Investigations into inherited genetic variations in the DNA code (known as polymorphisms) in the field of oncology have provided preliminary support for an association with cancer risks and outcomes. Early studies have highlighted several genes with this potential predictive and prognostic power. However, these studies have had methodological limitations and have produced inconsistent results, making impractical as yet the routine evaluation of such genetic polymorphisms in general clinical practice. Continued research in this area is essential if we are to be able to soon use genetic polymorphisms to better select patients for targeted anticancer interventions. This review discusses the role of genetic polymorphisms and their association with esophageal cancer risk and prognosis. The article also highlights future directions in this new, emerging field of molecular epidemiology.
Genetic polymorphisms, which are inherited DNA sequence variations observed among different individuals in a population, are common enough to affect at least 1% of a specific population. Single nucleotide polymorphisms (SNPs) are the simplest and one of the most common polymorphisms, accounting for a large proportion of the possible genetic code variations that occur within the human genome. SNPs may fall within coding sequences of genes, noncoding regions of genes, or in the intergenic areas between genes. Of the SNPs that occur within a coding sequence, only a portion of these polymorphisms ultimately contribute to changes in the amino acid sequence of the protein that is being produced. Such SNPs are known as nonsynonymous SNPs and may culminate in phenotypic changes. Conversely, those that do not result in amino acid alterations are synonymous SNPs. Among the SNPs located in noncoding regions of genes (eg, splice sites, promoter regions, or transcriptional binding sites), which are not always actively transcribed, some may still exert a phenotypic effect on the host through either indirect interactions with other genes or regulation of downstream proteins.
SNPs appear to influence the risks and outcomes of certain diseases. In some instances, SNPs predict the degree of response to particular therapeutic interventions. SNP research has been especially prominent in the field of oncology for two main reasons. The first is because prognosis for many cancers remains poor. It is thus hoped that SNP research will lead to better outcomes. SNP research has also received attention in the field of oncology because the therapeutic index of treatments is often narrow and the risks of life-threatening toxicities can be substantial, making the ability to predict risks and outcomes of cancers particularly appealing. Because of the potential predictive and prognostic power that SNPs may offer, many molecular biologists and epidemiologists hope that SNPs will soon make “personalized medicine” possible, enabling interventions to be tailored to specific patient populations. Thus far, most current research has been devoted to comparing SNPs of matched cohorts with the disease of interest to the SNPs of matched cohorts without the disease of interest. These risk analyses have shown promising but inconclusive results. Meanwhile, researchers face the challenge of deciding just which SNPs are relevant enough to study. Only a small proportion of the more than 3 million SNPs in the human genome have demonstrated actual clinical relevance. Even so, there are still many from which to choose.
The presence of other forms of genetic variations, including microsatellite instability and chromosomal insertions, deletions, and duplications, adds another layer of complexity to our understanding of SNPs. Microsatellites are short repetitive DNA sequences that are scattered throughout the genome; variations in the frequency of sequence repetitions cases instability in these sequences—microsatellite instability—is associated with many sporadic and familial cancers. For example, CACACACACA or (CA) 5 is a conventional dinucleotide repeat in a normal cell. By comparison, in the case of an intron 1 EGFR polymorphism, individuals can have between 14 and 23 repeats (ie, [CA] 14 to [CA] 23 ). These unusual patterns correlate with cancer risk. In addition, there exist chromosomal insertions and deletions that can be as small as only a few base pairs in length, but can also be as large as an entire gene, as is the case with glutathione S-transferases (eg, GSTM1 and GSTT1 deletions). Copy number variation refers to duplication (often by many fold) or deletion of a large segment of DNA (>1 kilobases) that can encompass one or more genes. Like SNPs, each of these different germline genetic variations can affect the risk of developing cancer and the prognosis following cancer diagnosis.
The genetic factors that contribute to the pathogenesis of cancer can be either tumor-specific (eg, somatic p53 mutation) or inherited (eg, germline p53 mutation in Li-Fraumeni syndrome). This review focuses on the germline/inherited variations. Inherited genetic factors, in conjunction with their interactions with other clinical variables, can alter the efficacy of the various intracellular pathways that ultimately result in carcinogenesis. For esophageal squamous cell carcinomas (ESCCs), important clinical risk factors include alcohol and tobacco exposure and local physical trauma. Meanwhile, for esophageal adenocarcinomas (EACs), clinical risk factors consist of gastroesophageal reflux disease (GERD), Barrett’s esophagus, and obesity. The interplay between genes and these clinical variables are referred to as gene–environment interactions. Genes and the factors affecting genes represent the foundation of the emerging field of molecular epidemiologic research.
