Therapeutic modulation of epigenetic drivers of drug resistance in ovarian cancer

Abstract

Epigenetic changes in tumours are associated not only with cancer development and progression, but also with resistance to chemotherapy. Aberrant DNA methylation at CpG islands and associated epigenetic silencing are observed during the acquisition of drug resistance. However, it remains unclear whether all of the observed changes are drivers of drug resistance, causally associated with response of tumours to chemotherapy, or are passenger events representing chance DNA methylation changes. Systematic approaches that link DNA methylation and expression with chemosensitivity will be required to identify key drivers. Such drivers will be important prognostic or predicitive biomarkers, both to existing chemotherapies, but also to epigenetic therapies used to modulate drug resistance.

Introduction

Although there have been substantial advances in current chemotherapeutic strategies, clinical drug resistance remains a major obstacle to successful cancer treatment and is still a limiting factor in patient survival [Broxterman et al. 2009]. This problem is particularly obvious in the treatment of ovarian cancer [Agarwal and Kaye, 2003]. Although around 80% of ovarian cancer patients initially respond to primary chemotherapy, the majority will relapse and eventually develop resistance to currently available treatment options. Conventionally, standard treatment consists of platinum-based drugs (carboplatin/cisplatin) which are either given as a single agent or in combination with the taxane, paclitaxel [Greenlee et al. 2001]. The chemotherapeutic activity of platinum drugs is mainly based on their ability to form DNA adducts and the 1,2-d (GpG) intrastrand crosslink has been particularly implicated. Here, the platinum coordinates the N7 atoms of adjacent guanosines in the DNA strand, which is believed to be poorly repaired in cells, with persistence of the lesion ultimately interfering with replication and transcription leading to cell death [Wang and Lippard, 2005; Kartalou and Essigmann, 2001]. Ovarian cancer is defined clinically as being ‘platinum resistant’ if the tumour recurrence occurs less than 6 months after completion of platinum-based first-line chemotherapy. An improved response to second-line treatment is seen if the tumour recurs later than 6 months following the end of first-line treatment and chances for a ‘platinum-sensitive’ response are greatly enhanced if the relapse occurs after 12 months.

CpG island methylation and drug resistance

A number of genetic alterations have been suggested to underlie the phenomenon of drug resistance, such as alterations in genes involved in DNA repair, drug uptake, apoptosis and cell cycle checkpoints [Broxterman et al. 2009; Luqmani, 2005]. However, in recent years it has become increasingly apparent that aberrant epigenetic mechanisms may also play a crucial role in drug resistance (for a summary of epigenetic mechanisms, see Figure 1). DNA methylation is one of the major epigenetic mechanisms controlling gene expression and cell differentiation [Bird, 1996]. Regions of high CpG dinucleotide density, called CpG islands, are frequently located in the promoters of house-keeping genes and are usually free of methylation in normal cells [Bird, 2002]. In cancer, CpG islands can become hypermethylated, contributing for example to silencing of tumour suppressor genes. This has been demonstrated for cancer susceptibility genes such as BRCA1 [Press et al. 2008]. Furthermore, methylation of the CpG island linked to BRCA1 is associated with good response to platinum-based chemotherapy in ovarian cancer [Teodoridis et al. 2005]. Mutation of the BRCA1/BRCA2 genes is also associated with good response to platinum-based chemotherapy [Cass et al. 2003]. Conversely, reversion mutations at BRCA2 have been shown to be associated with platinum resistance [Edwards et al. 2008]. It remains to be established whether reversal of BRCA methylation is a mechanism of acquired resistance. In ovarian cancer cell lines, methylation of the FANCF gene has been observed to be associated with increased sensitivity to cisplatin [Taniguchi et al. 2003]. FANCF is crucial for the activation of a DNA repair complex containing BRCA1 and BRCA2. Treatment with the demethylating agent 2′-deoxy-5-azacytidine led to demethylation of the FANCF gene and reduced sensitivity towards cisplatin in these cell line models [Taniguchi et al. 2003]; however, again, the relevance of FANCF methylation to clinical outcome following chemotherapy is still to be established.

