Advances in epigenetic therapeutics with focus on solid tumors

Epigenetic (“above genetics”) modifications can alter the gene expression without altering the DNA sequence. Aberrant epigenetic regulations in cancer include DNA methylation, histone methylation, histone acetylation, non-coding RNA, and mRNA methylation. Epigenetic-targeted agents have demonstrated clinical activities in hematological malignancies and therapeutic potential in solid tumors. In this review, we describe mechanisms of various epigenetic modifications, discuss the Food and Drug Administration-approved epigenetic agents, and focus on the current clinical investigations of novel epigenetic monotherapies and combination therapies in solid tumors.


Background
Carcinogenesis is a complex process that involves both genetic and epigenetic changes, leading to the transformation of normal cells into malignant cells. The aberrant genetic and epigenetic alterations are the hallmark of cancer. Epigenetic modifications are responsible for cellular plasticity, differentiation and reprogramming without altering the underlying DNA sequence of the organism [1]. Normal cell development depends on regulated transcription of critical proteins, and individual cells within specific tissues and organs maintain their unique biological functions based on heritable and evolutionary differences in the DNA packaging. Histone proteins (two copies of histones H2A, H2B, H3 and H4) wrap around 147 base pairs of DNA to form a nucleosome. Nucleosomes are further compacted by additional proteins to form chromatin. Epigenetic modifications, including acetylation and methylation (histone marks), can alter DNA accessibility and chromatin structure and regulate gene transcription activation or silencing. Acetylated histones are less compact, thereby enabling gene transcription by making the DNA more accessible to RNA polymerase and the transcriptional machinery. On the other end, methylated histones can be either repressive or activating, depending on the site and degree of methylation. Methylation of histone H3 at lysine 4, 36 and 79 is generally considered as an activation mark, whereas methylations on histone H3 lysine 9, 27 are linked to transcriptional repression [2]. In general, enzymes that add acetyl or methyl groups to the histone or DNA are referred to as "writers", whereas enzymes that remove histone marks are called "erasers". Proteins that recognize histone and DNA modifications are the chromatin "readers" [1].
The complex balance of normal and abnormal epigenetic regulation is an area of intense interest in cancer research, including therapeutic development in cancer [3]. This article will illustrate aberrant changes in DNA methylation, histone acetylation and histone methylation (summarized in Fig. 1) in cancer, discuss the epigenetic agents in both hematological malignancies and solid tumors, and highlight the recent novel combination grouped into two major categories: broad spectrum reprogrammers and narrowed spectrum reprogrammers [4]. An argument can be made for the potential effectiveness of both broad and targeted epigenetic therapies. Broad-spectrum reprogrammers include the inhibitors of DNA methyltransferase (DNMT), histone deacetylase (HDAC) and the bromodomain and extra-terminal motif proteins (BETs). These drugs cause genome-wide cancerspecific gene expression alterations. In contrast, narrowed spectrum epigenetic modifying agents targeting lysine-specific histone demethylase 1 (LSD1), enhancer of zeste homolog 2 (EZH2), DOT1-like histone lysine methyltransferase (DOT1L), to achieve precise inhibition of epigenetic regulatory proteins.

Broad spectrum reprogrammers DNMT (DNA methyltransferase-"writer") inhibitors
DNA methylation affects the transcription of genes without altering the DNA sequence. In eukaryotic DNA, cytosine is methylated and then converted into 5-methylcytosine by DNMTs [5]. Hypermethylation of specific regions, such as the CpG islands of tumor suppressor genes, plays an important role in carcinogenesis for many types of cancers [6][7][8]. There are 3 primary DNMTs-DNMT1, DNMT3A and DNMT3B [9][10][11]. DNMT1 is predominantly involved in maintaining the preexistent methylation pattern during DNA replication. DNMT3A and DNMT3B are involved in facilitating de novo DNA methylations at loci that were previously unmethylated [12]. Tumorigenesis often involves an interplay among all 3 DNMTs [13][14][15][16]. DNMT inhibitors act as cytidine analogs and induce loss of DNA methylation. There are two main classes of hypomethylating agents, the nucleoside analogs (such as 5-azacitidine that incorporates into DNA and RNA and 5-aza-2′-deoxycytidine, or decitabine, that incorporates into DNA) and the anti-sense DNA methyltransferase inhibitors (such as MG98) that do not require incorporation into DNA. The ability of azacitidine to be incorporated into DNA and RNA can lead to broad biological effects in resting and dividing cells [17]. DNMT inhibitors have shown to be particularly effective in targeting DNA methylation in leukemic cells [18,19].

