High-Throughput Detection of Glutathione S-Transferase Polymorphic Alleles in a Pediatric Cancer Population

Polymorphisms of glutathione S-transferase (GST) proteins are correlated with altered risk of many cancers (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20), as well as altered response and toxicity (5, 14, 15, 21, 22, 23, 24, 25, 26, 27, 28, 29) from the treatment of cancer. One of the major functions of GSTs is the detoxification of a broad range of environmental and nonenvironmental carcinogens, including polyaromatic hydrocarbons from second-hand cigarette smoke, chemotherapy such as alkylating agents and anthracyclines, inflammation-associated reactive oxygen species, and metabolism-derived lipid peroxides (1). Marked hereditary differences in substrate specificity and enzyme activity have been identified for four subclasses of the GST family (1): GSTM1; GSTM3; GSTP1; and GSTT1. Polymorphic alleles for each of these genes are summarized in Table 5. These GST polymorphic alleles occur at frequencies that range from at least 2.4–20% to as much as 40–84% (1).

This article demonstrates an inexpensive high-throughput assay for the detection of GST polymorphisms for use with large clinical pharmacogenetic trials. To determine the sensitivity and specificity of this high-throughput assay, we also developed a non-high-throughput genotyping assay using single nucleotide extension (SNE) that allows the identification of polymorphic alleles for GSTM1 and GSTP1. Using these SNE assays, we demonstrate the overall sensitivity and specificity of the high-throughput assay to be 92 and 97%, respectively. Using the high-throughput assay, we demonstrate an increased risk of acute lymphoblastic leukemia (ALL), glial brain tumors, and osteosarcoma for patients carrying non-null alleles of GSTM1 and GSTT1.

Informed consent was obtained from the subject or his/her guardian for peripheral blood DNA, mouthwash DNA, or buccal swab DNA collection on a protocol approved by the University of Utah Institutional Review Board. Guthrie card samples (from which the patient identifiers had been removed) were obtained from the Utah Department of Health under the guidelines of the above-mentioned Institutional Review Board-approved protocol.

Peripheral blood was obtained either through venipuncture or via an indwelling central venous catheter. Patients with WBC counts <1000 cells/mm² were excluded. As an alternative, children above the age of 5 years could participate by donating buccal epithelial cells by way of vigorously swishing a commercial mouthwash (FreshBurst Listerine; Warner Lambert, Morris Plains, NJ) for 60 s before expectoration into a collection container (30). In children <5 years of age, buccal epithelial cells were obtained by swabbing their cheeks with four independent buccal brush strokes (CytoSoft Cytology Brushes, GentraSystems, Inc., Minneapolis, MN). In total, 109 peripheral blood samples, 14 mouthwash samples, 4 buccal swab samples, and 340 Guthrie Card samples were collected and analyzed for the determination of sensitivity and specificity. For determination of pediatric cancer risk, the study accrued 171 of 234 (73%) of eligible pediatric cancer patients ages 0–18 years presenting to the pediatric oncology clinic in the 27 month study period from November 8, 1999 to February 7, 2002. The self-identified ethnicities for this group were Caucasian (88%), Hispanic (11%), and African American (1%). Patients with low-grade gliomas and astrocytomas were collectively analyzed as glial brain tumors.

All biological samples were processed for genomic DNA by the Huntsman General Clinical Research Center (GCRC) at the University of Utah. DNA was isolated from peripheral blood, buccal epithelial cells, or Guthrie cards using commercial protocols and reagents (PureGene Blood, PureGene Buccal Cell, and Generation Capture Card protocols, respectively) purchased from Gentra Systems (Minneapolis, MN). Of note, DNA from buccal swabs and Guthrie cards purification was present in an off-white lysate form, the spectrophotometric measurements of which did not correspond to the samples’ ability to serve as a PCR template using 1–2 μl of the lysate.

GSTM1 alleles were detected by fluorescent, allele-specific PCR using one forward primer, M1F, and two reverse primers, M1RA and M1RB. GSTM1 PCR primer sequences are given in Table 1. Polymorphic nucleotides were placed at the 3′-side of the reverse primers to achieve sequence specificity of PCR amplification. The β-actin gene was coamplified as a reaction control. β-Actin forward and reverse primers sequences are also presented in Table 1. Reaction and thermal cycling conditions for the coamplification reaction are presented in Table 2.

GSTM3 and GSTT1 were coamplified by PCR. GSTM3 alleles were detected directly as a size polymorphism. GSTT1 alleles were detected by the presence or absence of a PCR product. GSTM3 and GSTT1 primer sequences and reaction/thermal cycling conditions are given in Tables 1 and 2, respectively.

