Epigenetic Basis of Cancer Health Disparities: Looking Beyond Genetic Differences

Abstract

Despite efforts at various levels, racial health disparities still exist in cancer patients. These inequalities in incidence and/or clinical outcome can only be explained by a multitude of factors, with genetic basis being one of them. Several investigations have provided convincing evidence to support epigenetic regulation of cancer-associated genes, which results in the differential transcriptome and proteome, and may be linked to a pre-disposition of individuals of certain race/ethnicity to early or more aggressive cancers. Recent technological advancements and the ability to quickly analyze whole genome have aided in these efforts, and owing to their relatively easy detection, methylation events are much well-characterized, than the acetylation events, across human populations. The early trend of investigating a pre-determined set of genes for differential epigenetic regulation is paving way for more unbiased screening. This review summarizes our current understanding of the epigenetic events that have been tied to the racial differences in cancer incidence and mortality. A better understanding of the epigenetics of racial diversity holds promise for the design and execution of novel strategies targeting the human epigenome for reducing the disparity gaps.

1. Introduction

Health disparities in cancer patients of different racial and ethnic backgrounds have attracted a lot of attention in recent years. Because of the increased awareness and scrutiny, some progress has been made but there is irrefutable evidence to support that racial health disparities still exist in a vast majority of cancer patients [1–4]. When discussing racial cancer health disparities, black patients, most frequently African Americans (AA), are the most common racial group compared to the Caucasian Americans (CA) or white population of European heritage, referred as European Americans. Data on races other than AA and CA, such as Hispanics and Native Americans, is emerging, and points to existence of racial cancer health disparities in these groups as well [5]. Besides racial disparities in cancer incidence and overall clinical outcome, there is incongruence in diagnosis at the initial presentation and the time between the diagnosis and initiation of treatment as well [6, 7]. Even the low participation of minorities in clinical studies is a point of concern [8]. While there are studies that suggest narrowing cancer health disparities between different populations because of concerted efforts of health and local authorities [9], there is still overwhelming data to support existence of racial disparities across almost all human cancers [10–14].

In addition to the non-biological factors (such as socioeconomic, cultural, etc.), biological factors are also believed to play important roles in cancer health disparities [15–18]. Recent data has connected inherent genetic differences, such as those resulting in increased tumor heterogeneity in AA breast cancer patients, with more aggressive tumor biology [19]. AA breast cancer patients have substantially poorer outcomes if they present with a hormone receptor-positive, HER-2 (human epidermal growth factor receptor-2)-negative or the triple negative phenotype [20]. While the genetic basis of disparity is being actively pursued [17, 21], recent findings provide strong suggestion for an epigenetic basis of racial health disparities in cancer, which is discussed in the following sections.

2. Epigenetics in Cancer Health Disparities

Epigenetics is the study of changes in gene expression that are heritable and caused by events other than the change in DNA/gene sequence [22] (Figure 1). Epigenetic events have largely been identified as covalent modifications of either DNA or the histones [23]. However, regulation of gene expression by microRNAs (miRNAs) and other non-coding RNAs (ncRNAs) is also within the broader definition of epigenetics [24]. Covalent modifications include DNA methylation and various modifications of histones viz. acetylation, methylation, phosphorylation, ubiquitination and sumoylation.

Figure 1. Epigenetic events.

Epigenetic events such as methylation and acetylation have profound effect on the transcription of individual genes. These events serve as ‘ON’ / ‘OFF’ switches for transcription of genes and the eventual expression of gene products. While generally tightly regulated in ‘normal’ cells, these epigenetic changes are altered in cancer cells to favor proliferation, anti-apoptotic and pro-metastatic functions. The evidence suggesting a role of epigenetic events in cancer racial disparity is emerging, and more detailed studies are needed to fully understand the complex role of epigenetic events in cancer health disparities.

The research on health disparities has traditionally focused on socioeconomic factors [25]. Socio-economically disadvantaged and low-income populations suffer from disproportionately high incidence and/or mortality rates of cancer [26] as well as other chronic diseases [27, 28]. Social, economic and cultural inequalities, including unsafe living conditions and other stress-inducing conditions, can result in a distinct epigenetic signature with altered epigenetic markers [29–31] and even an accelerated epigenetic aging [32]. Further, lifestyle changes, such as diet and weight loss, have also been reported to influence global DNA methylation levels [33]. The epigenetic changes thus function as a liaison between social/cultural/economic/environmental factors and the genome. Since these inequalities persist for prolonged periods of time, often lifetime, the resulting epigenetic changes keep accumulating and affecting the epigenome in various racial/ethnic groups, resulting in poorer health outcomes, including cancer.

A role of epigenetics in cancer progression, metastasis, angiogenesis and drug resistance is well appreciated, and targeting of epigenome for cancer control is one of the latest areas of focus in cancer research [34, 35]. The information on the role of epigenetics in racial cancer health disparities is scattered, and through the next few sections we will attempt to comprehensively summarize all such information to make a case for the importance of epigenetic events in racial diversity observed in cancer patients.

2.1. DNA Methylation in Cancer Health Disparities

Among all the epigenetic events investigated so far for possible implications in cancer health disparities, methylation is evidently the subject of most interest [36]. Phosphate-linked cytosine and guanine nucleotides, the ‘CpG’ islands, are regions in the DNA with high frequency of CpG repeats. Methylation at CpG islands is an emerging biomarker for cancer diagnosis [37]. There are ~45000 CpG islands per haploid genome in humans [38]. A majority of these (80–90%) are methylated but the rest, found in the promoters of more than half of the genes, remain unmethylated. Methylation status is intricately linked to induction and suppression of genes, wherein reduced methylation switches ‘on’ the expression of genes and increased methylation results in switching ‘off’. Breast, prostate, colorectal and endometrial cancers are the only cancers where a link between DNA methylation and racial cancer health disparities has been suggested (Table 1). A discussion of such reports is presented in the sub-sections below.

Table 1.