HDAC (histone deacetylase-"eraser") inhibitors
Histone modification occurs via acetylation of lysine residues. Two families of enzymes, histone acetyltransferases (HATs) and histone deacetylases (HDACs), operate in an opposing manner. HATs acetylate lysines within the amino-terminal tails of histone proteins, resulting in relaxation of chromatin structure and facilitating gene activation. Conversely, HDACs remove acetyl groups from hyperacetylated histones and make the chromatin condensed and transcriptionally silent. There are four classes of HDAC enzymes based on their structures and functions: class I (HDAC 1-3 and 8), IIa (HDAC 4,5,7,9), IIb (HDAC 6, 10), III (Sir-2 related-SIRT1-7) and IV (HDAC 11) [20,21]. Class I HDAC proteins are mainly localized in the nucleus, whereas class II HDACs are expressed in a more tissue-restricted manner [22]. Sharing significant homology with both Class I and Class II HDACs, class IV HDAC does not possess a nuclear localization signal and its function is largely unknown [23]. HDACs are key elements in the regulation of gene expression, differentiation and development, and the maintenance of cellular homeostasis. HDAC inhibition causes global gene upregulation (potential oncosuppressors) and leads to arrest of tumor cell growth, apoptosis and anti-angiogenesis [24,25]. In addition, HDAC facilitates the binding of elongation factors to acetylated promoters and enhancers for efficient elongation. Therefore, HDAC inhibitors block gene elongation and inhibit gene expression, especially in highly expressed genes (oncogenes) [26]. Many HDAC inhibitors are non-specific and can be used to inhibit multiple isoforms of HDACs.

BET (bromodomain and extra-terminal motif proteins-"reader") inhibitors
BET proteins are known to recognize acetylated lysine in chromatin [27]. The BET family of proteins include BRD2, BRD3, BRD4, and the testes-specific BRDT [28,29]. Bromodomains can specifically bind acetylated lysine residues of histone proteins, and are involved with histone modifications, chromatin remodeling and transcriptional activation via recruitment of other proteins [30,31]. BRD2 and BRD3 facilitate the passage of RNA Pol II to elongate the DNA transcripts through hyperacetylated nucleosomes [32]. BRD4 enhances the recruitment of positive transcription elongation factor b (P-TEFb), leading to the release of Pol II from a pause in transcription elongation in the promoter-proximal region [33]. In particular, aberrant BRD4 expression contributes to carcinogenesis by mediating hyperacetylation of the chromatin associated with cell proliferation-promoting genes [34]. Suppression of BRD4 led to anti-leukemic effects in acute myeloid leukemia (AML) mouse models and revealed a potential epigenetic target for AML [35]. In addition, BRD4 and BET proteins also regulate enhancer (a short region of DNA that can be bound by transcription factors to enhance the transcription of a particular gene) function and, in particular, large clusters of enhancers (superenhancers), which drive oncogene expression, such as BCL-2 and c-MYC [36,37]. Interestingly, the pathogenic fusion product of NUT (nuclear protein in testis) with BRD4 or BRD3 (BRD4-NUT or BRD3-NUT) causes NUT midline carcinoma (NMC), which is a rare but poorly differentiated and highly aggressive cancer of the squamous cell lineage that arises in midline structures [38]. BET bromodomain blockade using small-molecule inhibitors leads to selective repression of the transcriptional network driven by c-MYC [39].

METTL3 (methyltransferase like-3-"writer") inhibitors
In addition to the epigenetic modifications on either DNA or histones, methylation is also observed in eukaryotic RNAs, including messenger RNA (mRNA), micro-RNA (miRNA) and long non-coding RNA (lncRNA), etc. Methylation modification impacts RNA processing, nuclear export, translation initiation and degradation [40]. In particular, N 6 -methyladenosine (m 6 A) modification of mRNA is most abundant, which occurs in two consensus sequence motifs including G(m 6 A)C primarily and A(m 6 A)C to a lesser extent [41,42]. m 6 A is installed by a multiprotein writer complex that consists of methyltransferase-like protein 3 (METTL3), methyltransferaselike protein 14 (METTL14) and other accessory subunits. m 6 A modification is reversible and it can be erased by ALKBH5 (alkB homolog 5) [43] and FTO (fat-mass and obesity associated protein) proteins ( Fig. 1) [44]. In addition, METTL3 and METTL14 are also identified as key actors of adenosine methylation of miRNAs [45,46], whereas FTO is recognized as a key actor of adenosine demethylation of miRNAs [47]. m 6 A reader proteins can specifically bind to m 6 A transcripts and regulate the metabolism of mRNA [48]. For example, YTHDF2 (YTH domain family 2) binds to m 6 A in mRNA and targets mRNA degradation, whereas YTHDF1, YTHDF3, and eukaryotic initiation factor 3 (eIF3) promote translation of mRNA transcripts [49]. METTL3 has been found to be upregulated with increased m 6 A levels in cancer compared with those in normal tissues, suggesting a potential oncogenic role in different cancer types including AML, renal cell carcinoma, non-small cell lung cancer (NSCLC) and gastric cancer [50][51][52][53]. The studies show that loss of either METTL14 or METTL3 in AML cell lines and primary leukemic blasts led to induction of differentiation [50,54]. In addition, METTL3 has been associated with multiple cell signaling pathways, including tumorigenesis, proliferation, invasion, migration, cell cycle, differentiation and cell viability [55]. Currently, multiple METTL3 inhibitors are under investigation in both AML and solid tumors, with pending clinical trials in the near future [56].
Besides the role of METTL3 in m 6 A modification on mRNAs and miRNAs, recent study suggested that DNMT3A methylates miRNA at cytosine residues and inhibits the formation of miRNA/mRNA duplex, leading to the loss of their repressive function in gene expression [57]. Therefore, using demethylating agent to block miRNA methylation may broaden its therapeutic potentials.

