Esophageal Cancer (EC) has limited treatment options resulting in poor 5-year survival rates of 15% [1] and it holds one of the highest cancer mortality rates [2]. The leading global histological subtype used to be esophageal squamous cell carcinoma (ESCC), but in Western countries a rapid increase in adenocarcinoma has occurred over the past decennia [3]. To a large extent this development is due to the increase in prevalence of gastroesophageal reflux disease (GERD) and obesity, two significant determinants of esophageal adenocarcinoma (EAC) [3]. Risk factors as smoking and alcohol consumption are significantly related to ESCC and contribute to the still dominating position of this subtype in Asia [3]. Exposure to environmental risk factors alone cannot explain all cases of esophageal carcinoma, genetic predisposition appears to play a role as well. Detoxification enzymes biotransform carcinogens and toxic agents into less active and water-soluble compounds ready for excretion with bile, urine or faeces. The activity of phase I and II enzymes determine the rate of detoxification of carcinogens in cells and tissues.

Glutathione S-transferases (GSTs) are important phase II biotransformation enzymes catalyzing the nucleophilic addition of glutathione to xenobiotics, oxidative stress products and phase I electrophilic and carcinogenic metabolites [4]. Allelic variation in GST genes affects in vivo enzyme activity and subsequently decrease the rate of carcinogen detoxification. The GST family has a widely variable organ distribution and four classes of GSTs; Alpha, Mu, Theta and Pi, are known to have esophageal expression [5]. The high level of esophageal expression, polymorphic nature of the GST genes and the wide range of carcinogenic substrates, make it all very plausible that this superfamily of detoxification enzymes may influence esophageal carcinoma susceptibility.

Many studies addressing this issue originate from Asia and as a result ESCC has been widely studied as summarized in a recent meta-analysis [6]. To a lesser extend EAC cases have been included in studies, along with Barrett’s esophagus (BE). BE is suggested to be the histological precursor of EAC, displaying metaplastic columnar epithelium and sharing GERD and obesity as risk factors [7]. It was demonstrated that there is a decreased GST enzyme activity or expression following the sequence; normal esophageal epithelium, Barrett’s metaplasia, dysplasia, adenocarcinoma [8, 9], suggesting an early etiological role for this enzyme system. Studies examining the role of functional polymorphisms in the GST genes expressed in the esophagus (classes Alpha, Mu, Theta and Pi) in relation to EAC risk are lacking. This study was set out to examine whether (combined) GST genotypes with altered predicted enzyme activities, modified EAC and ESCC risk in Caucasians. We conducted a case–control study between 2002 and 2012 on 440 patients with EC and 592 age and sex matched controls of the Caucasian race.

This case–control study reveals no associations between modified activity GST genotypes and EAC or ESCC susceptibility in Caucasians. Only GSTA1 low or intermediate genoytpe was associated with a non-significant protective effect for both tumour types, whereas the remaining GST classes showed contradictory effect sizes for EAC and ESCC.

Although many case–control studies investigated GST polymorphisms and EC risk, only seven originated from Western populations and reported on EAC risk. These studies were from Europe [16–19], the USA [20] and Canada [21, 22]. Our results contrasts with the findings of Casson et al. [21] and Abbas et al. [16] that the GSTP1 Ile105Val and GSTT1*1*1 genotype increased EAC risk, ORs are 2.5; 1.0 – 6.3 and 13.3; 1.7 – 106.9, respectively. Furthermore van Lieshout et al. [18] and Zendehdel et al. [19] reported on both ESCC and EAC susceptibility and found the GSPP1 variant allele to increase the risk for EAC and ESCC, respectively. However, all studies offer inconclusive results due to very low number of EAC and ESCC cases (n < 100), except for the study of Murphy et al. which included 207 EAC cases [17]. Our study includes a relatively large EAC group and consequently may offer a higher power.

