cellular hierarchy were consistent with a leukemia tiparametric flow cytometry in 74 patients with MDS and low propagating cell phenotype. blast count AML undergoing azacitidine therapy. Pretreatment Conclusions: Our findings provide a novel link among Stat3/5 þ Stat3/5 signaling profiles in CD34 cells were grouped by unsu- signaling and MDS pathobiology and suggest that the Stat3/5 pervised clustering. The differentiation stage and the molecular signaling biosignature may serve as both a response biomarker þ properties of the CD34 G-CSF–inducible Stat3/5 double-posi- and treatment target. Clin Cancer Res; 22(8); 1958–68. 2015 AACR.

Introduction 1 Department of Hematology, Democritus University of Thrace, Alex- The introduction of azacitidine has radically transformed the androupolis, Greece. 2Department of Hematology, Laikon General Hospital, National and Kapodistrian University of Athens, Athens, therapeutic approach and improved the outcome of patients with Greece. 3Second Department of Internal Medicine, Hematology Unit, high-risk myelodysplastic syndrome (MDS). Nevertheless, the exact 4 Attikon University General Hospital, Athens, Greece. Department of mechanism of action remains to be established, while both primary Clinical Hematology, G. Gennimatas Hospital, Athens, Greece. 5Department of Biochemistry and Molecular Biology, National and and secondary resistance confer a grave prognosis, as there are Kapodistrian University of Athens, Athens, Greece 6Second Prope- currently no effective alternative therapies (1). In addition, there is deutic Department of Internal Medicine, Aristotle University of Thes- lack of a serviceable, widely accepted biomarker of response and/or saloniki, Hippokration Hospital, Thessaloniki, Greece. 7Department of Haematology, Aristotle University of Thessaloniki, AHEPA Hospital, outcome that can offer a timely and valid estimation of the expected Thessaloniki, Greece. 8Department of Hematology,Theageneion Hos- beneﬁt from azacitidine and help to tailor treatment (2, 3). pital of Thessaloniki, Thessaloniki, Greece. 9Department of Hematol- Stat3 and Stat5 regulate fundamental cellular processes and ogy, University Hospital of Heraklion, Heraklion, Greece. aberrant cell signaling via Stat3/5 is implicated in leukemogenesis Note: Supplementary data for this article are available at Clinical Cancer (4–9). Constitutive upregulation of Stat3 and, less often, Stat5 Research Online (http://clincancerres.aacrjournals.org/). molecules has been reported in acute myeloid leukemia (AML), but Corresponding Author: Ioannis Kotsianidis, Department of Hematology, Demo- its prognostic impact is contentious (10). This is because measure- critus University of Thrace Medical School, Dragana, Alexandroupolis 68100, ment of basal Stat3/5 levels in heterogeneous cell populations by Greece. Phone: 3025-5103-0320; Fax: 3025-5103-0439; E-mail: using conventional proteomic assays does not address the cytokine [email protected] regulation of signaling cascades of malignant hematopoiesis and doi: 10.1158/1078-0432.CCR-15-1288 thus cannot portray the overall picture of signaling events at the 2015 American Association for Cancer Research. single cell level (11). Single cell network proﬁling using

1958 Clin Cancer Res; 22(8) April 15, 2016

Stat3/5 Signaling Architecture in High-Risk MDS

10 donors with nonclonal myelopoiesis (i.e., lymphomas, solid Translational Relevance tumors, and immune thrombocytopenia) were obtained before Azacitidine is the main option for high-risk myelodysplastic treatment initiation and at the indicated time points. Informed syndrome (MDS) patients, but mechanisms of resistance are consent was obtained in accordance with the Declaration of largely unknown and there is paucity of a serviceable bio- Helsinki. All patients received azacitidine in a nonclinical trial marker of response. While abnormal Stat3/5 signaling in setting at an initial dose of 75 mg/m2 s.c. for 7 days on 28-day hematopoietic stem/progenitor cells has been implicated in cycles. Dose reductions of 25% to 50% and/or treatment delays acute myeloid leukemia pathobiology, the architecture and were considered for severe myelotoxicity or myelosuppression- the effect of azacitidine on Stat3/5 signaling in MDS have not related complications. Granulocyte colony-stimulating factors yet been evaluated. Utilizing functional phenotyping by phos- (G-CSF) were used at the discretion of the treating doctor, whereas pho-protein flow cytometry, we explored the alterations of no erythropoiesis-stimulating agents were administered to any Stat3/5 signaling in high-risk MDS patients treated with aza- patient. Response to therapy was evaluated using the Internation- citidine. We demonstrate that the Stat3/5 signaling biosigna- al Working Group Response Criteria for MDS (23). Heavily ture can predict response and outcome of azacitidine treat- transfused patients were defined as those requiring 4 RBC ment. Moreover, we identified a prognostically relevant units/8 weeks (24). Mononuclear cells were isolated after density þ CD34 G-CSF–inducible Stat3/5 double-positive subpopula- centrifugation, frozen in liquid nitrogen, and processed within tion whose kinetics paralleled disease activity and were ame- 6 months after cryopreservation. nable to modulation by azacitidine in responding patients. Our data identify Stat3/5 signaling aberrations as predictors Antibodies, data acquisition, and analysis of resistance to azacitidine and set the scene for the therapeutic The following antibodies were used: CD34 (clone 8G12), targeting of the Stat3/5 signaling network in MDS. pStat3 (Y705), pStat5 (Y694), CD2 (RPA-2.10), CD3 (HIT3a), CD4 (RPA-T4), CD8 (RPA-T8), CD19 (HIB19), CD20 (2H7), GPA (GA-R2), CD45(2D1), CD114 (LMM741), p53 (DO7) and Ki-67 (B56) all from BD Biosciences; CD38 (LS198.4.3) from multiparametric phospho-specific flow cytometry can identify Beckman Coulter; Bcl-2 (124) from Dako; CD123 (6H6), CD90 phosho-Stat3/5 biosignatures at the hematopoietic stem/progen- (5E10) and CD45RA (HI100) from Biolegend. Fluorescence itor cell (HSPC) level which reflect the biologic behavior of AML minus one (FMO) was employed as negative control. Data were and can distinguish patient subgroups with worse prognosis (11– acquired on a 5-color FC-500 (Coulter) and a 4-color FACSCa- 13). In their seminal article, Irish and colleagues (12) have iden- libur (BD Biosciences) cytometers and analyses were performed tified a G-CSF–inducible Stat3/5 double-positive (DP) subpopu- þ with Flowjo software (Treestar). lation of CD34 cells in aggressive AML, without howevertesting its prognostic value. An analogous DP subset has also been observed Single-cell phospho-specific flow cytometry in Ph myeloproliferative neoplasms (14), raising the possibility of Thawed cells were washed once in RPMI to remove the a shared signaling biosignature among myeloid neoplasms. residual DMSO and allowed to rest for 1 hour in serum-free Both aberrant methylation and cell signaling deregulation RPMI medium at 37C. Cells were then distributed at 2–5 105 contribute to the pathogenesis of Myelodysplastic syndromes cells/well in 4 aliquots. Two aliquots remained unstimulated (MDS), while epigenetic defects of genes involved in cell signaling and used as FMO control and untreated sample controls and are frequently encountered in MDS patients, particularly in the the others were stimulated for 15 minutes at 37C with either late stages of the disease (15–18). Moreover, in addition to the human recombinant G-CSF or granulocyte macrophage stim- bidirectional interplay among the epigenetic machinery and cell ulating factor (GM-CSF, Miltenyi Biotec GmbH) at a final signaling (19), hypomethylating agents may indirectly affect concentration of 20 ng/mL for both cytokines. Stimulation signal transduction (20–22). Despite the above, a comprehensive was halted by fixation with Cytofix Fixation Buffer (BD Bios- view of Stat3/5 signaling alterations in late-stage MDS is missing ciences) and cells were permeabilized with Perm Buffer III (BD and the effect of hypomethylating therapy on leukemic signaling Biosciences) and stained with phospho-Stat3 (clone Y705), has not been addressed yet. phospho-Stat5 (clone Y694), and combinations of the afore- By using phospho-specific flow cytometry, we investigated phos- mentioned antibodies for 30 minutes at room temperature. pho-Stat3/5 signaling profiles in the HSPCs from 74 MDS and low Basal phosphorylation levels were expressed as the log2 ratio of blast count AML patients during azacitidine therapy. We show that mean fluorescence intensity (MFI) of unstimulated pStat3 and the pretreatment Stat3/5 biosignature in MDS HSPCs was strongly pStat5 divided by the FMO control, namely log2[MFI (unsti- associated with response status and patient outcome. We further mulated)/MFI (FMO)] and potentiated levels as log2[MFI identified a G-CSF–inducible phospho-Stat3/5 DP subpopulation þ (stimulated)/MFI (unstimulated)]. in the CD34 cell compartment (hereafter referred to as DP subset) whose pretreatment levels were inversely associated with response. Delineation of the differentiation stage and the leukemia stem The cellular hierarchy and the molecular properties of the DP subset þ þ þ cell characteristics of the CD34 G-CSF–inducible pStat3 /5 were consistent with a leukemia stem cell phenotype, whereas its DP subset kinetics followed the disease course and response to treatment. þ To map the cellular hierarchy of the CD34 DP subset, cells expressing mature lineage markers were depleted from mono- Patients, Materials, and Methods nuclear cells after staining with an antibody cocktail consisting Patients of anti-CD2, CD3, CD4, CD8, CD19, CD20, and GPA (25) and Following Institutional Review Board approval, peripheral anti-PE immunomagnetic MicroBeads (Miltenyi Biotec). Puri- blood and bone marrow mononuclear cells from 74 patients and fied Lin cells were then subjected to positive selection of

