Methylation age as a correlate for allele burden, disease status, and clinical response in myeloproliferative neoplasm patients treated with vorinostat

Abstract

•DNA methylation age varies depending on MPN phenotype.•Increased JAK2V617F mutation burden correlates with increased DNA methylation age.•Vorinostat therapy can alter DNA methylation age in PV and ET patients. The myeloproliferative neoplasms (MPNs) are a heterogeneous group of clonal neoplastic disorders. Driver mutations in JAK2, CALR, and MPL genes have been identified in the majority of cases. Alongside these, an increasing number of genes are repeatedly identified as mutated in MPN. These, including ASXL1, TET2, DMNT3A, and EZH2, have key roles in epigenetic regulation. Dysregulation of epigenetic processes is therefore a key feature of MPN. Vorinostat is a pan histone deacetylase inhibitor (HDACi) that has been investigated in MPN. DNA methylation (DNAm) is a well-defined epigenetic mechanism of transcription modification. It is known to be affected by ageing, lifestyle, and disease. Epigenetic ageing signatures have been previously described allowing calculation of a methylation age (MA). In this study we examined the effect of vorinostat on MA in MPN cell lines and in patients with polycythaemia vera (PV) and essential thrombocythaemia (ET) treated with vorinostat as part of a clinical trial. An older MA was observed in patients with a higher JAK2 V617F allele burden and those with a longer duration of disease. PV patients had a MA older than that predicted whilst MA was younger than predicted in ET. Treatment with vorinostat resulted in a younger MA in PV patients and older MA in ET patients, in both cases a trend towards the normal chronological age. When MA change was compared against response, nonresponse was associated with a younger than predicted MA in ET patients and a higher than predicted MA in PV patients. The link between MA and JAK2 mutant allele burden implies that allele burden has a role not only in clinical phenotype and disease evolution in MPN patients, but also in the overall methylation landscape of the mutated cells. The myeloproliferative neoplasms (MPNs) are a heterogeneous group of clonal neoplastic disorders. Driver mutations in JAK2, CALR, and MPL genes have been identified in the majority of cases. Alongside these, an increasing number of genes are repeatedly identified as mutated in MPN. These, including ASXL1, TET2, DMNT3A, and EZH2, have key roles in epigenetic regulation. Dysregulation of epigenetic processes is therefore a key feature of MPN. Vorinostat is a pan histone deacetylase inhibitor (HDACi) that has been investigated in MPN. DNA methylation (DNAm) is a well-defined epigenetic mechanism of transcription modification. It is known to be affected by ageing, lifestyle, and disease. Epigenetic ageing signatures have been previously described allowing calculation of a methylation age (MA). In this study we examined the effect of vorinostat on MA in MPN cell lines and in patients with polycythaemia vera (PV) and essential thrombocythaemia (ET) treated with vorinostat as part of a clinical trial. An older MA was observed in patients with a higher JAK2 V617F allele burden and those with a longer duration of disease. PV patients had a MA older than that predicted whilst MA was younger than predicted in ET. Treatment with vorinostat resulted in a younger MA in PV patients and older MA in ET patients, in both cases a trend towards the normal chronological age. When MA change was compared against response, nonresponse was associated with a younger than predicted MA in ET patients and a higher than predicted MA in PV patients. The link between MA and JAK2 mutant allele burden implies that allele burden has a role not only in clinical phenotype and disease evolution in MPN patients, but also in the overall methylation landscape of the mutated cells. The myeloproliferative neoplasms (MPNs) are a group of clonal hematological disorders in which there is a change from the polyclonal hematopoiesis seen in health to an abnormal monoclonal proliferation of blood cells. Polycythemia vera (PV) and essential thrombocythemia (ET) are characterized by the excess production of red blood cells and platelets, respectively. Identification of the JAK2 V617F driver mutation, in 95% of PV cases and 50% of ET cases, causing constitutive activation of the JAK/STAT pathway has revolutionized our understanding of the pathogenesis of these conditions [1Baxter EJ Scott LM Campbell PJ et al.Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative disorders.Lancet. 