Figure 1.

Possible targets for epigenetic therapy approaches in chemoresistant cells. Aberrant DNA hypermethylation (dark circles) at CpG islands can lead to transcriptional inactivation of genes and is frequently observed in tumours. Inhibition of the enzymes catalysing DNA methylation (DNMTs) leads to a genome-wide decrease of DNA methylation levels, thereby potentially re-activating vital anticancer genes. Each histone modification is established via specific enzymes (HDACs and HKMTs) catalysing the addition or removal of certain marks. Targeting aberrant hypoacetylation via HDACi can result in the re-expression of former transcriptionally incompetent chromatin. Similarly, the inhibition of certain aberrantly active histone methyltransferases (HKMTs) prevents methyl marks which may lead to repression of genes, as seen following the binding of HP1 (orange circles) to H3K9me3. Next to the two major epigenetic mechanisms of DNA methylation and histone modification, aberrant expression of miRNAs (small non-protein-coding RNAs of 21–23 nucleotides) has been correlated with tumourigenesis. Here, miRNAs potentially act on two pathways, the transcriptional silencing mechanism and the translational silencing mechanism. If in either case the miRNAs promiscuously bind sequences, for instance via mutation, the former specific gene regulation is out of control, possibly contributing to tumourigenesis and chemoresistance. In ovarian cancer, for example, downregulation of the miRNA let-7i increases resistance to cisplatin and is associated with shorter progression-free survival time of patients with late-stage ovarian cancer [Yang et al. 2008]. Although not yet available, gene therapy may be a tool to either re-establish lost endogenous miRNA expression or silence aberrant miRNA expression thereby complementing existing epigenetic therapies.

In contrast to BRCA and FANCF where epigenetic or genetic inactivation of the gene is associated with drug sensitivity, inactivation of genes involved in engaging an apoptotic response would lead to drug resistance [Teodoridis et al. 2005]. For instance, methylation of the DNA mismatch repair gene MLH1 and transcriptional silencing occurs in cisplatin-resistant ovarian cell line models. MLH1 has been shown to be necessary for engagement of a variety of downstream cellular responses to alkylating agents and cisplatin-induced DNA damage [Papouli et al. 2004; Stojic et al. 2004]. It has been argued that, since mismatch repair (MMR) proteins can recognize and bind to certain types of damage in DNA, this is necessary for MMR-dependent engagement of DNA damage responses such as activation of p53, p73 and other downstream apoptosis signalling pathways [Stojic et al. 2004; Shimodaira et al. 2003; Duckett et al. 1999]. Hence, loss of MLH1 expression may lead to reduced engagement of apoptosis either due to reduced cycles of futile repair [Karran and Hampson, 1996], reduced stalling (or increased bypass) of lesions in DNA during DNA replication [Moreland et al. 1999] or direct signalling of cell death pathways [Yoshioka et al. 2006]. Acquired methylation at the MLH1 locus has been observed following platinum-based chemotherapy in ovarian cancer, which has been shown to be associated with poor patient survival [Gifford et al. 2004].

Increased methylation of DAPK might also have an implication in chemoresistance. Hypermethylation of DAPK occurs frequently in tumours such as colon and breast tumours [Yamaguchi et al. 2003; Lehmann et al. 2002] and is an indicator of poor clinical outcome in lung cancer patients [Tang et al. 2000]. The calcium/calmodulin-regulated serine/threonin kinase DAPK has a pro-apoptotic role which is mediated via interferon (IFN)-γ, Fas and tumour necrosis factor (TNF)-related apoptosis-inducing ligand (TRAIL) [Tang et al. 2004; Cohen et al. 1999; Inbal et al. 1997]. Highlighting the potential role of DAPK in chemoresistance, it has been shown that TRAIL-resistant lung cancer cell lines can be resensitized by using DNMT inhibitors leading to demethylation and re-expression of DAPK [Tang et al. 2004]. Following onto this, in gastric cancer patients the promoter methylation of DAPK was also shown to correlate with shorter progression-free survival and lower response rates to chemotherapeutic treatment with 5-fluorouracil [Kato et al. 2008]. Consequently, the methylation of members of the apoptotic and anti-apoptotic machinery such as DAPK have the potential to influence apoptosis and hence chemosensitivity.