Narrowed spectrum reprogrammers LSD1 (histone demethylase-"eraser") inhibitors
LSD1 (lysine-specific histone demethylase 1, also known as KDM1A) is the first discovered histone lysine demethylase with the ability to erase the mono-methyl and dimethyl chromatin marks on histone H3, predominantly at lysines 4 and 9 (H3K4 and H3K9) [58][59][60]. It can also demethylate non-histone proteins, including DNMT1 and TP53 [59]. Moreover, LSD1 is a multifunctional subunit of both repressive and activating histone-modifying complexes and can therefore act as both a transcriptional repressor or activator in a context-dependent manner [61]. LSD1 regulates the balance between self-renewal and differentiation of stem cells, and LSD1 inhibition in mixed lineage leukemia (MLL)-rearranged leukemia has been shown to downregulate expression of some leukemia associated genes and cause apoptosis and cell differentiation [62]. In addition, LSD1 is overexpressed in various solid tumors including prostate, breast, lung and colorectal cancers, and neuroblastoma [63][64][65][66][67]. Pharmacological inhibition of LSD1 leads to inhibition of proliferation, differentiation, invasion, and migration in vitro and in vivo [68]. Thus, LSD1 inhibitors might be promising potential therapeutic options in a variety of cancers.
Recently, it has been demonstrated that the effects of LSD1 inhibitors are particularly robust for small cell lung cancer (SCLC) through promotion of differentiation of tumor-enriched stem-like cells [69].

DOT1L (histone methyltransferase-"writer") inhibitors
Disruptor of telomeric silencing 1 (DOT1) is a novel class of HMT that was first identified to dysregulate gene silencing near telomeres in yeast [84]. DOT1like (DOT1L) is the only known methyltransferase that deposits mono-, di-, and trimethyl marks on histone H3 lysine 79 (H3K79) in mammals. It participates in the regulation of transcription, differentiation and proliferation of normal cells. DOT1L has been shown to be critical for transformation by MLL fusion proteins in AML [85,86]. Preclinical models demonstrate that MLL-driven leukemia is particularly sensitive to inhibition of DOT1L activity, and DOT1L inhibitors have been shown to specifically reduce H3K79 methylation marks and expression of MLL-fusions target genes in leukemic cells [87]. In addition, a recent study demonstrated the role of DOT1L in breast cancers that do not harbor a MLL translocation. DOT1L plays an important role in the initiation and progression of breast cancer by targeting the gene expression of EMT-promoting factors, suggesting DOT1L to be a therapeutic target for aggressive breast cancer [88]. While the pre-clinical studies showed promising activity of DOT1L inhibitors, the phase I study of DOTlL inhibitor, pinometostat, in adult and pediatric patients with relapsed or refractory leukemia demonstrated limited clinical response [89,90].

IDH (isocitrate dehydrogenase) inhibitors
Mutations in genes encoding enzymes of the tricarboxylic acid (TCA) cycle can disrupt cell metabolism and alter the epigenetic landscape. For example, IDH1/2 enzymes metabolize isocitrate to α-ketoglutarate (α-KG) in the TCA cycle. α-KG serves as a co-factor for α-KGdependent dioxygenases, including the ten-eleven translocation (TET) family of DNA demethylases and Jumonji family of histone demethylases. TET family of DNA methylases act on methylated DNA sequences, convert 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC), which will ultimately remove methyl groups and ensure the correct DNA methylation in the cell [91]. IDH1/2 mutations are found in several cancer types, including AML, gliomas, chondrosarcoma and intrahepatic cholangiocarcinoma [92,93]. IDH mutations (gain-of-function) result in further processing of α-KG to 2-hydroxyglutarate . This leads to the production of "oncometabolite" 2-HG, which inhibits TET family of DNA demethylases and Jumonji family of histone demethylases [94] and promotes tumorigenesis [95]. Accumulation of 2-HG in leukemic cells leads to increased DNA and histone methylation and results in blocked cell differentiation [96,97]. Several small molecule inhibitors of both IDH1 and IDH2 have demonstrated reduction of 2-HG levels and differentiation of leukemic cells that carry the specific IDH mutations [98][99][100]. These effects also correlate with global changes in DNA methylation/ histone modification state, suggesting that the phenotypic effects are, to some extent, secondary to rewiring transcriptional programs in the leukemic cells [101].
The aforementioned RNA demethylases, FTO and ALKBH5 which demethylate m 6 A, are α-KG-dependent dioxygenases [102][103][104]. m 6 A destabilizes transcripts and controls expression of key transcription factors in hematopoietic stem cells (HSCs) and human embryonic stem cells (ESCs) [105]. 2-HG suppresses FTO activity in leukemia cells, leading to decreased expression of the lineage transcription factor CCAAT enhancer binding protein α (C/EBPα) that enforces normal HSC quiescence and myeloid differentiation [106]. Therefore, the inhibition of IDH may lead to the changes in metabolic activities in TCA cycle such as α-KG and 2-HG, coordinating the cell fate in HSCs and ESCs.