www.aacrjournals.org Clin Cancer Res; 22(8) April 15, 2016 1959

Miltiades et al.

þ CD34 cells by means of CD34 microbeads (Miltenyi Biotec). þ Results Isolated Lin-CD34 cells (purity always 98%) were then The pretreatment Stat3/5 signaling biosignature strongly stained with pStat3, pStat5, CD38, CD90, CD123, and correlates with clinical and biologic parameters CD45RA according to the above phospho-flow protocol and Patients' characteristics are listed analytically in Table 1. Two analyzed as previously described (25). For the characterization þ (2.7%) patients achieved marrow complete remission (mCR) of the molecular/LSC properties of the CD34 DP subset, with both erythroid and platelet hematologic improvement, 17 BMMNC were stained with pStat3, pStat5, and CD34 along attained CR (23%), 11 (14.9%) patients showed hematologic with one of bcl-2, Ki-67, or p53. improvements only in platelet count (HI), 17 (23%) remained stable (stable disease, SD) and 27 (36.4%) failed (failure, F) Assessment of G-CSF receptor (CSF3R) protein and mRNA azacitidine. The representativeness of our cohort was validated expression by the successful application of the prognostic score proposed by For the determination of CSF3R protein expression samples Itzykson and colleagues (24) in 72 evaluable patients. Three were analyzed by flow cytometry, by staining with CD114, CD34, groups with different OS (P ¼ 0.027) were identified (Supple- CD45, and the corresponding isotype controls. Data are expressed þ mentary Fig. S1). Unsupervised clustering of pretreatment signal- as the ratio of the CD114 MFI of CD34 cells to the MFI of isotype fi þ fi þ ing pro les in CD34 cells of MDS patients identi ed 2 signaling control. Immunomagnetically purified CD34 cells (purity clusters (SC), SC-1 and SC-2. The two clusters displayed similar 98%) from the above samples were used for mRNA extraction levels of constitutive Stat3/5 phosphorylation, whereas potenti- (RNAqueous Micro kit, Ambion) and reverse transcription (RET- ated responses of Stat3/5 to G-CSF and GM-CSF stimulation were ROscript, Ambion). Quantification of the CFS3R mRNA was weak in SC-1 and powerful in SC-2 (Fig. 1A). No differences were performed by Real-Time PCR using SYBR Green (Invitrogen) and observed among the two clusters regarding age, sex, WHO sub- 0 the following primers: human CSF3R forward, 5 -CATCACAG- type, transfusion burden, WPSS, IPSS, IPSS-R, and TET2 mutation 0 0 CCTCCTGCATCATC-3 , human CSF3R reverse, 5 -CTGAA- status (Table 1). In contrast, patients in SC-1 achieved better 0 0 GCTCTGCTCCCAGTCTC-3 , human GAPDH forward, 5 -ACTC- response to azacitidine (P ¼ 0.01), had worse cytogenetics both 0 0 CACGACGTACTCAGCG-3 , human GAPDH reverse, 5 -GGTC- by IPSS (P ¼ 0.01) and IPSS-R (P ¼ 0.02) and enjoyed longer 0 GGAGTCAACGGATTTG-3 . PCR conditions were as following: median EFS (12.5 vs. 7.8 months, respectively, P ¼ 0.01) than the 50 C for 2 minutes, 95 C for 10 minutes, 40 cycles of 95 C for 15 ones in SC-2 (Fig. 1 and Table 1), while median OS was also seconds and 60 C for 1 minute, followed by melting curve prolonged in patients of SC-1, without, however, reaching statis- analysis from 65 Cto90 C. Reactions were carried out in dupli- tical significance (13.5 vs. 10.4 months, respectively; P ¼ 0.08). cate in a PTC 200 Peltier Thermal Cycler with Chromo4 Real-Time Multivariate analysis confirmed the independent prognostic pow- PCR Detector. Data acquisition and analysis were performed by er of the pretreatment Stat3/5 signaling biosignature for EFS (P ¼ Chromo4 Real-Time PCR Detector and Opticon Monitor 3. þ 0.017), whereas heavy transfusion requirements was the other Relative CSF3R expression of CD34 cells was calculated by P ¼ DDC independent prognostic factor for both OS ( 0.004) and EFS 2 t method. (P ¼ 0.029, Supplementary Tables S1 and S2). These findings are in line with prior observations in AML (12, 27), and strongly Mutations analysis of TET2 and TP53 coding regions suggest involvement of aberrant signaling via Stat3/5 in MDS DNA was extracted from bone marrow mononuclear cells or pathobiology. We further performed unsupervised clustering peripheral blood samples collected before the initiation of of patient and nonclonal samples (Supplementary Fig. S5). All azacitidine. Mutational analysis of the coding region of TET2 patients with nonclonal myelopoiesis clustered with responding and TP53 genes in 30 and 11 samples, respectively, was per- patients, further supporting the existence of an aberrant signaling formed using next-generation sequencing (detailed method biosignature in nonresponders to azacitidine. provided in the Supplementary Methods). For the analysis of TET2 , a 10% cutoff of allele fraction was used as suggested Identification of a prognostically relevant G-CSF–inducible þ previously (26). Stat3/5 DP subpopulation of CD34 cells The complementary cytometric analysis of pretreatment Stat3/ Statistical analysis 5 signaling profiles in our patients revealed an identical to the Comparisons were performed by using c2,Mann–Whitney, G-CSF–inducible DP subpopulation previously described in AML Kruskal–Wallis, Wilcoxon signed-rank, and Friedman tests, as (12) and Ph myeloproliferative neoplasms (14) The median appropriate, and survival analysis with Kaplan–Meier and log- pretreatment levels of the DP subset were significantly lower in þ rank test. Overall survival (OS) was defined as the time from patients who achieved CR or marrow CR (38.6% of total CD34 azacitidine initiation to death from any cause and event-free cells, range 0.13%–83.5%) compared to those with stable disease survival (EFS) as the time from azacitidine initiation to disease (75.4%, 1%–90%, P ¼ 0.008) and failure to azacitidine (74.7%, progression, relapse, or death. Surviving patients were censored 12.1%–90%, P ¼ 0.006), whereas patients with HI did not show at last follow-up. Multivariate survival analysis was based on any significant differences with the other groups (HI, 55.3%, Cox proportional hazards model using a backward stepwise 0.5%–88.2%, Fig. 2A and B). Also, the levels of the DP subset selection procedure with entry and removal criteria of P ¼ 0.05 were significantly lower in patients with poor-risk cytogenetics by and P ¼ 0.10, respectively. Multiple Experiment Viewer soft- IPSS (38.6%, 0.5%–75.4%) compared to those with intermediate ware (MeV, http://sourceforge.net/projects/mev-tm4/) was (68.1%, 1%–85.5%, P ¼ 0.02) and good-risk (66.9%, 0.13%– employed for unsupervised hierarchical cluster analysis with 90%, P ¼ 0.005) karyotype. Similarly, poor-risk cytogenetics by complete linkage algorithm and Euclidean distance as distance IPSS-R were associated with lower percentage of the DP subset metric (12). (38.4%, 1%–75.4%) compared with good-risk disease (70.8%,