Epigenetic drug resistance drivers

It has been proposed that drug resistance is a polygenic phenotype caused by the cumulative effect of multiple drug resistance or sensitivity associated genes, rather than phenotypic effects on drug sensitivity of a few loci [Glasspool et al. 2006]. Indeed large numbers of gene expression changes are observed in tumours and during acquired drug resistance [Konstantinopoulos et al. 2008], as well as extensive changes in DNA methylation, which then may lead to the drug-resistant phenotype. One of the current challenges is to identify the key changes which drive a cell towards a drug-resistant phenotype among the vast majority of epigenetic changes caused, for instance, by global demethylation or localized hypermethylation of CpG islands [Ehrlich, 2006]. Many of these global changes could have occurred by chance either as part of a methylator phenotype or simply as random silencing events [Issa, 2004]. In other words: how do we discriminate a ‘driver’ which, in analogy to tumourigenesis, has a functional effect that provides the cell with a selective advantage from a ‘passenger’ that has no functional impact but was incidentally acquired [Greenman et al. 2007]?

For driver mutations of cancer it has been estimated that only 8–16% of all reported mutations within protein kinase genes represent true drivers [Carter et al. 2009; Greenman et al. 2007]. Moreover, a substantial number of these changes are only present at a low frequency [Wood et al. 2007]. By analogy, it might be only a very specific subset of epigenetic changes transforming a sensitive tumour cell into a resistant one. Few large-scale studies have addressed the issue of acquired drug resistance in ovarian cancer on an epigenome-wide basis. Dai et al. [2008] analysed methylation changes associated with acquired cisplatin resistance in isogenic ovarian cancer cell lines and showed that hypermethylation of CpG islands is prevalent during the acquisition of drug resistance. To take this further, Li and colleagues [2009] combined methylation with expression data and also identified extensive hypermethylation associated with epigenetic repression of loci implicated in cell adhesion as well as hypomethylation associated with activation of genes involved in the PI3K/AKT, transforming growth factor (TGF)-beta and cell cycle progression pathways. However, to pinpoint the epigenetic changes driving chemoresistance more precisely, it will be necessary to systematically investigate combined methylation, gene expression and resensitization in order to identify key drivers of the resistance phenotype. As an example, MLH1 epigenetic silencing is selected for during platinum treatment of ovarian tumour cell lines and re-expression of MLH1 either by demethylation or gene re-introduction resensitizes ovarian tumour cells to subsequent chemotherapeutic treatment [Plumb et al. 2000] and might, therefore, represent one of the key genes driving chemoresistance in the cell line models examined. However, this gives no indication of how many other genes may be key drivers of resistance, either in this model or in tumours in general. The actual proportion of epigenetically altered genes driving chemoresistance is still elusive. For future studies, systematic approaches will be required confirming a substantial effect of these methylation changes in terms of their ability to reverse drug resistance. If a small subset of genes drives drug resistance, this would be a valuable set of biomarkers for stratifying patients who may benefit from epigenetic approaches to resensitize tumours to chemotherapy.

Epigenetic modulators of drug resistance

It is widely recognized that a wide variety of epigenetic changes are prevalent in cancer [Jones and Baylin, 2007]. Importantly, unlike genetic mutations, epigenetic alterations are reversible and, therefore, the executing enzymes provide attractive drug targets for new therapies. Growing understanding of the mechanisms and enzymes governing epigenetic regulation has resulted in a variety of new drugs and possible drug targets for cancer treatment. These epigenetic therapies have the potential to re-activate or silence aberrantly regulated genes, thereby reversing many aspects of cancer phenotypes, including drug resistance. Examples of epigenetic therapies undergoing preclinical and clinical evaluation or registered for use in certain cancers are shown in Table 1.