Altered epigenetic events, relevant to racial disparity, as reported in cancer patients

Cancer	Gene(s)	Observation	Reference
Breast	CDH13	Differential methylation of CDH13	[56]
	-	ER-negative tumors are differentially methylated in AA	[43, 56, 57]
	DSC2, KCNK4, GSTM1, AXL, DNAJC15, HBII-52, TUSC3 and TES	Differential regulation of gene panel	[63]
Prostate	AR, RARβ2, SPARC, TIMP3 and NKX2-5	Differential regulation of 5 genes from a panel of 6 genes tested	[76]
	CD44	Hypermethylation of CD44 alone in AA, from a panel of 3 genes tested	[93]
	CD44	Hypermethylation of CD44 alone in AA, from a panel of 8 genes tested	[94]
	GSTP1	Hypermethylation of GSTP1 in AA	[92]
	TIMP3 and NKX2-5	Possible differential acetylation	[76]
	SNRPN, MST1R and ABCG5	Increased frequency of differentially mediated genes in AA	[101]
	TMS1	Increased methylation in CA cancer patients, relative to healthy controls	[107]
Colorectal	-	Significantly increased differentially methylated regions. CHL1, NELL1, GDF1, ARHGEF4 and ITGA4 hypermethylated in AA	[110]
	MGMT	Differential methylation of MGMT in AA patients	[129]
Endometrial	-	Differential ribosomal DNA methylation	[132]

2.1.1. DNA methylation in Breast Cancer Health Disparity

Breast cancer is the most frequently diagnosed cancer in females in the United States [1, 39]. The data for the year 2013 from Surveillance, Epidemiology and End Results (SEER) database puts breast cancer disparity ratio at 39.14% (28.19 AA deaths Vs. 20.26 CA deaths, per a population size of 100,000). Interestingly, the overall incidence rate of breast cancer among AA women is slightly less than CA women in the US (124.3 vs. 128.1, per a population size of 100,000) [1]. Combined with the mortality rate, this indicates that the breast cancer in AA women is comparatively more aggressive, an idea that is well supported by published literature [17, 19, 40]. Triple-negative breast cancers, the aggressive breast cancer subtype with no targeted therapy, are diagnosed at relatively younger age in AA women [41], and have worse clinical outcomes [42], compared to CA women. A number of studies investigating epigenetic basis of breast cancer disparity have looked at a pre-determined subset of genes, with a focus on the methylation status.

In one of the early studies on the subject [43], AA breast cancer patients were found to harbor higher frequency of methylation in four of five genes studied. The study looked at promoter methylation of 5 select genes viz. HIN-1 (high in normal-1), Twist, Cyclin D2, RAR-β (retinoic acid receptor-beta) and RASSF1A (ras associated domain family 1 isoform A) in 67 AA vs. 44 CA breast tumors. The five genes were selected based on the reports on their hypermethylation in breast cancers, which provided a rationale for this analysis. HIN-1’s selective hypermethylation in breast cancer epithelial cells has been linked to substantial loss of expression [43–45]. Moreover, reintroduction of HIN-1 to breast cancer cells has been demonstrated to reduce the growth of breast cancer cells, suggesting its potent tumor suppressor properties [44]. Further, the hypermethylation-induced inhibition of HIN-1 in majority of AA patients suggests a possible mechanism of methylation-induced disparate cancer health outcomes [43]. Twist is methylated in ~40% primary breast cancers [43, 45], and the study by Mehrotra et al. [43] demonstrated an increased promoter methylation of Twist in the AA cancer patients. In a study by Gort et al. [46], no correlation between Twist promoter methylation and protein/RNA expression was observed even though Twist promoter methylation was observed to be significantly more prevalent in malignant, as compared to healthy tissues. The observations regarding methylation of Twist are intriguing. Twist is a basic helix-loop-helix transcription factor that is considered to play important role in EMT (epithelial-to-mesenchymal transition) and metastasis; thus, correlating with poor overall survival [47, 48]. At the same time, methylation of its promoter has also been reported in breast cancers [49], which is counterintuitive because methylation silences expression. As a possible explanation of how promoter methylation as well as gene product can correlate with cancer progression, it has been suggested that either the proximal part of Twist1 promoter does not influence Twist1 expression or Twist1 methylation is probably an early event that precedes compensatory Twist1 over-expression [46, 50]. Expression of Cyclin D2 is not observed in a majority of breast cancer cells lines [51] and its promoter is methylated in ~50% primary breast cancers [43, 45]. RAR-β is methylated in ~50% invasive breast cancers and was observed to be hypermethylated in 76% of the AA cases vs. only 29% of the CA cases [43, 45]. Restoration of RAR-β expression in breast cancer cell lines has been associated with cell cycle arrest and induction of apoptosis [52]. In other studies, promoter methylation-induced silencing of the RAR-β gene has been reported not only in breast, but cervical cancers as well [53, 54]. Similarly, RASSF1A gene, methylated in 50–60% primary breast cancers, is a potent tumor suppressor, which is found to be hypermethylated in significantly greater number of AA cases than CA cases [43, 45, 55].

In the study by Mehrotra et al. [43], only the ER (estrogen receptor)-negative/PR (progesterone receptor)-negative breast tumors differed in the methylation patterns in AA vs. CA samples. Additionally, age of the patients below 50 was also an important determining factor in differential methylation of genes, suggesting that inherent epigenetic differences could possibly be linked to documented earlier onset of the breast cancer in most AA women, as compared to CA women. Similarly, the study by Wang and co-workers [56] also found the differential methylation status in AA vs. CA patients in a subset of patients of age <50 with ER-negative breast tumors. The finding that the ER-negative breast tumors are more differentially methylated than ER-positive breast tumors in AA, relative to CA patients, was confirmed by a genome-wide methylation study [57].