Genetic polymorphisms may act as molecular markers that can provide important predictive and prognostic information about cancers. In fact, numerous polymorphisms in several oncological settings have already been identified, including those involving breast, lung, colon, and ovarian cancers. In this review article, we specifically focus on the clinical and research implications that genetic polymorphisms may pose for esophageal cancers. In particular, we discuss the current state of the literature with regards to genetic polymorphisms and their association with esophageal cancer risk and prognosis, and also highlight some of the potential future directions in this novel field.
Polymorphisms and esophageal cancer risk
Currently, ESCC and EAC represent the predominant tumor types that comprise the majority of esophageal cancer cases. Interestingly, while the incidence of ESCC is declining, the incidence of EAC continues to increase steadily. EAC is now the most common malignant histology affecting the esophagus. In North America, the annual rate of EAC has increased dramatically three- to four-fold in the last 3 decades alone. The prognosis of EAC remains poor, however, with 5-year overall survival rates approximating 10% to 15% only despite the use of local endoscopic treatments, new surgical techniques, and aggressive multimodality approaches that incorporate chemotherapy and radiation.
In addition to the many established clinical risk factors for esophageal cancers, such as alcohol, smoking, and GERD, familial aggregations of esophageal cancers have frequently been described. Similar familial clustering of Barrett’s esophagus, a known determinant of esophageal cancer, has also been observed. Likewise, those with a family history of esophageal cancer have an increased risk of the disease. Whether these patterns simply reflect common environmental exposures among family members or actual inherited predispositions is uncertain, but the patterns suggest that genetic factors likely contribute in part to esophageal tumorigenesis. Moreover, alterations in certain key genes that govern DNA maintenance and repair have already been linked to elevated risks of developing various cancers, including malignancies of the esophagus. The real uncertainty, however, lies in the precise interrelationship between genetic and environmental variables, which is purportedly more complex because only a small proportion of people with either genetic or environmental factors ultimately develop EAC. This observation suggests that there are most likely additional parameters and interactions that are crucial to esophageal carcinogenesis.
There is mounting interest in clarifying the role that genetic factors play in cancer susceptibility. Much of this interest has arisen because of recent advances and more widespread availability of high-volume, low-cost genetic analysis programs as well as easier accessibility to detailed genomic information from the Human Genome Project and other related databases. It is hoped that better understanding of the molecular epidemiologic mechanisms that underlie the risk of developing esophageal cancer will enable clinicians to identify the most susceptible patients. In turn, this group of individuals would be expected to derive the most clinical benefit from strategies aimed at risk reduction, screening, and chemoprevention. An understanding of the genetic basis of esophageal cancer will ultimately facilitate the development of novel therapeutic approaches.
Most molecular epidemiologic research to date has focused on large-scale genome-wide studies exploring the simple association between genes and disease. These studies have been instrumental in the elucidation of specific alleles and gene loci that confer elevated risks for a diverse array of medical conditions. Most of these studies have been developed and performed under the auspices of international consortia or large collaborative groups that permit the inclusion of a large and representative patient population. This design allows for greater statistical power and, consequently, makes generalizations of the results more valid. Unfortunately, studies of genetic risk factors specific to the esophageal cancer setting have been noticeably more limited. Published literature in this area has been primarily confined to small, case-control studies. While some of these preliminary investigations have been quite successful in identifying genetic polymorphisms that warrant further study and in provoking hypotheses-generating findings, only a few putative genes have been analyzed to date. A recent systematic review indicated that fewer than 100 publications and only three small meta-analyses available in the published literature describe the genetic polymorphisms and their associations with esophageal cancer risk. Esophageal cancer is also unique in that the two main histologic subtypes, ESCC and EAC, are quite distinct in most respects, including their epidemiology, clinical features, and pathogenesis. Therefore, prior papers have typically chosen to analyze these histologies separately with the majority devoted preferentially to the study of ESCC.
