Genomic Landscape of HCC

Abstract

Introduction

Hepatocellular carcinoma (HCC) is a leading cause of cancer related mortality in the world and it has limited treatment options. Understanding the molecular drivers of HCC is important to develop novel biomarkers and therapeutics.

Purpose of Review

HCC arises in a complex background of chronic hepatitis, fibrosis and liver regeneration which lead to genomic changes. Here, we summarize studies that have expanded our understanding of the molecular landscape of HCC.

Recent Findings

Recent technological advances in next generation sequencing (NGS) have elucidated specific genetic and molecular programs involved in hepatocarcinogenesis. We summarize the major somatic mutations and epigenetic changes have been identified in NGS-based studies. We also describe promising molecular therapies and immunotherapies which target specific genetic and epigenetic molecular events.

Summary

The genomic landscape of HCC is incredibly complex and heterogeneous. Promising new developments are helping us decipher the molecular drivers of HCC and leading to new therapies.

INTRODUCTION

The global incidence of hepatocellular carcinoma (HCC), the most common primary liver malignancy and the 6th most common cancer worldwide, is expected to significantly increase over the next 10 years [1,2]. Unfortunately, HCC survival still remains dismal, with 5-year survival rates of 32.6%, 10.8%, and 2.4% for localized, regional, and distant stages of disease respectively [3]. Hepatitis B (HBV) and hepatitis C (HCV) are the most common risk factors for HCC, and the incidence of HCC has historically mirrored the incidence of these infectious diseases [4]. With the advent of the HBV vaccine and HCV antiviral therapy, there is hope that the burden of hepatitis-related HCCs will decrease. However, viral hepatitis is still expected to drive increased incidences of HCC over the next 10 years [2,5]. Furthermore, alcoholic liver disease (ALD), obesity, and non-alcoholic fatty liver disease (NAFLD) remain important risk factors for HCC, and these etiologies are actually increasing in incidence [6,7].

Even though risk factors for HCC vary from region to region, the mechanisms of hepatocellular carcinogenesis mostly converge on the processes of chronic liver inflammation and regeneration. Chronic liver injury secondary to either virus-induced inflammation, alcohol-induced hepatocellular damage, or lipotoxicity-induced oxidative stress lead to a vicious cycle of regeneration and fibrosis that increases the risk of genomic instability and hepatocarcinogenesis [8,9]. These pathogenic mechanisms underscore why 80–90% of HCC arise in a cirrhotic liver [10]. However, HCC can also arise in the non-cirrhotic liver, especially in patients with HBV or NAFLD [11,12]. Thus, the genomic landscape in which hepatocarcinogenesis occurs is incredibly complicated.

Currently, the only definitive therapeutic cure for HCC is liver transplantation (LT), and even LT is associated with recurrence rates of 10–15% [13]. ●For unresectable tumors, few therapies exist. The oral multi-kinase inhibitor sorafenib has been the therapeutic workhorse for unresectable HCCs ever since it was approved by the Food and Drug Administration (FDA) in 2007. However, its therapeutic efficacy has been greatly limited by rapid drug resistance and toxicities [14,15]. ●Despite the recent approval of new first line therapies like atezolizumab/bevacizumab or Lenvatinib and second line therapies like regorafenib, nivolumab, and cabozantinib, there still remains a pressing need for effective therapeutics that can significantly improve long term survival [16–18].

Genomic Landscape of HCC

The dearth of therapeutic options for HCC continues to propel research into the mechanisms of hepatocarcinogenesis. Over the last five years, significant progress has been made in the identification of somatic mutations, copy number variations (CNVs), and epigenetic modifications that drive hepatocarcinogenesis and contribute to disease outcomes (Table 1). Improvements in genome wide screening and high-throughput genomics have led to the identification of new gene signatures and proteomic targets that can help to diagnose and prognosticate HCC. Recent developments in single-cell RNA-sequencing (scRNA-Seq) are opening a window into the pathophysiology of tumor heterogeneity. This review summarizes the genomic landscape of HCC (Figure 1), and identifies studies that have recently expanded our understanding of hepatocarcinogenesis in a meaningful way.

