GSTT1 and GSTM1 polymorphisms and myelodysplastic syndrome risk: a systematic review and meta-analysis


  • Issa J. Dahabreh,

    1. Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, Tufts University School of Medicine, Boston, MA, USA
    Search for more papers by this author
  • Stavroula Giannouli,

    1. Department of Hematology, Medical School, National University of Athens, Athens, Greece
    Search for more papers by this author
  • Vaia Gota,

    1. Haematology Research Laboratory, Biomedical Research Foundation, Academy of Athens, Athens, Greece
    Search for more papers by this author
  • Michael Voulgarelis

    Corresponding author
    1. Department of Pathophysiology, Medical School, National University of Athens, Athens, Greece
    • Department of Pathophysiology, Medical School, National University of Athens, 75 Mikras Asias St., 11527 Athens, Greece
    Search for more papers by this author
    • Tel: +30 210 7462648, Fax: +30 210 7462664


Glutathione-S-transferace polymorphisms may make hematopoietic lineage cells susceptible to genotoxicity following exposure to heavy metals or benzene. We conducted a systematic review and meta-analysis to define the effect of GSTM1 and GSTT1 null polymorphisms on MDS risk. We searched the PubMed and SCOPUS databases to identify peer-reviewed published case-control studies investigating the association between GSTT1 and/or GSTM1 null genotypes and development of MDS. Between-study heterogeneity was assessed using Cochran's Q statistic and the I2 statistic. Odds ratios from individual studies were pooled using fixed and random effects models. Thirteen studies were considered eligible for the GSTT1 meta-analysis (1471 cases, 1907 controls) and 10 were considered eligible for the GSTM1 meta-analysis (1161 cases, 1668 controls). For the GSTT1 polymorphism, there was moderate between study heterogeneity (pQ = 0.01; I2 = 52.3%) and the null genotype was significantly associated with increased risk of MDS development, random effects OR = 1.43 (95% CI, 1.09–1.89); p = 0.01. For the GSTM1 polymorphisms there was moderate between-study heterogeneity (p = 0.07; I2 = 43.1%) and the random effects OR = 1.02 (95% CI, 0.82–1.28) was non-significant (p = 0.85). The GSTT1 null genotype is a significant risk factor for MDS development. Gene-environment interactions need to be further explored.

The myelodysplastic syndromes (MDS) are clonal myeloid disorders characterized by ineffective hematopoiesis, bone marrow dysplastic changes and an increased risk of transformation to acute leukemia. The generally accepted mechanism of primary MDS pathogenesis involves an initial deleterious genetic event within a hematopoietic stem cell, subsequent development of an excessive cytokine/inflammatory response leading to a pro-apoptotic/proliferative state, and resultant peripheral cytopenias despite the presence of a hypercellular bone marrow. The presence of detectable cytogenetic abnormalities in approximately 40–70% of patients with primary MDS and over 80% of those with secondary MDS, as well as the validated prognostic value of cytogenetic data have been considered to support the theory of an inciting genetic event.1

Recent studies, have demonstrated that the recurrent cytogenetic abnormalities associated with MDS, previously considered the “primary” cause of disease, are actually “secondary” events that arise due to cytogenetically cryptic initiating lesions in an established clonal hematopoietic stem cell population.2 Such initiating events are likely to be heterogeneous and could either be inherited or result from acquired DNA damage, genomic instability, defective DNA repair, or perturbations in cellular signaling pathways that give rise to stem cell clones with a proliferative or survival advantage. While genetic and familial mapping studies have clearly demonstrated that inherited mutations in specific genes, such as AML1, NF1, or genes mediating DNA repair, can predispose to the acquisition of secondary cytogenetic abnormalities and MDS, it is likely that highly penetrant inherited mutations will account for only a minority of MDS cases.3–5 Furthermore, the majority of MDS cases are “sporadic” and several epidemiologic investigations have demonstrated associations between MDS and exposure to a variety of environmental toxins.6 In light of these findings, a complex MDS pathogenetic model has been proposed, whereby genetic predisposition to the disease, mediated by low penetrance polymorphisms in DNA repair genes, cooperates with environmental carcinogens.7, 8