Table 1.

Classes and names of DNA methylation inhibitors and histone deacetylase inhibitors, their targets and clinical status.

Type of epigenetic therapy	Class of compound	Compound	Target	Development stage
DNA methylation inhibitor	Nucleoside analogue	5′-azacytidine (Vidaza)	DNMTs	Approved for MDS, CMML and AML
		5-aza-2′-deoxycytidine (Decitabine/ Dacogen)	DNMTs	Approved for MDS
		Zebularine	DNMTs	Preclinical
	Non-nucleoside analogue	RG 108	DNMT1	Preclinical
		Procaine	DNMT1	Preclinical
		Procainamide	DNMT1	Preclinical
		Hydralazine	DNMT1	Phase II
	Antisense	MG98	DNMT1	Phase II
Histone deacetylase inhibitor	Hydroxamate	Suberoylanilide hydroxamic acid (SAHA, Vorinostat)	Class I, II	Approved for advanced CTCL
		PXD101	Class I, II	Phase II
		LAQ824, LBH589	Class I, II	Phase I
		Trichostatin A	Class I, II	Preclinical
		Oxamflatin, Scriptaid, SBHA	N/A	Preclinical
		Pyroxamide	Class I, unknown effect on class II	Phase I
		SK-7041, SK-7068	HDAC1, 2	Preclinical
		Tubacin	HDAC6	Preclinical
	Aliphatic acid	Valproic acid (VPA)	Class I, II	Phase II, III
		Phenylbutyrate	Class I, II	Phase I, II
		Savicol	N/A	Phase I, II
		AN-9 (prodrug)	N/A	Phase I, II
		Baceca	Class I	Phase I, II
	Benzamide	MS-275 (SNDX-275)	HDAC1, 2, 3 and slightly 8	Phase I, II
		MGCD0103	HDAC1, 2, 3, 11	Phase I, II
	Cyclic tetrapeptide	Depsipeptide (FK228)	Class I	Phase I, II
		Trapoxin A	Class I, II	Preclinical
		Apicidin	HDAC1, 3	Preclinical
		CHAPs	Class I	Preclinical

AML, acute myeloid leukaemia; CHAP, cyclic hydroxamic-acid-containing peptide; CMML, chronic myelomonocytic leukaemia; CTCL, cutaneous T-cell lymphoma; DNMT, DNA methyltransferase; DNMTi, DNMT inhibitor; HDAC, histone deacetylase; HDACi, HDAC inhibitor; MDS, myelodysplastic syndrome; N/A, not available; SBHA, suberic bishydroxamic acid. Adapted from Bolden et al. [2006], Yoo and Jones [2006] and Xu et al. [2007].

DNA methylation inhibitors

So far, DNA methylation inhibitors (DNMTi) represent the most widely studied class of epigenetic therapeutic agents. The US Food and Drug Administration (FDA) approved 5-azacytidine (Vidaza) and Decitabine (5-aza-2′-deoxycytidine, Dacogen), cytidine ribose and deoxyribose nucleoside analogues, respectively [Issa, 2007]. Both drugs exert their effect by becoming incorporated into DNA, and inhibiting the DNA methyltransferase (DNMT) by forming covalent adducts with the enzyme, leading to its sequestration and cellular depletion. As a result of reduced DNMT activity during subsequent cell division DNA methylation is increasingly diminished. Consequently, cytosine methylation is reduced in newly replicated DNA, but not in the DNA of resting or non-dividing cells. In addition, Decitabine has been shown to induce depletion of DNMT1 through proteosomal degradation [Ghoshal et al. 2005]. In cell line models, these nucleoside DNA methylation inhibitors proved to effect methylation and to re-activate epigenetically silenced tumour suppressor genes [Mund et al. 2006]. Importantly, Decitabine induced re-expression of genes known to mediate drug response, for example MLH1, resulting in enhanced chemosensitivity to cytotoxic drugs [Plumb et al. 2000].

