http://orcid.org/0000-0001-8376-0344
Figures
Abstract
The Philadelphia chromosome-negative myeloproliferative neoplasms (MPN) share similar molecular characteristics in that they frequently harbor hotspot mutations in JAK2, CALR or MPL, leading to activated JAK/STAT signaling. However, these MPN have distinct symptoms, morphology, and natural histories, including different tendencies to progress to fibrosis. Although the role of inflammation in tissue fibrosis is well recognized, inflammatory gene expression in bone marrows involved by MPN has been understudied. We analyzed the expression of inflammatory genes by directly measuring RNA transcript abundance in bone marrow biopsies of 108 MPN patients. Unsupervised analyses identified gene expression patterns that distinguish prefibrotic (grade 1–2) MPN from overtly fibrotic (grade 2–3) MPN with high sensitivity and specificity in two independent cohorts. Furthermore, prefibrotic and overtly fibrotic MPN are separable into subsets with different activities in biological pathways linked to inflammation, including cytokines, chemokines, interferon response, and toll-like receptor signaling. In conclusion, this study demonstrates that MPN with overt fibrosis is associated with significant inflammatory gene upregulation in the bone marrow, revealing potential roles for multiple pro-inflammatory signaling networks in the development of myelofibrosis and suggesting potential therapeutic mechanisms to alleviate this process in the bone marrow microenvironment.
Citation: Wong WJ, Baltay M, Getz A, Fuhrman K, Aster JC, Hasserjian RP, et al. (2019) Gene expression profiling distinguishes prefibrotic from overtly fibrotic myeloproliferative neoplasms and identifies disease subsets with distinct inflammatory signatures. PLoS ONE 14(5): e0216810. https://doi.org/10.1371/journal.pone.0216810
Editor: Kevin D. Bunting, Emory University, UNITED STATES
Received: February 8, 2019; Accepted: April 29, 2019; Published: May 9, 2019
Copyright: © 2019 Wong et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This study was supported by a Brigham and Women’s Hospital Faculty Career Development Award (O.P.) and by the Brigham Research Institute and the Center for Faculty Development and Diversity’s Office for Research Careers Microgrant Program (O.P.). The funders had no roles in study design, data collection and analysis, decision to publish or preparation of the manuscript.
Competing interests: The study was supported by a Brigham and Women’s Hospital Faculty Career Development Award (O.P.) and by the Brigham Research Institute and the Center for Faculty Development and Diversity’s Office for Research Careers Microgrant Program (O.P.). Dr. Kit Fuhrman is an employee of Nanostring Technologies, Inc; no funding was received from Nanostring Technologies for this study. Dr. Fuhrman’s role was limited to help with data validation. This does not alter our adherence to PLOS ONE policies on sharing data and materials. Nanostring Technologies did not play a role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Other authors (WJW, OP, MB, AG, JCA and RHP) do not have commercial affiliation with the Nanostring Technologies, along with any other relevant declarations relating to employment, consultancy, patents, products in development, or marketed products, etc.
Introduction
Essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF) are a group of Philadelphia (Ph)-negative myeloproliferative neoplasms (MPN) characterized by overlapping clinical and laboratory features, as well as common phenotypic driver mutations in the JAK2, CALR and MPL tyrosine kinase genes. Despite these similarities, disease progression and clinical outcome vary greatly between disease types. For example, 12% of prefibrotic PMF patients develop overtly fibrotic disease within 10 years of diagnosis, whereas post-ET myelofibrosis is rarer, occurring in only 0.8% of ET patients [1]. The biological basis of bone marrow fibrosis in MPN remains unclear, but likely involves aberrant growth factor and cytokine signaling in neoplastic hematopoietic cells [2]. These proinflammatory molecules, such as transforming growth factor beta (TGF-β), platelet derived growth factor (PDGF), and fibroblast growth factor (FGF), elicit a secondary response in stromal fibroblasts and endothelial cells resulting in bone marrow fibrosis. However, the study of cytokine gene expression levels in the microenvironment has been technically difficult, and has sometimes produced contradictory findings [3].