Approved epigenetic therapies
To date, the FDA-approved epigenetic agents are mostly limited in treating hematologic malignancies. Two DNMT inhibitors are approved for the treatment of myelodysplastic syndrome (MDS)-azacitidine and decitabine. Clinical trials with azacitidine and its deoxy derivative, decitabine, demonstrated that 15% or more of the patients with AML or intermediate to high-risk MDS showed improvement in blood cell counts and survival [107,108]. Several HDAC inhibitors are approved for the treatment of hematologic malignancies, including belinostat for peripheral T cell lymphoma (PTCL), panobinostat for multiple myeloma, vorinostat for cutaneous T cell lymphoma (CTCL) and romidepsin for both CTCL and PTCL. IDH inhibitors, enasidenib and ivosidenib, have been approved for relapsed or refractory AML with IDH mutations [109][110][111]. EZH2 inhibitor, Tazemetostat, has been approved for patients with relapsed or refractory follicular lymphoma (R/R FL) with EZH2 mutation and who have received at least 2 prior systemic therapies, and for adult patients with R/R FL who have no satisfactory alternative treatment options [112].
Clinical trials are ongoing in solid tumors with agents from multiple drug classes. In January 2020, tazemetostat has been granted accelerated approval by FDA in treating epithelioid sarcoma, for which we will discuss later in this article [113]. These FDA-approved agents are summarized in Table 1.

Monotherapies in solid tumors
Historically, the first generation DNMT inhibitors (azacytidine and decitabine) showed limited activity in solid tumor, in part due to their toxicity. Biomarker studies demonstrated evidence of DNA methylation changes associated with drug administration; however, the responses were short-lived and treatment resistance developed early [114][115][116][117]. A phase I study of decitabine was conducted in patients with stage IV lung cancer, esophageal cancer, and malignant pleural mesothelioma. No objective response was observed and severe toxicities occurred. Grade 4 neutropenia was observed in 43% (15 out of 35) of the patients and grade 3 hepatotoxicity were seen in two patients with extensive liver metastases [118].
The second-generation DNMT inhibitors, such as guadecitabine (SGI-110), have been undergoing investigation. Guadecitabine is a novel hypomethylating prodrug of decitabine with a prolonged half-life. This novel compound is an oligonucleotide consisting of decitabine linked through a phosphodiester bond to the endogenous nucleoside deoxyguanosine. The dinucleotide configuration provides protection from drug clearance [119]. Guadecitabine [119] has been demonstrated to be safe and well tolerated as a single agent, with evidence of promising activity in heavily pretreated MDS and AML patients [120]. A phase II trial of SGI-110 monotherapy in patients with HCC who progressed on sorafenib (NCT01752933) was completed. The single agent SGI-110 demonstrated disappointing PFS in this trial.
Similar to DNMT inhibitors, HDAC inhibitors have shown limited single agent activity, and responses have been rare in solid tumors [121,122]. A phase II study of vorinostat in relapsed non-small cell lung cancer (NSCLC) showed no objective response in 14 evaluable patients, and severe toxicities were reported including neutropenia, lymphopenia, fatigue and pulmonary embolisms [123]. A phase III trial of vorinostat as second-line monotherapy in advanced mesothelioma was conducted in patients who had previously received chemotherapy, and it showed that single agent vorinostat did not improve overall survival (OS) compared with placebo [124]. Representative recent clinical trials of single agent DNMT inhibitors and HDAC inhibitors in solid tumors are summarized in Table 2.  To further explore the efficacy of epigenetic monotherapy, newer epigenetic agents have been investigated beyond HDAC and DNMT inhibitors, targeting more specific patient population with a narrowed spectrum epigenetic modulation. Among them, tazemetostat is the first FDA-approved epigenetic therapy in the solid tumor, epithelioid sarcoma [112]. ES is a rare soft tissue sarcoma that is characterized by the loss of expression in INI1/SNF5/SMARCB1. SMARCB1 (SWI/SNF related, matrix associated, actin dependent regulator of chromatin, subfamily b, member 1), a subunit of SWI/ SNF (SWItch/Sucrose Non-Fermentable) chromatin remodeling complex, can repress EZH2 transcription [125]. The loss of INI1 function leads to elevated expression and recruitment of EZH2 to target genes, resulting in the upregulation of several oncogenic signaling pathways [126]. The accelerated approval of tazemetostat was based on the results of a single arm cohort in patients with metastatic or locally advanced ES who are not eligible for complete resection (NCT02601950). Nine out of sixty two patients with INI1-negative ES (15%) had partial response (PR) and six out of those nine patients (67%) had a duration of response lasting 6 months or longer. Tazemetostat was generally well tolerated [127] in the study.
In addition, early phase studies demonstrated BET inhibitors had clinical activities in patients with NMC. NMC is a rare and aggressive squamous cancer, which is commonly driven by the BRD4-NUT or BRD3-NUT fusion oncoprotein. A phase Ib study of birabresib Only select studies within the past 5 years have been included due to extent of clinical trials AE adverse events, CRC colorectal cancer, CRPC castrate-resistant prostate cancer, DCR disease control rate, DNMT DNA methyltransferase, HCC hepatocellular carcinoma, HDAC histone deacetylase, HR hormone receptor, ITT intention-to-treat, NPC nasopharyngeal carcinoma, NSCLC non-small cell lung cancer, ORR objective response rate, OS overall survival, PD progressive disease, PFS progression-free survival, PR partial response, SAE serious adverse event, SCLC small cell lung cancer, SD stable disease, TEAE treatment-emergent adverse event Results: Stable disease was the best response in 27 patients. Median PFS and stable disease duration were both 11.4 months and median OS has not been reached. Grade 3 AEs that occurred in more than 1 patient included lymphopenia (n = 5), hypertension (n = 3), oral pain (n = 2), thromboembolic event (n = 2) and fatigue (n = 2). Eleven patients required dose reduction due to drug related AEs [187] (MK-8628/OTX015) was conducted in patients with NMC. Three out of ten patients (30%) with NMC had a PR with duration of response of 1.4 to 8.4 months [128].
In another phase I study of molibresib (GSK525762), out of nineteen NMC patients, four (21%) achieved either confirmed or unconfirmed PR and eight patients (42%) had stable disease as best response [129]. These results have demonstrated that targeting BRD4-NUT and BRD3-NUT with BET inhibitors resulted in strong antitumor activity in this rare patient population.
Another new epigenetic agent targeting a specific genetic defect in epigenetic pathways has been investigated. The phase III ClarIDHy trial (NCT02989857) evaluated the IDH1 inhibitor ivosidenib in 185 previously treated patients with IDH1-mutated advanced cholangiocarcinoma. Ivosidenib improved PFS from 1.4 months with placebo to 2.7 months (hazard ratio [HR] = 0.37; P < 0.001). Although the objective response rate was low (2.4%), clinical benefit was observed with stable disease (SD) in 50.8% of patients. Median OS was 10.8 months with ivosidenib versus 9.7 months with placebo (HR = 0.69; P = 0.06), including 57% of patients who crossed over from placebo group [130]. As a side note, the benefit of IDH1 inhibitors in patients with chondrosarcoma is controversial [131,132], in part due to the different histological subtype with various disease aggressiveness and clinical outcome [133].
Summarized clinical trials investigating novel epigenetic drugs (single agent) in solid tumors are listed in Table 3.