2005; 365: 1054-1061Abstract Full Text Full Text PDF PubMed Scopus (2300) Google Scholar]. In JAK2 V617F-negative cases, driver mutations in MPL and CALR have been identified in the majority of remaining ET cases [2Nangalia J Massie CE Baxter EJ et al.Somatic CALR mutations in myeloproliferative neoplasms with nonmutated JAK2.N Engl J Med. 2013; 369: 2391-2405Crossref PubMed Scopus (1199) Google Scholar,3Pikman Y Lee BH Mercher T et al.MPLW515L is a novel somatic activating mutation in myelofibrosis with myeloid metaplasia.PLoS Med. 2006; 3: e270Crossref PubMed Scopus (997) Google Scholar]. There is now evidence that MPNs are the result of combined genetic and epigenetic dysregulation, with mutations in cooperating genes increasingly reported [4Grinfeld J Nangalia J Green AR Molecular determinants of pathogenesis and clinical phenotype in myeloproliferative neoplasms.Haematologica. 2017; 102: 7-17Crossref PubMed Scopus (38) Google Scholar]. These include genes involved in cell signalling pathways (LNK, CBL, NRA, NF1), epigenetic regulation (ASXL1, EZH2, TET2, DNMT3A, IDH1, IDH2), transcriptional regulation (TP53, RUNX1), and mRNA processing (SF3B1, SRSF2, U2AF1, ZRSR2). These mutations dictate the way genes are expressed, and are not MPN specific, being found in a wide variety of myeloid disease phenotypes and clonal hematopoiesis of indeterminate potential [5Tremblay D Sokol K Bhalla S Rampal R Mascarenhas JO Implications of mutation profiling in myeloid malignancies: Part 1.Myelodysplastic syndromes and acute myeloid leukemia. Oncology (Williston Park). 2018; 32: e38-e44PubMed Google Scholar]. In addition, they are not mutually exclusive, making the hierarchy complex and unpredictable. Increasing evidence suggests that the order of acquisition of mutations can determine the phenotype of the disease [6Ortmann CA Kent DG Nangalia J et al.Effect of mutation order on myeloproliferative neoplasms.N Engl J Med. 2015; 372: 601-612Crossref PubMed Scopus (244) Google Scholar]. Other mechanisms of epigenetic dysregulation have been identified in MPN. JAK2 V617F protein localises to the nucleus mediating phosphorylation of histone H3 and the arginine methyltransferase PRMT5 [7Liu F Zhao X Perna F et al.JAK2V617F-mediated phosphorylation of PRMT5 downregulates its methyltransferase activity and promotes myeloproliferation.Cancer Cell. 2011; 19: 283-294Abstract Full Text Full Text PDF PubMed Scopus (176) Google Scholar,8Dawson MA Bannister AJ Göttgens B et al.JAK2 phosphorylates histone H3Y41 and excludes HP1alpha from chromatin.Nature. 2009; 461: 819-822Crossref PubMed Scopus (449) Google Scholar]. Overexpression of the transcription factor in NFE2 in PV results in elevated levels of JMJD1C, a histone demethylase, and subsequent global reductions in H3K9me1 and H3K9me2 levels [9Peeken JC Jutzi JS Wehrle J et al.Epigenetic regulation of NFE2 overexpression in myeloproliferative neoplasms.Blood. 2018; 131: 2065-2073Crossref PubMed Scopus (17) Google Scholar]. In addition, DNA methylation patterns have been reported to be abnormal in chronic-phase MPN and change further in transformation to blast-phase disease [10Pérez C Pascual M Martín-Subero JI et al.Aberrant DNA methylation profile of chronic and transformed classic Philadelphia-negative myeloproliferative neoplasms.Haematologica. 2013; 98: 1414-1420Crossref PubMed Scopus (32) Google Scholar]. Best available therapies (BATs) have remained unchanged for PV and ET patients for many years and include low-dose aspirin and cytoreductive agents such as hydroxycarbamide. These therapies have no effect on modifying the underlying disease process. Recent developments include the use of specific JAK inhibitors including ruxolitinib, a direct JAK1 and JAK2 inhibitor. Studies have indicated improved hematocrit control, spleen volume reduction, and modest but sustained allele burden reduction in PV patients resistant or intolerant to HU [11Vannucchi AM Kiladjian JJ Griesshammer M et al.Ruxolitinib versus standard therapy for the treatment of polycythemia vera.N Engl J Med. 2015; 372: 426-435Crossref PubMed Scopus (400) Google Scholar,12Vannucchi AM Verstovsek S Guglielmelli P et al.Ruxolitinib reduces JAK2 p.V617F allele burden in patients with polycythemia vera enrolled in the RESPONSE study.Ann Hematol. 2017; 96: 1113-1120Crossref PubMed Scopus (41) Google Scholar]. However, no benefit over BATs was established for ET in the same second-line setting [13Harrison CN Mead AJ Panchal A et al.Ruxolitinib vs best available therapy for ET intolerant or resistant to hydroxycarbamide.Blood. 2017; 130: 1889-1897Crossref PubMed Scopus (71) Google Scholar]. As the role of epigenetic dysregulation in MPN becomes increasingly established, epigenetic therapies have been trialed in MPN. Vorinostat (MK-0683) is a pan histone deacetylase inhibitor (HDACi) which has been reported to induce tumour cells to undergo growth arrest, differentiation, or apoptotic cell death [14Atadja P Gao L Kwon P et al.Selective growth inhibition of tumor cells by a novel histone deacetylase inhibitor, NVP-LAQ824.Cancer Res. 2004; 64: 689-695Crossref PubMed Scopus (119) Google Scholar, 15Bradbury CA Khanim FL Hayden R et al.Histone deacetylases in acute myeloid leukaemia show a distinctive pattern of expression that changes selectively in response to deacetylase inhibitors.Leukemia. 