Vidaza and Decitabine have been clinically tested and have shown substantial therapeutic potential in haematological cancers such as leukaemias [Fenaux et al. 2009; Stewart et al. 2009; Issa et al. 2005; Issa et al. 2004]. A randomized phase III trial with myelodysplastic syndrome (MDS) patients supported the potency of 5-azacytidine to improve overall survival time of patients when compared with other therapies [Fenaux et al. 2009]. However, the effectiveness of DNMTi in solid tumours has been limited so far [Graham et al. 2009]. Several reasons may account for this, including their short half-life in plasma combined with the relatively low proliferation rate of solid tumours cells, limiting the amount of aza-nucleosides becoming incorporated into DNA. To overcome this problem, treatments with prolonged exposures to demethylating agents at lower doses may be necessary, in order to reverse methylation and restore gene activity rather than triggering cytotoxicity as seen with higher doses of DNMTi [Issa et al. 2004].

Furthermore, recent clinical trials have aimed to optimize currently available treatments by combining those with DNMTis to enhance susceptibility to conventional chemotherapeutic agents. For example, Decitabine has been studied as a modulator of resistance in combination with carboplatin in solid tumours in a phase I clinical trial [Appleton et al. 2007]. Encouragingly, demethylation rates were reported in peripheral blood cells as indicated by MAGE1A promoter methylation levels. Following on from the initial success, a randomized phase II trial was started in patients with recurrent ovarian cancer that had progressed within 6–12 months following the first round of platinum therapy [Glasspool et al. 2009]. However, the study was closed after a planned interim analysis due to poor feasibility and lack of efficacy of the combination. One possibility for the lack of efficacy seen in the phase II trial may be due to reduced dose intensity of the carboplatin and/or due to the additional myelosuppression associated with the Decitabine treatment. In order for these agents to progress within the solid tumour setting, it will be important to develop agents that are less myelosuppressive in combination with an altered schedule. For example, studies in haematological malignancies have used a 1 h infusion of Decitabine over 5–10 days. Regarding ovarian cancer, studies are currently underway combining platinum chemotherapy with similar schedules [Matei and Nephew, 2010]. Another feasible explanation may be that the demethylation of certain genes leads to an adverse effect counteracting the carboplatin sensitizing effect of demethylation. Again highlighting the importance of patient selection based on their tumour methylation profile for treatments and future studies.

More recently, much attention has been focused on developing small molecule inhibitors of DNMTs. RG 108 is one example of such a nonnucleoside compound which inhibits DNA methylation by directly blocking the active site of the enzyme [Brueckner et al. 2005]. RG 108 has been shown to restore the activity of epigenetically silenced tumour suppressor genes such as p16^Ink4 in human colon cancer cell lines, notably without the drug-induced toxicity usually observed with traditional DNMTi [Stresemann et al. 2006; Brueckner et al. 2005]. However, further improvement of the drug is highly desirable in terms of its cell uptake and efficacy [Brueckner et al. 2007]. The potential of non-nucleoside inhibitors to more specifically inhibit DNMTs, without causing non-specific DNA damage associated with nucleoside DNMT inhibitors, suggests that this class of compounds may have potential to be less-toxic therapies. However, clinical studies are needed to reveal more about their clinical applicability.

Histone modulators

Histone deacetylase inhibitors

The second intensively studied class of drugs targeting epigenetic silencing mechanisms are the histone deacetylase inhibitors (HDACis). Histone deacetylases (HDACs) can regulate the chromatin conformation through the removal of acetyl groups from the lysine residue of histone tails. Inhibiting these enzymes promotes accumulation of the acetylated form of histone proteins, ultimately leading to less-condensed packaging of genes in chromatin which may lead to the re-expression of silenced tumour suppressor genes. However, HDACis also target various other proteins, including key molecules regulating tumour cell growth, which may be responsible for the observed induction of different phenotypes in transformed cells such as proliferation arrest, differentiation and apoptosis [Egger et al. 2004]. Although the actual mechanism of growth inhibition might not necessarily be due to acetylation of histones per se, the fact that normal cells are usually not affected by HDACi-induced cell death at low concentrations supports their suitability as specific anticancer agents [Warrener et al. 2003]. Currently, a variety of up to 15 different HDACis are under investigation in phase I–III clinical trials [Marks and Xu, 2009].