In another study that compared DNA methylation in 32 AA vs. 33 CA breast cancer patients, CpG islands were evaluated for the genes CDKN2A, RASSF1A, RARβ2, ESR1 (estrogen receptor-1), LINE1 (long interspersed nuclear element-1), CDH13 (cadherin-13), HIN1 and SFRP1 (secreted frizzled related protein-1) [56]. CDKN2A is a known tumor suppressor encoding for the protein p16. ESR1 codes for estrogen receptor 1 which is important for hormone binding and activation of transcription. Mutations in ESR1 have been implicated in resistance to endocrine therapy [58] and metastasis [59] in breast cancer patients. LINE1 is a global methylation marker and LINE1 repeats make up roughly 17% of human genome and, therefore, its promoter methylation is widely used as a surrogate marker for global DNA methylation [60]. Other six genes studied are known growth suppressor genes, whose methylation and consequential suppression of expression is relevant to cancer progression. The one gene that stood out in this study was CDH13, which exhibited significantly differential methylation status in AA patients, compared to the CA patients. More importantly, CDH13 methylation was particularly increased in triple negative breast cancers. Other studies have also demonstrated CDH13 to be frequently silenced in breast tumors because of its promoter methylation [61] and associated its promoter methylation with increased breast cancer risk [62]. More recently [63], 216 AA breast tumors were compared with 301 non-AA breast tumors for differential DNA methylation at the CpG islands of select cancer-related genes. The genes that were observed to be particularly differentially methylated were DSC2 (desmocollin-2), KCNK4 (potassium channel subfamily K member 4), GSTM1 (glutathione S-transferase mu-1), AXL, DNAJC15 (DnaJ heat shock protein family member C15), HBII-52 (also known as SNORD115@ small nucleolar RNA, C/D box 115 cluster), TUSC3 (tumor suppressor candidate 3) and TES (testin). Methylation-induced silencing of a number of these genes is well documented in different human cancers. For instance, DSC2’s [64] and TES’s [65] methylation-induced silencing has been reported in breast cancer. There has been an interest in methylation of AXL because of the observation that it is an imprinted gene whose methylation-influenced expression can be maternally transmitted [66]. As a consequence of this ‘heritability’ of AXL methylation, it has been shown that fetal exposure of children, predominantly girls, to maternal smoking can induce methylation of AXL [67]. Methylation of GSTM1 [68] and TUSC3 [69] has been linked to poor disease-free survival of acute myeloid leukemia and ovarian cancer patients, respectively. Benevolenskaya and colleagues [70], analyzed 75 paraffin-embedded breast samples from 21 CA, 31 AA, and 23 Hispanic patients, and reported hypermethylation of promoters of genes FZD9 (fizzled-9), MME (membrane metalloendopeptidase), BCAP31 (B-cell receptor-associated protein-31), HDAC9 (histone deacetylase-9), PAX6 (paired box-6), SCGB3A1 (secretoglobin family 3A member 1), PDGFRA (platelet-derived growth factor receptor alpha), IGFBP3 (insulin like growth factor binding protein-3), and PTGS2 (prostaglandin-endoperoxide synthase 2), and they all correlated with receptor (ER and PR)-positive disease.

2.1.2. DNA methylation in Prostate Cancer Health Disparity

Prostate cancer is the most common non-cutaneous malignancy in men in the United States [39]. According to the data from US SEER database for year 2013, prostate cancer had a disparity ratio of 118.52% (39.05 AA deaths Vs. 17.87 CA deaths, per a population size of 100,000). There is disparity in both the incidence and mortality rates of prostate cancer, with the numbers being disproportionately higher for AA men; nearly two-thirds higher incidence and more than double mortality [1, 71–75]. In a study that investigated molecular basis of racial disparities in prostate cancer, a set of six genes (GSTP1, glutathione S-transferase pi-1; AR, androgen receptor; RARβ2; SPARC, secreted protein acidic and rich in cysteine; TIMP3, tissue inhibitor of metalloproteinase 3 and NKX2-5, NK2 homeobox-5) was chosen based on the reports on their hypermethylation in patient samples [76]. AR signaling, central to prostate cancer progression [77], is known to be epigenetically regulated [78]. The tumor suppressor RARβ2, similar to its hypermethylation in breast cancer [54, 79], cervical cancer [53] and leukemia [80], is significantly more hypermethylated in prostate cancer patients, and correlates with increased tumor risk [81]. Role of SPARC has been investigated in multiple cancer models [82] and its function largely depends on whether it is produced by cancer cells or the stromal cells in the tumor microenvironment [82]. It is frequently down-regulated in cancer cells [83] and suppresses prostate cancer cells’ proliferation and migration [84]. It is silenced through promoter methylation in metastatic and aggressive prostate cancer cells, with its promoter hypermethylation inversely correlating with prostate cancer patients’ disease-free survival [85]. Hypermthylation-induced silencing of SPARC has also been reported in pancreatic cancer [86], ovarian cancer [87] hepatocellular carcinoma [88] and laryngeal and hypopharyngeal carcinomas [89]. Low TIMP3 expression in many cancers, including prostate cancer, has been demonstrated to facilitate metastatic potential of tumor cells, due to an increase in the activity of matrix metalloproteinase activity in the tumor microenvironment [90]. Similarly, the gene product of NKX2-5 has been proposed as a tumor suppressor in prostate cancer, while its precise biological activity is yet to be determined [91]. Hypermethylation of all of these genes in prostate cancer samples was observed, relative to matched normal samples [76]. However, only five of them, with the exception of GSTP1, were found to be relevant to racial cancer health disparity, exhibiting increased methylation in AA prostate cancers, compared to CA prostate cancers. There was interest in GSTP1 hypermethylation because of an earlier observation where GSTP1 hypermethylation was proposed as a sensitive biomarker in AA prostate cancer patients, compared to CA prostate cancer patients [92]. However, some other studies [93, 94] also failed to find a utility of GSTP1 in racial disparity in prostate cancer patients. One study looked at hypermethylation of GSTP1, CD44 and E-cadherin [93] and, while it failed to find differential hypermethylation of GSTP1 in AA patients, it did find such correlation for CD44, with its 1.7 folds hypermethylation in AA prostate cancer patients, compared to the CA counterparts. Another study by the same group looked at an even larger set of genes viz. GSTP1, RASSF1A, RARβ2, CD44, EDNRB (endothelin receptor type B), E-cadherin, Annexin-2 and Caveolin-1, and again found a possible differential CD44 hypermethylation in AA prostate cancer patients [94]. CD44 is a critical marker of prostate cancer stem cells [95] and its differential expression, through epigenetic regulation, can influence metastasis as well as resistance to therapy. Interestingly, loss of CD44 in primary prostate cancer has also been linked to unfavorable clinical behavior [96]. In addition to prostate cancer [97], hypermethylation-induced regulation of CD44 has been reported in other cancers, including breast [98], colorectal [99] and neuroblastoma [100].