Combination therapies in solid tumors
Due to the limited efficacy of epigenetic monotherapy as described previously, and the complexity of epigenetic modification in cancer, many trials are investigating combination therapies in solid tumors. Recent clinical trials include epigenetic modifier combinations as well as combinations of epigenetic agents with cytotoxic chemotherapy, hormonal therapies, and immune checkpoint inhibitors (ICIs).

Combination of DNMT inhibitors and HDAC inhibitors
Preclinical studies demonstrated that DNMT inhibitor enhances apoptosis in cancer cells induced by HDAC inhibitors, suggesting the potential synergism of DNMT in combination with HDAC inhibitors [134]. A phase I/ II trial of azacitidine and entinostat in NSCLC yielded some promising results with durable responses [135]. This trial included heavily pre-treated patients who had received a median of three prior therapies. Clinical efficacy was observed with one complete response (CR) for 14 month duration, one PR for eight month duration, and ten patients with SD lasting at least 12 weeks. One of these patients had stable disease for 18 months and another for 14 months. The prolonged clinical benefit in certain patients in this trial prompted a correlative biomarker study to predict treatment response. The study collected and examined the promoter methylation status in circulating DNA from patient plasma collected before therapy (day 0) and after 1 cycle of therapy (day 29). Of these, ten out of 26 patients demonstrated a decrease in methylation during the first four weeks of treatment compared to their baseline. There was a higher response rate and improvement in overall survival in the patients with methylation changes ("methylation signature"-positive) compared to patients without methylation change ("methylation signature"-negative). The median OS and PFS were 10.42 months for the methylation signaturepositive cohort versus 6.54 months for the methylation signature-negative (P = 0.035). This suggests a potential role of epigenetic therapy in NSCLC, and the important role of biomarkers to predict response and benefit in patients.