Polymorphisms at codon 104 of GSTP1 were detected by fluorescent, allele-specific PCR. Polymorphic nucleotides were placed at the 3′-side of the forward primers to achieve sequence specificity of PCR amplification. Polymorphisms at codon 113 of GSTP1 were detected similarly. Primer sequences and reaction/thermal cycling conditions are given in Tables 1 and 2, respectively.

Through the design of product size differences and fluorescent tag differences for each of the PCR reactions mentioned above, all of the PCR reaction products could be combined and loaded into a single lane on an ABI PRISM 373/377 Sequence Detection System (Applied Biosystems, Foster City, CA). Fragments were sized against the Life Technologies, Inc. 50-500 Tamra size standard (Invitrogen Corporation, Carlsbad, CA) using ABI GeneScan v3.1.2 software. Alleles were automatically scored using ABI Genotyper v2.5 software based on size and color. Allele size and peak area information was captured and exported into Microsoft Excel for peak area analysis. Peaks were scored only if the peak height was >150 relative fluorescent units; otherwise, the peak was considered to be absent (GSTM1, GSTT1) or the reaction was considered to have failed (GSTM3, β-actin, GSTP1). Specificity of GSTM1 and GSTP1 genotyping results was determined by comparing high throughput results to corresponding SNE assays that are described below.

To determine the specificity of the high-throughput assay for nonnull GSTM1 polymorphic alleles, we used a SNE assay that uses the ABI Prism SnaPshot ddNTP Primer Extension kit (Applied Biosystems) to extend a primer 1 bp at the single nucleotide polymorphism (SNP) site. The first step of the SNE assay was to generate by PCR a specific 262 bp fragment from genomic DNA that contains the GSTM1 SNP site. The nucleotide sequence of this PCR product has been confirmed by sequencing. The second step of the SNE reaction was an extension (mini-sequencing) reaction performed on the PCR product using the SnaPshot kit. The SnaPshot chemistry allows a primer immediately 5′ of the SNP site to incorporate one dye-labeled dideoxynucleotide triphosphate (ddNTP), thereby terminating the reaction after one nucleotide incorporation at the SNP site. Both sense and antisense extension reactions were performed as internal validation for the GSTM1 SNP result.

Primer sequences, reaction conditions, and thermal cycling conditions for the GSTM1 PCR reaction are given in Table 3. An aliquot of the PCR product from the first SNE reaction was monitored by electrophoresis on a 3% agarose gel. Subsequently, the nonelectrophoresed PCR product was cleaned of free nucleotides and any single-stranded DNA using ExoSAP-IT (USB Corporation, Cleveland, OH). The cleaned product of the GSTM1-specific PCR reaction was used as a template in a second SNaPshot reaction using a second set of sense and antisense primers that flank the SNP site. The unique dyes for terminator nucleotides A, C, G, and T were dR6G, dTAMRA, dR110, and dROX, respectively. The forward and reverse primers used in the GSTM1 SNaPshot extension reaction yielded 43 and 36 bp products, respectively. Of note, the sense and antisense SNaPshot reactions for GSTP1-104 and GSTP1-113 were combined with the GSTM1 SNaPshot extension reaction for cost savings (the error rate for multiplexed versus singleplex reactions were assumed to be negligible because of the concordance for SNE results and the high throughput assay results.) After the SNaPshot extension reaction was complete, a second clean up was performed to remove all free dye-labeled ddNTPs. Products of both sense and antisense GSTM1 polymorphism SNE were detected on an ABI Prism 373/377 Sequence Detection System (Applied Biosystems) in the same 5% polyacrylamide gel lane used for the GSTP1-104 and GSTP1-113 SNE products, multiplexed by size and color. Results were scored for each sample only if sufficient DNA was present to successfully amplify both the GSTM1 SNE second reaction products (when present); otherwise, a single forward or reverse SNE result reaction was considered to be a failed genotype. A reaction for which an SNE product was not present when a high-throughput GSTM1 product was present was also considered to have failed. Details for clean-ups and extension reactions used in the GSTM1 SNE assay are given in Table 3.

The principals of these SNE reactions are the same as described for the GSTM1 SNE reaction. In the first SNE reactions, the 57-bp region surrounding the GSTP1-104 SNP or the 41-bp region surrounding the GSTP1-113 SNP was amplified by PCR in separate reactions. Aliquots of PCR products were monitored by electrophoresis on a 3% agarose gel. Subsequently, the non-electrophoresed PCR product was cleaned of free nucleotides and any single-stranded DNA using ExoSAP-IT (USB Corporation).

