Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Thorac Oncol. Author manuscript; available in PMC 2010 December 6.
Published in final edited form as:
PMCID: PMC2998042

Evaluation of Glutathione Metabolic Genes on Outcomes in Advanced Non-small Cell Lung Cancer Patients after Initial Treatment with Platinum-Based Chemotherapy

An NCCTG-97-24-51 Based Study



We evaluated the role of glutathione-related genotypes on overall survival, time to progression, adverse events, and quality of life (QOL) in stage IIIB/IV non-small cell lung cancer patients who were stable or responding from initial treatment with platinum-based chemotherapy and subsequently randomized to receive daily oral carboxyaminoimidazole or a placebo.


Of the 186 total patients, 113 had initial treatment with platinum therapy and DNA samples of whom 46 also had QOL data. These samples were analyzed using six polymorphic DNA markers that encode five important enzymes in the glutathione metabolic pathway. Patient QOL was assessed using the Functional Assessment of Cancer Therapy-Lung and the UNISCALE QOL questionnaires. A clinically significant decline in QOL was defined as a 10% decrease from baseline to week-8. Multivariate analyses were used to evaluate the association of the genotypes on the four endpoints.


Patients carrying a GCLC 77 genotype had a worse overall survival (hazard ratio (HR) = 1.5, p = 0.05). Patients carrying the GPX1-CC genotype had a clinically significant decline in the UNISCALE (odds ratio (OR): 7.5; p = 0.04), total Functional Assessment of Cancer Therapy-Lung score (OR: 11.0; p = 0.04), physical (OR: 7.1; p = 0.03), functional (OR: 5.2; p = 0.04), and emotional well-being constructs (OR: 23.8; p = 0.01).


Genotypes of glutathione-related enzymes, especially GCLC, may be used as host factors in predicting patients' survival after platinum-based chemotherapy. GPX1 may be an inherited factor in predicting patients' QOL. Further investigation to define and measure the effects of these genes in chemotherapeutic regimens, drug toxicities, disease progression, and QOL are critical.

Keywords: Gluthathione metabolic genes, Non small cell lung cancer, Platinum-based chemotherapy

Platinum-based compounds remain the most commonly used chemotherapeutic agents in the treatment of advanced-stage lung cancer patients, and the glutathione (GSH) metabolic pathway is directly involved in the inactivation of platinum compounds.1 Specifically, elevated GSH and/or GSH-dependent enzyme levels (denoted as high or positive) may correlate with inferior treatment response, and inhibition or reduction of GSH and/or GSH-dependent enzymes increases treatment response rates. Our earlier work indicated that 1-year survival rates were 60 to 78% for patients with high and/or positive genotypes compared with 89 to 100% for other types; the survival advantage was greater among advanced-stage patients who carried other types than low-stage patients.2 Particularly, in late-stage patients who undertook platinum-based chemotherapy, a 3-nucleotide repeat polymorphism in γGCS affected survival significantly; those with a homozygous 77 repeat genotype had the worst survival outcome, even after 5 years.3 However, these results have not been replicated in prospective cohort studies nor clinical trials. In the setting of a phase III trial (NCCTG-97-24-51), we evaluated the role of GSH-related genotypes on clinical outcomes and patient quality of life (QOL) in stage IIIB/IV non-small cell lung cancer patients (NSCLC) who were stable or responding to one initial treatment with a platinum-based chemotherapy regimen.

Materials and Methods


Results from the NCCTG-97–24-51 trial have been reported previously.4 Briefly, the trial drug, Carboxy-amidotriazole (CAI), NSC 609974, is a synthetic inhibitor of both nonvoltage- and voltage-gated calcium pathways; more detailed information of this drug and the trial have been described in our earlier report,4 and summarized as follows. CAI inhibits angiogenesis, tumor cell motility, adhesion, and metastatic potential; it also decreases matrix metalloprotein-ase-2/gelatinase A and basic epidermal growth factor in vitro; and it reduces vascular endothelial growth factor and interleukin-8 production in cancer animal models. In a phase I clinical trial, disease stabilization and improvement in performance status was reported in patients with refractory cancers including NSCLC, who were treated with CAI. Patients were 18 years of age or older, histologically or cytologically confirmed stage III or IV NSCLC, and had completed one and only one chemotherapy regimen (with or without thoracic radiation) within the previous 6 weeks with a resulting complete response, partial response, or stable disease. Women who were pregnant, nursing, or not utilizing adequate contraception; patients with untreated brain metastases; and those with planned additional chemotherapy, radiotherapy, or immunotherapy or participation in another phase III trial were not eligible for this trial. All patients signed a written informed consent that had been approved by the Institutional Review Board of the treating institution. Patients were assigned randomly to receive either oral CAI at a dose of 250 mg daily or a placebo after completion of at least 3 and no more than 6 months of one chemotherapy regimen and demonstration of disease stability or regression/response by the World Health Organization criteria. Treatment with the study agent continued until disease progression, unacceptable adverse event (AE), or patient refusal. All AEs were graded according to the National Cancer Institute's Common Toxicity Criteria (version 2.0).5 The UNISCALE6 and the Functional Assessment of Cancer Therapy-Lung (FACT-L)7 were used to assess patient QOL. Patients completed the questionnaires at baseline and after every 8 weeks (2 cycles) of treatment before the assessment by the treating physician.