2005; 19: 1751-1759Crossref PubMed Scopus (312) Google Scholar, 16Butler LM Zhou X Xu WS et al.The histone deacetylase inhibitor SAHA arrests cancer cell growth, up-regulates thioredoxin-binding protein-2, and down-regulates thioredoxin.Proc Natl Acad Sci USA. 2002; 99: 11700-11705Crossref PubMed Scopus (443) Google Scholar]. In PV and ET, vorinostat has demonstrated efficacy in MPN. Discontinuation of therapy over a 6-month treatment phase was high, with only 48% of patients completing the treatment course. A majority of patients had decreased leucocyte or platelet counts on treatment, with a reduction in the prevalence of splenomegaly and pruritus observed. Very modest reductions in the JAK2 V617F burden in positive patients were observed with no relation to response [17Andersen CL McMullin MF Ejerblad E et al.A phase II study of vorinostat (MK-0683) in patients with polycythaemia vera and essential thrombocythaemia.Br J Haematol. 2013; 162: 498-508Crossref PubMed Scopus (59) Google Scholar]. DNA methylation (DNAm) is known to be altered by ageing and can reflect the effect of diet, lifestyle, or disease on cellular processes [18Horvath S. DNA methylation age of human tissues and cell types.Genome Biol. 2013; 14: R115Crossref PubMed Scopus (2172) Google Scholar]. Changes in DNAm influence the relative transcription profile of the cell by activating or inactivating gene transcription. “Methylation age” (MA) may be a more accurate reflection of disease than chronological age (CA). Using an ageing signature composed by Weidner et al. to generate individual MA [19Weidner CI Lin Q Koch CM et al.Aging of blood can be tracked by DNA methylation changes at just three CpG sites.Genome Biol. 2014; 15: R24Crossref PubMed Scopus (438) Google Scholar], we hypothesised that DNAm may be altered in MPN patients, resulting in a change in MA. Further, we hypothesised that the use of an epigenetic modifier would alter MA in PV and ET patients. Therefore, we set out to investigate the effect of vorinostat on MA in a clinical trial setting. UKE-1, SET-2, and HEL cell lines were cultured in line with standard practices. Vorinostat was solubilized in dimethyl sulfoxide (DMSO). Specified concentrations were added to a cell suspension obtained at 2 × 105 cells/mL. Cells were then incubated at 37°C for required time frames. To validate the DNAm ageing signature previously described, the granulocyte fraction of whole blood obtained in EDTA from healthy volunteers was obtained by centrifugation, separation of buffy coat, addition of phosphate-buffered saline (PBS), and further centrifugation. DNA extraction was performed using the Quick-gDNA as of DNA was performed using the as to The was to DNA for and genes using with of the DNA was using the and DNAm levels were the epigenetic ageing signature previously described by Weidner et al. [19Weidner CI Lin Q Koch CM et al.Aging of blood can be tracked by DNA methylation changes at just three CpG sites.Genome Biol. 2014; 15: R24Crossref PubMed Scopus (438) Google Scholar]. were available from PV and ET patients in a phase II study of vorinostat trial patients had to the and of blood for use in associated with the trial. of JAK2 V617F was performed as previously described [17Andersen CL McMullin MF Ejerblad E et al.A phase II study of vorinostat (MK-0683) in patients with polycythaemia vera and essential thrombocythaemia.Br J Haematol. 2013; 162: 498-508Crossref PubMed Scopus (59) Google Scholar]. were to response as previously described [17Andersen CL McMullin MF Ejerblad E et al.A phase II study of vorinostat (MK-0683) in patients with polycythaemia vera and essential thrombocythaemia.Br J Haematol. 2013; 162: 498-508Crossref PubMed Scopus (59) Google Scholar]. was to including and using the as not In Weidner et al. performed a of CpG in blood [19Weidner CI Lin Q Koch CM et al.Aging of blood can be tracked by DNA methylation changes at just three CpG sites.Genome Biol. 2014; 15: R24Crossref PubMed Scopus (438) Google Scholar]. described the of DNAm levels at three key the of a MA that CA in normal using an calculation To the ageing signature calculation in the was of CA in normal in our were obtained with from healthy The age of the volunteers was years and and 2 The granulocyte fraction was obtained from blood obtained in EDTA with subsequent DNA of gene of was performed in of the DNA and of CpG methylation was three and MA was for of the volunteers using the ageing signature Using the of DNAm at and the of DNAm at in and the DNAm for the in resulted in a MA that with the chronological age of the volunteers as a whole as in Vorinostat was as a in a group of PV and ET patients as part of an phase II study study patients from Vorinostat was at a of for to vorinostat response and response on an was with a in the of splenomegaly and There a of and or of response [17Andersen CL McMullin MF Ejerblad E et al.A phase II study of vorinostat (MK-0683) in patients with polycythaemia vera and essential thrombocythaemia.Br J Haematol. 