Table 1:

Overview of Major Genomic Landscape and Clinicopathologic Correlations in HCC

Type of Genomic Change	Class	Main Gene Target	References	Clinicopathologic Associations	Therapeutic Relevance
Somatic Mutations	Tumor Suppressors	TP53 (p63, p73, p14ARF, mdm2, mdm4, ATM)	Teramoto et al, 1994 Chan et al, 2004 Villanueva et al, 2011 Chittmittrapap et al 2013 Zhan et al, 2013 Zhan et al, 2014 Torrecilla et al, 2017 Long et al, 2019 Li et al, 2020	Clinical: HBV-HCC (32) Histopathology: increased stem cell markers (30), high Edmonson gade (29, 32), decreased in areas of cholestasis and tumor infiltrating lymphocytes (32), increased in areas of necrosis (32) Survival: high recurrence rates, poor OS (30–33) and DFS (31,32), downregulated immune responses (39)	Resistance: doxirubicin (34–35)/paclitaxel (34) (in vitro) Therapeutics: p53 MVA vaccine + pembrolizumab (Phase I, 40), recombinant adenovirus p53 (Phase II, 41, 43, 44)
		ARID1A ARID1B	Schulze et al, 2015 Nahon et al, 2017 Hu et al, 2018 Qu et al, 2019 Moore et al, 2019 Li et al, 2020 Peng et al, 2020	Clinical: ALD-HCC (45), HBV-HCC (45–47), tobacco use (45) Histopathology: hepatic steatosis (48–49), larger more aggressive tumors (51–52), increased angiogenesis (53), poor OS (51, 52), mismatch repair (52)	--
	Telomerases	TERT promoter	Nault et al, 2013 Nault et al, 2016 Pezzuto et al, 2016 Torrecilla et al, 2017 Yu et al 2017 Pezutto et al 2017 Kawai-Kitahata et al, 2017 Li et al, 2018 Cheng et al, 2019 Li et al, 2020	Clinical: Older age, African or European ancestry, HCV-HCC, HBV-HCC (61–63) Histopathology: low/high grade dyspastic nodules (56–57) Survival: shorter DFS (60, 64–65), late intrahepatic recurrence (60)	Therapeutics: INO-1400 or INO-1401 ± INO-9012 (Phase I, 66)
	Growth Factor Pathways	CTNNB1 (Wnt/β-catenin) (APC, AXIN1, AXIN2, PRC1, AKIP1, TXNDC12, DICER1)	Zulehner et al, 2010 Schulze et al, 2015 Torrecilla et al, 2017 Ataide et al, 2017 Li et al, 2020	Clinical: ALD-HCC (46) Histopathology: micro- and macrovascular invasion, increased pathologic grade, increased tumor size, and multifocal disease (73–74) Survival: HCC recurrence (73–74)	Therapeutics: DKK1 antibody (Phase I/II, 79)
	Signal Transduction	RPS6KA3 (MAP/ERK)	TCGA, 2017 Calderaro et al, 2017 Doycheva et al, 2019	Histopathology: poor differentiation, macrovascular invasion, high proliferation, and chromosomal instability, multifocal disease (80–82)	--
		STAT3 (JAK/STAT)	Kan et al, 2013	Histopathology: promotes proliferation, angiogenesis, metstasis, Warburg effect, inhibition of apoptosis	Therapeutics: STAT inhibitors: OPB-111077 (Phase I, 97), OPB-31121 (Phase I, 98), Napabucasin (Phase I, 99), and AZD9150 (Phase 1b, 100)
	Other	KEAP1	Zheng et al, 2019	--	Resistance: acquired resistance to sorafenib, lenvatinib, and regorafenib (83)
		HNF1A	Reznik et al, 2004 Rebouissou et al, 2007 Zeng et al, 2011 Hetchman et al, 2019	Clinical: negative viral status, female sex, no cirrhosis (84–86) Histopathology: reduced expression associated with proliferation, migration, and invasion in HCC cells (87) forced expresioon leads to differentiation into mature hepatocytes (87–88)	--
CNVs	Gene Amplification	MYC	Chan et al, 2004 Poon et al, 2006 Schlaeger et al, 2008 Pedica et al, 2013 Ao et al, 2016 Yu et al, 2017	Clinical: viral hepatitis, ALD (105–106) Histopathology: large undifferentiated tumors (107–110) Survival: poor DFS, metatastasis, HCC recurrence (107–110)
		E2F1/E2F3	Kent et al, 2017	Histopathology: advanced HCC (111)	--
		MMP9	Yu et al, 2020	Clinical: AFP level (112) Histopathology: tumor size, differentiation, invasion, stage (112)	--
		UBE2Q1	Zhang et al, 2018	Survival: Poor OS and DFS (114)	--
		EXT1	Dong et al, 2018	Clinical: AFP level (115) Survival: Poor DFS (115)	--
		WNK2	Zhou et al, 2019	Histopathology: increased tumor growth, macrophage infiltration (116) Survival: early tumor recurrence, metastasis (116)	--
		FGF19	Kaibori et al, 2016	Histopathophysiology: agressive, undifferentiated tumors (in vivo) (Sawey et al, 2012), increased proliferation/invasion (in vitro) (Miura et al, 2012) Survival: poor overall and disease free survival (Miura et al, 2012)	Resistance: complete response after sorafenib treatment
	Gene Deletions	CDKN2A	Kawai-Kitahata et al, 2016 Cancer Genome Atlas Research Network et al, 2017	Histopathology: Cell cycle arrest in G1/G2 Survival: Poor Prognosis, advanced aggressive cancer	--
		RB1	Kawai-Kitahata et al, 2016 Cancer Genome Atlas Research Network et al, 2017 Sun et al, 2019	Clinicals: HBV/HCV HCCs (113)	--
Epigenetic Modifications	DNA Methylation	APC	Um et al, 2011 Shen et al, 2013	--	Treatment: 1st-Line: azacytadine and decitabine, 2nd Line: guadecitabine and zebularine (Phase I/II, 126–127)
		GSTP1	Shen et al, 2013	Clinicals: HBV
		RASSF1a	Shen et al, 2013	Clinicals: Aflatoxin B1
		p16	Um et al, 2011	Clinicals: Aflatoxin B1
		COX2	Um et al, 2011	Histopathophysiology: Dysplastic nodules (123)
		CDH1	Wu et al, 2019	Survival: Poor OS(124)
	miRNAs	mIR21	Xu et al, 2013	Histopathology: HCC proliferation and migration (129–134)	--
		mIR221/222	Pineau et al, 2010		Treatment: Miravirsin (Phase II, 137–138)
		mIR224	Lin et al, 2016		--
		mIR-26	Yang et al, 2014	Histopathology: Suppression of HCC proliferation and angiogenesis (129–134)	--
		mIR-122	Coulouarn et al, 2009		--
		mIR-199	Guo et al, 2015		--
		mIR 200a	Chen et al, 2017	Histopathology: inhibit cell growth, migration and invasion (135, 136)	--
	lncRNAs	HULC	Li et al, 2016	Histopathology: proliferation, EMT, angiogenesis (140)	--
		HOTAIR	Fujisaka et al, 2017	Histopathology: no impact on cell proliferation, migration (141), actively recruited macrophages	Resistance: chemoresistance (140,142)
	HDAC	HDAC3	Lu et al, 2018 Ji et al, 2018	Histopathology: HCC proliferation (143–144) Survival: HCC recurrence, reduced DFS (143)	Treatment: Panabinostat, Belinostat, Resminostat, CUDC-101 (Phase I/II, 148–150)
		HDAC1/2	Ler et al, 2015 Yang et al, 2019	Survival: predict mortality (145, 146)	Resistance: regulation doxorubicin sensitivity (145)

