Pancreatic ductal adenocarcinoma (PDAC) has an incidence of 495,773 cases per year worldwide and accounts for 466,003 deaths, making it the 14th most common cancer and the 7th most likely cause of a cancer death as of 2020 (Ushio et al., 2021). PDAC incidence specifically within the USA is expected to be 62,210 cases in 2022 (Siegel, Miller, Fuchs, & Jemal, 2022). The median survival time of an individual with PDAC is 10–12 months with a 5-year survival rate of only 8–11% (Orth et al., 2019, Principe et al., 2021). PDAC long term survival is often only achieved in cases where surgical resection is possible which occurs in less than 20% of cases (Deshwar et al., 2018, Gillen et al., 2010, Gress et al., 1999, Singhi et al., 2019, Strobel et al., 2022). The efficacy of palliative systemic chemotherapy has increased over the last ten years, but lags behind many other cancers. Both the complex genetic landscape, dense stroma, absence of targetable mutations and limited response to immunotherapy have contributed to the inability to improve efficacy. Epigenetics may offer a different treatment strategy that can impact several of these tumor components.

Epigenetics is the study of “changes in gene function that are mitotically and/or meiotically heritable and does not entail a change in DNA sequence” (Ct & Morris, 2001). Some of the best understood and targetable epigenetic changes occur through mechanisms that involve histone acetylation/deacetylation and DNA and histone methylation through covalent modifications of DNA bases (Dupont, Armant, & Brenner, 2009). Within the process of tumorigenesis, the epigenome undergoes alterations including loss of DNA methylation in the genome, regional hypermethylation, global changes in histone modification and deregulation of non-coding RNA engagement. These epigenetic changes allow for altered cell function, drug resistance and changes to the tumor microenvironment (TME) or the immune system (Cheng et al., 2019, Esteller, 2008, Fernandez et al., 2012, Fraga et al., 2005, Kanwal and Gupta, 2012, Liz and Esteller, 2016, Lu et al., 2020, Roberti et al., 2019) and reprogram cancer cells to a more treatment responsive state.

In Table 1 we summarize the different classes of epigenetic drugs, their targets and broad impact on cancer biology, specifically DNA Methyltransferase inhibitors (DNMTi), Histone Deacetylase inhibitors (HDACi), Bromodomain and Extra-terminal inhibitors (BETi) and Enhancer of Zeste Homolog 2 inhibitors (EZH2i). These epigenetic therapies are often utilized to activate silenced genes through hypomethylation, increase histone acetylation to result in a more favorable open chromatin conformation for drug targeting (Kang et al., 2020), regulation of MYC proteins and cell proliferation and improved efficacy of immunotherapy (Giri and Aittokallio, 2019, Hu et al., 2021, Li et al., 2021a, Pérez-Salvia and Esteller, 2017, Shorstova et al., 2021, Suraweera et al., 2018).

In addition to summarizing the broad effects of epigenetic drugs, here we also present specific strategies in PDAC that have resulted in subtype switching, affecting the tumor microenvironment and enabling renewed immune response for improved patient outcomes. Table 2 provides a summary of the current clinical trials using different classes of epigenetic agents. As epigenetic changes are fundamental to pancreatic carcinogenesis, we also present current data and advances in the use of epigenetic changes as biomarkers (Adams et al., 2019, Lomberk et al., 2019, Roalsø et al., 2022).

Several recent studies have employed progressively more advanced next-generation sequencing technologies on large cohorts of PDAC tumors (Dennaoui et al., 2021, Grady, 2021, Martinez-Useros et al., 2021, Søreide et al., 2020, Yu et al., 2021) to improve the histopathological classification of PDAC with the ultimate goal of predicting treatment response (Alexandrov et al., 2015, Bettaieb et al., 2017, Politi and Herbst, 2015). As a result of these studies, not only signature mutations in KRAS, TP53, CDKN2A, and SMAD4, have been confirmed, but the identification of several new genes either mutated or transcriptionally deregulated has been possible. These studies further highlighted the molecular heterogeneity of PDAC, which, in turn, led to the identification of different PDAC subtypes with distinct molecular and phenotypic characteristics that directly impact prognosis and response to therapies (Dennaoui et al., 2021, Martinez-Useros et al., 2021, Roalsø et al., 2021, Yu et al., 2021).

The first attempts to stratify PDAC tumors were based on genetic mutations (Jones et al., 2008, Waddell et al., 2015, Witkiewicz et al., 2015), however, despite the high number of mutation and rearrangement identified, their clinical impact has been modest. Other genomic parameters like the chromosomal instability index and copy number aberrations were not associated with any specific PDAC subtype (Nicolle et al., 2017). Consequently, efforts have been directed towards a molecular-based stratification. The first proposed PDAC molecular subtyping was developed in 2011 based on the results obtained through array-based mRNA expression analysis of resected PDAC by epithelium microdissection with stroma exclusion. It included three subtypes: classical, exocrine-like and quasi-mesenchymal (Collisson et al., 2011). A following research analyzed samples from PDAC primary tumors and metastasis using hybridization arrays and RNA-seq. By digitally separating tumor, stromal and normal pancreatic gene expression the authors identified two tumor subtypes (classical and basal-like) and two stromal subtypes (normal and activated) (Moffitt et al., 2015). Shortly after Bailey et al. performed a comprehensive integrated genomic analysis of 456 PDACs and their histopathological variants and described four subtypes: squamous, pancreatic progenitor, immunogenic, and aberrantly differentiated endocrine exocrine (ADEX) (Bailey et al., 2016). A subsequent transcriptome and epigenome based classification identified basal-like (also called squamous) and classical (Dennaoui et al., 2021, Grady, 2021, Martinez-Useros et al., 2021, Roalsø et al., 2021, Yu et al., 2021) subtypes. The classical subtype shows higher differentiation grade, increased sensitivity to chemotherapy and better prognosis (Dennaoui et al., 2021, Roalsø et al., 2021, Yu et al., 2021). On the other hand, basal-like subtype tumors show higher tumor grade, chemoresistance and worse prognosis (Dennaoui et al., 2021, Roalsø et al., 2021, Yu et al., 2021).