The only HDACi approved by the FDA is Vorinostat (SAHA), a hydroxide acid derivative, registered for treatment of cutaneous T-cell lymphoma (since 2006). Vorinostat has shown proven efficacy in the treatment of haematologic malignancies [Garcia-Manero et al. 2008; Duvic et al. 2007; Olsen et al. 2007]. It has also been tested as a single agent in clinical phase I/II trials in various solid tumours including head and neck, breast and thyroid cancer. However, response rates suggested that the activity of this HDACi is low in solid tumour types [Batty et al. 2009]. Recently, Vorinostat has been shown to enhance the efficacy of the cytotoxic agents carboplatin and paclitaxel in patients with advanced non-small cell lung cancer in the setting of a clinical phase II trial [Ramalingam et al. 2010]. However, due to increased toxicity optimization of the schedule seems to be required for this combination. The synergy of HDACi and conventional chemotherapeutic agents might be a promising route for the treatment of solid tumours.

Despite the initial encouraging results, monotherapy with DNMTis is limited due to the observed toxicity and the eventual re-methylation of genes [Plumb et al. 2004]. Now, there is growing interest in combined epigenetic therapies involving DNMTis and HDACis. For example, the DNMTi treatment can act synergistically with an HDACi in restoring gene expression [Cameron et al. 1999]. It has been proposed that DNA hypermethylation can lead to compact nucleosomes resistant to acetylation, thereby dominating silencing. A sequential administration of DNMTi prior to the HDACi administration appears to be important for sufficient efficacy in the treatment of solid tumours. Preclinical studies and early clinical trials are exploring the combination of epigenetic remodelling agents [Griffiths and Gore, 2008]. For example, the sequential treatment with Decitabine and the hydroxamate HDACi, PXD101 (Belinostat), has been tested in a cisplatin-resistant human ovarian tumour xenograft model [Steele et al. 2009; Plumb et al. 2004]. Combined treatment resulted in a marked increase in expression of epigenetically silenced MLH1 and MAGE1A compared with treatment with Decitabine alone supporting the idea that combinatorial epigenetic therapy might improve sensitivity to chemotherapeutic agents. A recent clinical phase I/II trial combining the short fatty acid Valproic acid (VPA) (HDACi) with Decitabine was done in patients with leukaemia [Garcia-Manero et al. 2006]. Here, it emerged that this combination led to a transient reversal of epigenetic marks such as methylation of the p15 promoter and deacetylation of histone H3 and H4. However, neither the level of DNA demethylation nor the level of acetylation could be correlated to clinical response.

Histone methyltransferase inhibitors

The discovery of histone methyltransferases (HKMT) has opened up another promising avenue to target aberrant epigenetic mechanisms [Lachner et al. 2003]. HKMTs are chromatin-modifying enzymes which establish methyl marks on the lysine of histone tail proteins. Depending on the position of the methyl mark the histone code can be read from binding proteins which then lead to transcriptional activation or repression of genes. Deregulation of HKMT activity has been linked to tumour development, especially high levels of the H3K27 methyltransferase EZH2 (polycomb group protein, enhancer of zeste homologue) have recently been observed in a variety of tumours including prostate, breast and melanoma and are associated with poor prognosis [Yu et al. 2007; Bachmann et al. 2006; Varambally et al. 2002]. Subsequent efforts to find appropriate inhibitors for EZH2 led to the discovery of 3-deazaneplanocin A (DZNep). DZNep is an S-adenosylhomocysteine hydrolase inhibitor capable of depleting EZH2 and other components of the Polycomb Repressive Complex 2 (PRC2) from cells in vitro and induces apoptosis in breast cancer cells [Tan et al. 2007]. However, subsequent studies showed that DZNep is not specific to EZH2 but is generally inhibiting methyltransferase activity thereby affecting global histone methylation levels [Miranda et al. 2009]. This is most likely due to its indirect mechanism of inhibition which blocks S-adenosyl methionine (SAM)-dependent methyltransferases through byproduct inhibition.