AFP-alpha-fetoprotein; ALD-alcoholic liver disease; HBV-hepatitis B virus; HCC-hepatocellular carcinoma; HCV-hepatitis C virus; HDAC-histone deacetylases; OS-overall survival; DFS-disease-free survival, lncRNAs-long non-coding RNAs; miRNAs-microRNAs

Figure 1: Major Genetic Alterations in Human Hepatocellular carcinoma (HCC).

Somatic Mutations

The normal aging liver is thought to acquire 30–40 somatic mutations per year, either induced by genotoxic stress or random mutations arising from DNA replication [19,20]. Hepatic stem cells and differentiated hepatocytes have both been shown to acquire these mutations, with a recent study revealing that mature hepatocytes in the normal liver have twice the rate of somatic mutations as hepatic stem cells [21].● Still, compared to other tissues heavily dependent on stem-cell regeneration, the overall mutational burden and malignant potential of the normal adult liver remains low [19,20]. In the setting of chronic liver disease and inflammation, however, hepatocytes are susceptible to additional proliferation-induced mutagenesis by way of mitochondrial damage and oxidative/endoplasmic reticulum stress [22]. Another mechanism of mutagenesis is via genomic viral integrations as seen in HBV-HCCs, which typically have the highest rates of somatic mutation [23]. HCV, on the other hand, typically promotes HCC carcinogenesis through double strand breaks that result in missense mutations. Lastly, somatic mutations in NAFLD, ALD, and toxin-driven HCCs are typically caused by direct DNA damage through chronic inflammation and reactive oxygen species.