1960 Clin Cancer Res; 22(8) April 15, 2016 Clinical Cancer Research

Stat3/5 Signaling Architecture in High-Risk MDS

Table 1. Baseline patient characteristics and clinical information (n ¼ 74) Total pts (n ¼ 74) Cluster 1 (n ¼ 37) Cluster 2 (n ¼ 37) P value Age (median, range) 73.2 (48.6–83.7) 73.8 (52–83.5) 72.7 (48.6–83.7) P ¼ 0.28 >65 58 (78.4%) 31 (83.8%) 27 (73%) <65 16 (21.6%) 6 (16.2%) 10 (27%) Sex P ¼ 0.14 Male 48 (64.9%) 27 (73%) 21 (56.8%) Female 26 (35.1%) 10 (27%) 16 (43.2%) Baseline blood counts Hemoglobin (g/dl) 8.7 (5.7–12.8) 8.8 (6.1–12.8) 8.7 (5.7–10.4) P ¼ 0.53 ANC(109/L) 1.55 (0.04–31) 2.26 (0.05–31) 0.92 (0.04–15.5) P ¼ 0.09 Platelets (109/L) 53.5 (9–383) 54 (9–383) 53 (9–300) P ¼ 0.375 Number of completed cycles P ¼ 0.046 Median (range) 6 (1–36) 6 (2–33) 5 (1–36) WHO classiﬁcation P ¼ 0.1 RCMD 3 (4%) 3 (8.1%) 0 (0%) RAEB-I 4 (5.4%) 1 (2.7%) 3 (8.1%) RAEB-II 31 (42%) 11 (29.8%) 20 (54.1%) CMML-II 13 (17.6%) 8 (21.6%) 5 (13.5%) AML-LBC 20 (27%) 13 (35.1%) 7 (18.9%) MDS/MPD 3 (4%) 1 (2.7%) 2 (5.4%) IPSS P ¼ 0.79 Intermediate-2 30 (40.5) 14 (37.8%) 16 (43.2%) High 34 (46%) 17 (45.9%) 17 (45.9%) N/A 10 (13.5%) 6 (16.2%) 4 (10.9%) WPSS P ¼ 0.64 High 23 (31.1%) 10 (27%) 13 (35.1%) Very high 14 (18.9%) 5 (13.5%) 9 (24.3%) N/A 37 (50%) 22 (59.5%) 15 (40.6%) IPSS-R P ¼ 0.15 Intermediate 5 (6.7%) 4 (10.8%) 1 (2.8%) High 25 (33.8%) 9 (24.3%) 16 (43.2%) Very high 34 (46%) 18 (48.7%) 16 (43.2%) N/A 10 (13.5%) 6 (16.2%) 4 (10.8%) IPSS-R Cytogenetic risk P ¼ 0.024 Good 35 (47.3%) 13 (35.1%) 22 (59.5%) Intermediate 19 (25.7%) 9 (24.3%) 10 (27%) Poor 11 (14.9%) 9 (24.3%) 2 (5.4%) Very poor 6 (8.1%) 5 (13.5%) 1 (2.7%) N/A 3 (4%) 1 (2.7%) 2 (5.4%) IPSS Cytogenetic risk P ¼ 0.010 Good 33 (45%) 13 (35.1%) 21 (56.8%) Intermediate 21 (28%) 9 (24.3%) 11 (29.7%) Poor 17 (23%) 14 (37.9%) 3 (8.1%) N/A 3 (4%) 1 (2.7%) 2 (5.4%) PB blasts P ¼ 0.27 Present 38 (51.4%) 17 (45.9%) 21 (56.8%) Absent 31 (41.9%) 18 (48.7%) 13 (35.1%) N/A 5 (6.7%) 2 (5.4%) 3 (8.1%) GFM prognostic score 0.067 Low 8 (11%) 5 (13%) 3 (8%) Intermediate 49 (67%) 20 (54%) 29 (78%) High 15 (20%) 11 (30%) 4 (11%) N/A 2 (3%) 1 (3%) 1 (3%) Transfusions 4 per month P ¼ 0.45 Yes 23 (31.1%) 13 (35.1%) 10 (27%) No 51 (68.9%) 24 (64.9%) 27 (73%) TET2 mutations (all) Yes 19/30 (63.3%) 5 (45%) 14 (74%) P ¼ 0.12 No 11/30 (36.7%) 6 (55%) 5 (26%) TET2 mutations (VAF 10%) Yes 5/30 (16.7%) 1 (9%) 4 (21%) P ¼ 0.4 No 25/30 (83.3%) 10 (91%) 15 (79%) Best response P ¼ 0.017 CR þ mCR 19 (25.7%) 15 (40.6%) 4 (10.8%) Hematologic improvement 11 (14.9%) 4 (10.8%) 7 (18.9%) Stable disease 17 (23%) 9 (24.3%) 8 (21.6%) Failure 27 (36.4%) 9 (24.3%) 18 (48.7%) (Continued on the following page)

www.aacrjournals.org Clin Cancer Res; 22(8) April 15, 2016 1961

Miltiades et al.

Table 1. Baseline patient characteristics and clinical information (n ¼ 74) (Cont'd ) Total pts (n ¼ 74) Cluster 1 (n ¼ 37) Cluster 2 (n ¼ 37) P value Median follow up Months 47.7 Treatment after azacitidine failure total, n ¼ 8 P ¼ 0.4 Intensive chemotherapy 7 4 3 Allo-SCT 1 1 0 NOTE: Numbers in bold indicate P < 0.05. Abbreviations: CR, complete response; mCR, complete marrow response with incomplete blood count recovery; GFM, Groupe Francophone des Myelodysplasies; N/A, not applicable/not available; VAF, variant allele frequency.

0.13%–90%, P ¼ 0.03). Of note, consistent with the results transfusion requirements (data not shown). We also tested the þ obtained by the clustering of signaling profiles, nonclonal rest GM- and G-CSF–inducible or not CD34 subpopulations, for patients had identical levels of the DP subpopulation with example, single positive Stat3 or Stat5 as well as Stat3/5 DP responders to azacitidine (CR and HI), but significantly lower subsets, for correlations with clinical and biologic parameters, compared with nonresponders (F and SD, Supplementary but we were unable to find any associations except significantly Fig. S5). higher basal Stat3 levels in patients with poor-risk cytogenetics No differences in the levels of the DP subset were observed compared with the good-risk ones by both IPSS (P ¼ 0.002) and regarding age, gender, MDS subtype, TET2 mutation status, and IPSS-R (P ¼ 0.008, Supplementary Fig. S2).