We previously demonstrated that the mutational profiles of PMF, ET, and PV correlate with histomorphologic characteristics in a cohort of Ph-negative MPN patients [4]. In this study, we expand this cohort of PMF, ET, and PV patients to characterize levels of inflammatory gene expression in the bone marrow. Using a technique that permits direct measurement of transcript levels in clinical bone marrow biopsies, we demonstrate a strong correlation between myelofibrosis and inflammatory gene expression in the bone marrow. Gene expression profiles were identified that distinguish prefibrotic MPN from overtly fibrotic MPN and define MPN subsets with different inflammatory pathway activities. These results emphasize the central role of the inflammatory microenvironment in the initiation and persistence of myelofibrosis and suggest that distinct MPN phenotypes may be functionally categorized by differences in proinflammatory signals.
Materials and methods
Study population
The pathology archives at Brigham & Women’s Hospital (BWH) and Massachusetts General Hospital (MGH) was queried to identify patients diagnosed with PMF, ET, PV, or MPN, unclassifiable (MPN-U) on bone marrow biopsy with concurrent hematologic data obtained between 2005 and 2016. Patients diagnosed with myelodysplastic syndrome/MPN overlap disease and those who had progressed to acute leukemia, received treatment with chemotherapeutic agents for prior cancer diagnoses, or had undergone stem cell transplantation were excluded. BWH specimens were fixed in Bouin’s fixative and decalcified in RapidCal Immuno (BBC Biochemical) for 15 minutes, followed by routine processing. Specimens from MGH were fixed in B-plus fixative for a minimum of 4 hours and decalcified in RapidCal Immuno (BBC Biochemical) for 30 minutes, followed by routine processing. Histologic review and fibrosis grading were performed by W.W., R.P.H. and O.P. based on consensus, using the 2016 WHO Revised Classification of Myeloid Neoplasms [5]. The study was conducted in accordance with the principles set forth by the Declaration of Helsinki and the requirement for informed consent was waived by the institutional review board.
Mutational analysis
Targeted sequencing of 95 commonly mutated genes in myeloid neoplasms was performed on DNA isolated from peripheral blood or bone marrow aspirates of 83 of the patients as part of their clinical evaluation. Amplicon library generation (TruSeq Custom Amplicon, Illumina, San Diego, CA) and next generation sequencing (MiSeq, Illumina, San Diego, CA) were performed as described [6]. Data processing and analysis were performed using MuTect for single-nucleotide variants with subsequent manual review and annotation (including evaluation of allele frequencies). Likely pathogenic variants were defined as frameshift, nonsense, splice-site mutations, insertions-deletions, or known pathogenic missense alterations.
Gene expression analysis
RNA was isolated from 50 μm sections prepared from fixed, decalcified and paraffin-embedded bone marrow biopsies using Qiagen RNeasy Kit (Germantown, MD). Briefly, 50 μm histologic sections were scraped into deparaffinization solution and RNA was isolated according to manufacturer’s instructions. RNA concentration was measured using a Nanodrop spectrophotometer (Thermo Fisher, Waltham, MA). Multiplexed mRNA quantification was performed using Nanostring nCounter GX Human Inflammation Kit (Seattle, WA), which contains color-coded hybridization probes against 249 inflammation-related genes, including cytokines, chemokines, pattern recognition receptors, cell adhesion molecules, and regulators of lymphocyte activation. Gene expression analysis was carried out using nSolver software (Nanostring, Seattle, WA) and RStudio (Boston, MA). For each gene, transcript count was normalized to the geometric mean of five housekeeping genes (GAPDH, GUSB, HPRT1, PGK1, TUBB) and six synthetic positive control RNA probes. Baseline threshold expression was defined as two standard deviations above the mean of six synthetic negative control RNA probes. Target genes with raw counts below the baseline threshold in more than two-thirds (67%) of samples were excluded from analysis. Samples with more than 75% of raw transcript counts below the baseline threshold, or requiring a housekeeping gene normalization factor of greater than 20.0, were considered to be of inadequate quality and excluded from analysis.