Clinically significant driver mutations in HCC have been shown to involve these major pathways: tumor suppressor genes (TP53, ARID 1/2, RB1, TSC1/2), telomerases (TERT/TERC), the WNT-beta catenin pathway (CTNNB1, AXINI, AXIN2), PI3K/Akt/MTOR pathways, MYC pathway, JAK/STAT pathways (JAK1, IL6R, IL6ST), oxidative stress pathways (KEAP1, NFE2L2), RAS/RAF/MAP kinases (RPS6KA3), and the MET pathway. We will now discuss the mutations in these major pathways in further detail.

Tumor Suppressor Genes

As the “guardian of the genome”, the TP53 tumor suppressor gene is responsible for the regulation of cellular processes like cell death and angiogenesis. It is the most frequently altered gene in human cancer, with the International Agency for Research on Cancer (IARC) reporting over 29,000 TP53 mutations in human cancers [24]. TP53 is altered or inactivated in 30–50% of all HCCs- 80% of aflatoxin B1-HCCs, 45% of HBV-HCCs, and 13% of HCV-HCCs [24–26]. Notable gain of function (GOF) mutations in HCC include TP53 V157F and TP53 R249S, the latter being associated with aflatoxin and hepatitis B exposure [24,27,28].

Clinically, mutations in the TP53 family (including p63 and p73), activator p14ARF and inhibitors MDM2 and MDM4, are associated with higher expression of stem-cell like markers, high Edmonson grade, high rates of recurrence, lower disease-free survival, and therapy resistance [29–32]. TP53 may also downregulate the immune response [33],●● making these HCCs a potential target for immunotherapy [34]. ● Therapies targeting TP53 are aimed at supplementing wild-type p53 or blocking its interaction with cytoplasmic partners. Palbociclib, an oral cyclin-dependent kinase 4/6 inhibitor, has been shown to inhibit p53 DNA-damage partner Ataxia telangiectasia mutated (ATM) and to increase radiosensitivity of HCC cell lines, with the potential implication that gain-of-function p53 expression can be suppressed [35]. ● Furthermore, recombinant adenovirus p53, when combined with transarterial chemoembolization (TACE), has been shown to increase overall survival (OS) and disease-free survival (DFS) in patients with HCC [36].

The tumor suppressors ARID1A and ARID1B are components of SWI/SNF complexes that allow DNA repair machinery to access chromatin. Loss-of-function mutations in these genes have historically been associated with alcoholic liver disease and HBV infection [37,38], and recent studies suggest ARID1 may contribute to the development of hepatic steatosis [39,40]. ●ARID mutations typically occur in the later stages of HCC development, and result in larger, more aggressive tumors [41–43] with high tumor mutational burdens and increased angiogenesis, both of which potentially make them susceptible to immunotherapy and anti-angiogenic therapies [44,45]. ●●

Telomerases

Telomeres are repetitive nucleotide sequences that provide a docking location for the DNA polymerase complex during replication, and also protect chromosomes from deterioration or fusion. When telomeres shorten beyond a critical length after successive rounds of DNA replication, the telomerase complex, composed of a core catalytic telomerase reverse transcriptase (TERT) and RNA template telomerase RNA component (TERC) is activated to lengthen telomeres and restore the liver’s regenerative capacity. The TERT promoter mutation is one of the most common genetic alterations in HCC, with an overall frequency of 30–60% [46]. Its presence in low-grade and high dysplastic nodules reveal its role in the early stages of hepatocarcinogenesis [47,48], which is to drive malignant transformation by selecting for those HCC precursors that escape apoptosis with indefinite telomerase activity. Ironically, loss of function of telomerase gene variants also predispose hepatocytes to malignant transformation by impairing hepatocyte response to chronic injury and accelerating cirrhosis [49].

Clinically, TERT promoter mutations are associated with shorter DFS and late intrahepatic recurrence after surgical resection [50]. They are also more frequent in patients with older age, African or European ancestry, and HCV-HCC [51–53]. Other clinically meaningful TERT gene alterations include HBV viral integrations at the TERT gene promoter locus and TERT gene amplifications [50], both of which are associated with decreased OS [54,55]. ● The therapeutic promises of TERT mutations are yet to be realized. There are currently no approved therapies targeting TERT mutations, although a Phase I clinical trial with an immunotherapeutic agent against hTERT in solid tumors is currently in progress [56]. TERT mutations in circulating DNA may also be a novel way of screening for patients at high risk patients for HCC [57,58]. ●