Epigenetic therapy with cytotoxic chemotherapy
Preclinical studies suggested that DNMT and HDAC inhibitors have the greatest efficacy when combined with chemotherapy in an attempt to re-sensitize cancers to the standard cytotoxic agents [136,137]. Acquired resistance to the chemotherapy agents might be reversed when combined with DNMT and/or HDAC inhibitors, especially in ovarian cancers [138]. A phase I trial of low-dose decitabine combined with carboplatin was conducted in patients with recurrent platinum-resistant ovarian cancer. The low dose decitabine was tolerated and demonstrated biological activity in DNA hypomethylation. However, the clinical benefit was modest [139]. Another phase II randomized study compared guadecitabine in combination with carboplatin against secondline chemotherapy in patients with platinum-resistant ovarian cancer. It does not meet the primary endpoint and there is no difference in either median PFS or OS between the two groups [140,141]. Similarly, in a phase I trial in patients with metastatic colorectal cancer who were exposed to irinotecan previously, guadecitabine in combination with irinotecan showed modest clinical activity with stable disease as the best response [142]. As a note, the challenge in epigenetic agents in combination with cytotoxic chemotherapies include the side effects of additive toxicities needing dose reduction of epigenetic agents. In addition, the chemotherapies cause G1/S cell cycle arrest, which may interfere with incorporation of hypomethylating agents into the DNA and RNA.

Epigenetic therapy with immune checkpoint inhibitors
ICIs have recently changed the cancer treatment landscape in many types of cancers. The combination of epigenetic agents with ICIs is an area of investigation in a variety of solid tumors [143]. In the clinical trial involving 45 patients with advanced-stage NSCLC who were treated with azacitidine and entinostat, five patients who had disease progression during the trial were subsequently enrolled in trials of anti-PD-1 therapy [135]. Three of the five patients achieved an objective response and the other two had SD for 24 weeks before disease progression. This clinical observation has led to pre-clinical research to understand the mechanism of epigenetic therapies in modulating immune responses. Treatment of tumor cells with DNMT inhibitors can induce the transcription of endogenous retrovirus (ERVs), which are normally silenced in most somatic tissues [144]. The reactivation of ERVs result in the formation of cytoplasmic double-stranded RNAs [145,146], the cognate ligand of the retinoic acid inducible gene I (RIG-I)-like receptors (RLR), including RIG-I and melanoma differentiation associated gene 5 (MDA5) [147]. Activation of the RLR family (innate immune sensors) initiates signaling cascades leading to the production of type I and III interferons, which elicit an antitumor immune response (virial mimicry) by activation of CD8+ T cells [148,149]. Also, epigenetic therapy can lead to the re-expression of tumor antigens, such as cancer testis antigens (CTAs) and melanoma-associated antigen 1 (MAGE1), increasing immunogenicity [150][151][152]. Therefore, both pre-clinical and clinical studies suggests that these epigenetic therapies might augment antitumor immune response through various mechanisms, enhancing tumor antigen expression and infiltration of cytotoxic T cells, and reversing T cell exhaustion with a concurrent increase in the abundance of effector and/or memory T cells, among others [153]. These observations are being translated into clinical trials that focus on the combination of ICIs with epigenetic drugs in a variety of solid tumors. A phase I/Ib trial of pembrolizumab plus oral vorinostat (HDAC inhibitor) has been completed in patients with advanced/metastatic NSCLC [154]. Thirty-three patients were treated, including thirteen in phase I and twenty in phase Ib. In phase I, both ICI-naïve and ICIpretreated patients were enrolled to determine dose-limiting toxicities (DLTs). No DLTs were observed, and the recommended phase II dose was pembrolizumab 200 mg and vorinostat 400 mg/day. The most common adverse events of any grade included fatigue (33%) and nausea/ vomiting (27%). Among those 6 ICI-naïve patients, there was 1 case (16.7%) of confirmed PR, 4 cases (66.7%) of SD, and 1 case (16.7%) of PD. Of 24 ICI-pretreated patients evaluable for response, there were 3 cases with (13%) PR (1 confirmed), 11 cases with (46%) SD and 10 cases (42%) with progressive disease (PD). The results suggested the combined therapy of pembrolizumab and vorinostat is feasible with a manageable safety profile and active in both ICI-naïve and -exposed NSCLC patients. The presence of CD8+ T-cell in tumor stroma in pretreatment samples, not CD8+ T-cell in tumor bed, was associated with treatment benefit. In addition, on-treatment biopsies showed the increase in CD8+ T cells in the stroma was correlated with clinical benefit (with SD or PR for a period of ≥ 24 weeks). It would be crucial to investigate whether the combination is better than ICI alone in ICI-naïve patients in the front line setting and/or if the combination is superior to the standard of care in ICI-exposed patients in the later line treatment setting. An ongoing randomized phase 2 trial is examining pembrolizumab +/− vorinostat in ICI-naive advanced/metastatic NSCLC patients (NCT02638090).
Similarly, a phase II study is investigating azacitidine and entinostat with concurrent nivolumab in patients with metastatic NSCLC, in both ICI-naïve and ICIresistant patient populations (NCT01928576) and a phase I study is investigating pembrolizumab in combination with guadecitabine and mocetinostat for patients with advanced lung cancer who progressed on prior ICIs (NCT03220477). These on-going trials include correlative studies to evaluate induced viral mimicry, interferon induction, and T cell function phenotypes [153].
The newer epigenetic agents in combination with ICIs are also under investigation. A phase I/II trial is evaluating a BET inhibitor, INCB057643, in combination with pembrolizumab and epacadostat (indoleamine 2, 3-dioxygenase or IDO-1 inhibitor) in patients with advanced or metastatic solid tumors (NCT02959437). Additionally, trials of EZH2 inhibitors in combination with ipilimumab (CTLA-4 inhibitor) or pembrolizumab are recruiting the patients with advanced solid tumors (NCT03525795 and NCT03854474).