Nevertheless, DZNep is being explored to improve the epigenetic effect of HDACi at an early preclinical stage. Results seem encouraging, for example, treatment of an acute myeloid leukaemia cell line with combined DZNep and the HDACi panobinostat highly increased rates of apoptotic cell death in comparison with treatment with either drug alone [Fiskus et al. 2009]. However, there is a need for the development of novel, more selective inhibitors of HKMT which is supported by studies performed in glioblastoma multiform, an aggressive form of malignant glioma. Here, it was demonstrated that pharmacologic disruption of EZH2 by DZNep strongly impaired glioblastoma multiforme cancer stem cell self-renewal in vitro and tumour-initiating capacity in vivo due to repression of c-myc, a proto-oncogene mediating cell growth [Suva et al. 2009]. Therefore, the development of novel inhibitors with greater biological specificity is highly sought and may provide valuable tools for more targeted therapies.

Also in light of combined epigenetic therapy the current efforts to develop compounds targeting more distinct classes of epigenetic enzymes should greatly enhance their efficacy. The pleiotropic effects observed with anticancer agents already in the clinic can often be a drawback and limit their value in therapy. More selectively acting compounds should aid in overcoming toxic effects currently seen. Combination studies indicate that sequential administration of drugs against different classes of enzymes might also increase efficacy of treatment. Although still at an early stage, future inhibitors of HKMT may represent a new class of compounds which could offer more selectivity compared with the currently available options.

Conclusions

It is becoming increasingly clear that epigenetic changes occurring during the acquisition of drug resistance are substantial and complex and might even outnumber genetic alterations. There are clear examples of the potential of epigenetic alterations at specific loci affecting drug resistance. However, in order to successfully target abnormal epigenetic changes in the clinic, it will be necessary to comprehensively identify the key events driving this process. Consequently, a therapy specifically targeting epigenetically deregulated genes could help in overcoming resistance. Longitudinal studies monitoring tumours acquiring resistance during the course of chemotherapy could provide an opportunity to map genome-wide epigenetic changes which may allow the correlation to acquisition of drug resistance. However, the magnitude of changes and heterogeneity among tumours will make such analyses challenging. Here, it will be crucial to perform epigenetic profiling of an appropriately sized panel of tumours pre- and postchemotherapy in order to achieve sufficient power for statistical analysis. In addition, those large-scale studies will also greatly expand the repertoire of available biomarkers which are vital for stratification of patients and for monitoring efficacy of existing epigenetic therapies.

Current efforts to isolate and characterize so-called cancer sustaining (stem) cells may yield new insights into genetic and epigenetic mechanisms governing chemoresistance [Ferrandina et al. 2008; Zhang et al. 2008; Agarwal and Kaye, 2003]. One of the features of such tumour sustaining cells is their inherent resistance to a number of chemotherapeutics, making them a suitable model for drug-resistance studies. Cancer sustaining cells have stem-cell-like properties in the way that they have the capacity to recapitulate the original tumour and promote recurrence.

Although successfully used in haematological malignancies DNMTi and HDACi work on a multitude of targets. The concern is that the genome-wide re-expression of aberrantly and normally regulated genes could lead to conflicting effects concerning response to chemotherapy. Therefore, the development of more targeted (epigenetic) therapies might be a prerequisite of successfully preventing drug resistance. The idea of reversing the malignant epigenetic marks of a small subset of genes driving drug resistance in order to sensitize tumours to chemotherapy will require novel strategies based on our increased understanding of the complexities of epigenetic regulation.

Funding

This research received funding from Cancer Research UK.

Conflict of interest statement

The authors declare they have no conflicts of interest.