Esophageal Squamous Cell Carcinoma Risk
Considering that the incidence of ESCC is particularly high in parts of China and Japan, the majority of molecular epidemiologic studies have understandably taken place in the Asian population. Our previous knowledge of the environmental risk factors for ESCC, including smoking, alcohol, nitrosamine exposure, micronutrient-deficient diets, and human papillomavirus infection, has helped to shape many of these molecular epidemiologic studies. For instance, studies have generally evaluated genetic polymorphisms involved in enzymatic processes or intracellular pathways that modulate the effects of these environmental exposures. Table 1 summarizes these findings.
|Positive Association Studies
|Positive & Negative Association Studies
|Negative Association Studies
|Phase I enzymes
|CYP1A1 ∗1/∗2 CYP3A5 ∗1/∗3
|Phase II enzymes
|NQO1 T609C GSTM3 A/B
|GSTM1 A/B GSTT1 Ile/Val GSTP1 Ile/Val mEH His113Tyr
|Enzymes involved in DNA repair, cell cycle, apoptosis
|Fas A670 G Fas-L T844C MDM2 T309 G ECRG1 Arg290Gln ECRG1 TCA4/TCA3 p21 Arg31Ser
|p53 Arg72Pro XRCC1 Arg280His XPD Asp312Asn
|XRCC1 Arg194Trp hOGG1 T911C CCND1 G870A
|ALDH2 ∗1/∗2 ADH2 ∗1/∗2 MTHFR C677 T MTRR A66 G COX2 G765C BRCA2 G203A MMP7 A181 G MMP2 C735 T SHMT1 C420 T TAP2 G379A LMP7 C145A EGFR Arg497Lys with EGF A61 G
|MTHFR C677 T
Enzymes of particular interest have included those that participate in either carcinogen activation or detoxification, such as phase I enzymes (eg, cytochrome P450 [ CYP ] family of enzymes), phase II enzymes (eg, glutathione S-transferase [ GST ] family of enzymes), nicotinamide adenine dinucleotide phosphate hydrogen (eg, quinine oxidoreductase 1 [ NQO1 ] enzymes), microsomal epoxide hydrolase (eg, mEH ); and those that participate in alcohol metabolism (eg, aldehyde dehydrogenases [ ALDH2 ] and alcohol dehydrogenase [ ADH2 ]) and folate metabolism (eg, thymidylate synthase [ TS ] and methylenetetrahydrofolate reductase [ MTHFR ]). Another group of candidate genes that have generated tremendous scientific interest are those involved in cell cycle regulation, DNA repair, and apoptosis (eg, p53 , cyclin D1 CCND1 ), nucleotide excision repair (NER) genes, and base excision repair (BER) genes. All of these genes have been shown to play a critical role in governing susceptibility to carcinogenic exposure, cellular response to these exposures, or subsequent ability for cellular repair and cell programmed cell death following DNA damage. However, none have been consistently shown to affect ESCC risk.
The matrix metalloproteinases (MMP) genes serve as good examples of a pathway that plays important roles in tumor invasion and metastasis through degrading extracellular matrix components. Variations in the DNA sequence in MMP genes may contribute to altered MMP production or activity. Therefore, these genetic alterations may in turn modulate an individual’s susceptibility to cancer, such as ESCC. Indeed, studies in the Chinese population have found significant differences in the genotype and allele distributions of the P574R polymorphism of MMP-9 among ESCC cases and controls. For instance, the P574R GG genotypes were consistently associated with a significantly increased risk of ESCC as compared with the CC genotypes (odds ratio [OR] 4.08).
In a similar manner, epidermal growth factor (EGF) has been implicated in cell proliferation and differentiation, and alterations or overexpression of EGF has been shown to be associated with a higher risk of thoracic and gastrointestinal malignancies. Specifically, genetic polymorphisms in EGFR 497 Arg > Lys and EGF +61A > G have been shown to influence cell cycle progression, apoptosis, angiogenesis, and metastasis. Interestingly, while the EGF +61A/A genotype has been noted to be significantly associated with risk of ESCC (OR 1.70), there is no clear association for the EGFR 497Arg/Arg genotype and ESCC risk. This observation highlights the complexity of genetic polymorphisms in that only a subset of them is ultimately responsible for an effect or association even though all of the genes are involved in the particular carcinogenesis pathway.
One of the most consistent pieces of evidence in support of the significance of polymorphisms in ESCC risk involves the aldehyde dehydrogenase gene ( ALDH2 ), which is the gene responsible for the elimination of acetaldehyde, a product of alcohol metabolism, from the body. Individuals with the ALDH ∗1/∗2 heterozygous genotype experience a substantial increase in ESCC risk with an estimated OR of 3.2. This particular polymorphism, which is prevalent among East Asians but otherwise rare in other populations, codes for an inactive enzyme that results in elevated serum acetaldehyde levels after consumption of alcohol. Clinically, it is also associated with a flushing reaction following alcohol use. More importantly, this polymorphism provides proof of a gene–environment interaction since the increased ESCC risk observed with ALDH ∗1/∗2 heterozygous genotype is strongly dependent on the degree of alcohol consumption. Conversely, the homozygous ∗2/∗2 variant is associated with a lower risk of ESCC, which has been attributed to the complete intolerance to alcohol that occurs among individuals with this genotype.