In the study discussed above [76] that investigated the methylation of a six gene panel in prostate cancer patients, the effect of demethylating agent, 5′aza-dC, was also compared with that of histone deacetylase inhibitor, TSA (trichostatin A). First, the findings from methylation part of the study were validated. With a hypothesis that the demethylating agent reduces methylation, resulting in re-expression of genes, it was no surprise to find such effect on almost all genes in prostate cell line pNT1a. However, an effect of TSA was also observed on a few genes namely, TIMP3 and NKX2-5, suggesting that the expression of these two genes is controlled by acetylation as well. While the earlier study [76] employed pyrosequencing to quantitate methylation, a latter study [101] by the same researchers used genome-wide methylation analysis of prostate cancer tissues from AA vs. CA patients. A clear discrepancy in frequency of differentially methylated genes was observed with samples from AA patients exhibiting greater rate of differentially methylated genes. 3303 probe sets showed more than 1.5 fold changes in methylation in AA patients, while only 1075 probe sets showed more than 1.5 fold changes in the CA patients. Of these differentially methylated probes, only 330 probe sets were common to both; 2973 probe sets were unique to AA patients and 745 were unique to CA patients. The top 25 differentially methylated genes in AA vs CA patients were further subjected to pyrosequencing in 7 prostate cancer cell lines representing the two different racial origins. The three genes that stood out with possible differential methylation status in AA vs CA were SNRPN (small nuclear ribonucleoprotein polypeptide N), MST1R (macrophage stimulating 1 receptor) and ABCG5 (ATP binding cassette subfamily G member 5). SNRPN is down-regulated in cancer samples [102], and higher methylation of SNRPN’s promoter CpG islands in prostate cancer tissues, relative to normal tissues, has been confirmed by others as well [103]. The SNRPN encoded protein has a role in the processing of pre-mRNA and possibly governs alternative splicing events in tissue-specific manner [104]. MST1R’s promoter methylation leading to decreased expression has also been reported in cancers other than prostate [105]. Further, stromal cell expression of MST1R within the prostate tumor microenvironment has been observed to drive tumor growth [106]. Knock-down of genes in prostate cancer cell lines of CA origin, to mimick the effect of hypermethylation in AA tissues, resulted in significant inhibition of proliferation and invasion [101].

In a study that looked at the tumor suppressor, TMS1 (target of methylation-induced silencing-1, also known as ASC, apoptosis-associated speck-like protein containing a CARD), an evidence for its differential racial methylation was seen [107]. Prevalence wise, no difference in methylation of this gene was observed in AA (66.7%) vs. CA (62.2%) prostate cancer patients. However, when these numbers were compared to those in healthy individuals, it was observed that TMS1 methylation was more prevalent in CA prostate cancer patients than in the CA healthy controls. AA populations, on the other hand, had similar prevalence in healthy individuals as well as prostate cancer patients. Based on all the available evidence on differentially-methylated genes in AA vs. CA prostate cancer patients, it appears that there are some valid leads that need to be investigated further.

2.1.3. DNA methylation in Colorectal Cancer Disparity

Similar to breast and prostate cancers discussed above, colorectal cancers also have racially disparate mortality rates. SEER database 2013 suggests a mortality disparity ratio of 34.65% among US females and 46.97% among US males, per a population size of 100,000. Colorectal cancer in AAs is suggested to be more aggressive, with 28% higher incidence and 60% increased mortality associated with advance stage colorectal cancers, as compared to CA [108, 109]. There is also evidence supporting epigenetic basis of colorectal cancer disparity [110]. In this study performed on DNA and RNA samples from 6 AA and 7 CA patients, a large number of differentially-methylated regions (DMRs) were identified. More than 27K CpG sites were detected in 1688 differentially-methylated regions in the samples from AA patients, while only 764 methylated CpG sites were detected in 113 DMRs in the samples from CA patients. This is a very significant difference in the status of DMRs in AA patients, compared to CA patients. In this study, hypomethylation vs. hypermethylation in AA vs. CA colorectal cancer patients was also analyzed. 100 DMRs were found to be hypomethylated, while 1588 DMRs were hypermethylated in AA patients whereas only 4 DMRs were hypomethylated and 109 DMRs hypermethylated in CA patients. When extended to differentially methylated genes, the study revealed 23 hypermethylated genes in AA samples and 29 hypermethylated genes in CA samples. Further, 4 genes were observed to be hypomethylated in AA and 1 gene hypomethylated in CA samples. Genes CCDC178 and FLI1 were hypermethylated in both AA and CA tumor samples. However, differential methylation status of these two genes between AA and CA specimens remains unclear as this particular analysis made comparisons of methylation of AA/CA tumor specimens with their respective matched normal tissues. Still, this suggests that these two genes, in general, might be involved in cancer progression. Indeed, FLI1 (Friend leukemia integration 1) has recently been demonstrated to be hypermethylated and, consequently, down-regulated in gastric cancers, relative to non-metaplastic mucosa [111]. FLI1 is relatively abundant in adult hematopoietic tissues, compared to non-hematopoietic tissues, and its activation has been linked to malignant transformation [112]. FLI1 plays an important role in normal development and homeostasis, but its aberrant expression can result in cancer onset [113] as well as metastasis [114].

The study on DMRs in colorectal cancer [110] also identified 108 down-regulated genes and 34 up-regulated genes in AA samples, when compared directly with CA patient samples. Among the top down-regulated genes in AA patient samples, compared to CA samples, the most significantly differentially expressed gene, RPL13 (ribosomal protein L13) has actually been reported as an oncogene in gastrointestinal cancers [115]; its knockdown causes cell cycle arrest and increases sensitivity to DNA damaging agents. The next gene in the list, HMGCS2 (3-hydroxy-3-methylglutaryl-CoA synthase) is involved in cell differentiation and there is conflicting evidence in the literature regarding its role in tumorigenesis. It has been positively correlated with tumor progression, resulting in poor prognosis [116] but has also been observed to be down-regulated in several intestinal cancers in a myc-dependent way [117]. Other top down-regulated genes in AA samples with a reported oncogenic function include CES2 (carboxylesterase 2; overexpressed in pancreatic cancer and suggested as a determinant of response to FOLFIRINOX therapy [118]), KRT19 (keratin 19: influences invasiveness of breast cancers [119] and radioresistance and cancer stem cell phenotype in colorectal cancers [120]) and RPS2 (ribosomal protein S2: over-expressed in prostate cancer tissues and cell lines [121]). Further, TFF3 (trefoil factor 3) is expressed in normal breast lubules and ducts, benign lesions as well as in invasive carcinomas [122]. Given the comparatively more aggressive disease in AA, this apparent disconnect and the lack of enough information on individual genes is indicative of the enormous work that needs to be done to fully understand the genetic and epigenetic basis of cancer health disparities. On a positive note, the top differentially up-regulated genes in AA samples, compared to CA samples, have also been reported as tumor-promoting genes. These include THBS2 (thrombospondin-2; associated with poor disease-free and overall survival of colorectal cancer patients [123]), PCA3 (prostate cancer antigen-3: over-expressed in prostate cancer cells [124] and a suggested prostate cancer prognostic marker [125]), CYP1B1 (cytochrome P450 1B1; overexpressed in tumor tissues [126]) and BCAT1 (branched chain amino-acid transaminase 1; over-expressed in ovarian tumors [127], possibly through a mechanism involving hypomethylation of the gene [128]) In a study that did not specially focus on cancer health disparities, a differential-methylation of MGMT (O-6-methylguanine-DNA methyltransferase) in AA colorectal patients, particularly the elderly population, was noted [129]. Silencing of MGMT has been shown to precede and associate with the appearance of G-to-A point mutations in the KRAS gene during colorectal tumorigenesis [130]. These early reports on epigenetic basis of racial disparity in colorectal cancer patients have provided encouraging preliminary data, however, more such studies are urgently needed.