Wnt/β-catenin pathway

The Wnt/β-catenin pathway, which in normal tissues is critical for embryonic body axis patterning, cell migration and cell fate specification, is commonly exploited in hepatocarcinogenesis. CTNNB1, which encodes β-catenin, a multifunctional protein that links the intracellular actin cytoskeleton to adherens junctions and also serves as a key nuclear effector of canonical Wnt signaling [59], is mutated in 20–40% of all HCCs [60,61]. Missense mutations in CTNNB1 result in higher nuclear and cytoplasmic β-catenin expression in HCCs compared to normal liver, para-carcinoma tissue, and cirrhotic liver [62]. Nuclear expression in HCCs has been associated with more aggressive histopathologic features, such as micro- and macrovascular invasion, increased pathologic grade, increased tumor size, multifocal disease, and tumor recurrence [52,63]. Furthermore, mutations in any of the proteins responsible for the activation or destruction of β-catenin can lead to aberrant nuclear accumulation. For example, loss-of-function mutations in the APC, AXIN1 and AXIN2 genes result in the sustained activation of the Wnt pathway by disruption of the multiprotein destruction complex that tags β-catenin for degradation. Other recently identified members of the Wnt/β-catenin pathway upregulated in HCC include protein regulator of cytokinesis 1 (PRC1), AKIP1, and thioredoxin protein TXNDC12 [64–66].● Germline mutations in the microRNA processing gene DICER1 have also been associated with CTNNB1 mutations in familial HCC, although the mechanistic relationship remains unclear [67]. So far, Wnt pathway proteins and genes have not proved to be druggable targets. An antibody that currently targets DKK1, a protein regulator of the Wnt pathway, is currently in phase I clinical trials for HCCs [68]. ●●

Other Notable Mutations

Several other notable pathways have been implicated in hepatocarcinogenesis. Mutations in RPS6KA3, a MAP/ERK pathway kinase that was recently shown to be mutated in 4–10% of HCCs, were associated with poor differentiation, macrovascular invasion, high proliferation, and chromosomal instability [69–71].●● Genome-wide screening revealed mutations in KEAP1, a master regulator and ubiquitinator of antioxidant gene NFR2, to be the top cause of acquired resistance to sorafenib, lenvatinib, and regorafenib in HCC cell lines [72]. ●HNF1A is a liver-enriched transcription factor (TF) that regulates cellular homeostasis and metabolism. Inactivated or mutated HNF1A has been found in HCCs in patients with negative viral status, female sex, and no cirrhosis [73–75]. A specific point mutation (c.A1532 > T/p.Q511L) causes reduced expression, proliferation, migration, and invasion in HCC cells, while forced expression induces differentiation of these cells into mature hepatocytes [76]. Additionally, the combinatorial transduction of TFs NF4A, HNF1A and FOXA3 was shown to suppress cellular proliferation of HCC cells [77]. ●Lastly, the transmembrane receptor Janus kinases (JAKs) and the signal transducers STATs are commonly deregulated in HCC [78]. STAT3 mutations promote a number of cancer hallmarks, such as proliferation, angiogenesis and metastasis [79–83]. ●●Several small molecule STAT inhibitors, including Stattic, OPB-111077, OPB-31121, Napabucasin, and AZD9150 are currently in preclinical or Phase I clinical trials for HCC [84–88].●

Thus, recent NGS based studies have allowed a comprehensive understanding of the somatic mutation landscape of HCC. Although none of the major mutations are directly druggable at present, there are several promising candidates in the pipeline.

Copy Number Alterations

Somatic copy number alterations (SCNAs) result from the gain or loss of individual genes, or more commonly, entire chromosomal arms. The molecular consequence of SCNAs is the potential activation of oncogenes and loss of tumor suppressors, both of which drive carcinogenesis. Multiple studies have shown that copy number gains in chromosomes 1q and 8q, and losses in 8p and 17p, are the most frequent chromosomal arm level alterations in HCC [69,89,90]. Apart from the arm-level changes, gene-level changes are also important to identify. The well-known driver oncogenes CCND1, FGF19, MYC, MET, VEGFA, MCL1, and TERT, were recently shown to be significantly amplified in HCCs [69]. ●●Amplification of MYC, a transcription factor known to regulate all of the programs that are hallmarks of cancer, is thought to be an early genomic event in liver carcinogenesis. It has been found in both chronic liver disease, and in 70% of viral and alcohol-related HCCs [91]. MYC amplification at 8q24.1 has been repeatedly associated with large undifferentiated liver tumors, poor prognosis, metastasis and HCC recurrence [92–94]. Other well-known CNV amplifications include RB-regulated transcription factors E2F1 and E2F3. Amplification of these genes resulted in spontaneous HCC in murine models, and queries of the Cancer Genome Atlas (TCGA) datasets revealed a significant increase in the E2F family gene dosage in tumors of patients with advanced HCC [95]. ●Furthermore, copy number increases in matrix metalloproteinase-9 (MMP9), which promotes tumor metastasis via the breakdown of the extracellular matrix, were also shown to be associated with key clinicopathological features of HCC such as alpha-fetoprotein (AFP) level, tumor size, differentiation, invasion, and stage [96]. ●The selective presence of MMP9 CNVs in tumor tissue over normal tissue makes it a potentially promising diagnostic biomarker for HCC. On the other hand, analysis of HCC tumors has also revealed significant gene deletions in CDKN2A and tumor suppressors like ERRFI1, NCOR1, and RB1 [97], the latter of which is a common mechanism for the development of HBV and HCV HCCs [98]. ●