Statistical analysis
Cluster analysis was performed using stats R package. The k coefficient for k-means clustering was determined by the elbow method. Stepwise selection of differentially expressed genes was performed using the stepAIC function in MASS. Logistic regression modeling and leave one out cross validation were performed using caret. The predictive score cut-off was determined by recursive partitioning using rpart R package. Gene set enrichment analysis was performed using GSEA software and Molecular Signature Database (Broad Institute, Boston, MA). Differentially expressed genes were also analyzed for enrichment in Gene Ontology pathways (http://www.geneontology.org/page/go-enrichment-analysis, accessed August 2018). Statistical significance (p < 0.05) was determined using log2 transformed values by one-way ANOVA or Student’s t test, as appropriate. Adjusted false discovery rate (FDR; q < 0.05) was calculated using the Benjamini-Yekutieli method.
Results
We identified 142 bone marrow biopsies from unique individuals diagnosed with MPN. 115 samples were from BWH; 27 samples were from MGH; tissue for gene expression analyses was available in 135 specimens. Of 135 specimens with available tissue, 108 (80.0%) yielded gene expression profiles of adequate quality and constituted the final study sample. Low quality samples excluded from the analysis were associated with older specimen age and fixation in B-Plus fixative. There was no significant association between sample exclusion and MPN disease type (p = 0.9) or fibrosis grade (p = 0.9). The final cohort of 108 MPN patients included 36 (33.3%) with PMF, 31 (28.7%) with ET, 25 (23.1%) with PV, and 16 (14.8%) with MPN-U (Table 1). Of the 36 PMF cases, 13 (34.2%) were prefibrotic (WHO MF grade 0–1), whereas 23 (65.8%) showed overt fibrosis (WHO MF grade 2–3). Eight cases (32%) of PV had progressed to post-PV myelofibrosis and 10 (32.2%) cases of ET had progressed to post-ET myelofibrosis at the time of the analyzed biopsy. Seven (43.8%) out of 16 MPN-U cases represented advanced stages of MPN in which fibrosis obscured the underlying diagnosis. In total, 60 cases (55.6%) represented MPN with WHO MF grade 0–1, whereas 48 cases (44.4%) demonstrated overt (WHO MF grade 2–3) fibrosis. These specimens were randomly divided into training (n = 76) and test (n = 32) cohorts for further analysis (Table 1).
Unsupervised clustering of inflammatory gene expression was performed on the training cohort using both hierarchical and agglomerative methods. Hierarchical clustering separated overtly fibrotic MPN from prefibrotic disease with a sensitivity of 72% and a specificity of 89% (Fig 1A). However, this analysis did not distinguish specific MPN disease types (ET, PV, prefibrotic PMF, versus MPN-U with MF grade 0–1 in the prefibrotic cluster; and post-ET MF, post-PV MF, overtly fibrotic PMF, and MPN-U with MF grade 2–3 in the overtly fibrotic cluster) from each other. K-means agglomerative clustering (k = 4) demonstrated that both prefibrotic and overtly fibrotic MPN can be further partitioned into two groups, each with distinct gene expression patterns (Fig 1B). Cluster 1 and cluster 3 represented mostly prefibrotic MPN, whereas clusters 2 and 4 contained mostly fibrotic disease. The overall sensitivity was 95% for prefibrotic MPN and 72% for overtly fibrotic disease respectively, with an overall accuracy of 83%. Cluster 1 contained most of MF grade 0–1 MPN-U; only three cases (30%) were present in cluster 3, which was relatively enriched in ET instead (Fig 1B). In contrast, overtly fibrotic PMF was concentrated in cluster 2, whereas cluster 4 showed no disease-specific patterns. These differences in disease type were not statistically significant (p > 0.05), suggesting that in this analysis, pro-inflammatory responses represent a final common pathway leading to end-stage fibrosis in MPN. Therefore, we focused on inflammatory genes with altered expression in myelofibrosis independent of disease type.