A Signaling profiles pSTAT3/ basal pSTAT5/ basal pSTAT3/ G-CSF pSTAT5/ G-CSF pSTAT5/GM-CSF Phosphorylation scale

B Min Max Parameters SC-1 SC-2 Response Karyo IPSS Karyo IPSS-R Response (P = 0.017) Cytogenetics IPSS-R (P = 0.024) Cytogenetics IPSS (P = 0.010) CR/mCR Good Good HI Intermediate Intermediate SD Poor Poor Failure Very poor N/A N/A C

SC-1 SC-1 SC-2 SC-2

P = 0.08 P = 0.01 Overall survival Event free survival

Time (months) Time (months)

Figure 1. Association of clinical parameters with pretreatment signaling biosignatures. A, heatmap of pretreatment signaling profiles. Basal and potentiated phosphorylation levels are represented with a double gradient color scale (green-black-red) displaying underexpression relative to the mean as green, overexpression þ as red, and spots where there is little differential expression as black. Unsupervised clustering of pretreatment basal and potentiated Stat3/5 levels of CD34 cells in 74 patients distinguished two signaling clusters (SC), SC-1 (left) and SC-2 (right). B, patients in SC-1 (n ¼ 37) showed significantly higher rates of CR/mCR (P ¼ 0.017) and had worse karyotype according to both IPSS (P ¼ 0.010) and IPSS-R (P ¼ 0.024) compared to those in SC-2 (n ¼ 37), whereas there were no differences among the two SCs regarding sex (P ¼ 0.14), WHO subtype (P ¼ 0.1), transfusion burden (P ¼ 0.45), WPSS (P ¼ 0.6), IPSS (P ¼ 0.8) and IPSSR (P ¼ 0.15, data shown in Table 1). Each box in the color chart represents a single patient. C, overall (OS) and event free (EFS) survival of patients in SC-1 (n ¼ 37) and SC-2 (n ¼ 37). The former group enjoyed significantly longer EFS (P ¼ 0.011), whereas OS was also prolonged in patients of SC-1, although not significantly (P ¼ 0.08).

1962 Clin Cancer Res; 22(8) April 15, 2016 Clinical Cancer Research

Stat3/5 Signaling Architecture in High-Risk MDS

Untreated GCSF A 4 104 10 0.7% 0.1% 0.7% 3.1% 3 103 10 99% 0.2% 93% 3.2%

2 CR 102 10

pSTAT5 1 101 10

0 100 10 0 1 2 3 100 101 102 103 10 10 10 10 10 10

104 104 1.7% 0.1% 4.2% 86% 3 3 10 98% 0.2% 10 8.4% 1.4%

Failure 102 102

1 1 pSTAT5 10 10

100 100 100 101 102 103 10 100 101 102 103 10 pSTAT3

B Response IPSS Cytogenetics IPSS-R Cytogenetics

P = 0.03 P = 0.006 P = 0.005 P = 0.008 P = 0.02 cells + % of CD34

CR HI SD F Good Int Poor Good Int Poor Very + poor mCR

Figure 2. Identification of a G-CSF-inducible Stat3/5 double-positive subpopulation of CD34þ cells, which is adversely associated with response to azacitidine. A, representative flow cytometric analysis of pretreatment samples of a patient who achieved CR (top, CR) and one who failed azacitidine (bottom, Failure). A CD34þ double-positive (DP) Stat3/5 subpopulation (DP subset) is strongly induced in the nonresponding patient after G-CSF stimulation. Plots are gated on CD34þ cells. B, the median pretreatment levels of the DP subset were inversely associated with response and cytogenetic risk. Patients who achieved complete remission or marrow CR (CRþmCR, n ¼ 19) had significantly lower levels of the DP subset compared with those with stable disease (SD, n ¼ 17, P ¼ 0.008) and failure to azacitidine (F, n ¼ 27, P ¼ 0.006), whereas patients with hematologic improvement (HI, n ¼ 11) did not show any significant differences with the other groups. Also, patients with poor-risk cytogenetics by IPSS (n ¼ 17) had significantly lower levels of the DP subset compared with those with intermediate (n ¼ 21, P ¼ 0.02) and good-risk (n ¼ 33, P ¼ 0.005) karyotype. Likewise, poor-risk cytogenetics by IPSS-R (n ¼ 11) were associated with lower levels of the DP subset compared with good-risk disease (n ¼ 35, P ¼ 0.03). P values by Kruskal–Wallis test.

The kinetics of the DP subset follow the disease course and ﬁrst cycle only in patients with CR. The pretreatment median þ response to azacitidine percentage in these patients (38.6% of total CD34 cells, range We next sought to investigate the effect of azacitidine and 0.13%–84%) downregulated to 9.6% (0.3%–70.4%, P ¼ 0.01), disease course upon the kinetics of the DP subset. The latter whereas patients with HI (47.6%, 0.5%–82.2% changed to 32%, subpopulation was downregulated signiﬁcantly on day15 of the 0.2%–72.2%, P ¼ 0.12), SD (75.4%, 1%–87.6% changed to

www.aacrjournals.org Clin Cancer Res; 22(8) April 15, 2016 1963

Miltiades et al.

76.1%, 1%–83.7%, P ¼ 0.23), and failure (63%, 12.1%–87.2% achieving CR was significantly reduced on day 15 of the first changed to 58%, 23.8%-87%, P ¼ 0.26), retained unaltered levels cycle and remained at low levels until disease relapse, which of the DP subset (Fig. 3A). Of note, day 15 of the first cycle was was accompanied by a marked expansion of the DP subset (P ¼ chosen because the hypomethylating effect of azacitidine usually 0.007). Similar kinetics of the DP subset were observed in peaks 15 days after its administration (18), while clonal hema- patients with HI, without however, reaching statistical signif- topoiesis still dominates in bone marrow as shown by the icance (P ¼ 0.13). þ identical percentage of bone marrow CD34 cells before (15%, Thus, it appears that the kinetics of the DP subset are following range 4%–29%) and 15 days after first azacitidine administration the disease course and response to azacitidine, indicating poten- (14.5%, 4%–26%, P ¼ 0.9) in our patients. tial involvement of the former subset in mechanisms underlying In 19 patients, the alterations of the DP subset were studied disease progression and resistance to azacitidine. longitudinally during the disease course (Fig. 3B and C). Measurements were performed on days 0 and 15 of the first The DP subpopulation is enriched in cells with a leukemia stem cycle, at response evaluation after 6 cycles and at disease cell phenotype progression or relapse. The levels of the DP subset remained Recent findings in both mice and humans challenge the unaffected throughout the disease course in patients with SD leukemia stem cell (LSC) model and suggest that in approximate- þ and failure to azacitidine. In contrast, the DP subset in patients ly 90% of CD34 AML cases LSCs reside in the lymphoid-primed