Epigenetic therapy with other anticancer therapies
New approaches combining epigenetic agents with other anticancer therapies, including hormonal therapy, have been explored as an approach to overcome treatment resistance. In the phase II study ENCORE301, entinostat was added to exemestane (steroidal aromatase inhibitor [AI]) in patients with hormone receptor (HR)-positive advanced breast cancer with disease progression after prior non-steroidal AI. The study demonstrated a significant improvement in PFS (HR = 0.73; p = 0.06) and also in OS (HR = 0.59; p = 0.036). The combination was well tolerated, with neutropenia (13%) and fatigue (11%) being the most frequent grade 3 or 4 toxicities in entinostat-treated patients [155]. Therefore, entinostat, when added to exemestane, was designated by the FDA as breakthrough therapy for postmenopausal women with HR-positive advanced breast cancer whose disease has progressed after nonsteroidal AI therapy. Based on the ENCORE301 study, a phase III trial (E2112) is ongoing to investigate entinostat versus placebo in combination with exemestane in patients with locally advanced or metastatic breast cancer who have experienced disease progression after a non-steroidal AI [156] Everolimus, a sirolimus (formerly called rapamycin) derivative, inhibits phosphatidylinositol 3-kinase (PI3K)/ Akt/(158)mammalian target of rapamycin (mTOR) signaling pathway, which is one of the mechanisms of endocrine resistance in HR-positive breast cancer [157,158]. In preclinical studies, the use of everolimus in combination with aromatase inhibitors results in synergistic inhibition of the proliferation and induction of apoptosis [159]. The BOLERO-2 trial showed that everolimus in combination with exemestane improved PFS compared to exemestane alone in post-menopausal women with advanced HR+/Her2-negative breast cancer [160]. However, recent data suggested that the combination of exemestane and everolimus did not yield a durable clinical response, indicating a need for alternative combinations and therapeutic strategies [161]. The pre-clinical studies showed that resistance to everolimus was mediated by overexpression of MYC in ER-positive cancers, which can be reversed by BET inhibitors [162]. Also, a combination of BET inhibitor with fulvestrant (ER degrader) showed long-lasting antitumor effect in a tamoxifen (selective ER modulator)-resistant breast cancer xenograft mouse model [163].
Similarly, the combination of BET inhibitors with AR antagonists is able to subvert resistance in castrateresistant prostate cancer (CRPC) in preclinical experiments [164]. Other studies combining BET and PARP inhibition show mitotic catastrophe (cell death related to premature entry of cells into mitosis) with induction of apoptosis, causing synergistic effect in suppressing BRCA1/2 wild-type ovarian cancer. This study also suggests that BET inhibitors re-sensitize PARP-inhibitorresistant BRCA mutant epithelial ovarian cancer cells to PARP inhibition [165]. DNMT inhibitors create a "BRCAness" phenotype through downregulating expression of key homologous recombination and nonhomologous end-joining (NHEJ) genes, and promote synergism with PARP inhibitors in the setting of BRCA-proficient NSCLC in animal models. These pre-clinical data support the expansion of therapeutic studies of PARP inhibitors and various epigenetic agents in patients with BRCA-proficient cancer [166].
There are also ongoing clinical trials with BET inhibitors in combination with PARP inhibitors, ER antagonists, and AR antagonists. A phase I trial is accruing patients to investigate AZD5153 in combination with olaparib for platinum-resistant/refractory ovarian cancer. Other accruing studies include a phase II trial of ZEN003694 in combination with talazoparib in TNBC (NCT03901469); a phase I/II trial to test GSK525762 in combination with fulvestrant in advanced HR-positive breast cancer (NCT02964507); and a phase Ib study combining GSK525762 with abiraterone or enzalutamide in advanced CRPC (NCT03150056). In addition, several early phase trials are investigating EZH2 inhibitors in combination with enzalutamide or abiraterone in metastatic CRPC, given the synergistic effect of EZH2 inhibitors in combination with AR antagonists.
Ongoing clinical trials of combination therapies of epigenetic drugs with chemotherapy or other agents including ICIs in solid tumors are listed in Table 4.