Among the most extensively studied inherited genetic risk factors for MDS are variants of glutathione S-transferases (GSTs, Enzyme Commission (EC) number Among the GST substrates, there are several environmental carcinogens found in food, air, or medications, such as polycyclic aromatic hydrocarbons, found in combustion products, diet, and tobacco smoke. Polycyclic aromatic hydrocarbons are activated by members of the phase 1 cytochrome P-450 supergene family to epoxide-containing metabolites (e.g., benzo[a]pyrene-7,8-diol-9,10-oxide), which are substrates for the mu, alpha, theta and pi GST classes.9GSTT1 and GSTM1 are good candidate cancer susceptibility genes because of their involvement in the metabolism of chemicals such as methylating agents, pesticides, and industrial solvents. In vitro studies suggest that both GSTT1 and GSTM1 enzymes protect cells from the toxic products of phase 1 detoxification reactions.10 However, GSTT1-catalyzed reactions can also increase the toxicity of some compounds, such as dichloromethane. GSTs also conjugate isothiocyanates, which are potent inducers of enzymes that detoxify environmental mutagens. The conjugation process diverts the isothiocyanates from the enzyme induction pathway into excretion, leading to elimination of these anticarcinogenic substances and thus decreasing their potential chemopreventive effect. The pleotropy of GST enzyme functions indicates that their role in human carcinogenesis is complex.

The most common polymorphism in GSTT1 consists of a deletion of the whole gene (null genotype), resulting in the lack of active enzyme.11, 12 Complete deletion at the GSTT1 locus was hypothesized by observing the phenotypic variation in glutathione-related detoxification of halomethanes by human erythrocytes, resulting in “conjugator” and “nonconjugator” phenotypes. Individuals with homozygous deletions of either the GSTM1 locus or the GSTT1 locus have no enzymatic functional activity of the respective enzyme. This has been confirmed by phenotype assays that have demonstrated 94% or greater concordance between phenotype and genotype. Many genotoxins, including benzene and cytotoxic drugs, act directly in vivo on bone marrow cells.13 Bone marrow cell protection from these insults depends upon intact detoxification pathways and DNA repair systems to prevent cellular mutagenesis/apoptosis.14 Exposure to environmental toxins, such as benzene or heavy metals, has been implicated in the etiology of MDS.15 Therefore, it has been proposed that the decreased production of GSTM1 and GSTT1, characteristic of the null genotypes, may be associated with an increased MDS risk in the presence of environmental mutagens and/or carcinogens exposure.

Individual genetic association studies, such as those that have evaluated selected GST polymorphisms and risk of MDS, are frequently extremely underpowered and often report small or variable effects. We therefore conducted a systematic review and meta-analysis to more accurately define the effect of GSTM1 and GSTT1 polymorphisms on risk for MDS.


FAB: French-American-British; MDS: myelodysplastic syndrome; OR: odds ratio; RA/RARS: refractory anemia/refractory anemia with an excess of blasts; RAEB/RAEB-t: refractory anemia with an excess of blasts/refractory anemia with an excess of blasts in transformation; WHO: World Health Organization

Material and Methods

Study identification, eligibility criteria, and data extraction

We used the PubMed (MEDLINE) and SCOPUS databases to identify genetic association studies published before April 30th, 2009, investigating the association between the GSTT1 and/or GSTM1 null genotypes and MDS. We also used the National Institutes of Health (NIH) Genetic Association Database (GAD, last search: April 30th, 2009). Computer searches were conducted using keywords relevant the genes of interest (e.g., “GSTT1,” “GSTM1” and “glutathione S-transferase”) in combination with words related to “myelodysplastic syndrome” and “myelodysplasia.” To increase the yield of our search, we also scanned the reference lists of all relevant studies and review articles, manually searched relevant journals and consulted with experts in the field of MDS. Case-control studies that employed validated genotyping methods and examined the prevalence of GSTT1 and/or GSTM1null genotypes were eligible for inclusion. Family based studies were excluded owing to different design considerations. The search was limited to studies published in English.