(A) Hierarchical clustering separates MF grade 0–1 MPN from MF grade 2–3 MPN in a training cohort of 76 patients. (B) K-means clustering (k = 4) separates prefibrotic (MF grade 0–1) disease from overtly fibrotic disease (MF grade 2–3). Four clusters with differential gene expression are seen: clusters 1 and 3 consist of mostly prefibrotic MPN, whereas clusters 2 and 4 consist of overtly fibrotic MPN. Each row represents a single gene (n = 199). Each column represents 1 of 76 patients. The relative abundance (log2 count) of each gene transcript is indicated by the color bar.
Out of 199 genes expressed above background, the transcript levels of 123 genes were significantly altered in the bone marrow of patients with overt fibrosis (q < 0.05; S1 Table). Only three transcripts (HMGB2, ALOX15, DEFA1) were downregulated, while the remainder were upregulated. Five genes (CCL2, MX1, IFIT1, CXCL10, C1R) were upregulated by greater than 3-fold; 28 genes were upregulated by 2- to 3-fold (S1 Table). Notably, CCL2 is upregulated in other disorders marked by abnormal fibrosis, including fibrosing diseases of the lung, liver, and kidney. Other upregulated transcripts include additional mediators of the fibrogenic response [7], such as TGFB3 (2.6-fold upregulated), TGFB1 (1.9-fold upregulated), PDGFA (2.2-fold upregulated), and HIF1A (1.3-fold upregulated). Cytokines implicated in tissue fibrosis, such as TNF (1.9-fold upregulated) and IL1B (1.5-fold upregulated), were also significantly upregulated. Other upregulated transcripts encode toll-like receptors and their downstream effectors, such as TLR3, TLR5, RIPK2, and IRF7. In addition, some of the most highly upregulated genes in overtly fibrotic MPN included downstream targets of interferon signaling (IFIT1 and MX1; both upregulated 3.5-fold).
Gene set enrichment analysis (GSEA) confirmed that each MPN cluster is enriched in genes associated with distinct expression modules (Table 2 and Fig 2). Cluster 1 showed no gene expression profile enrichment by GSEA and is termed “cytokine low”. Cluster 3 showed moderately increased transcript levels of genes involved in cytokine signaling, including a significant enrichment in interferon gamma gene expression signature (“IFNγ moderate”). Compared to clusters 1 and 3, clusters 2 and 4 were characterized by high inflammatory cytokine gene expression. Cluster 2 showed abundant expression of interferons and tumor necrosis factors (“IFN/TNF high”), whereas cluster 4 demonstrated a gene expression signature associated with mature dendritic cells (“DC high”). Interestingly, although clusters 2, 3 and 4 demonstrated increased expression of PPARG2 associated genes, only cluster 2 (“IFN/TNF high”) showed increased expression of PPARG1 pathway genes, indicating a differential contribution of these closely-related inflammatory mediators towards certain types of myelofibrosis. We further carried out gene ontology (GO) term enrichment analysis (S2 Table), which identified pathways similar to those highlighted by GSEA. These pathways included JAK/STAT signaling, Toll-like receptor signaling, NFκB signaling, TNF signaling, interferon signaling, cytokine production and chemotaxis. In addition, pathways involved in VEGF signaling and osteoclast differentiation were significantly enriched in MPN demonstrating overt fibrosis (MF grade 2–3).
For each MPN subgroup identified by cluster analysis, an example of a positively or negatively correlated gene set from the Molecular Signature Database is shown. (A) Cluster 1 shows no significant enrichment in genes involved in innate immunity and other inflammatory pathways (q = 0.98). (B) Gene expression in cluster 2 is enriched for IFNB1 and TNF signaling (q < 0.05). (C) Cluster 3 shows moderate enrichment in interferon-γ pathway genes (q < 0.05). (D) Cluster 4 shows enrichment by genes associated with mature dendritic cells (q < 0.05).
Only a few genes showed a significant correlation with distinct driver mutations. TGFB2 was significantly upregulated 3.1-fold in MPL-mutant disease, whereas PTGER3 was downregulated 2.2-fold in MPN lacking JAK2, MPL, or CALR mutations (‘triple negative’ disease). The number of genes that are differentially expressed in various MPN disease types were small and no statistically significant pathway enrichment was detected.