Although there is a degree of inconsistency and population sizes are relatively small (n = 45 and n = 234 [23, 24]), ESCC risk is generally addressed by Asian studies and reported increasing ESCC risk for the homozygous GSTP1105Val and homozygous GSTM1*0 genotypes whereas some studies even conclude them to be independent risk factors for ESCC in China [25–27] and Brazil [28]. The largest studies however (n = 562 & n = 245; [29, 30]), originated from South Africa with conflicting results, as mixed ancestry in African populations are a difficulty for genetic investigators. So although results differ globally, our finding is in accordance with several meta-analyses [6, 31–33], concluding that GST genotypes do not seem EC risk factors, the variant GSTP1 105 genotypes excluded [33]. Zendehdel et al. conducted a Swedish case–control study as well as a meta-analysis, including only studies with Caucasians, and stratified according to histology [19]. Interestingly, this meta-analysis consisted of the Caucasian studies mentioned above [16–18, 20–22] and found that the GSTP1 I105V polymorphism was associated with an increased risk for ESCC (1.4; 1.0 – 2.2) and not for EAC (1.2; 0.9 – 1.6). However, their own data contributed largely to this effect, as their patient numbers consisted 60% of the pooled numbers of the meta analysis.

So according to the literature so far, only the GSTP1 Ile105Val polymorphism seems involved in ESCC etiology. Our genotype-genotype interaction analyses confirmed this premise, as combinations of low and intermediate GSTA1, or low GSTT1, with low and intermediate GSTP1 105 genotypes showed a trend for a decreased ESCC risk. Other studies on interactions mostly examined the GSTM1 null genotype in combination with either CYP2E1[25, 26, 34] or CYP1A1 genotypes [27]. Moaven et al. reported that interactions between GST polymorphisms were not associated with a modifying effect for ESCC in an Iranian population [35], while Wang et al. reached similar conclusions in a Chinese cohort [36]. Both studies however had small patient numbers (n = 148 and n = 107, respectively). Larger study sizes are warranted to investigate the effect of combined GST genotypes, more explicitly combinations with GSTP1 genotypes, and the risk for ESCC. Moreover, GSTP1 is the main GST enzyme expressed in the esophagus [9] and the genetic variants express proteins with a large reduction of enzyme activity [37].

Another important finding clearly depicted by this study, is the contrasting effects for ESCC and EAC risk, possibly related to their different etiologies. An increasing risk for EAC by GST genotypes may be due to a lower detoxification rate of carcinogens. But studies originating from South Africa [30] and Iran [38] correspond with our results and also report protective effect sizes of GST variants for ESCC risk. Matejic et al. explains this by the prevention of glutathione (GSH) depletion, due to a decreased conjugation activity of the GSTs. GSH is a powerful antioxidant and thus an optimal GSH level to protect against oxidants is maintained [30]. However, it remains difficult to clarify the difference in effect size between EAC and ESCC. Substrate overlap between GST isozymes can compensate the decreased detoxification of mutagens attributable to a single enzyme deficiency. Moreover, the risk for EAC and ESCC is probably substrate dependent. Although substrate overlapping occurs for oncogenetic compounds such as Benzo A-Pyrene-Diol-Epoxide, 1,2-epoxy-3-butene and 1,2:3,4-diepoxybutane, there is a degree of specificity per GST isoenzyme. The Mu, Alpha and Pi GSTs conjugate several carcinogenic epoxides like aflatoxin B1 epoxide (a naturel carcinogen) and polycyclic aromatic hydrocarbons diol epoxide such as benzo[a]pyrene (BaP). The Theta class GST is involved in the detoxification of small dihaloalkanes such as dichloromethane which is an important compound used in paint strippers, plastics and pharmaceutical drugs, and also dibromoethane, an anti-knocking agent in gasoline.

Limitation of our study are the relatively small number of cases involved, which after stratification according to tumor histology, results in even lower subgroup numbers and therefore limits the power of the study. For instance, interactions with the GSTP1 variants showed a tendency to modify ESCC risk, but genotype combinations failed to reach statistical significance, which mainly may be a power issue as the especially the ESCC group is small in our Dutch Caucasian population. Another limitation is the lack of data on exposure to mutagenic compounds (smoking, alcohol consumption, etc.) in patients and controls, to counterbalance confounding effects (and additionally examine potential gene-environment effects).

We conducted the largest case–control study so far on GST variant genotypes and esophageal cancer risk in a Western population of Caucasian ethnicity. Although this study did not detect significant associations between altered predicted enzyme activity GST genotypes and EAC or ESCC risk, our results indicate that gene-gene interactions between GSTP1 variants could play a role in EC susceptibility.