A CR HI SD Failure Figure 3. P = 0.01 P = 0.12 P = 0.23 P = 0.26 The kinetics of the GCSF-inducible Stat3/5 DP subpopulation follow the disease course and response to azacitidine. A, the DP subset was significantly downregulated on day 15 of the first azacitidine cycle only in patients who achieved CR (n ¼ 13, P ¼ 0.01), whereas it remained unaltered in % of CD34+ cells patients with hematologic improvement d0 d15 d0 d15 d0 d15 d0 d15 (HI, n ¼ 8, P ¼ 0.12), stable disease (SD, n ¼ 9, P ¼ 0.23) and failure (F, n ¼ 16, P ¼ 0.26). B, kinetics of the DP B subpopulation in patients with CR (n ¼ 8), HI (n ¼ 5), SD (n ¼ 3), and F (n ¼ 3) were assessed longitudinally during azacitidine treatment. Measurements were performed on days 0 and 15 of the first cycle, at response evaluation after 6 cycles and when the disease progressed or relapsed after an CR initial response. In patients who P = 0.007 achieved CR the kinetics of the DP % of CD34+ cells subset paralleled disease severity and response to azacitidine (P ¼ 0.007 by d0 d15 Response Progression Friedman test), while in patients with SD evaluation and F the DP subpopulation persisted at high levels throughout the disease C course. The kinetics of the DP subset in Day 0 15Day 6 months Progression patients with HI also displayed a trend to follow the disease course, which, 30,3 30,3 7.4 38.6 1.3 19.2 3.3 24.6 7 51.7 however, did not reach statistical significance (P ¼ 0.13). C, representative 9.5 30.3 29.4 11.9 flow cytometric plots of serial measurements in a responder (CR, top) and a nonresponder (F, bottom) to CR CR CR CR azacitidine. In the responder, the DP subset was downregulated on day 15 of the first cycle, remained at low levels during remission (6 months from 1.7 77.4 9.9 73.1 5 89,6 1.2 90.6 treatment initiation) and expanded 9.9 6.5 3,2 5.7 when the patient lost response to azacitidine and the disease progressed. In contrast, stable, high-level expression of the DP subset was observed in the F FF F nonresponder throughout the disease

pSTAT5 course. Plots are gated on G-CSF þ pSTAT3 stimulated CD34 cells.

1964 Clin Cancer Res; 22(8) April 15, 2016 Clinical Cancer Research

Stat3/5 Signaling Architecture in High-Risk MDS

A Gated on Gated on proliferation (32), respectively, and were therefore used to Lin-CD34+CD38- cells Lin-CD34+CD38+ cells

103 103 address the molecular properties of the DP subset. After we G-CSF treated HSC:0.7% CMP:24% GMP:74% conﬁrmed the compatibility of the phospho-ﬂow protocol with 102 102 CD34+ cells MPP:25% LMPP:74.1% MEP:1.1% the intracellular staining for the above markers (not shown), we 3 DP 1 1 10 10 10 found that the DP subset, compared with the DN one, exhibited

0 0 102 DP 10 10 – 57.3% decreased pretreatment levels of Ki-67 (52%, 22% 81.7% vs. – P ¼ 1 10 0 1 2 3 0 1 2 3 66.5%, 33% 87%, respectively, 0.01) and increased Bcl-2 10 10 10 10 10 10 10 10 DN 3 3 34.5% 10 10 – – P < pSTAT5 HSC:1% CMP:55 % GMP:39.3% MFI (9.5, 6.9 25.5 vs. 6.1, 3.7 24.1, 0.001) and p53 MFI 100 102 102 MPP:54.7% LMPP:44.1% MEP:4% (3.67, 2–14.3 vs. 2.7, 1.6–8.7, P ¼ 0.03), indicating quiescence

DN 101 101 100 101 102 103 and increased antiapoptotic and oncogenic properties, respec- pSTAT3 100 100 tively (Fig. 5A and Supplementary Fig. S4; refs. 33, 34). Of note, the DP subset displayed signiﬁcantly increased p53 levels com- CD90 CD123 100 101 102 103 100 101 102 103 n ¼ CD45RA CD45RA pared with the DN one in both p53 mutated ( 3) and unmutated cases (n ¼ 8, data not shown), whereas the expression B of the above molecules in both the DP and DN subsets remained unaltered on d15 after azacitidine initiation (Fig. 5A), suggesting * * that the above signaling subsets represent distinct cellular entities with stable characteristics.

HSC MPP LMPP A * * * ** * * * % or DN subsets of DP * * p53 MFI %Κι-67+ Bcl-2 MFI CMP GMP MEP

Figure 4. d0 d15 d0 d15 d0 d15 Delineation of the position of the GCSF-inducible Stat3/5 DP subpopulation in B the hematopoietic hierarchy. A, assessment of the cellular hierarchy of Lin- CD34þ DP subset by phosphospecific flow cytometry. Representative P = 0.95 P = 0.8 contour plots of HSPC compartments in the Lin-CD34þ G-CSF inducible Stat3/5 double positive (DP) and double negative (DN) subsets of patient #39. B, the DP subpopulation contained significantly higher levels of granulocyte-macrophage progenitor (GMP)-like and lymphoid-primed multipotent progenitor (LMPP)-like cells and lower levels of multipotent MFI CD114 progenitor (MPP)-like, common myeloid progenitor (CMP)-like, and CSF3R mRNA megakaryocyte-erythroid progenitor (MEP)-like cells, compared with the G- DPhigh DPneg DPhigh DPneg CSF–unresponsive, Stat3/5 DN subset, whereas the HSC-like progenitors did not differ among the two subsets. Five patient samples with predominance of C LMPP/GMP-like progenitors were analyzed. P < 0.05 by Wilcoxon signed- P = 0.27 rank test. P = 0.56 multipotent progenitor (LMPP)-like and granulocyte-macrophage progenitor (GMP)-like compartments (25, 28). Interest- MFI CD114 þ CSF3R mRNA ingly, the same CD34 cell compartments are clonally expanded in late-stage MDS (29). By combining surface with phospho- d0 d15 d0 d15 þ staining in purified Lin CD34 cells of 5 patients, we assessed Figure 5. the cellular hierarchy of the DP subset (Fig. 4A and Supplementary Molecular properties of the GCSF-inducible Stat3/5 DP subpopulation. A, þ Fig. S3). We observed that, in comparison with the other major expression of Ki-67 (n ¼ 12), p53 (n ¼ 11) and Bcl-2 (n ¼ 13) on the CD34 DP þ signaling subset, namely the CD34 G-CSF–unresponsive, Stat3/ and DN subsets. The former subpopulation expresses significantly higher 5 double negative (DN) cells of the same patient, the DP subset pretreatment (d0) levels of Bcl-2 and p53 and lower Ki-67. Identical findings was enriched in LMPP-like (70.8%, range 46%–98.5%, vs. 43%, are observed 15 days after azacitidine initiation (d15), indicating that the DP and DN subsets represent separate cellular entities with distinct molecular 33%–69%, respectively) and GMP-like cells (79.5%, 53%–98% characteristics. B, the interpatient variability and the azacitidine-induced – vs. 43%, 25% 76%), whereas it contained less multipotent alterations of the DP subset are not due to quantitative changes of the G-CSF þ progenitor (MPP)-like (28%, 1.6%–44% vs. 52%, 28.7%- receptor (CSF3R). Protein and mRNA levels of CSF3R in unstimulated CD34 60%), common myeloid progenitor (CMP)-like (19%, 0.9%– cells were identical in patients with either very high (>87%, DPhigh, white neg 41% vs. 44%, 6.1%–57%), and megakaryocyte-erythroid progen- boxes, n ¼ 4) or null (<4%, DP , gray boxes, n ¼ 4) expression of the DP itor (MEP)-like cells (1.5%, 0.1%–6.4% vs. 8.5%, 0.6%–14.5%, subset (left). C, likewise, in a separate group of four patients, CSF3R expression remained unaltered 15 days (d15) after first azacitidine P ¼ 0.04 for all comparisons, Fig. 4B). administration (d0), despite the significant downregulation of the DP subset Bcl-2, p53, and Ki-67 are well established molecular indicators on day 15 in each of these patients (not shown). P < 0.05 and P < 0.001 by of resistance to apoptosis (30), oncogenesis (31), and cellular Mann–Whitney and Wilcoxon signed-rank test as appropriate.

www.aacrjournals.org Clin Cancer Res; 22(8) April 15, 2016 1965

Miltiades et al.