The following information was abstracted from each study: first author, journal, year of publication, study design, matching, geographical location, ethnicity of participants, definition and numbers of cases and controls, source of controls, DNA extraction and genotyping methods, frequency of GST null and normal genotypes, MDS classification criteria (French-American-British, FAB or World Health Organization, WHO), morphological classification (refractory anemia with or without ringed sideroblasts, RA/RARS; refractory anemia with an excess of blasts, RAEB; refractory anemia with an excess of blasts in transformation, RAEB-t).

Evidence synthesis

For GSTM1 and GSTT1, since only two genotypes are possible, we compared genotype frequencies (null versus normal) between cases and controls. All associations were presented as odds ratios (OR) with the corresponding 95% confidence interval (CI). Between-study heterogeneity was assessed using Cochran's Q statistic and inconsistency quantified using the I2 statistic, which describes the proportion of variation in the log odds ratios that is attributable to genuine differences across studies rather than to random error.16 Pooled OR was estimated based on the individual OR using fixed- (Mantel-Haenszel) and random-effects (DerSimonian and Laird) models.17–20 Cumulative meta-analysis was carried out for each polymorphism to evaluate the trend of random effects OR over time.21, 22

Assessment of bias and subgroup analysis

Source of controls (e.g., population versus hospital based) and ethnicity (individuals of East Asian origin versus all other) were pre-specified as characteristics for assessment of heterogeneity by subgroup analysis. The differential magnitude of effects in large versus small studies was assessed using the Egger and Begg tests.23, 24 A test for interaction was used to compare the OR of the first study with the pooled OR of subsequent studies, as well as the pooled ORs between subgroups.25 The following subgroup comparisons were pre-specified: source of controls (healthy versus hospital), MDS classification (FAB versus WHO), inclusion of secondary MDS cases (yes versus no), ethnicity of participants (Caucasian versus East Asian) and morphological diagnosis (RA/RARS versus RAEB/RAEB-t).

Statistical analyses were performed using STATA (version 10/SE, Stata Corp., College Stage Station, Texas). Studies in the figures are listed by the year of publication.


Eligible studies

Out of 98 studies retrieved by our initial search strategy, 13 were considered eligible for this systematic review and were included in the meta-analysis; all studies investigated the GSTT1 and 10 also investigated the GSTM1 polymorphism (Table 1).7, 26–37

Table 1. Characteristics of studies included in the meta-analysis
  1. Studies are listed by year of publication.

  2. Abbreviations: FAB, French-American-British classification; MDS, myelodysplastic syndrome; NA, not applicable; NR, not reported; PCR, polymerase chain reaction; RFLP, restriction fragment length polymorphism; WHO, World Health Organization classification.

inline image

Studies had been conducted over a period of thirteen years (1996 to 2009). The vast majority of the studies (11 of 13) used the French-American-British (FAB) criteria to classify patients and only two used the more recent World Health Organization Criteria (WHO). Ten of the studies were conducted in Caucasian populations, two in East Asian populations, and one in a Latin American population. Hospital controls were used in seven studies and healthy controls (healthy volunteers, blood donors or newborns) were used in six studies. In general, studies were relatively small: mean number of cases was 113.15 (SD= 80.4, range: 13–323) and the mean number of controls was 146.7 (SD = 89.57, range: 43–330). Nine studies did not match cases to controls; controls were matched to cases for age and gender and in three studies, for age, gender and smoking in one study. All studies used validated polymorphism detection methods, such as PCR or PCR-RFLP, with the majority employing multiplex PCR protocols to identify the GSTT1 and GSTM1 null genotypes along with control genes for reaction quality control.