An independent validation cohort demonstrated similar changes in gene expression between non-fibrotic and fibrotic MPN. Similar to the training cohort, hierarchical and k-means clustering of differentially expressed genes in the test cohort segregated overtly fibrotic (grade 2–3) disease from MPN with grade 0–1 fibrosis, with a sensitivity of 89% and specificity of 78% in both clustering algorithms (Fig 3A and 3B). We constructed a generalized logistic regression model using stepwise selection to identify the most distinctive gene expression patterns between non-fibrotic and fibrotic MPN. This predictive signature consists of five differentially expressed genes in fibrotic MPN: DDIT3, ALOX15, TCF4, MAPK14, and MAPKAPK5. Using this model, we accurately predicted bone marrow fibrosis in the test cohort with a sensitivity of 78% and a specificity of 91%, with an overall accuracy of 88% (kappa = 0.69). We also performed leave-one-out cross validation in a combined dataset comprising both training and test cohorts, which generated an overall prediction accuracy of 86% (kappa = 0.72) in assigning fibrosis grade.
(A) Hierarchical clustering separates MF grade 0–1 MPN from MF grade 2–3 MPN in a test cohort of 32 patients. (B) K-means clustering (k = 4) separates prefibrotic (MF grade 0–1) disease from overtly fibrotic disease (MF grade 2–3) into four clusters. Clusters 1 and 3 consist of mostly prefibrotic MPN, whereas clusters 2 and 4 consist of overtly fibrotic MPN. Each row represents a single gene (n = 199). Each column represents 1 of 32 patients. The relative abundance (log2 count) of each gene transcript is indicated by the color bar.
We and others have reported that overtly fibrotic PMF is enriched in ASXL1 mutations [4,8]. In 83 cases from both training and test cohorts with sequencing data, ASXL1 mutation was present in 21 (25.3%) cases. The presence of ASXL1 mutation correlated with the upregulation of 58 genes (S3 Table). All but five of these genes (KEAP1, DAXX, MAPKAPK5, TLR9, RIPK1; 8.6%) belong to the larger set of 123 differentially expressed genes in overtly fibrotic MPN (S1 Table). GO term enrichment analysis of these 58 genes demonstrates that the overrepresented biological processes in ASXL1 mutant disease are similar to those observed in overtly fibrotic MPN (S4 Table). However, multiple growth factor pathways, such as MAPK/JNK, PI3K/AKT, and VEGF signaling show comparatively greater enrichment in ASXL1 mutant disease than in overtly fibrotic MPN, suggesting that the lack of ASXL1 specifically enhances receptor tyrosine kinase signaling in MPN (Fig 4).
Select GO terms are shown that demonstrate greater fold enrichment in ASXL1 mutant MPN compared to MPN with grade 2–3 fibrosis. The color intensity reflects absolute fold enrichment. q < 0.05 for all data shown.
Many of the therapies given to MPN patients, such as JAK inhibitors or interferon, would be expected to impact inflammatory gene expression. 16 out of 108 patients (14.8%) had prior (n = 8) or current (n = 8) exposure to JAK inhibitor or interferon (Table 1). In a comparison with 38 control patients matched for fibrosis grade, no differentially regulated genes were identified (data not shown).
Discussion
The study of MPN is challenging due to the wide degree of overlap in mutational landscape, histologic features, and clinical presentation [9]. Furthermore, the development of myelofibrosis is a shared characteristic in all MPN, albeit with varying frequencies depending on the disease type. Our results suggest that overt myelofibrosis coincides with inflammatory gene upregulation in the bone marrow of MPN patients. The inflammatory gene expression pattern differs in various subsets of prefibrotic and overtly fibrotic MPN, suggesting that distinct inflammatory pathways give rise to the common phenotype of bone marrow fibrosis in advanced MPN. These findings suggest a new taxonomy of MPN based on transcriptional patterns in inflammation-related pathways. These gene expression profiles do not correlate with MPN disease type. Instead, the high degree of overlap between differentially regulated transcripts in JAK, MPL and CALR-mutant MPN supports the concept that MPN pathogenesis converges on a common JAK/STAT axis [2].