Taken together, the increased levels of LMPP and GMP-like cells SC-P2 reported by Irish and colleagues in adult AML (12), along with the high expression of Bcl-2 and p53 and the lower Ki- which was also associated with disease resistance, highlights the 67 levels imply that the DP subset is enriched in cells with a LSC biologic relationship of AML with high-risk MDS and suggests a phenotype. common signaling biosignature of aggressive leukemic HSPCs in adult AML and high-risk MDS. In contrast, increased pStat3 response to G-CSF ligation correlated with superior outcome in The expression of CSF3R and its modulation by azacitidine are pediatric AML, potentially reflecting the biologic differences not responsible for the interpatient variability and treatment- among adult and pediatric myeloid malignancies (27). Nota- induced alterations of the DP subpopulation bly, patients with nonclonal myelopoiesis and responders to Myeloid blasts display variable expression of CSF3R and a azacitidine shared an identical Stat3/5 signaling biosignature heterogeneous response to G-CSF stimulation (12, 27, 35). Sur- characterized by little or no potentiated G-CSF responses, face CSF3R could not be assessed concomitantly in the DP and þ further corroborating the role of Stat3/5 signaling aberrations DN subsets of G-CSF–stimulated CD34 blasts because it is in azacitidine resistance. downregulated after G-CSF ligation (36). Therefore, to determine Further analysis of Stat3/5 signaling profiles revealed a G-CSF– whether CSF3R levels and/or their modulation by azacitidine inducible Stat3/5 DP subpopulation of CD34þ cells whose levels contribute to the interpatient variability and azacitidine-induced correlated inversely with response to azacitidine. A phenotypi- modifications of the DP subpopulation, we employed two dif- cally identical, chemoresistant subpopulation has been previous- ferent approaches. In the first, we evaluated CSF3R expression on ly reported in adult AML and Ph myeloproliferative neoplasms 8 samples that showed either absence (<4%, DPneg) or very high (12, 14), but no further investigation of the kinetics and the expression (87%, DPhigh) of the DP subset, while in the second molecular properties of the DP subset was conducted. We we measured CSF3R levels on day 0 and day 15 of the first cycle of observed a remarkable downregulation of the DP subset in azacitidine in 4 other patients who displayed significant down- responding patients on day 15 after the first azacitidine admin- regulation of the DP subset on day 15. Using the first approach, we istration, whereas nonresponders exhibited no changes. More found that patients with either high or null expression of the DP important, the kinetics of the DP subset in responders mirrored subset showed identical pretreatment mRNA and protein levels of those of the tumor burden and treatment response, as the former CSF3R (Fig. 5B). Likewise, CSFR3 protein and transcript expres- subpopulation remained at low levels during CR, whereas a sion remained unaltered despite the significant downregulation parallel development of resistance to azacitidine with an expan- of the DP subset from 66.1% (range, 23%–80%) on day 0 to sion of the DP subset occurred. These findings strongly suggest, on 45.5% (1%–59%, P ¼ 0.008) on day 15 in 4 patients (Fig. 5C), one hand, that azacitidine can restore and partially control the indicating that quantitative changes of CSFR3 are not implicated pathologic Stat3/5 signaling in responding patients and on the in the generation of the G-CSF–inducible DP subset. other involvement of the DP subset in mechanisms underlying disease progression and azacitidine resistance. Of note, though Discussion the distinction of clonal from normal hematopoiesis is often Abnormal hematopoietic stem/progenitor cell (HSPC) signal- problematic in MDS, the alterations of the DP subset pertained to ing via Stat3/5 is typically observed in leukemic hematopoiesis. In clonal HSPCs as shown by two findings. First, the blast percentage both adult and pediatric AML, functional phenotyping of Stat3/5 on day 15 of the first cycle was identical to the pretreatment one signaling networks by phospho-protein flow cytometry provides and second, three responding patients (2 with CR and one with important prognostic information and pathobiologic insights HI) who downregulated significantly the DP subset after 6 cycles (12, 27, 37). Yet, despite the reciprocal interactions of DNA had still abnormal karyotype indicating persistence of clonal methylation with Stat3/5 signaling (19, 29, 38), the high rate of hematopoiesis. In addition, it has been clearly demonstrated that mutations (39) and aberrant methylation of genes involved in cell even patients in CR have residual MDS HSPCs in substantial signaling (18) in high risk MDS, no current study addresses the numbers (25, 29). Stat3/5 signaling alterations at the single HSPC level in such Although previous studies linked the DP subset to aggressive patients. Only a recent study explored signaling abnormalities disease, there is currently no detailed phenotypic and molecular in MDS, but it was mainly focused on erythropoietin-induced characterization of the former subpopulation. First, we confirmed Stat5 phosphorylation in erythroid progenitors in early disease the compatibility of the phospho-flow techniques with the intra- stages (6). In the current work, we demonstrate an abnormal cellular measurement of Bcl-2, p53, and Ki-67 and the accurate Stat3/5 signaling biosignature of HSPCs in high-risk MDS, which assessment of HSPC subsets as has been previously shown is amenable to modulation by azacitidine and can predict treat- (refs. 40, 41; Supplementary Fig. S3). We then observed a signif- ment response and outcome. icant enrichment of the DP subset in LMPP-like and GMP-like þ Two signaling clusters were identified by hierarchical clus- cells compared with the other dominant CD34 signaling subset, tering of pretreatment basal and potentiated Stat3/5 and Stat5 the G-CSF–unresponsive Stat3/5-negative subpopulation. Also, phosphorylation patterns. Patients in SC-1 displayed better Bcl-2 and p53 levels were significantly higher in the DP subset, response to azacitidine and longer EFS, whereas OS was also whereas Ki-67 expression was lower compared with the DN prolonged, though not statistically significant. Intriguingly, subset. Considering the negative role of Bcl-2 and p53 in the although correlated favorably with prognosis, SC-1 was inverse- pathobiology of MDS (33, 42, 43), the downregulation of Ki-67 in þ ly associated with cytogeneticriskandshowedno correlation CD34 cells when MDS progresses to overt leukemia (33) and with TET2 mutation status, emphasizing that signaling profiles recent data regarding the hierarchy of LSCs (25), we surmise that are not merely a surrogate for the underlying molecular abnor- the DP subset potentially possesses properties of leukemia-prop- malities, but instead provide additional information. More- agating cells, which may account for its association with poor risk over, the characteristic similarity of SC-2 in our study with the disease and resistance to azacitidine.