2.1.4. DNA methylation in Endometrial Cancer Disparity

For endometrial cancer, the racially disparate mortality ratio is 92.68 % (7.9 AA deaths Vs. 4.1 CA deaths, per population size of 100,000, according to US SEER database). Similar to breast cancer discussed above, while the death rate is higher in AA women, incidence rate is higher in CA women by 9.8% [1]. Endometrial cancers are diagnosed at later stage, and with higher grade disease, in AA women [131]. In addition, the incidence rate of aggressive endometrial cancer subtypes is particularly high in AA women [13]. In endometrial cancer, epigenetic changes in DNA region that codes for ribosomal RNA (referred as ribosomal DNA or rDNA) have been tied with racial cancer health disparity [132]. A comparison of methylation of rDNAs from 176 CA and 39 AA patients-derived tumors revealed that the tumors with high levels of rDNA methylation were the ones with relatively favorable prognosis. Conversely, the tumors with low levels of rDNA methylation correlated with poor prognosis and the AA patients seemed to have relatively higher percentage of such low rDNA methylations (46% vs. 22% in CA). The role of rDNA methylation in endometrial tumorigenesis remains unclear. It has been suggested that rDNA methylation might be a surrogate for overall aberrant methylation of key genes [132]. No specific genes have yet been identified that are differentially methylated between AA and CA endometrial cancers. The AA women have seen an annual percent change of 2.5% in the incidence of endometrial cancer and the 5-year survival rate of AA endometrial cancer patients is significantly less than that of CA patients [13]. While the factors responsible for this disparity are under investigation, the role of epigenetic events, such as the differential rates of rDNA methylation, needs further evaluation.

2.2. Histone modifications in Cancer Health Disparities

As discussed so far, DNA methylation has been shown to be different to some extent in AA vs CA populations (Figure 2). Another important epigenetic event is the modifications of histones, with acetylation and methylation of histones as two well-studied histone modifications. In contrast to DNA methylation, there has been little progress on our understanding of histone modifications in the context of cancer racial health disparities. This is best exemplified by a very recent report that looked at 59,089 men of African and European ancestries and did not find any significant differences in the histone acetylation [133]. The study specifically looked at H3k27 acetylation as a biomarker. The one positive indication for a role of histone modification in cancer racial disparity came from the study in prostate cancer model where an effect of TSA was seen on TIMP3 and NKX2-5 [76]. TSA is an inhibitor of histone deacetylases and thus influences the acetylation of histones. The effect of TSA was seen when it was used to treat prostate cancer cell lines. This is an indirect evidence, which needs to be corroborated directly in patient samples. The lack of convincing evidence in support of histone acetylation, and even histone methylation, is not necessarily a verdict on the relevance of these histone modifications in cancer racial disparity. We need to be cognizant that DNA methylation is relatively easy to detect and thus the reports on DNA methylation are much more forthcoming. This will surely change with time, and with the advances in technological capabilities.

Figure 2. Differential epigenetic events that underline racial cancer health disparities.

Differences in methylation patterns have been proposed as underlying causes for disparate breast, prostate, colorectal and endometrial cancers in AA vs. CA populations. There is also evidence for differential expression of individual microRNAs across racial groups. AA, African American; DMRs, Differentially Methylated Regions; rDNA, ribosomal DNA.

2.3. miRNAs and ncRNAs in Cancer Health Disparities

Gene regulation involving miRNAs and ncRNAs is yet another important epigenetic event, and a role of miRNAs and ncRNAs in cancer health disparities has started emerging (Table 2). Interestingly, expression of miRNAs may, in turn, also be regulated by epigenetic events. In this section, a discussion on the differential expression of miRNAs/ncRNAs, and the resulting functional consequence, in racially disparate cancers is provided.

Table 2.

microRNAs in racial cancer health disparities

miRNA (s)	Observation	Cancer	Reference
miR-9	Hypermethylated in AA	Colorectal	[110]
miR-26a	Up-regulated in AA	Prostate	[143]
miR-31	Up-regulated in CA	Thyroid	[154]
miR-34	Hypermethylated in CA	Colorectal	[110]
miR-124	Hypermethylated in AA	Colorectal	[110]
miR-137	Hypermethylated in AA	Colorectal	[110]
miR-152	Hypermethylated	Prostate	[144]
miR-182	Up-regulated in AA	Colorectal	[153]
miR-212	Down-regulated in AA	Prostate	[135]
miR-221	Up-regulated in CA	Thyroid	[154]
miR-337	Down-regulated in CA	Endometrial	[157]
miR-548	Hypermethylated in AA	Colorectal	[110]
miR-663	Hypermethylated in AA	Colorectal	[110]
miR-1207	Increased expression in AA	Colorectal	[151]
miR-1279	Up-regulated in AA, compared to CA	Colorectal	[110]
miR-2682	Hypermethylated in AA	Colorectal	[110]
miR-6130	Hypermethylated in AA	Colorectal	[110]