The broadest distinction in inflammatory gene expression lies between MF grade 0–1 (prefibrotic) and MF grade 2–3 (overtly fibrotic) MPN (Fig 1). Prefibrotic MPN is characterized by overall low cytokine gene expression. In contrast, overtly fibrotic MPN activate multiple inflammatory pathways. These pathways include Toll-like receptor (TLR) signaling and its downstream components--including NF kappa-B, MAPK, and TNF signaling (Table 2). TLR signaling in fibroblasts and tissue macrophages has been implicated in hepatic, cardiac and pulmonary fibrosis [10]. Our findings suggest that sustained proinflammatory cues in the bone marrow similarly lead to TLR-mediated fibrosis. Furthermore, two of the downregulated transcripts identified in this study, HMGB1 and HMGB2, are putative endogenous TLR ligands, suggesting that there is a negative modulatory response to abnormally activated TLR signaling in the microenvironment of overtly fibrotic MPN.
In contrast, other upregulated genes are likely to affect both the neoplastic myeloid clone as well as bone marrow inflammatory cells. For example, activated JAK/STAT signaling is a defining feature of clonal myelopoiesis in MPN, but JAK/STAT activity in stromal cells can also lead to fibroblast activation and myelofibrosis [11,12]. Another example is the IL1 family of cytokines, which exert a direct effect on myeloid development and are associated with fibrotic transformation in PV and PMF [13–15]. IL1RAP, a gene encoding the IL1 coreceptor, is upregulated in overtly fibrotic MPN and may represent a potential therapeutic target against myelofibrosis via anti-IL1RAP antibodies [16]. In a subset of MPN with grade 2–3 fibrosis, GSEA identified a gene expression pattern characteristic of dendritic cells ("DC high") (Table 2). Dendritic cells are a major source of pro-inflammatory cytokines in MPN and are efficiently targeted by ruxolitinib [17]. Dendritic cells also present antigens to T cells, priming them toward effector or regulatory responses. A recent study showed that defective CD4 T cell activation by dendritic cells causes a myeloproliferative phenotype in mice, highlighting the contribution of stromal inflammatory cells to MPN pathogenesis [18].
Importantly, the pattern of inflammatory gene expression did not generally differ between MPN disease types in our study, except in rare specific circumstances. MPN-U with MF grade 0–1 was absent from the “cytokine low” cluster and was instead enriched in the “IFNγ moderate” cluster, demonstrating that MPN-U in the prefibrotic phase possesses elevated pro-inflammatory activity compared to other prefibrotic MPN. Conversely, ET with MF grade 0–1 was relatively enriched in the “cytokine low” cluster, suggesting minimal inflammatory pathway activity compared to pre-fibrotic PMF, PV, and MPN-U. This gene expression profile may be related to the histological characteristics of ET, which displays relatively lower cellularity and lacks myeloid and/or erythroid hyperplasia in comparison to PV and prefibrotic PMF [5].
In overtly fibrotic MPN, increased marrow fibrosis in MPN is accompanied by an alteration in the bone marrow cellular composition. These histologic changes include increased stromal elements as well as myeloid and megakaryocytic hyperplasia [5,19]. Although our experiments do not provide spatial resolution of transcriptional changes within the bone marrow environment, we observed upregulation of genes important in myofibroblast proliferation in overtly fibrotic MPN, such as PDGFA and MAFK [20,21]. Furthermore, the upregulation of specific transcripts that are highly expressed in distinct hematopoietic cell lineages (e.g. CD163 and CD40LG) suggests that inflammatory cell populations, including macrophages and T lymphocytes, are increased in the bone marrow of overtly fibrotic MPN. GO analysis also demonstrated enrichment of osteoclast differentiation and angiogenic pathway genes in overtly fibrotic MPN. Therefore, multiple cell types contribute to the proinflammatory nature of a fibrotic marrow [22,23]. Intriguingly, a subset of MPN with grade 0–1 fibrosis demonstrate increased interferon gamma gene expression signature (cluster 3), suggesting that increased cytokine activity in the bone marrow microenvironment predates the onset of overt fibrosis [23].