1966 Clin Cancer Res; 22(8) April 15, 2016 Clinical Cancer Research

Stat3/5 Signaling Architecture in High-Risk MDS

Hypersensitivity of HSPCs to G-CSF may predispose to leuke- inhibitors induced apoptosis and decreased colony formation in mic transformation and propagation via overexuberant Stat3/5 HSPCs from both MDS (29) and AML (45) patients. Importantly, activation, but there is considerable interpatient variability of G- in all studies, the pharmacologic blockade of Stat3/5 selectively CSF responsiveness in AML (12, 44, 45). We found that the targeted clonal but spared normal HSPCs. Further corroborating differential response of Stat3/5 to G-CSF in our patients is not these observations, our findings may serve as a guidepost for the related to the protein and mRNA levels of CSF3R, suggesting that ongoing investigation of Stat3/5 inhibition as a therapeutic other mechanisms such as qualitative defects of CFS3R, abnormal strategy to overcome azacitidine resistance (4, 48). receptor-associated proteins, dysfunctional downstream signaling elements, or abnormalities of positive or negative regulators of Disclosure of Potential Conflicts of Interest signaling via Stat3/5 are potentially responsible for the variability T.P. Vassilakopoulos is a consultant/advisory board member for Genesis in G-CSF sensitivity (46). Of note, azacitidine downregulated the Pharma. S.G. Papageorgiou reports receiving speakers bureau honoraria from DP subset only in responding patients, implying that diverse Genesis. I. Kotsianidis reports receiving speakers bureau honoraria and com- mercial research grants from Genesis Pharma. No potential conflicts of interest molecular defects are responsible for the induction of the DP were disclosed by the other authors. subset in MDS patients, only a part of which is susceptible to epigenetic reprogramming by azacitidine. Authors' Contributions In contrast to our results, Redell and colleagues observed a Conception and design: I. Kotsianidis positive correlation of CSF3R expression with the magnitude of Development of methodology: P. Miltiades, E. Lamprianidou, S.G. Papageor- pStat3 induction in pediatric AML samples. However, no associ- giou, E. Nakou, I. Kotsianidis ation with simultaneous potentiation of Stat3 and Stat5 with Acquisition of data (provided animals, acquired and managed patients, CSF3R levels was reported, whereas several samples with high provided facilities, etc.): T.P. Vassilakopoulos, S.G. Papageorgiou, A. Galano- levels of CSF3R failed to induce a Stat3/5 response. Moreover, the poulos, S. Vakalopoulou, V. Garypidou, E. Hatjiharissi, H.A. Papadaki, E. Spanoudakis, C. Tsatalas, I. Kotsianidis fact that CSF3R knockout mice can still mobilize effectively Analysis and interpretation of data (e.g., statistical analysis, biostatistics, hematopoietic progenitors provide further support for a computational analysis): P. Miltiades, E. Lamprianidou, T.P. Vassilakopoulos, CSF3R-independent mechanism of induction of the DP subset C.K. Kontos, P.G. Adamopoulos, M. Papaioannou, E. Spanoudakis, in MDS patients (44). I. Kotsianidis Collectively, we report for the first time a disturbed Stat3/5 Writing, review, and/or revision of the manuscript: P. Miltiades, E. Lampria- signaling architecture in high-risk MDS, which is amenable to nidou, T.P. Vassilakopoulos, S.G. Papageorgiou, C.K. Kontos, V. Pappa, A. Scorilas, I. Kotsianidis modulation by azacitidine therapy in responding patients and its Administrative, technical, or material support (i.e., reporting or organizing alterations parallel disease activity. Aside from furnishing critical data, constructing databases): P. Miltiades, E. Lamprianidou, I. Kotsianidis insights in MDS biology, our findings have obvious translational Study supervision: V. Pappa, I. Kotsianidis implications. There is paucity of a serviceable biomarker of Other (carried out part of the mutational analysis): C.K. Kontos, outcome in MDS patients treated with azacitidine, whereas the P.G. Adamopoulos mechanisms of resistance to azacitidine are largely unknown and there is currently no effective treatment after azacitidine failure. Grant Support The prognostic relevance of the Stat3/5 biosignature of MDS This work was supported in part by an educational grant from Genesis HSPCs in our study may help to identify which patients benefit Pharma Hellas (to I. Kotsianidis). The costs of publication of this article were defrayed in part by the payment of most from azacitidine, while its alterations during the disease page charges. This article must therefore be hereby marked advertisement in course can provide a tool for early detection of disease progres- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sion. Also, small-molecule JAK inhibitors and siRNA-mediated þ knockdown of Stat3/5 decreased the growth of CD34 cells from Received June 2, 2015; revised October 8, 2015; accepted November 26, 2015; high-risk AML patients both in vitro and in vivo (47), whereas Stat3 published OnlineFirst December 23, 2015.

References 1. Ades L, Santini V. Hypomethylating agents and chemotherapy in MDS. Best malignancies with specific clinical and biologic correlates. Cancer Cell Pract Res Clin Haematol 2013;26:411–9. 2008;14:335–43. 2. Santini V. Novel therapeutic strategies: hypomethylating agents and 8. Han L, Wierenga AT, Rozenveld-Geugien M, van de Lande K, Vellenga E, beyond. Hematology Am Soc Hematol Educ Program 2012;2012:65–73. Schuringa JJ. Single-cell STAT5 signal transduction profiling in normal and 3. Bejar R, Steensma DP. Recent developments in myelodysplastic syn- leukemic stem and progenitor cell populations reveals highly distinct dromes. Blood 2014;124:2793–803. cytokine responses. PLoS One 2009;4:e7989. 4. Dorritie KA, McCubrey JA, Johnson DE. STAT transcription factors in 9. Padron E, Painter JS, Kunigal S, Mailloux AW, McGraw K, McDaniel JM, hematopoiesis and leukemogenesis: opportunities for therapeutic inter- et al. GM-CSF-dependent pSTAT5 sensitivity is a feature with therapeu- vention. Leukemia 2014;28:248–57. tic potential in chronic myelomonocytic leukemia. Blood 2013;121: 5. Gaipa G, Bugarin C, Longoni D, Cesana S, Molteni C, Faini A, et al. Aberrant 5068–77. GM-CSF signal transduction pathway in juvenile myelomonocytic leuke- 10. Benekli M, Baumann H, Wetzler M. Targeting signal transducer and mia assayed by flow cytometric intracellular STAT5 phosphorylation activator of transcription signaling pathway in leukemias. J Clin Oncol measurement. Leukemia 2009;23:791–3. 2009;27:4422–32. 6. Spinelli E, Caporale R, Buchi F, Masala E, Gozzini A, Sanna A, et al. Distinct 11. Krutzik PO, Irish JM, Nolan GP, Perez OD. Analysis of protein phosphor- signal transduction abnormalities and erythropoietin response in bone ylation and cellular signaling events by flow cytometry: techniques and marrow hematopoietic cell subpopulations of myelodysplastic syndrome clinical applications. Clin Immunol 2004;110:206–21. patients. Clin Cancer Res 2012;18:3079–89. 12. Irish JM, Hovland R, Krutzik PO, Perez OD, Bruserud O, Gjertsen BT, et al. 7. Kotecha N, Flores NJ, Irish JM, Simonds EF, Sakai DS, Archambeault S, et al. Single cell profiling of potentiated phospho-protein networks in cancer Single-cell profiling identifies aberrant STAT5 activation in myeloid cells. Cell 2004;118:217–28.

www.aacrjournals.org Clin Cancer Res; 22(8) April 15, 2016 1967

Miltiades et al.