As stated for the GSTM1 SNE reactions, the GSTP1-104 and GSTP1-113-specific PCR reaction products were used as templates in the subsequent SNaPshot extension reactions. The primers used for GSTP1-104 and GSTP1-113 extension reactions were the same as those used in the antecedent PCR reaction. As described above in the section for GSTM1 SNE assay, the GSTP1 extension reactions are multiplexed with the GSTM1 extension reaction and detected in the same electrophoresis gel lane. The size of the fragments produced in the final SNaPshot reaction for 104F, 104R, 113F, and 113R are 31, 27, 19, and 23 bp, respectively. Details for primer sequences, reaction conditions, and thermal cycling conditions used in the SNE assays are given in Table 4. Results for GSTP1-104 and GSTP1-113 SNE products are scored as described earlier for the GSTM1 SNE assay.

Neither the high-throughput assay nor the SNE assays was able to distinguish GSTM1*A or GSTM1*B homozygotes (i.e., GSTM1*A/A and GSTM1*B/B) from corresponding heterozygotes, GSTM1*A/null and GSTM1*B/null, respectively. Therefore, to determine gene dosage for GSTM1, a long-range PCR to confirm or negate the presence of the GSTM1*0 allele was developed by modification of a previously reported protocol (31). Using the Expanded Long Template PCR System (Roche, Mannheim, Germany) and primers M2F10 (5′-AAGACAGAGGAAGGGTGCATTTGATA-3′) and M5R16A (5′-ACAGACATTCATTCCCAAAGCGAC-3′), a 4.8-kb product can be amplified in the presence of a GSTM1*0 allele. Specificity of this 4.8 kb product has been confirmed by sequencing. In this PCR reaction, a 5.2-kb product (or in some cases a polymorphic 5.4 kb product) of the human tissue plasminogen activator gene is coamplified as a positive control using primers TPA7SF primer (5′-GGAAGTACAGCTCAGAGTTCTG-3′) and the TPAREV primer (5′-AGCGGGACGAATCCGATTTCAG-3′). The 25 μl of reaction mixture contained 2–20 ng template DNA, 0.24 μm primers (TPA7SF, TPAREV, M5R16A, and M2F10), 1.9 units of DNA polymerase mix, 350 μm deoxynucleoside triphosphates, 2.75 mm MgCl₂, and 1× Roche Long Range Buffer System no. 2. After a 2-min incubation at 94°C, we performed 10 cycles of 94°C (10 s), followed by 67°C (30 s), followed by 68°C (5 min), then 15 cycles of 94°C (15 s), followed by 67°C (30 s plus an additional 5 s/cycle), followed by 68°C (5 min), then 10 min at 68°C. PCR products were visualized on a 0.65% agarose gel stained with 0.5% ethidium bromide via UV light illumination.

To determine gene dosage for GSTT1, a long-range PCR was developed by modification of a previously reported protocol (31). Two sets of primers were used, one specific for the GSTT1*0 allele and the other nonspecific for the GSTT1/GSTT2 alleles. The latter set served as an internal control for the PCR reaction.

The GSTT1*A-F and GSTT1*A-R primers with sequences 5′-AATGCTTTGTGGACTGCTGAGG-3′ and 5′-TGATGCATGTGAGTGCTGTGG-3′, respectively, generated a 455-bp product in the presence of GSTT1*1 or GSTT2 alleles. The primers GSTT1*0-F and GSTT1*0-R, have sequences 5′-TACAGTTGTGAGCCACCGTACC-3′ and 5′-ATAGTTGCTGGCCCCCTCATT-3′, yielded a 1460 bp product when the GSTT1*0 allele was present. Specificity of this 1460 bp product has been confirmed by nucleotide sequencing. The PCR mixture contained 0.01 μm GSTT1*A-F, 0.01 μm GSTT1*A-R, 0.02 μm GSTT1*0-F, 0.02 μm GSTT1*0-R, 0.25 units of Taq Platinum Polymerase (Invitrogen Corporation), 0.2 nm spermidine HCl, 1.5 mm MgCl₂, 40 mm NaCl, 10 mm Tris-HCl (pH 8.3), 200 μm deoxynucleoside triphosphates each, and 9% DMSO. Reaction volume was 20 μl. After a 5-min incubation at 94°C, we performed 6 cycles of 94°C (30 s) followed by 68°C (30 s with a decrement of 1°C/cycle), followed by 72°C (2 min), then 30 cycles of 94°C (30 s) followed by 63°C (30 s), followed by 72°C (2 min), then 10 min at 72°C. The resulting PCR products were visualized by electrophoresis on a 1% agarose gel stained with 0.5% ethidium bromide via UV light illumination.