2.3.1. miRNAs in Prostate Cancer Health Disparity

As mentioned above, prostate cancer is one of the most racially disparate cancers, with disproportionally higher mortality of AA patients, as compared to CA patients. Apart from genetic and epigenetic events, the role of miRNAs in prostate cancer health disparities is being appreciated [134]. In prostate cancer patients, miR-212, a negative regulator of splicing factor heterogeneous nuclear ribonucleoprotein H1 (hnRNP H1), was found to be down-regulated in AA patients, compared to CA patients [135]. Such down-regulation of miR-212 correlated with aberrant expression of hnRNP H1, as well as the androgen receptor, resulting in castration-resistance. In a more recent article that focused on TMPRSS2 (transmembrane protease, serine-2) translocations, a panel of 18 differentially altered miRNAs was identified [136]. These miRNAs were associated with DNA CpG methylation, and the resulting aggressive disease in TMPRSS2 fusion-negative prostate cancers. Among these, miR-125b, miR-17, miR-29 and miR-200b have been studied in prostate cancer [136]. miR-125b has been demonstrated to facilitate the development of castration-resistant prostate cancer by targeting the repressor of AR signaling, NCOR2/SMRT (nuclear receptor corepressor 2/silencing mediator of retinoic acid and thyroid hormone receptor) [137]. Similarly, in prostate cancer, miR-17 suppresses the levels of PTEN [138] and miR-29 suppresses the metastatic cascade at multiple steps [139]. miR-29 also regulates the expression of anti-apoptotic and matrix molecules after up-regulation by MBP-1 (c-myc promoter binding protein) in prostate cancer [140]. Williams et al. [141] demonstrated that miR-200b inhibits spontaneous metastasis in an orthotopic mouse model through reversal of EMT with increased E-cadherin, a phenomenon similar to one reported for this miRNA earlier in breast cancer [142]. In another study, miR-26a was found to be significantly overexpressed in AA prostate cancer cells, as compared to CA-derived cell lines [143]. While the precise mechanism of action of miR-26a in prostate cancer has not been delineated, it was observed to modulate apoptosis and cell survival, partly through the inactivation of caspase 3/7 [143]. Theodore et al. [144], observed decreased levels of miR-152 in aggressive prostate cancer cell lines, concomitant with increased promoter methylation, which could be reversed by the treatment of 5-aza-2-deoxycytidine. Clinically, a statistically significant down-regulation of miR-152 was found in about 50% of the AA cancer tissue samples, compared to 35% of the CA samples. A reciprocal inhibitory loop of miR-152-DNMT1 (DNA methyltransferase-1) has been deduced to be the underlying mechanism of action of miR-152 in prostate [144], ovarian [145] and breast cancer [146], underlying the epigenetic connection of this miRNA.

2.3.2. miRNAs and ncRNAs in Colorectal Cancer Health Disparity

Colorectal cancer is a relatively better studied malignancy in terms of miRNA regulation of cancer health disparity. For instance, when extended to differential epigenetic regulation of miRNAs, the study on cancer health disparity in colorectal cancer, discussed above [110], found a small set of differentially-hypermethylated miRNAs, with no miRNA being hypomethylated. A total of 7 miRNAs (miR-9, miR-124, miR-137, miR-548, miR-663, miR-6130 and miR-2682) were hypermethylated in AA patients’ tumors, relative to their adjacent normal tissue, while only miR-34 was hypermethylated in CA patients. Hypermethylation of the promoter region of these miRNAs leads to a decrease in the expression of these miRNAs [147–150]. Almost all these miRNAs, hypermethylated in AA tumors, are known tumor suppressors, with no reports available for the function of miR-6130 and miR-2682. Interestingly, miR-34 has been demonstrated as a potential regulator of metastasis in several cancers, but was observed to be the only miRNA hypermethylated in the CA patients. Also, miR-1279 was identified as the only miRNA dysregulated between AA and CA patient samples [110]. This miRNA was up-regulated in AA samples, however, the functional relevance of miR-1279 is not known. Another study with colorectal cancer patients (53 AA and 47 CA), not only found an evidence of miRNAs, but also of cancer stem cells and long noncoding RNA PVT1, the host for miR-1207-5p, in cancer racial disparity. AA patients had significantly high levels of miR-1207-5p, which was mechanistically linked to increased ‘stemness’, as evidenced by increase in various markers of cancer stem cells and EMT [151]. This increase in cancer stem cells can possibly explain the relatively aggressive colorectal cancers in AA patients.

Identification of differential epigenetic regulation of miRNAs is only the first step in determining a role of miRNAs in cancer health disparity. Since a single miRNA can affect the expression of a number of target genes, epigenetic regulation of just one miRNA can potentially affect the expression and function of many key genes. This underlines the impact of miRNAs. At the same time, it presents a challenge to the researchers because the identification of exact targets, among the plethora of putative targets, is very important to understand the mechanistic details as well as to exploit this knowledge for possible intervention and therapy. In the colorectal study [110], RNA sequencing was done to list the differentially-expressed genes. When the tumor samples from AA and CA patients were compared, 34 genes were found to be up-regulated, while 108 genes were down-regulated in AA samples, compared to the CA samples. The researchers compared the two different sub-sets: the list of epigenetically differentially-expressed miRNAs, and the differentially-expressed genes. This was done to possibly locate differentially expressed miRNA(s) and the target gene(s). This led to the realization that in this particular study, hypermethylation of miR-124-3p correlated with the up-regulation of two of its target genes, POLR2B (RNA polymerase II subunit B) and CYP1B1 (cytochrome P450 family 1 subfamily B member 1). Interestingly, miR-124-3p was differentially hypermethylated in AA patients and consequently these two target genes were up-regulated only in the AA patients, and not in CA patients. This observation calls for further scrutiny of POLR2B and CYP1B1 in racial disparity among colorectal AA vs. CA patients.

As mentioned above, methylation of genes is an epigenetic regulation, which controls their eventual expression. Wang et al. [110] also observed CHL1 (cell adhesion molecule L1 like) to be hypermethylated in colorectal tumors from AA patients. Such hypermethylation should result in suppression of this gene. As cross-referenced by the authors of this study, there is evidence for down-regulated CHL1 in up to 48% of colorectal cancers [152]. This suggests that CHL1, which is down-regulated in a large number of colorectal patients, is particularly hypermethylated in AA patients. Whether or not such down-regulation of CHL1 is significantly more prevalent in AA patients, is something that needs to be investigated further. Interestingly, miR-182, a miRNA that targets CHL1, was previously reported by the same group to be differentially up-regulated in AA patients [153], relative to CA patients. While this provides a complete loop that the up-regulated miR-182 in AA patients can possibly be linked to the down-regulated CHL1 in AA colorectal patients, the observed hypermethylation of CHL1 needs to be explored further. The study that identified miR-182 to be up-regulated in AA patients, also identified FOX01 (forkhead box O1) and FOXO3A (forkhead box O3) as its two gene targets that themselves were differentially-expressed in AA patients [153].