Other recently identified clinically significant CNVs include UBE2Q1, EXT1, WNK2, and JAGGED1. UBE2Q1 is an E2 ubiquitin-conjugating enzyme thought to promote carcinogenesis via the β-catenin/EGFR-PI3K-Akt-mTOR signaling pathway. Copy number gains in this gene are associated with poorer OS and DS [99]. ●EXT1, which encodes an endoplasmic reticulum glycosyltransferase, has previously been shown to prognosticate breast cancer, cholangiocarcinoma, and acute lymphoblastic leukemia (ALL). EXT1 mRNA was recently shown to be correlated with serum AFP, and its upregulation was found to be associated with worse DFS [100]. ●Analysis of 736 primary HCC samples revealed copy number loss of WNK2, a potential tumor suppressor, to be associated with early tumor recurrence, macrophage infiltration, tumor growth, and metastasis, likely via ERK1/2 signaling activation [101]. ●Amplifications of JAGGED1, which encodes a NOTCH pathway ligand, were also recently shown to be associated with poor OS and early HCC recurrence. Lastly, copy number mutations can also have positive therapeutic consequences. Increased FGF19 copy numbers were associated with a complete response after sorafenib treatment [102].

Epigenetic modifications

Epigenetic changes, which can occur via DNA methylation, histone modification, chromatin remodeling and non-coding RNAs, alter the way that genetic code is expressed, rather than directly affecting the nucleotide sequence. Dysregulated DNA methylation has been shown to be an important early event in the pathogenesis of HCC. Studies have noted greater global hypomethylation in HCC tumor tissue, particularly CpG dinucleotides within CpG islands, compared to adjacent tissue, with anywhere from 500 to 684 CpG sites being significantly hypermethylated in matched HCC and normal adjacent tissue comparisons [103,104]. A 2012 study suggested that these hypermethylated genes may be good early biomarkers for HCC, and five randomly selected genes (CDKL2, STEAP4, HIST1H3G, CDKN2A and ZNF154) from the top 18 hypermethylated genes in their study were detectable in the plasma of 63% of patients [103]. A more recent study identified 6 hypermethylated genes (NEBL, three FAM55C sites, GALNT3, and DSE) from 375 HCC samples that, when used as biomarkers for HCC, achieved a 98% specificity for HCC [105]. ●Other individual genes historically found to be hypermethylated in HCC include APC (81.7%), GTP1 (33.3%), RASSF1a (66.7%), p16 (48.3%), COX2 (35.0%), and Cadherin-1 (CDH1)(33.3%)[103,106]. A meta-analysis of 12 relevant HCC studies covering 981 patients showed that CDH1 hypermethylation was significantly higher in HCC tissues compared to normal liver, and was correlated with worse OS [107]. ●

Epigenetic changes in HCC can potentially be targeted using small molecule inhibitors of DNA methytransferases (DNMTs), which have historically been used in the treatment of myelodysplastic syndrome. Several first generation DNMTs like azacytadine and decitabine have been shown to reduce tumor formation by inducing hepatic cell differentiation and increasing cell sensitivity to sorafenib in preclinical studies, with decitabine Phase I/II clinical trials revealing acceptable safety and toxicity [108,109]. ●●Second generation DNMTs like guadecitabine and zebularine were created to improve upon the short half-lives of first generation DNMTs, and are also being tested in Phase I/II clinical trials [110].