From a molecular perspective, the two subtypes are characterized by distinct gene signatures and epigenetic profiles (Agostini et al., 2022, Søreide et al., 2020, Yu et al., 2021). Specifically, the classical one shows a gene signature coherent with epithelial differentiation, while the basal-like expression profile is more mesenchymal (Dennaoui et al., 2021, Grady, 2021, Martinez-Useros et al., 2021, Roalsø et al., 2021, Søreide et al., 2020, Yu et al., 2021). However, more than 90% of the analyzed tumors contained cells co-expressing classical and basal markers (Williams et al., 2023) suggesting that, these cells, may constitute an intermediate state between subtypes and that most exist on a basal-classical polarization continuum.

It has been demonstrated that both basal and classical PDAC subtypes could be classified by specific alterations in their DNA methylation patterns which correlate with transcriptomic alterations (Bailey et al., 2016, Collisson et al., 2011, Kumar Mishra and Guda, 2017, Moffitt et al., 2015, Nicolle et al., 2017, Nones et al., 2014, Zhao et al., 2021). More specifically, from the characterization of the epigenetic landscape and the resulting gene expression, it has been established that tumor cells belonging to the basal-like subtype show higher expression of genes related to cell cycle and glycolysis pathways, as well as hyperactivation of the WNT pathway, while pathways upregulated in the classical phenotype are also active in normal pancreatic cells or other types of gastrointestinal cells (Aguilera and Dawson, 2021, Torres and Grippo, 2018). Furthermore, they display differential activity of different Super-enhancers (SEs) and their upstream regulatory elements (Andricovich et al., 2018). SEs are large clusters of transcriptional enhancers that control cell identity by influencing target gene expression (Allis & Jenuwein, 2016).

Subtype-specific SEs and transcription programs are regulated by transcription factors (TFs) differentially expressed among subtypes. Specifically, GATA6, PDX1, HNF1A and HNF4A are the main TFs found in classical tumors, while MET, MYC, GLI2 and the ΔN isoform of the transcription factor TP63 (ΔNp63) are specific for the basal-like subtype (Agostini et al., 2022, Andricovich et al., 2018, Bates, 2020, McDonald et al., 2017, Waddell et al., 2015). There are several examples where epigenetic mechanisms impact subtype-specific gene expression: (1) ΔNp63 increases H3K27ac levels at the enhancers of basal-like lineage genes leading to increased expression of genes that promote tumor growth and metastasis, such as KRT5/6, TRIM29, and PTHLH (Somerville et al., 2018); (2) histone H3K27me2/3-specific lysine demethylase 6A (KDM6A) is frequently mutated in the basal-like subtype (Andricovich et al., 2018) and loss of KDM6A in PDAC can directly induce the basal-like subtype by altering the chromatin distribution (Hyun et al., 2017, Lavery et al., 2020); (3) FOXA1 promotes metastasis by increasing H3K27ac (Roe et al., 2017) at specific enhancer regions and, therefore, expressing foregut developmental genes that promote invasion and anchorage-independent cell growth.

Perhaps the most important transcription factor involved in the regulation of PDAC subtype identity is the GATA binding protein 6 (GATA6) (Collisson et al., 2011, Kloesch et al., 2022, Lomberk et al., 2018). It has been shown that its depletion favors the basal-like subtype by inducing ΔNp63 (Kloesch et al., 2022), thus making GATA6 both a good biomarker for discriminating between classical (GATA6high) and basal-like (GATA6low) (Aung et al., 2018), and an obvious target for therapeutic strategies aimed at inducing or stabilizing it. GATA6 overexpression through different approaches, such as inhibition of basal-like subtype-specific superenhancer regulator MET or loss of the GLI Family Zinc Finger 2 (GLI2), transcription factor leads to a shift from basal-like gene signature to a classical one (Adams et al., 2019). Inhibition of the histone methyltransferase enhancer of zeste homolog 2 (EZH2) can also induce GATA6 expression (Patil et al., 2020), which was perhaps the first epigenetic approach towards class switch (Fig. 1). EZH2 binds to the transcription start site of GATA6 repressing its transcription, therefore inhibition of EZH2 was found to be sufficient to rescue GATA6 expression. Currently the EZH2 inhibitor tazemetostat is being evaluated in a clinical trial that encompass multiple kind of solid tumors including PDAC (NCT04705818).

As we previously mentioned, loss of KDM6A activity also favors basal-like subtype through aberrant SEs activation (Andricovich et al., 2018). KDM6A-deficient PDAC tumors are selectively sensitive to inhibitors targeting bromodomain and extra-terminal motif (BET) proteins, which reversed squamous differentiation and reduced tumor growth both in vivo and in vitro (Andricovich et al., 2018) (Fig. 1). The combination BET-proteins/HDAC inhibitors showed promising activity in preclinical PDAC models (He et al., 2020, Mazur et al., 2015).

Overall, the plasticity and heterogeneity of PDAC subtypes and their regulation by epigenetic mechanisms make several of the epigenetic treatment strategies important adjuncts to PDAC treatment. A further understanding of the therapeutic benefit of subtype switching will be needed to understand the role of these agents.