2.3.3. miRNAs in Health Disparity among other Cancers

In a study involving thyroid cancer patients [154], miRNA-array profiling identified a panel of miRNAs that was differentially-regulated in formalin-fixed paraffin-embedded tissue specimens from AA vs. CA patients. Two miRNAs (miR-221 and miR-31) were found to be of particular interest with respect to racial disparity. miR-221 was up-regulated in 92% of CA tumors, but only in 40% of AA tumors. miR-31 was up-regulated in all CA tumors, but only in 75% AA tumors. miR-31 and miR-221/222 have been associated with induction of chemoresistance in cancers [155, 156]. In an array-based determination of differentially-expressed miRNAs [157], miR-337-3p was found to be frequently down-regulated in CA endometrial tumors, compared to AA endometrial tumors. A comparison of samples from 9 patients each led to the identification of this differentially-expressed miRNA and it was further validated in an independent set of 24 AA and 23 CA patients. miR-337-3p inhibits tumor progression through the repression of matrix metalloproteinase 14 [158, 159].

The emerging knowledge on epigenetic regulation of cancer racial disparity is complex (Figure 3). As discussed above, miR-182 is up-regulated in AA colorectal cancer patients [153], compared to CA patients, and its target CHL1 is suppressed by hypermethylation in AA patients as well [152]. Thus, CHL1 seems to be regulated at multiple levels in AA patients; through its own methylation and through miR-182. miR-212, on the other hand, is down-regulated in AA prostate cancer patients [135], compared to CA patients, resulting in attenuated repression of its target hnRNP H1. Interestingly, both miR-182 and miR-212 have been reported to be regulated by methylation [160, 161], albeit in different cancer models, and not in the context of cancer racial disparity. Clearly, the information on complex epigenetic regulation, particularly the one involving miRNAs, is fragmented and warrants further thorough investigations.

Figure 3. Implications of differential miRNA expression on racial cancer health disparities.

Expression of miR-182 and miR-212 has been shown to be influenced by methylation. In AA prostate cancer patients, lower miR-212 expression results in abrogation of suppression of its target hnRNP H1. Conversely, a higher miR-182 expression in AA colorectal cancer patients leads to suppression of its target CHL1 which can, additionally, itself be regulated through differential methylation of its own CpG islands. This model describes the complex interplay of multiple epigenetic events, namely, methylation and regulation through miRNAs, in making an impact on the racially disparate cancer outcomes.

3. Race-associated epigenetic changes in healthy individuals

Some individuals might be pre-disposed to developing cancer, and there might be a role of epigenetic changes in this predisposition. Hypermethylation of ADAMTS14 (a disintegrin and metalloproteinase with thrombospondin motifs 14) was observed in the normal colon mucosa of AA colorectal cancer patients [129]. Although preliminary, this observation lands credibility to the hypothesis that differential methylation of individual or a set of genes could predispose individuals to the risk of developing cancer. Since the analysis in this study was limited to ‘healthy’ regions of colon in colorectal cancer patients, it is difficult to ascertain whether or not such hypermethylation events are actual determinants of cancer disposition, or they are just the correlated epigenetic events that associate with tumorigenesis in the larger vicinity.

Cancer racial disparity in methylation patterns in healthy individuals has been a subject of quite a few investigations [162]. This has resulted in some preliminary evidence, often indirect, supporting a differential methylation status of specific genes across ethnic populations. Examples include hypermethylated NKX2 and TIMP3 in healthy prostates of AA individuals [76], lower levels of colorectal methylated ERα and SFRP1 in healthy AA individuals [163] and hypermethylated p16(INK4) in healthy breasts of CA women [164]. In a study that looked at racial differences in DNA methylation in healthy AA vs. CA women [165], a number of distinct hypermethylated CpG sites were observed – 282 in AA and 203 in CA. Interestingly, there was a noticeable difference in the genome regions that were hypermethylated. Whereas CpG hypermethylations in AAs were more common in intergenic regions, the hypermethylations in CA were more prominent in promoter regions. It thus appears that not only is there a differential hypermethylation in AA vs CA populations, there also seem to be differential regions within the genome that are differentially methylated.

A number of epigenetic differences among races can possibly be attributed to true genetic variations, as suggested by evaluation of cytosine modifications in individuals of African ancestry (Yoruba people from Nigeria) vs. those of Caucasian of European ancestry [166]. This study found approximately 13% differentially-methylated CpG sites between the two distinct populations. These observations confirmed a preceding work [167] that reported 13.7% differentially methylated autosomal CpG sites in AA vs. CA newborns. Interestingly, the earlier work [167] also found a number of cancer pathway genes, particularly those related to pancreatic, prostate, bladder and meningioma, among those with differentially-methylated CpGs. No further data is available on the topic, but it will be very interesting to see if the differential-methylations in ‘normal’ tissue(s) correlate with a progression to tumorigenesis. This will open up new avenues for the use of epigenetic biomarkers in prediction of human cancers.

4. Epigenetic differences among other ethnic groups

Not just the CA vs. AA populations, there seem to be epigenetic differences in other populations and ethnic groups as well. For example, it has been suggested that the Chinese bladder cancer patients may have distinct methylated genes, compared to the bladder cancer patients from US [168]. This study compared its own results with two previous ones [169, 170]. One of the earlier studies [169] reported methylation of 5′ regions of multiple apoptosis-associated genes, such as DAPK (death associated protein kinase-1), BCL2, TERT (telomerase reverse transcriptase), RASSF1A and TNFRSF25 (tumor necrosis factor receptor superfamily member 25) in the urine sediments of bladder cancer patients. The other study [170] observed promoter methylation in at least one of the four genes (CDKN2A, cyclin dependent kinase inhibitor 2A; ARF, ADP ribosylation factor; MGMT and GSTP1) of 175 bladder cancer patients. The study in Chinese patients found the predictive power of a set of 11 genes, methylation of which could confirm the existing diagnosis of 87% bladder cancers studied, with a sensitivity of 91.7%. While a number of genes from the earlier US studies were investigated, only BCL2 and RASSF1A managed to make it to the 11-gene list in Chinese patients. It was pointed out that the methylation markers in Chinese bladder cancer patients might be very distinct from the patients in the US.

The discrepancy in the validity of epigenetic events in Chinese vs. Western populations was further confirmed by evaluating the DNA methylation markers for the detection of bladder cancer, where it was found that only hypermethylation of SFRP1 was critical for detection of bladder cancer in Chinese population, as opposed to several other dysregulated methylations observed in other studies conducted on bladder cancer patients from western countries [171]. Looks like the observations from bladder cancer patients are true for lung cancer patients as well because the same research team found a set of genes methylated in Chinese non-small cell lung cancer patients that was exclusively different from the one previously recommended in the studies in western populations [172].