Non-coding RNAs, which include micro RNAs (miRNAs) and long non-coding RNAs (lncRNAs), constitute a well-studied class of epigenetic regulators in HCC. Upregulated expression of mIR 21, mIR 221/222, and mIR 224 have been associated with increased HCC proliferation and migration, while decreased expression of mIR-26, mIR 122, and mIR-199 have been shown to suppress HCC proliferation and angiogenesis [111].●● The downregulated expression of mIR 200a was also recently shown to inhibit cell growth, migration and invasion [112].● These miRNAs may prove to be promising therapeutic targets. Anti-mIR 221 is currently in pre-clinical trials, and treatment with Miravirsin, a mIR 122 inhibitor that has completed Phase IIa trials, resulted in a substantial and prolonged decrease in plasma mIR 122 in patients with HCV [113,114]. Notable upregulated lncRNAs include HULC and HOTAIR [115,116]. HULC is thought to promote HCC proliferation and carcinogenesis indirectly by activating the CREB transcription factor, and is also associated with the epithelial-to-mesenchymal transition and angiogenesis [115]. HOTAIR, on the other hand, is thought to maintain the tumor microenvironment via CCL2 expression [116] ● and its loss was shown to sensitize HCC cells lines to chemotherapy [115,117].

Histones are protein octamers that help to condense DNA. Modifications to the histone tails that protrude from the DNA/histone nucleosome structure play an important role in the regulation of gene transcription and expression. The placement and removal of acetyl groups from these histone tails by histone acetyltransferases (HATs) and histone deacetylases (HDACs) are often dysregulated in HCC. Even though increased expression of histone deacetylase 3 (HDAC3) has repeatedly been shown to promote HCC proliferation and predict HCC recurrence, deficiency of HDAC3 was recently shown to promote HCC carcinogenesis in a murine model via a defect in the H3K9ac/H3K9me3 transition [118,119]. ● Upregulated HDACs 1 and 2 have also been shown to predict mortality in patients with HCC, and they may regulate doxorubicin sensitivity in HCC cell lines [120]. Although they have been used in the treatment of hematological malignancies, HDAC inhibitors have unknown efficacy in HCC. Panabinostat is currently in the preclinical phase of investigation, whereas Belinostat, Resminostat, and CUDC-101 are currently in Phase I/II clinical trials [121–124].

Gene Signatures

Transcriptomic studies in HCC have helped identify gene signatures, or clusters of differentially expressed genes, that can diagnose and prognosticate HCC, and also predict therapeutic response. Historically, gene signatures in HCC have focused on hepatocyte proliferation gene clusters and the EPCAM-positive hepatic cancer stem cell (hCSC) gene clusters. The proliferation cluster, which is expressed across a broad spectrum of human malignancies, includes the A- and B- type cyclins that control the cell cycle at G1/S and/or G2/M transition (CCNA2, CCNB2), cell division cycle proteins (CDC2, CDC7, CDC14, CDC20), heterohexamer DNA helicase minichromosome maintenance protein complex (MCM3–7), proliferating cell nuclear antigen (PCNA), and DNA topoisomerase 2 α (TOP2A), among others [125,126]. HCCs that express this gene signature, which largely mirrors a c-MYC- regulated gene signature, are associated with poorer OS, and were more likely to also have decreased expression of liver-specific genes that promoted hepatocyte dedifferentiation.

The hCSC gene cluster, on the other hand, was more likely to have increased expression of cell adhesion molecule EpCAM, epithelial marker CK19, and AFP, and has clinically been associated with chemotherapeutic resistance [127,128]. A recent study determined that EpCAM-regulated intramembrane proteolysis helps to drive the hCSC signature, and can potentially be targeted for inhibition in HBV-HCCs [129]. The molecular consequence of an hCSC gene signature is de-differentiation of tissues and loss of epithelial morphology, both of which promote malignant tumor behaviors and predict a worse prognosis for patients with HCC [130]. ●●Moreover, hCSC signatures may also be correlated with chemokine networks thought to create a hospitable inflammatory niche for tumor progression and metastasis [131]. A recent study identified an eight gene signature (TK1, CTTN, CEP72, TRIP13, FTH1, FLAD1, CHRM2, AMBP) in HBV-HCC tumors similar to that found in a previous study that is controlled by transcription factor OCT4, which is abundantly expressed in pluripotent stem cells [132].

Several clinically meaningful gene signatures have recently been identified. Through the application of two different algorithms to screen for differentially expressed genes in paired and unpaired HCC, Zhang et al identified a 14 gene signature in the cell-cycle-related gene cluster (BIRC5, BUB1B, CDC45, DTL, GINS2, KIF23, KIF2C, MAD2L1, MCM4, OIP5, PLK4, PTTG1, and ZWINT) that predicts poor OS and HCC recurrence [133]. ●●Another study revealed evidence of network reprogramming in carbon metabolism and cancer pathway genes through the identification of a 22-carbon metabolism gene signature that predicts poorer OS and DFS [134]. Li et al identified a DNA repair-related prognostic signature of seven genes (ADA, FEN1, POLR2G, SAC3D1, UPF3B, SF3A3, and SEC61A1) that, when used to stratify patients into high-risk and low-risk groups, predicts survival in HCC [135].●● None of the gene signatures have yet been validated or approved for clinical use, but significant progress is being made towards this goal.