Disease risk was analyzed univariately by comparing disease-free patients (controls) versus a specific group of patients with respect to a particular genotype. The controls were not matched to the patients. χ² test was used to detect an association between genotype and disease. Whenever the cell counts were small, an exact test was used. No adjustment for multicomparisons has been made. Multivariate analysis was performed with a smaller subset of the data. The presence of a particular disease was used as a binary response variable, and logistic regression was used for the analysis. The set of explanatory variables was constructed using factorial coding of various genotypes. Backward variable selection procedures and likelihood ratio test were used to select significant variables.

Four representative electropherograms for the high-throughput assay are shown in Fig. 1. Differentiation of the alleles was based on size and fluorescent tag color. Potential peaks read from left to right correspond to the following polymorphisms or control genes: GSTP1-113T; GSTP1-113C; GSTM1*A; GSTM1*B; GSTM1*0 (absent allele); GSTP1-104A; GSTP1-104G, β-actin (a control); GSTT1*1; GSTT1*0 (absent allele); GSTM3*B; and GSTM3*A. As shown in Table 5, the GSTP1 genotype is determined from the combined detections of SNPs at the GSTP1-104 and GSTP1-113 codons.

The sensitivity and specificity of the high throughput assay are presented in Table 6. Sensitivity has been defined as the proportion of successful reactions for each polymorphic GST gene, whereas specificity (for GSTM1 and GSTP1) was defined as the level of concordance with the results of an SNE verification assay. For GSTM3 and GSTT1, specificity was not measured because the length polymorphisms of GSTM3 and the presence/absence polymorphisms of GSTT1 are less prone to errors of specificity than the SNPs of GSTM1 and GSTP1. Sensitivity and specificity for high throughput and SNE assays using mouthwash and buccal swab-derived samples were not calculated as percentages because of the small numbers of samples.

Overall, the sensitivity and specificity of the high throughput assay were 85–99 and 93–100%, respectively, depending on the source of genomic DNA. To our surprise, the high-throughput assay was more sensitive for Guthrie card-derived DNA samples (sensitivity 92–99%) than for peripheral blood-derived samples (85–89%). We speculate that Guthrie card-derived DNA may be more fragmented than peripheral blood-derived DNA and therefore may present an advantage for PCR amplification so long as the template regions are relatively short. Of note, peripheral blood-derived DNA sample results were slightly more specific (98–100%) than Guthrie card-derived samples (93–100%).

Four representative electropherograms for the SNE assay are presented in Fig. 2. SNPs are discriminated both by size and fluorescent tag. From left to right, the peaks correspond to the following SNE primers: GSTP1-113F; GSTP1-113R; GSTP1-104R; GSTP1-104F; GSTM1-M1R; and GSTM1-M1F. Results are determined from the color of each peak, as described in the figure legend.

The sensitivity and specificity of the combined SNE assay were 90–98 and 100%, respectively, depending on the source of genomic DNA. Unlike the high-throughput assay, sensitivity was significantly better for peripheral blood-derived DNA samples (98%) than for Guthrie card-derived DNA samples (90–98%). This degree of sensitivity of the SNE assay presents an opportunity to successfully genotype peripheral blood-derived DNA samples that have failed the high-throughput assay, albeit at a higher per sample cost. Nevertheless, the combination of a first-line high-throughput assay and a second line SNE assay provides a cost-effective means of GST genotyping large cohorts from peripheral blood at high levels of sensitivity and specificity.

Most clinical studies of GSTM1 and GSTT1 polymorphisms to date have been disadvantaged by the inability to differentiate homozygous nonnull GSTM1 or GSTT1 genotypes from heterozygous nonnull plus null genotypes. To overcome this problem, long-range PCR assays to determine whether a GSTM1*0 or GSTT1*0 allele was present, respectively, were developed to calibrate gene dosage determination for the high-throughput assay (Fig. 3). We originally expected to be able to determine gene dosage of GSTM1 and GSTT1 by examining the area under the curve ratios for GSTM1:β-actin and GSTT1:GSTM3 in the high-throughput assay. The long-range PCR assays for GSTM1 and GSTT1 were developed to provide a gold standard for the true gene dosage of GSTM1 and GSTT1. Although high-throughput assay area under the curve ratios corresponded to true gene dosage, area under the curve ratios did not provide a clear cutoff to distinguish between the presences of one versus two non-null alleles for either gene. Nevertheless, the long-range assays described herein are robust and can be used to supplement the high-throughput assay in the determination of GSTM1 and GSTT1 gene dosage.