Another example that illustrates the importance of gene–environment interactions involves the C677 T polymorphism of the gene for methylenetetrahydrofolate reductase ( MTHFR ). The TT and TC genotypes of the MTHFR C677 T were observed to significantly increase the risk of both esophageal squamous cell dysplasia (OR 2.25) and ESCC (OR 1.58) when compared with the CC genotype. Notably, a strong interaction was observed between the TT and TC genotypes and established risk factors for ESCC, such as alcohol drinking, smoking, and family history. As a result, individuals with these genotypes in combination with the risk factors possessed an even greater risk for dysplasia and ESCC. Despite the strength of some of these results, however, few genes and interactions have been consistently shown to correlate with ESCC susceptibility across different studies. Unfortunately, such contradictory findings hinder our ability to derive any definitive conclusions from these data.
Esophageal Adenocarcinoma Risk
In contrast to ESCC, relatively little has been published on molecular epidemiologic patterns for EAC. Results from positive studies and details of the specific polymorphisms are summarized in Table 2 . Approximately 24 genetic polymorphism associations with EAC risk have been reported to date. Consistent with the literature on the classic epidemiologic features (eg, incidence proportion, prevalence rates) for EAC, studies of molecular epidemiology have focused primarily on the North American and European patient populations. Again, the number of cases in the majority of these studies has been small, ranging from as few as 50 to approximately 300. As with ESCC, the polymorphisms selected for study have usually involved DNA repair genes (eg, XPC , XPD , and ERCC1 ), cell cycle control and p53 pathway genes (eg, cyclin D1 and p73 ), and phase I and II enzymes involved in carcinogen activation or detoxification (eg, GSTT1 , GSTP1 , and GSTM 3).
|Gene of Interest
|Single Nucleotide Polymorphism
|Predominant Ethnicity of Patient Population
|Intron 9 poly(AT) insertion
|5’UTR G4A + C14 T
|Deletion ( ∗1 → ∗2 )
|460C/T , +405C/G , and +936C/T
|rs12268840 and I143 V
|C8092A and 118C/T
Vascular endothelial growth factor (VEGF) has also generated substantial interest among researchers. VEGF is a major regulator of angiogenesis in the process of tumor growth and metastasis. Polymorphisms in the VEGF gene have been associated with altered VEGF expression and plasma VEGF levels. As expected, preliminary investigations have discovered that functional polymorphisms in the VEGF gene (eg, − 460C/T and +936C/T ) are correlated with EAC risk. Compared with the +936CC genotype, for instance, the +936CT and TT genotypes were significantly associated with increased risk of developing EAC (OR 1.49). Furthermore, the − 460CT +CC were associated with increased risk of EAC only in smokers (OR 1.57), while the − 460CT +TT were associated with decreased risk of EAC (OR 0.47) only in nonsmokers. Compared with nonsmokers with the +460TT , smokers with the +460CT +CC had significantly higher risk of EAC (OR 3.32), confirming the presence of a gene–environment interaction. Similarly, studies of the EGF A61 G polymorphism indicate that the G/G genotype poses an increased EAC risk with an OR of 1.81. This risk is even higher in the subgroup of EAC patients with concurrent GERD or Barrett’s esophagus (OR 2.18), again pointing to the presence of a gene–environment interaction.
The DNA repair protein O(6)-methylguanine-DNA methyltransferase (MGMT) is the major cellular defense against alkylating DNA damage. Each of the five different MGMT SNPs (eg, rs12269324 , rs12268840 , L84F , I143 V , K178R ) conferred increased risks of EAC. Strong associations were found for the two variant MGMT alleles, rs12268840 and I143 V . Providing additional support for the significance of gene–environment interactions, homozygous carriers of MGMT rs12268840 who also suffered frequent acid reflux had significantly higher risks of EAC (OR 15.5). There were similar interactions with smoking. Other DNA repair genes have also been implicated. The homozygous variant ERCC1 genotype has been associated with reduced risks of EAC, while the homozygous variant XRCC1 genotype has been observed to confer higher risks of EAC.