Conclusions
The development of epigenetic therapeutics has promise for cancer treatment, particularly with advancements in hematologic malignancies. In solid tumors, only one epigenetic agent (EZH2 inhibitor, tazemetostat) has been approved (ES). It is not fully understood why solid tumors are not as sensitive to epigenetic agents, even though there is profound aberrant epigenetic alterations in solid tumors. There may be a critical difference in cellular differentiation and epigenetic plasticity between solid tumors and hematological malignancies. Solid tumors arise from a more terminally differentiated state, which may be intrinsically more resistant to epigenetic reprogramming. In contrast, hematopoietic lineages are precisely controlled by epigenetic modulation. It is understandable that epigenetic agents demonstrated robust clinical activity in hematological malignancies in which cell differentiation is a key biological feature. The alternative explanation could be that altered epigenetic modulation may occur early in oncogenesis, however, it is not the "driver" event that controls the tumor cell proliferation and survival [167]. In the era of precision oncology, the broad impact of epigenetic treatment is both promising in "reprograming" solid tumor epigenetic dysfunction, as well as challenging in targeting particular epigenetic driving events. In recent years, the further development of next generation of broad spectrum agents and the emerging narrow spectrum agents as potential targeted epigenetic therapy have provided the new opportunities for solid tumor therapy. The approval of an epigenetic agent (EZH2 inhibitor, tazemetostat) in treatment of a rare soft tissue malignancy, epithelioid sarcoma, is a solid step towards the future breakthrough in the mechanism based solid tumor epigenetic treatment.
Various HDAC and DNMT inhibitors have been tested for treatment of both hematologic malignancies and solid tumors. Primary and secondary resistance to these therapies are common [168,169]. No clear clinical benefits have been observed as yet in solid tumors. The limited antitumor activity with DNMT and HDAC inhibitors as monotherapy in solid tumors may also be related to either the short half-lives of the S phase-specific drugs with low incorporation into DNA [115] or due to a lack of specificity. Combination therapies with dual DNMT and HDAC inhibitors are explored in clinical trials; the therapeutic rationale is that densely methylated DNA is usually accompanied by deacetylated histone (transcriptionally repressive) [170]. However, most of the dual-agent epigenetic therapy trials did not result in an obvious clinical benefit, except the observation of durable responses in select NSCLC patients [135].
Potential novel therapies are being investigated to target new epigenetic modulation, such as IDH mutation inhibition and LSD1 inhibition, in both hematologic and solid malignancies. Many of these agents are targeting specific genetic defects in epigenetic pathways. Ivosidenib showed improved PFS in patients with cholangiocarcinoma harboring IDH1 mutation [130]. Pre-clinical studies suggest targeted epigenetic therapy may be effective in specific patient subsets, such as LSD1 inhibitors in the treatment for SCLC [69]. Early phase studies demonstrated BET inhibitors had activities in NMC, which is driven by BET fusion proteins. Most recently, METTL3 inhibitors and other agents targeting RNA epigenetics are emerging as potential cancer therapies with pending clinical trials. Results: Although the recommended combination was tolerable, no significant anticancer activity was observed [194] Azacitidine + vorinostat (HDACi) Locally recurrent and metastatic NPC and nasal natural killer T-cell lymphoma DLTs include grade 4 thrombocytopenia, grade 3 nausea, vomiting and fatigue and grade 5 hepatic failure, and worsening of pre-existing Sweet's Syndrome. Common grade 1/2 AEs were fatigue (73%), cough (64%), anorexia (55%) and injection site reaction (45%). One minor response was seen and 5 patients had prolonged stable disease (> 16 weeks) [195] Combination       The exciting finding that epigenetic agents are able to modulate tumor microenvironment has been a focus of epigenetic research. The combination of these "reprogramming" effects with other approved or novel therapies are being extensively explored. One of the current focuses is the combined epigenetic and immune therapy. It may be speculated that epigenetic agents have a significant "reprogramming" activity in immune cell components in addition to cancer cell component. There are many ongoing clinical trials evaluating the combination of the epigenetic agents with ICI in solid tumors. DNMT, HDAC, and other epigenetic inhibitors may enhance the response to and/or reverse the resistance to ICIs, if these agents can modulate key components of the tumor microenvironment including tumor cells, stromal cells, and innate and/or adaptive immune cells.
Beyond the scope of the current review, there are also important implications of epigenetic biomarkers in cancer screening, diagnosis, prognosis, and prediction to treatment. The development in the epigenetic biomarkers field are addressed in other reviews, including this one by Berdasco et al. [171].
In summary, epigenetic drugs represent "genomic medicines" that do not require existing DNA mutations. Given the wide diversity of solid tumors, epigenetic therapy is attractive because of the potential to target and modify the cancer genome functions. It is likely that cancer cells exploit epigenetic modulation to activate cellular pathways in cancer cell survival, including drug resistance and immune surveillance. Thus, epigenetic agents may have great therapeutic potential in the future under the right contexts. It will be essential to continue fundamental research to better identify the underlying mechanism and to translate these findings into clinical trial of newer epigenetic agents and optimize combinatorial approaches with exploration of predictive biomarkers in solid tumors.