The availability of Guthrie card samples (newborn blood spotted onto filter paper for newborn screening tests) from which patient identifiers have been removed provides an ideal, unbiased cross-sectional resource for the determination of GST polymorphic allele frequencies. Table 7 shows the frequency of the different GST polymorphic alleles determined using the high-throughput assay and a random sampling of the population in the state of Utah of 340 Guthrie cards of infants born in the year 2001–2002. These allele frequencies differ interestingly from those previously reported for Caucasian population of British/Northern European decent (1, 8, 12, 32, 33, 34). Using Ordinal Logistic Regression, the GSTM1, GSTM3, and GSTT1 allele frequencies are comparable with Northern European populations, whereas GSTP1*B and GSTP1*C frequencies are significantly higher and GSTP1*A frequencies are significantly lower (P = 0.034). Most striking is a 3-fold higher frequency of the GSTP1*C allele [7.5 versus 2.4% (33)], the highest frequency of this allele in any population reported to date. The over-representation of the GSTP1*C allele may be, in part, because Utah is an outbred population, founded by a small cohort of Northern European descendents.

The GST polymorphic allele frequencies for pediatric cancer patients are shown in Fig. 4. Surprisingly, only GSTM1- and GSTT1-polymorphic alleles achieved significance by multivariate analysis in this study (Table 8). In each case, individuals with two null alleles of each gene were associated with the lower relative risk. Specifically, we found a 4.3-fold increased relative risk of acute lymphoblastic leukemia among subjects that carry one nonnull (GSTM1*A or GSTM1*B) allele of GSTM1 over individuals with two null alleles (95% confidence intervals 2.581–12.415 (P = 0.001) and 1.8–10.2 (P < 0.001), respectively]. The relative risk of acute lymphoblastic leukemia was also found to be 2.6-fold higher among individuals with one nonnull allele of GSTT1 as compared with individuals with two null alleles (95% confidence interval 1.1–6.3, P = 0.035). Similar outcomes were found among glial brain tumors and osteosarcoma. The relative risk of glial brain tumors was 4.9-fold higher for subjects carrying one nonnull (GSTM1*A) allele of GSTM1 in comparison to subjects with two null alleles (95% confidence interval 1.5–16, P = 0.009). For osteosarcoma the presence of GSTM1*A or GSTM1*B nonnull alleles resulted in 6.9–16-fold increased relative risk over genotypes with GSTM1*0 null alleles [95% confidence intervals 1.1–42.7 (P = 0.038) and 2.8–92.2 (P = 0.002), respectively].

Polymorphisms of GST enzymes have been correlated with altered risk of many cancers, as well as altered response to and toxicity from cancer chemotherapy. Clinically significant differences in cancer risk have been associated with polymorphic alleles of GSTM1 (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13), GSTM3 (2, 12), GSTP1 (14, 15), and GSTT1 (16, 17, 18, 19, 20), whereas differences in response, toxicity, and outcome of treatment for cancer have been associated with polymorphic alleles of GSTM1 (5, 21) and GSTP1 (14, 15, 22, 23, 24, 25, 26, 27, 28, 29). Most of these results have been retrospective studies; therefore, prospective examinations of risk and response in the context of national trials are likely to lead to the discovery of new relationships between GST genotypes and treatment outcomes.

GST polymorphic alleles occur at prevalences that make them medically and socioeconomically important. The GST polymorphic alleles examined for this study occur in mixed populations at frequencies of at least 2.4–20% and as much as 40–84% (1). The new high-throughput assay presented here provides a low cost but highly accurate GST allele detection method that can be performed not only for sizeable study cohorts using peripheral blood but also for large control populations using readily available materials such as discarded Guthrie cards. The sensitivity (92–99%) of our assay with Guthrie cards suggests that the assay could be used with filter-based, room temperature stable buccal swab DNA collection methods, available commercially (Isocode; Schleicher & Schuell, Keene, NH), which could be conveniently sent via mail without concern over therapy-related low blood counts.