Unfortunately, much like studies of ESCC, studies of EAC have shown conflicting results. The lack of consistency among the various studies that often have examined the same genetic polymorphisms cannot be easily dismissed and has made the integration of the new genetic knowledge difficult. While some postulate that the variations in findings are due to possible differences in patient populations across studies, they may also be artifacts of the small sample sizes that invariably lead to high rates of false positivity. Conversely, a number of genetic polymorphisms have been reported to show no association with EAC susceptibility, but these may represent false-negative results, given the modest samples sizes of these studies.
Polymorphisms and esophageal cancer prognosis
A technique to accurately estimate the outcomes of patients with esophageal cancer and to predict who might fare better with more aggressive therapies would be a powerful tool. Understandably, this possibility has generated tremendous interest among clinicians. Research into possible relationships between different genetic variants and survival among ESCC and EAC patients constitutes a new, emerging field of study. Current research aims to identify candidate genes that can be evaluated rigorously and then validated in multiple datasets. The ultimate hope is that these various genes can be used to develop prognostic models (eg, nomograms) that, in conjunction with clinical information from patients, can be summarized to provide patients and physicians meaningful prognostic information so better decisions can be made regarding management and supportive-care issues.
Overall survival rates for esophageal cancer continue to be in the range of 10% to 15% despite the aggressive use of local treatments, surgery, chemotherapy, radiation, or a combination of modalities. In addition, patients who undergo these treatments frequently suffer tremendous morbidity and complications, including local toxicities (eg, dysphagia, esophagitis, disfigurement) and systemic side effects (eg, weight loss, depression), all of which can diminish quality of life. Many people could benefit from a better understanding of genetic polymorphisms to potentially identify a priori individuals who might have the best chance at survival and therefore derive the most clinical benefit from treatment. Outcomes of particular scientific interest for molecular epidemiologic studies should include overall survival, recurrence and progression-free survival, response to treatment, and early and late toxicities stemming from chemotherapy and radiation.
The development of prognostic models based on genetic polymorphisms will be challenging. While a certain degree of overlap can be expected, it is possible for certain variants to have a positive effect on one outcome (eg, survival) but an opposite effect on another (eg, toxicity). For instance, nucleotide excision DNA repair pathway genes, which confer improved DNA repair capacity, may make cisplatin-based chemotherapy for non–small-cell lung cancer and ovarian cancer less effective. At the same time, those same DNA repair pathway genes may help reduce the severity of side effects from exposure to platinum agents.
The current literature related to polymorphisms and esophageal cancer prognosis is limited to a handful of studies ( Table 3 ). Similar to studies of esophageal cancer risk, studies of esophageal cancer outcomes have been generally divided into groups based on histology and ethnicity, namely (1) analysis of genetic polymorphisms and EAC outcomes among whites and (2) analysis of polymorphic variants and ESCC outcomes among Asians. Not surprisingly, most studies to date are again limited by small sample sizes in the form of case series and retrospective cohort studies. The majority of these studies selected overall survival and disease-free survival as the main outcome measures. In almost all of these cases, the choice of which polymorphism to analyze was determined by using preliminary information available from generic cancer-risk studies. In some cases, the decision was based on previous knowledge of potential intracellular pathways and targets believed to play important roles in carcinogenesis.
|Gene of Interest
|Single Nucleotide Polymorphism
|TSER/ 6 base pair del 3′ UTR
|2 or 3 homozygous variants of TSER , 6 base pair del 3’UTR , and Ile105Val had better prognosis
|Intron 2 long/short
|Short allele had poorer prognosis
|No prognostic significance
|No prognostic significance
|A/G and G/G more responsive to chemoradiation
|C/C had reduced survival
|6 base pair del 3′ UTR
|Deletion had nonsignificant improved prognosis
|G/A and A/A had better prognosis and combined variants had better prognosis
|TS and MTR
|“At-risk” allele combinations had worse prognosis
|C/ C and C/T had improved prognosis
|Decreasing number of “at-risk” alleles had better prognosis
|A/A and G/A had worse prognosis
|No prognostic significance
|No prognostic significance
|Ile/Val and Val/Val had worse prognosis
|G/C in TSER
|2R/3 G (vs 3R/3R ) genotype had 11-fold increase in lymph node metastasis in ESCC patients
|6 base pair del 3’UTR
|No prognostic significance
|No prognostic significance