In conclusion, we have found that overtly fibrotic MPN demonstrates marked upregulation of inflammatory genes within the bone marrow via multiple proinflammatory signaling networks. Clustering analysis and GSEA identified four subsets of MPN that are distinguished from each other by fibrosis grade and inflammatory gene expression (chemokines, cytokines, and innate immune response pathways) in the bone marrow. These results were validated in an independent cohort. By using bone marrow biopsies obtained as part of the routine care of patients with MPN, we have demonstrated a robust method of gene expression profiling that is applicable to Bouin’s-fixed, decalcified tissue. Furthermore, the upregulation of transcripts encoding targetable proteins in overtly fibrotic MPN suggests potential therapeutic mechanisms to alleviate myelofibrosis in MPN.
Supporting information
S1 Table. Differentially expressed genes in MPN with Grade 0–1 and Grade 2–3 fibrosis.
https://doi.org/10.1371/journal.pone.0216810.s001
(PDF)
S2 Table. Gene ontology terms enriched in MPN with Grade 2–3 fibrosis.
https://doi.org/10.1371/journal.pone.0216810.s002
(PDF)
S3 Table. Differentially expressed genes in ASXL1 mutant MPN.
https://doi.org/10.1371/journal.pone.0216810.s003
(PDF)
S4 Table. Gene ontology terms enriched in ASXL1 mutant MPN.
https://doi.org/10.1371/journal.pone.0216810.s004
(PDF)
Acknowledgments
This article is dedicated to the memory of Dr. Frank Kuo, who contributed substantially to this work until his untimely death.
References
- 1. Barbui T, Thiele J, Passamonti F, Rumi E, Boveri E, Ruggeri M, et al. Survival and disease progression in essential thrombocythemia are significantly influenced by accurate morphologic diagnosis: an international study. J Clin Oncol. 2011;29: 3179–3184. pmid:21747083
- 2. Agarwal A, Morrone K, Bartenstein M, Zhao ZJ, Verma A, Goel S. Bone marrow fibrosis in primary myelofibrosis: pathogenic mechanisms and the role of TGF-β. Stem Cell Investig. 2016;3.
- 3. Hasselbalch HC. The role of cytokines in the initiation and progression of myelofibrosis. Cytokine Growth Factor Rev. 2013;24: 133–145. pmid:23415024
- 4. Wong WJ, Hasserjian RP, Pinkus GS, Breyfogle LJ, Mullally A, Pozdnyakova O. JAK2, CALR, MPL and ASXL1 mutational status correlates with distinct histological features in Philadelphia chromosome-negative myeloproliferative neoplasms. Haematologica. 2018;103: e63–e68. pmid:29146710
- 5.
Swerdlow SH, Campo E, Harris NL, Jaffe ES. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues, Fourth Edition. Lucerne: World Health Organization; 2017.