This high-throughput assay has three advantages over existing high-throughput assays. In contrast to other PCR-based GST assays, the protocol does not require restriction endonuclease digestion of the PCR products. In addition, unlike most other assays, the high-throughput assay can reliably differentiate between the nonnull GSTM1 alleles, GSTM1*A and GSTM1*B. Finally, the assay presented here is the first comprehensive, high-throughput method of assessing all four GST polymorphic alleles known to date to be clinically relevant. The only other reported high-throughput PCR assay for GST genotypes, reported by Kristensen et al. (20), does not distinguish between GSTM1*A and GSTM1*B. In addition, the assay by Kristensen et al. (20) only determines GSTP1 polymorphisms at codon 104 of GSTP1, resulting in an incomplete and inconclusive GSTP1 genotype.

A major appeal of our high-throughput assay for detecting GST alleles is the low cost and time savings it offers over other methods. From a practical perspective, our methodology allows us to determine the polymorphic alleles of all four different GST genes for 96 patient DNA samples within ∼8 h at a current cost of approximately $11.46/sample. We have recently converted the assay to the use of capillary electrophoresis, which further speeds the assay without additional cost.

We demonstrate that our assay has sample source versatility because DNA can be derived from a variety of readily available sources, including peripheral blood, Guthrie cards, expectorated mouthwash, and mouth swabs. For samples that fail the high-throughput assay, the SNE verification assay can be performed at a cost of $8.30 or $16.58/SNP, depending upon whether all three or one SNP is assayed.

In a multivariate analysis of pediatric cancer patients that was limited by a small samples size, we have been able to demonstrate an increased risk of acute lymphoblastic leukemia, glial brain tumors, and osteosarcoma for patients carrying nonnull alleles of GSTM1 and/or GSTT1. Our results are consistent with a trend increase in GSTT1 nonnull alleles among pediatric ALL patients in a Portuguese population (35). These results are initially counter-intuitive because GSTM1 and GSTT1 are generally thought to be Phase II detoxifying enzymes responsible for the inactivation of carcinogens. However, GSTT1 also is known to have Phase I activity and the ability to activate carcinogens; Bruning et al. (19) have shown that GSTM1 and GSTT1 non-null alleles appear to increase the risk of trichloroethylene-induced renal cell carcinoma, which may be explained by the fact that GSTT1 is involved in the activation of some halogenated hydrocarbons rather than their inactivation (36).

Previously published reports examining GSTM1 and GSTT1 genotypes and acute lymphoblastic leukemia risk have been contradictory. GSTM1 and GSTT1 nonnull alleles have either had no significant association with risk of acute lymphoblastic leukemia (37), or GSTT1 nonnull alleles have been associated with reduced risk in general populations (38) or population subsets (1). In contrast to many of these studies, our patient population is derived predominantly from a small founder population of Northern European descent, and the homogeneity of our population may allow for the statistical detection for small inherited variations in metabolism. At first glance, one might attribute differences in these and our studies to small sample sizes, ethnic heterogeneity, or the use of case versus population controls; however, the relative importance of GSTT1 polymorphisms may in fact depend on complex interactions between GSTT1 genotype, an individual’s profile for many other polymorphic detoxification genes, and specific local environmental exposures. For these reasons, assessment of cancer risk by genotype may necessarily need to be restricted to specific geographic locations.

We believe that the assay presented in this report will expand the study of GSTs in human disease and will facilitate the incorporation of GST genotypes into the clinical management decision making. Pharmacogenetic applications of this assay include (a) detection of individuals at risk for specific diseases followed by genetic counseling and prevention strategies, (b) tailored therapy for patients likely to have a GST-based altered response to therapy such as worsened response or increased toxicity, and (c) patient-specific utilization of allele-specific small molecule inhibitors reversing chemotherapy resistance among cancers overexpressing certain GST polymorphic alleles.

Our assay has been designed for use in large, prospective clinical trials. An ongoing study at our institution is directed at determining how inheritance of common polymorphic alleles of the GST detoxifying enzymes affects a child’s susceptibility to common pediatric cancers and to determine whether inheritance of unfavorable GST alleles serves as predictors of a child cancer patient’s response to chemotherapy. By studying GST genotypes in the context of large trials, we hope to identify patient populations that will benefit from tailored, pharmacogenetically based preventative and therapeutic interventions.

We thank Dixie Thompson, Shawna Baker, Karen Osborne, and James Kushner from the University of Utah Clinical Research Center, as well as Lauren Reed, for their assistance in the implementation of this study. We also thank Mario Capecchi for his comments and suggestions in the preparation of this manuscript.