The frequency of the GSTT1 null genotype among controls ranged from 30.2% to 46% in studies of East Asian populations and from 6.7% to 25% in studies of non-East Asian populations. The control group frequency of the GSTM1 null genotype was 53.5% in the one East Asian study that provided information and ranged from 33.3% to 59.4%% in studies of non-East Asian populations.

Evidence synthesis

The studies investigating the GSTT1 polymorphism reported on a total of 1471 cases and 1907 controls. 376 cases (25.6%) had null genotypes compared to 340 (17.8%) controls. There was moderate between study heterogeneity (pQ = 0.01) and inconsistency (I2 = 52.3%). Overall, the GSTT1 null genotype was associated with increased risk of MDS, fixed-effects OR= 1.44 (95% CI, 1.21–1.72; p < 0.0001). Random effects models were also consistent with an association between the GSTT1 null genotype and MDS risk, OR = 1.43 (95% CI, 1.09–1.89; p = 0.01]. Figure 1 demonstrates the results of the meta-analysis of GSTT1 studies.

Figure 1.

Forest plot for GSTT1 null polymorphism and MDS risk; meta-analysis using a random effects model. Each study is shown by the point estimate of the odds ratio (square proportional to the weight of each study) and 95% confidence interval for the odds ratio (extending lines); the summary odds ratio and 95% confidence intervals by random effects calculations is depicted as a diamond. For all figures, values higher than 1 indicate that the null genotype is associated with increased risk of MDS development. [Color figure can be viewed in the online issue, which is available at]

The studies investigating the GSTM1 polymorphism reported on a total of 1161 cases and 1668 controls. There was moderate between study heterogeneity (pQ = 0.07) and inconsistency (I2 = 43.1%). Overall, the GSTM1 null genotype was not associated with increased risk of MDS, fixed effects OR = 0.97 (95% CI, 0.83–1.14); p = 0.69. Random effects models also did not demonstrate an association between GSTM1 genotypes and MDS risk, OR = 1.02 (95% CI, 0.82–1.28); p = 0.69. Figure 2 demonstrates the results of the meta-analysis of GSTM1 studies.

Figure 2.

Forest plot for GSTM1 polymorphism and MDS risk; meta-analysis using a random effects model. [Color figure can be viewed in the online issue, which is available at]

Subgroup analysis

The results of subgroup analysis by ethnicity, control selection and morphological characteristics are presented in Table 2. The relatively small number of studies per subgroup and small number of participants did not allow for accurate estimation of ORs. The point estimates appeared consistent between subgroups and nominal significance was achieved for several subgroups regarding the GSTT1 polymorphisms. In general, ORs for the GSTT1 polymorphisms were consistent across subgroups and favored an increased risk of MDS for patients carrying the null genotype.

Table 2. Results of meta-analysis and subgroup analysis. Significant results are highlighted in bold. All odds ratios (OR) are calculated using random effects models
  1. Abbreviations: FAB, French-American-British; MDS, myelodysplastic syndrome; OR, odds ratio; RA/RARS, refractory anemia/refractory anemia with an excess of blasts; RAEB/RAEB-t, refractory anemia with an excess of blasts/refractory anemia with an excess of blasts in transformation; WHO, World Health Organization.

inline image

Potential for bias

For the GSTT1 polymorphism there was no evidence of small study effects (Begg-Mazumdar's test p = 0.67 and Egger's test p = 0.79). For the GSTT1, the odds ratio of the first published study34 (OR = 4.46, 95% CI 2.53–7.89) was statistically significantly different from the pooled odds ratio of all subsequent studies (random effects OR = 1.27, 95% CI 1.06–1.54; interaction test p < 0.0001). Interestingly, removal of the first study also eliminated between-study heterogeneity (pQ = 0.679; I2 = 0%). Over time the random effects OR for GSTT1 followed a decreasing trend (Figure 3). For the GSTM1 polymorphism there was there was some evidence of small study effects (Begg-Mazumdar's test p = 0.03 and Egger's test p = 0.02) but there was no evidence that the first study differed significantly from subsequent evidence.