2013; 162: 498-508Crossref PubMed Scopus (59) Google Scholar]. DNA from PV and ET patients from this trial were available for The was patients and patients with a age of years at trial from to trial was available for but and was on patients a and were at the time of trial the of the trial relative JAK2 V617F allele and response trial from blood were at trial to of vorinostat and of In of the patients, DNA was also available of vorinostat MA was for at time A predicted MA was also for using known CA and the established in the healthy were then to MA with clinical including known disease and The of MA to CA at time is for patients in The trend line established in healthy volunteers the patients in the MA was older or younger than to be a trend towards a higher than predicted MA was older than for CA years younger to years of vorinostat had altered with a trend a than MA compared with CA years years younger to years the was disease group and was observed a towards a higher than MA at time in whilst the was in the ET was a CA ET PV patients, the MA was years at years at and years at ET patients MA was years at years at and years at The between the was at trial years older years but not years older years or years older years of vorinostat the change in age of the and disease In this patients at a CA of and MA as a of The change in MA of vorinostat was not and the change from normal chronological ageing change of to When the on the patients had available were a in MA was from to to When disease were the change from to was only ET patients to and not the PV group to An MA was by the predicted MA from the observed MA. this MA was for the for ET and for the PV of the MA was for the whole for the ET group and for the PV group An of was performed to between MA and the seen in the the of driver mutations or mutant JAK2, JAK2 mutant allele CALR, and cooperating mutations (ASXL1, TET2, EZH2, A link between MA and JAK2 allele burden was JAK2 allele burden was associated with MA at using with patients with JAK2 allele patients with JAK2 allele at had an older MA at years and therapy years PV and ET patients were examined PV patients with a JAK2 allele burden compared with the group had a MA of years versus years at and years versus years ET, patients compared with burden patients had a MA of years versus years at and years versus years However, of vorinostat this between allele burden and MA was not in the overall with a MA in patients of years versus years in the was also in the PV group with the MA years versus years in the There were remaining burden patients in the ET group to When patients were in the the results were phenotypes and time with only the ET time had not been these There was no between JAK2 burden at and change in MA on There was no between change in allele burden and MA on vorinostat examined the effect of time to in the study on JAK2 V617F allele burden and MA. from to was available for but There was a between allele burden and longer to the patients, time to was compared with in the group we investigated MA was with time to and a positive was and However, by the no When the MA the between and predicted was the same were Therefore, Vorinostat the MA and MA were now of time to cooperating mutations were in patients with patients more than effect was seen on MA or MA in relation to these MA was in with the known response of the response response and nonresponse In the compared with was associated with a younger MA of therapy years evidence patients with a younger MA with nonresponse was also the ET compared with was associated with a younger MA of therapy years and compared with was associated with a younger MA of therapy years When the effect of MA on response was further were the was by compared with was associated with an MA younger than that for CA a MA years was also the ET compared to was associated with an MA younger than for CA at years In addition, compared with was associated with an MA older than for CA at years and of therapy years The role of epigenetic dysregulation in MPN pathogenesis has been increasingly S K McMullin MF The pathogenesis and epigenetic dysregulation efficacy in myeloproliferative Med. 2018; PubMed Scopus Google Scholar]. Using an epigenetic ageing signature on DNAm at three genes for which has been for