13. Irish JM, Kotecha N, Nolan GP. Mapping normal and cancer cell signalling 31. Muller PA, Vousden KH. Mutant p53 in cancer: new functions and networks: towards single-cell proteomics. Nat Rev Cancer 2006;6:146–55. therapeutic opportunities. Cancer Cell 2014;25:304–17. 14. Oh ST, Simonds EF, Jones C, Hale MB, Goltsev Y, Gibbs KD Jr, et al. Novel 32. Scholzen T, Gerdes J. The Ki-67 protein: from the known and the unknown. mutations in the inhibitory adaptor protein LNK drive JAK-STAT signaling J Cell Physiol 2000;182:311–22. in patients with myeloproliferative neoplasms. Blood 2010;116:988–92. 33. Parker JE, Mufti GJ, Rasool F, Mijovic A, Devereux S, Pagliuca A. The role of 15. Tefferi A, Vardiman JW. Myelodysplastic syndromes. N Engl J Med apoptosis, proliferation, and the Bcl-2-related proteins in the myelodys- 2009;361:1872–85. plastic syndromes and acute myeloid leukemia secondary to MDS. Blood 16. Itzykson R, Fenaux P. Epigenetics of myelodysplastic syndromes. Leukemia 2000;96:3932–8. 2014;28:497–506. 34. Asai T, Liu Y, Bae N, Nimer SD. The p53 tumor suppressor protein regulates 17. Issa JP. Epigenetic changes in the myelodysplastic syndrome. Hematol hematopoietic stem cell fate. J Cell Physiol 2011;226:2215–21. Oncol Clin North Am 2010;24:317–30. 35. Sultana TA, Harada H, Ito K, Tanaka H, Kyo T, Kimura A. Expression and 18. Figueroa ME, Skrabanek L, Li Y, Jiemjit A, Fandy TE, Paietta E, et al. MDS functional analysis of granulocyte colony-stimulating factor receptors on and secondary AML display unique patterns and abundance of aberrant CD34þþ cells in patients with myelodysplastic syndrome (MDS) and DNA methylation. Blood 2009;114:3448–58. MDS-acute myeloid leukaemia. Br J Haematol 2003;121:63–75. 19. Mohammad HP, Baylin SB. Linking cell signaling and the epigenetic 36. Beekman R, Touw IP. G-CSF and its receptor in myeloid malignancy. Blood machinery. Nat Biotechnol 2010;28:1033–8. 2010;115:5131–6. 20. Yoo CB, Jones PA. Epigenetic therapy of cancer: past, present and future. 37. Nolan GP. Deeper insights into hematological oncology disorders via Nat Rev Drug Discov 2006;5:37–50. single-cell phospho-signaling analysis. Hematology Am Soc Hematol Educ 21. Sigalotti L, Fratta E, Coral S, Cortini E, Covre A, Nicolay HJ, et al. Epigenetic Program 2006:123–7, 509. drugs as pleiotropic agents in cancer treatment: biomolecular aspects and 38. Stevenson WS, Best OG, Przybylla A, Chen Q, Singh N, Koleth M, et al. DNA clinical applications. J Cell Physiol 2007;212:330–44. methylation of membrane-bound tyrosine phosphatase genes in acute 22. Cocco L, Finelli C, Mongiorgi S, Clissa C, Russo D, Bosi C, et al. An increased lymphoblastic leukaemia. Leukemia 2014;28:787–93. expression of PI-PLCbeta1 is associated with myeloid differentiation and a 39. Papaemmanuil E, Gerstung M, Malcovati L, Tauro S, Gundem G, Van Loo longer response to azacitidine in myelodysplastic syndromes. J Leukocyte P, et al. Clinical and biological implications of driver mutations in Biol 2015;98:769–80. myelodysplastic syndromes. Blood 2013;122:3616–27. 23. Cheson BD, Greenberg PL, Bennett JM, Lowenberg B, Wijermans PW, 40. Gibbs KD Jr, Gilbert PM, Sachs K, Zhao F, Blau HM, Weissman IL, et al. Nimer SD, et al. Clinical application and proposal for modification of the Single-cell phospho-specific flow cytometric analysis demonstrates bio- International Working Group (IWG) response criteria in myelodysplasia. chemical and functional heterogeneity in human hematopoietic stem and Blood 2006;108:419–25. progenitor compartments. Blood 2011;117:4226–33. 24. Itzykson R, Thepot S, Quesnel B, Dreyfus F, Beyne-Rauzy O, Turlure P, et al. 41. Irish JM, Anensen N, Hovland R, Skavland J, Borresen-Dale AL, Bruserud O, Prognostic factors for response and overall survival in 282 patients with et al. Flt3 Y591 duplication and Bcl-2 overexpression are detected in acute higher-risk myelodysplastic syndromes treated with azacitidine. Blood myeloid leukemia cells with high levels of phosphorylated wild-type p53. 2011;117:403–11. Blood 2007;109:2589–96. 25. Goardon N, Marchi E, Atzberger A, Quek L, Schuh A, Soneji S, et al. 42. Jadersten M, Saft L, Smith A, Kulasekararaj A, Pomplun S, Gohring G, et al. Coexistence of LMPP-like and GMP-like leukemia stem cells in acute TP53 mutations in low-risk myelodysplastic syndromes with del(5q) myeloid leukemia. Cancer Cell 2011;19:138–52. predict disease progression. J Clin Oncol 2011;29:1971–9. 26. Bejar R, Lord A, Stevenson K, Bar-Natan M, Perez-Ladaga A, Zaneveld J, et al. 43. Bejar R, Levine R, Ebert BL. Unraveling the molecular pathophysiology of TET2 mutations predict response to hypomethylating agents in myelodys- myelodysplastic syndromes. J Clin Oncol 2011;29:504–15. plastic syndrome patients. Blood 2014;124:2705–12. 44. Liu F, Kunter G, Krem MM, Eades WC, Cain JA, Tomasson MH, et al. Csf3r 27. Redell MS, Ruiz MJ, Gerbing RB, Alonzo TA, Lange BJ, Tweardy DJ, et al. mutations in mice confer a strong clonal HSC advantage via activation of FACS analysis of Stat3/5 signaling reveals sensitivity to G-CSF and IL-6 as a Stat5. J Clin Invest 2008;118:946–55. significant prognostic factor in pediatric AML: a Children's Oncology 45. Redell MS, Ruiz MJ, Alonzo TA, Gerbing RB, Tweardy DJ. Stat3 signaling in Group report. Blood 2013;121:1083–93. acute myeloid leukemia: ligand-dependent and -independent activation 28. Krivtsov AV, Twomey D, Feng Z, Stubbs MC, Wang Y, Faber J, et al. and induction of apoptosis by a novel small-molecule Stat3 inhibitor. Transformation from committed progenitor to leukaemia stem cell initi- Blood 2011;117:5701–9. ated by MLL-AF9. Nature 2006;442:818–22. 46. Marvin J, Swaminathan S, Kraker G, Chadburn A, Jacobberger J, Goolsby C. 29. Will B, Zhou L, Vogler TO, Ben-Neriah S, Schinke C, Tamari R, et al. Stem Normal bone marrow signal-transduction profiles: a requisite for enhanced and progenitor cells in myelodysplastic syndromes show aberrant stage- detection of signaling dysregulations in AML. Blood 2011;117:e120–30. specific expansion and harbor genetic and epigenetic alterations. Blood 47. Cook AM, Li L, Ho Y, Lin A, Li L, Stein A, et al. Role of altered growth factor 2012;120:2076–86. receptor-mediated JAK2 signaling in growth and maintenance of human 30. Czabotar PE, Lessene G, Strasser A, Adams JM. Control of apoptosis by the acute myeloid leukemia stem cells. Blood 2014;123:2826–37. BCL-2 protein family: implications for physiology and therapy. Nat Rev 48. Miklossy G, Hilliard TS, Turkson J. Therapeutic modulators of STAT Mol Cell Biol 2014;15:49–63. signalling for human diseases. Nat Rev Drug Discov 2013;12:611–29.

1968 Clin Cancer Res; 22(8) April 15, 2016 Clinical Cancer Research

The Stat3/5 Signaling Biosignature in Hematopoietic Stem/Progenitor Cells Predicts Response and Outcome in Myelodysplastic Syndrome Patients Treated with Azacitidine

Paraskevi Miltiades, Eleftheria Lamprianidou, Theodoros P. Vassilakopoulos, et al.

Clin Cancer Res 2016;22:1958-1968. Published OnlineFirst December 23, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-15-1288

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2016/04/15/1078-0432.CCR-15-1288.DC1

Cited articles This article cites 47 articles, 22 of which you can access for free at: http://clincancerres.aacrjournals.org/content/22/8/1958.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/22/8/1958. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.