- 6. Kluk MJ, Lindsley RC, Aster JC, Lindeman NI, Szeto D, Hall D, et al. Validation and Implementation of a Custom Next-Generation Sequencing Clinical Assay for Hematologic Malignancies. J Mol Diagn. 2016;18: 507–515. pmid:27339098
- 7. Eming SA, Wynn TA, Martin P. Inflammation and metabolism in tissue repair and regeneration. Science. 2017;356: 1026–1030. pmid:28596335
- 8. Guglielmelli P, Pacilli A, Rotunno G, Rumi E, Rosti V, Delaini F, et al. Presentation and outcome of patients with 2016 WHO diagnosis of prefibrotic and overt primary myelofibrosis. Blood. 2017;129: 3227–3236. pmid:28351937
- 9. Tefferi A, Pardanani A. Myeloproliferative Neoplasms: A Contemporary Review. JAMA Oncol. 2015;1: 97–105. pmid:26182311
- 10. Huebener P, Schwabe RF. Regulation of wound healing and organ fibrosis by toll-like receptors. Biochim Biophys Acta. 2013;1832: 1005–1017. pmid:23220258
- 11. Albrengues J, Bertero T, Grasset E, Bonan S, Maiel M, Bourget I, et al. Epigenetic switch drives the conversion of fibroblasts into proinvasive cancer-associated fibroblasts. Nature Communications. 2015;6: 10204. pmid:26667266
- 12. Čokić VP, Mitrović-Ajtić O, Beleslin-Čokić BB, Marković D, Buač M, Diklić M, et al. Proinflammatory Cytokine IL-6 and JAK-STAT Signaling Pathway in Myeloproliferative Neoplasms. Mediators Inflamm. 2015;2015: 453020. pmid:26491227
- 13. Panteli KE, Hatzimichael EC, Bouranta PK, Katsaraki A, Seferiadis K, Stebbing J, et al. Serum interleukin (IL)-1, IL-2, sIL-2Ra, IL-6 and thrombopoietin levels in patients with chronic myeloproliferative diseases. Br J Haematol. 2005;130: 709–715. pmid:16115126
- 14. Pourcelot E, Trocme C, Mondet J, Bailly S, Toussaint B, Mossuz P. Cytokine profiles in polycythemia vera and essential thrombocythemia patients: clinical implications. Exp Hematol. 2014;42: 360–368. pmid:24463275
- 15. Tefferi A, Vaidya R, Caramazza D, Finke C, Lasho T, Pardanani A. Circulating interleukin (IL)-8, IL-2R, IL-12, and IL-15 levels are independently prognostic in primary myelofibrosis: a comprehensive cytokine profiling study. J Clin Oncol. 2011;29: 1356–1363. pmid:21300928
- 16. Ågerstam H, Hansen N, von Palffy S, Sandén C, Reckzeh K, Karlsson C, et al. IL1RAP antibodies block IL-1-induced expansion of candidate CML stem cells and mediate cell killing in xenograft models. Blood. 2016;128: 2683–2693. pmid:27621309
- 17. Heine A, Held SAE, Daecke SN, Wallner S, Yajnanarayana SP, Kurts C, et al. The JAK-inhibitor ruxolitinib impairs dendritic cell function in vitro and in vivo. Blood. 2013;122: 1192–1202. pmid:23770777
- 18. Humblet-Baron S, Barber JS, Roca CP, Lenaerts A, Koni PA, Liston A. Murine myeloproliferative disorder as a consequence of impaired collaboration between dendritic cells and CD4 T cells. Blood. 2019;133: 319–330. pmid:30333120
- 19. Thiele J, Kvasnicka HM, Vardiman J. Bone marrow histopathology in the diagnosis of chronic myeloproliferative disorders: a forgotten pearl. Best Pract Res Clin Haematol. 2006;19: 413–437. pmid:16781481
- 20. Decker M, Martinez-Morentin L, Wang G, Lee Y, Liu Q, Leslie J, et al. Leptin-receptor-expressing bone marrow stromal cells are myofibroblasts in primary myelofibrosis. Nat Cell Biol. 2017;19: 677–688. pmid:28481328
- 21. Ruberti S, Bianchi E, Guglielmelli P, Rontauroli S, Barbieri G, Tavernari L, et al. Involvement of MAF/SPP1 axis in the development of bone marrow fibrosis in PMF patients. Leukemia. 2018;32: 438–449. pmid:28745329
- 22. Wen QJ, Yang Q, Goldenson B, Malinge S, Lasho T, Schneider RK, et al. Targeting megakaryocytic-induced fibrosis in myeloproliferative neoplasms by AURKA inhibition. Nat Med. 2015;21: 1473–1480. pmid:26569382
- 23. Schneider RK, Mullally A, Dugourd A, Peisker F, Hoogenboezem R, Van Strien PMH, et al. Gli1+ Mesenchymal Stromal Cells Are a Key Driver of Bone Marrow Fibrosis and an Important Cellular Therapeutic Target. Cell Stem Cell. 2017;20: 785–800.e8. pmid:28457748