changes in cellular between [19Weidner CI Lin Q Koch CM et al.Aging of blood can be tracked by DNA methylation changes at just three CpG sites.Genome Biol. 2014; 15: R24Crossref PubMed Scopus (438) Google Scholar], we set out to investigate the on DNAm resulting from the use of a histone deacetylase inhibitor, in clinical trial DNAm is the best described epigenetic mechanism of transcription regulation. The effect of ageing on DNAm levels has been to as the and is a that has been for more than years [18Horvath S. DNA methylation age of human tissues and cell types.Genome Biol. 2013; 14: R115Crossref PubMed Scopus (2172) Google Scholar]. The of a landscape in has been associated with neoplastic A et DNA methylation of genes that are in cells is a of Res. PubMed Scopus Google Scholar], and DNAm changes have also been in myeloid T C of DNA methylation in acute myeloid 2014; PubMed Scopus Google Scholar]. have reported an change in DNAm patterns in chronic-phase MPN compared with normal and a further change transformation to blast-phase disease [10Pérez C Pascual M Martín-Subero JI et al.Aberrant DNA methylation profile of chronic and transformed classic Philadelphia-negative myeloproliferative neoplasms.Haematologica. 2013; 98: 1414-1420Crossref PubMed Scopus (32) Google Scholar]. In with we have found a in MA between PV and ET, with PV patients a higher than predicted MA in to ET patients a than predicted MA. also found that patients with a higher JAK2 V617F allele burden have an increased MA for patients and the to the role of DNA methylation change in the pathogenesis of the MPN understanding of the of MPN phenotype JAK2 V617F allele are associated with an of the PV phenotype A et al.The clinical phenotype of and JAK2V617F in polycythemia 2006; PubMed Scopus Google Scholar]. Recent suggests that the order of acquisition of mutations may the resulting phenotype [6Ortmann CA Kent DG Nangalia J et al.Effect of mutation order on myeloproliferative neoplasms.N Engl J Med. 2015; 372: 601-612Crossref PubMed Scopus (244) Google Scholar], whilst cooperating somatic mutations are observed with between MPN a role in the The number of somatic mutations is also observed to be higher in myelofibrosis in to ET or PV and of myeloproliferative neoplasms.Blood. 2017; PubMed Scopus Google Scholar]. In this in DNAm between ET and we have not or The older MA may be a reflection of cellular processes the PV phenotype to the JAK2 allele The between time from to with JAK2 allele burden the that allele burden may as a for disease In JAK2 there is of the DG J et of cells is by JAK2V617F cell Biol. 2013; PubMed Scopus Google Scholar]. It be that of this cell in higher JAK2 allele burden or may have an on the MA of the resulting the change in DNAm which is by the MA observed may influence the of a phenotype regulation of gene transcription a or expression to these of the role of vorinostat in DNAm have been by the clinical trial the of the in the trial resulted in a with many patients to the of Using we observed a change in the MA of both PV and ET patients. with predicted in PV patients as be of a disease ageing older than predicted at trial and towards younger with In ET patients had a younger than MA at trial and older on towards the predicted MA from our normal over the of treatment, that there was a of DNAm patterns resulting from vorinostat These results are to resulting from the of with MA at the of the for When we at the of the in MA at of therapy in the ET was with a effect towards normal resulting from vorinostat not between JAK2 allele burden at and change in MA on therapy or between the change in allele burden on therapy and MA. have described a of MA to response in the clinical trial. was associated with a younger MA of therapy in to in the ET nonresponse was associated with an older MA than predicted at in to in the PV suggests that MA may to that in these vorinostat is to the mechanisms altered methylation patterns of disease in the methylation patterns are towards normal an with disease response It be to investigate this of of DNAm patterns with disease response in the setting of therapies for including the JAK study investigated the effect of the vorinostat on DNAm at three key genes and which have previously been to an epigenetic signature in observed an older MA at in patients with a higher JAK2 V617F allele with PV had an observed MA older than whilst patients with ET had an observed MA younger than Therefore, DNA methylation patterns may be or the resulting disease in ET patients was associated with a younger than predicted MA therapy in to patients with a in nonresponse was associated with an older than predicted MA in to to is that of the DNAm ageing pattern of cells towards normal may be of response more In an of vorinostat to DNAm in a number of cases a to