Proteomics

Genetic, epigenetic, and post-translational dysregulation in HCC ultimately results in changes in protein expression levels and protein-protein interactions. Early proteomic studies in HCC, while useful in identifying potential protein targets related to early HCC diagnosis, were limited by smaller sample sizes, lack of validation, and absence of additional functional characterization [136]. These limitations have largely been addressed by advances in high-throughput protein analysis techniques and have resulted in detailed proteomic maps of HCC. Recently, a proteomic and phospho-proteomic comparison between 110 paired HBV-HCC and non-tumor tissues revealed enrichment of cell cycle, integrin, PDGF signaling, MAPK, TNF, and MET pathways, as well as hyperphosphorylation of the p38, RHO, myosin, RB1, and IL1 pathways [137].●● Metabolic reprogramming was found to be a key feature of HBV-HCC l on paired tumor and adjacent non-tumor liver tissues of 316 patients [138]. ●●Forty-two proteins known to play a role in amino acid metabolism and oxidoreductase activity were dysregulated in HCC. Importantly, the study found that with the exception of a few key metabolic enzymes (SOAT1, SOAT2, GLS, GLUD2), most proteins in liver-specific pathways, including gluconeogenesis, detoxification, and ureagenesis-ammonia, were significantly decreased in tumors.

Many other proteomic studies are focused on identifying protein targets for either diagnostic or therapeutic purposes. Using an absolute quantitation-based multidimensional liquid chromatography-tandem mass spectrometry technique, Liu et al identified 27 differentially abundant proteins, mostly in the ERK1/2 and nuclear factor-κ beta (NF-KB) pathways, in the serum of patients post-radical resection that were associated with HCC early recurrence [139]. ●PGK1, a glycolysis enzyme that has been detected in the serum of patients with a broad spectrum of malignancies [140], was specifically identified as an independent predictor of HCC recurrence and OS. Zhao et al used a high throughout urinary proteome analysis platform to compare the urine of 74 HCC and 82 high-risk patients with HBV-HCC to identify seven features that distinguish HCC from the high risk control population in a non-invasive fashion [141]. ●●Thus, large scale proteomics are adding to our understanding of the functional pathways activated in HCCs and are identifying promising diagnostic and predictive biomarkers for HCC.

Conclusions

HCC is a heterogeneous disease with such a complex genomic landscape. Understanding the molecular drivers of HCC carcinogenesis is essential both to identify biomarkers and to develop molecular targeted therapies. Promising new developments in this field are enabling us to develop therapies that can target the various drivers of HCC and evolve personalized therapeutic strategies. Future genomic studies promise to advance our understanding of this malignancy in meaningful ways and will hopefully ultimately lead to improvement in clinical outcomes for patients with HCC.

Acronyms

2D LC-MS/MS

multidimensional liquid chromatography-tandem mass spectrometry

AFP

alpha-fetoprotein

ALD

alcoholic liver disease

ALL

acute lymphoblastic leukemia

ATM

Ataxia telangiectasia mutated

CDH1

Cadherin-1

CNV

copy number variations

DFS

disease-free survival

DNMTs

DNA methytransferases

FDA

Food and Drug Administration

GOF

gain of function

HATs

histone acetyltransferases

HBV

hepatitis B virus

HCC

hepatocellular carcinoma

hCSC

hepatic cancer stem cell

HCV

hepatitis C virus

HDACs

histone deacetylases

HDAC3

histone deacetylase 3

JAKs

Janus kinases

lncRNAs

long non-coding RNAs

LT

Liver Transplantation

MAT1A

methionine adenosyltransferase 1

MMP9

matrix metalloproteinase-9

micro RNAs

micro RNAs

NAFLD

nonalcoholic fatty liver disease

NGS

next generation sequencing

NF-KB

nuclear factor-κ beta

OS

overall survival

PLEC1

plectin 1

PRC1

protein regulator of cytokinesis 1

SCNAs

somatic copy number alterations

scRNA-Seq

single-cell RNA-sequencing

TACE

transarterial chemoembolization

TCGA

the Cancer Genome Atlas

TERC

telomerase RNA component

TERT

telomerase reverse transcriptase

TF

transcription factor

TOP2A

topoisomerase 2 α

JAKs

transmembrane receptor Janus kinases