Not just the comparison of Chinese populations with Western populations, there has also been an effort to look at the differential epigenetic markers in ethnic populations within China. In one such study [173], differential methylation of TFPI2 (tissue factor pathway inhibitor 2) was evaluated in Uygur vs. Han ethnicity cervical cancer patients. The incidence of cervical cancer is disproportionately high in Uygur women. However, no difference in TFPI2 methylation was observed between the two ethnicities. A correlation between increased methylation and HPV16 infection was, however, observed in Uygur women, which was absent in Han women. With the known importance of HPV in the etiology of cervical cancer, this comes across as preliminary evidence supporting the epigenetic basis of ethnic disparity, with a possibility of such disparity across races as well.

Currently additional reports on epigenetic differences between other racial/ethnic groups are not available; however, significant differences between the incidence of cancer and/or cancer-associated mortalities between ethnic populations in Asia and Europe Vs. those in US have been reported. These differences have been largely attributed to diet and life-style. For example, the consumption of soy in the East and the utilization of olive oil in the Mediterranean region have been argued as the basis of reduced prostate cancer-associated deaths in these populations [174, 175]. Genistein, a major constituent of soy with significant anticancer properties [176, 177], has been reported to enhance the expression of tumor suppressor genes in an epigenetic mechanism [178]. Similarly, an up-regulation of CB1 tumor suppressor to inhibit colon cancer cells via epigenetic mechanisms by extra-virgin olive oil has also been reported [179]. The consumption of tobacco and tobacco-related products in India is identified as the underlying cause of oral cancer, which contributes to ~26% of the global burden of oral cancer [180]. Tobacco constituents have been identified to cause epigenetic changes in a number of cancers including oral cancer [181]. Thus, these studies provide further in-direct suggestions to potential differences in epigenetic make-up of different racial/ethnic backgrounds, which need to be further investigated.

5. Conclusions and future perspectives

Research on racial health disparities in cancer is a rapidly emerging area of emphasis, and, sure enough, epigenetic basis of these disparities is gaining ground. Early detection is key to successful management of human cancers leading to improved overall survival of patients. Health guidelines call for periodical screening of select cancers, such as breast, prostate and colorectal cancers, all with the aim of diagnosing the disease early. It is envisioned that screening for epigenetic changes, such as DNA methylation status, might be a good strategy [182]. This also has potential of simultaneous screening for multiple cancers because there seems to be a few epigenetic events that are observed across different malignancies. The idea is still in its infancy; not yet well-tested, but worth trying.

As discussed in this article, methylation remains one of the better studied epigenetic events in the context of cancer health disparities. It is evident that a number of genes are differentially methylated in AA cancer patients, relative to CA cancer patients. Some of these genes have either already been proposed as biomarkers for specific cancers, or have been shown to be functionally involved in different stages of cancer progression. While epigenetic regulation of these genes can put individuals of any race/ethnicity at higher risk of cancer, AA populations with seemingly higher degree of basal epigenetic changes in these genes, might start at a disadvantage, and are, therefore, much more pre-disposed to cancer onset, and an even aggressive disease. In most of these studies on differential methylation in AA vs. CA cancer patient samples, the selection of genes is based on their previously reported methylation-dependent regulation of expression. While this provides a good rationale for their evaluation in studies focused on differential methylation, such as those on cancer racial disparity, it remains to be seen whether the differential methylation of genes in racially disparate populations, as observed in these studies, correlates with actual differential expression in these populations. This is an aspect that is not adequately addressed in most of the studies. Further, differential expression of these genes and their products needs to be tied to differential disease outcome in AA vs. CA populations, before the fundamental involvement of epigenetics in racial cancer health disparities can be established.

A number of factors can affect epigenetic events. Minority populations such as AAs and Hispanics often live in socio-economically disadvantaged and low income neighborhoods, and, as discussed above, social, cultural and economic inequalities leave a profound imprint on the epigenome through distinct epigenetic changes. The overall stress-full living conditions also contribute to these changes. Therefore, complex interactions among genes, environment, and the diseases, including cancer, require an examination of how epigenetic changes regulate susceptibility to various stressors. To better understand cancer disparities in disease susceptibility, future studies will need to assess the cumulative effect of various social, and environmental stressors on genetic substrates. It can also be envisioned that the populations that have migrated and inhabited a new geographical location for many generations can show distinct epigenetic traits, relative to the populations at the site of origin. This adds a new layer of complexity to the phenomenon that is already poorly understood. At the very least this means that any new observations have to be analyzed with caution. In the US, a lot of effort has traditionally been directed towards understanding the cancer health disparity between AA and CA populations. But slowly and steadily, data from Hispanic and Native American populations is also being made available. This is still not as robust as the one comparing AA vs. CA populations, but the changing ethnic proportions in the US will undoubtedly result in inclusion of other significant minorities in the studies in due course of future. When looked from the global perspective, the task seems even more challenging. Clearly, large collaborative projects with shared specimens, data and resources will be needed to address this. Alternatively, knowledge gained from simple comparisons of ethnic and racial groups in a population will need to be evaluated in entirely different population for possible comparison, overlap or exclusivity. Yet another challenge in the studies involving race is the way race/ethnicity is determined. Most often, designation of race is based on self-reporting. However, it has been suggested that a better approach would be to evaluate ‘genetic admixture’ which takes into account the intermixing of races that has happened over the last several centuries resulting in individuals/populations with admixed background [183]. This concept was tested recently in a pilot study involving endometrial cancer patients [183] and results from larger cohorts as well as other cancers will be interesting to look forward to.

The increasing interest and advancements in epigenetic connection of racial cancer health disparity are made possible by the recent advancements in techniques that have enabled a critical and quick look at epigenetic events throughout the genome. Emerging techniques such as single cell epigenomics make it possible to characterize epigenetic events within a single cell [184]. Hence, it is envisioned that further refining of techniques will make it possible to evaluate all epigenetic changes, especially those that have not yet been explored, for their relevance in cancer racial disparity.

Highlights.

• Racial health disparities exist among cancer patients

• Epigenetic basis of cancer health disparities is increasingly being recognized

• Racially disparate genes are often differentially methylated

• Regulation involving microRNAs is important for differential cancer racial outcomes