Figure 3.

Cumulative meta-analysis for GSTT1 polymorphism and MDS risk using a random effects model. Each point estimate with corresponding confidence intervals presents the meta-analysis results as each study was published. Pooled odds ratios are depicted as circles with extending lines representing 95% CI.


This report is the first meta-analysis examining the effect of GSTT1 and GSTM1 polymorphisms on the risk of MDS. We found the null variant of GSTT1 to be associated with an increased MDS risk. Although less evidence is available, there was no indication for an association between GSTM1 and MDS. These polymorphisms have been investigated as potential risk factors for the development of many solid tumors, and convincing associations have been demonstrated for lung, prostate, colorectal, bladder, head and neck cancer as well as hepatocellular carcinoma.38–43 A systematic review that investigated the association of GST null polymorphisms with acute leukemia, both myeloid and lymphoblastic found that GSTM1 and GSTT1, but not GSTP1 polymorphisms, appear to be associated with an increase in the risk of acute leukemia.44 Previous studies have established that the prevalence of the GSTT1 null genotype is lower among Caucasians (10–20 percent) compared with Asians (50–60 percent).45 MDS in Asian populations occurs at a younger age compared to Western populations, indicating a stronger genetic component.46–48 We speculate that this difference may be partly ascribed to the higher prevalence of the GSTT1 null genotype among Asian populations.

The association of environmental toxin exposure and MDS development strongly supports the existence of gene environment interactions. MDS and acute leukemia form a pathogenetic and clinical continuum, with pro-apoptotic effects and cytopenias on one end and increased proliferation, blocked differentiation and high blast counts on the other. Given its significant association with both acute leukemia and MDS, the GSTT1 null genotype appears as a potential initiating risk factor for both conditions. Based on our analysis, the association of the GSTT1 null genotype with MDS is strengthened by the consistency of results across ethnic and morphological subtypes of disease.

Functional studies provide further support for the association between the GSTT1 null genotype and MDS. Benzene exposure has been associated with a variety of disorders, including MDS.49 The effects of benzene may be mediated by intermediate genotoxic and cytotoxic metabolites that induce DNA damage, including chromosomal aberrations and sister chromatid exchanges.50, 51 Some benzene metabolites accumulate in the bone marrow where they undergo autoxidation, producing highly reactive oxygen species which are responsible for cell damage.52 Both GSTM1 and GSTT1 are involved in the detoxification of benzene oxide to s-phenylmercapturic acid.53 Several studies have examined the role of GSTM1 and GSTT1polymorphisms on biomarkers of benzene toxicity. Kim et al, reported an increased occurrence of chromosomal aberrations in subjects carrying the null genotype of both GSTM1 and GSTT1 relative to the heterozygotes or non-null individuals.51 In addition, the frequency of diepoxybutane-induced sister chromatid exchanges was significantly increased in GSTT1 null subjects.54

This study is limited by the unavailability of individual patient data that would allow the identification of gene-environment interactions as well as the potential association of GST polymorphisms with specific disease characteristics such as cytogenetic abnormalities or somatic mutations. Be that as it may, by pooling the results of all published independent studies this review has increased power to detect the association between GST polymorphisms and MDS risk. Also, by including studies published over a long period of time, it has been possible to demonstrate that, apart from the somewhat exaggerated effect estimate of the first published study, the available evidence supports an association between the GSTT1 null genotype and MDS risk.

In conclusion, this review demonstrates that null polymorphisms in GSTT1, but not in GSTM1, are associated with increased odds of MDS development. Given the strong association between GSTT1 null genotype and MDS risk, as well as the established role of GSTs in environmental carcinogen detoxification and acute leukemia pathogenesis, studies to investigate potential gene-environment interactions are necessary across the myelodysplasia-leukemia spectrum.