Between April 1999 and January 2004, 186 patients (CAI 94, placebo 92) were randomized to receive either CAI or a placebo. The primary study results demonstrated that the addition of CAI after chemotherapy did not provide clinical benefit or improvement in QOL over a placebo in advanced NSCLC. The current study was one of the ancillary correlative components to evaluate the effects of patients' inherited variations in drug metabolizing pathways on disease outcomes, either independent of or interactive with the chemotherapeutic agents. Although the trial was terminated too early to reach the anticipated sample size for evaluating potential genomic influence on responses to CAI, the available samples were used to validate our previous findings on the role of the GSH pathway genomic polymorphisms on survival of platinum-drug treated late stage lung cancer patients.

Of the total 186 randomized patients, 145 had initial treatment with platinum therapy (134 had carboplatin alone, 10 had cisplatin alone, and 1 had both), 148 had DNA samples available for genotyping, and 111 had QOL assessed at baseline and week-8 (<50% of patients were evaluable for QOL at the 16- and 24-week time points, due to ending treatment). A total of 113 (76% of 148) patients had initial treatment with platinum therapy and DNA samples genotyped (cohort I), and 46 patients had initial treatment with platinum therapy, DNA samples genotyped, and baseline and week-8 QOL data (cohort II) (Figure 1). All patients were off active treatment at the time of this analysis. In terms of subsequent treatment, 33% received no further treatment for cancer, 34% received chemotherapy (C) alone, 8% received radiation (RT) alone, and 25% received C along with either: RT, surgery and RT, antibody, antibody and RT, tyrosine kinase inhibitor (TKI), TKI and RT, and TKI, antibody, and RT.

Patient cohorts included in the various analyses.



DNA samples were analyzed using six polymorphic DNA markers that encode five important enzymes in the GSH metabolic pathway. Laboratory methods and assay descriptions have been previously reported,2,8 using a polymerase chain reaction-based Beckman single nucleotide polymorphism (SNPstream) genotyping system and TaqMan and ABI fragment analysis performed by the Mayo Clinic Genotyping Shared Resources Laboratory. The contrasting genotypes used in the analysis were GCLC (homozygous repeat 77 versus heterozygous 7* versus **), GPX1 (CC versus other), GSTP1-I105V (AA versus other), GSTP1-A114V (CC versus other), GSTM1 (null versus present), and GSTT1 (null versus present).

Quality of Life (QOL)

The UNISCALE6 and the FACT-L7 were used to assess patient QOL. The FACT-L includes four separate constructs of well-being namely physical, social/family, emotional, and functional, and additional concerns, dealing solely with tumor related symptoms. The questions within each FACT-L construct are summated to obtain a construct score, and a final total score is derived by adding these single summated scores.

Statistical Analysis

Fisher's exact test tests were used to compare the baseline and follow-up characteristics between the different cohorts (overall, cohort I, and cohort II—Figure 1). Overall survival (OS) was defined as the time from randomization to death from any cause. Time to disease progression (TTP) was defined as the time from randomization to the date of first documented disease progression. The distributions of OS and TTP for the genotypes at each locus were estimated using the Kaplan-Meier method.9 Multivariate Cox proportional hazards models adjusted for age, gender, treatment arm, ECOG performance score, stage, and response to initial treatment with platinum therapy were used to evaluate the prognostic significance of the genotypes at each locus on OS and TTP.10 Hazard ratios (HR) and 95% confidence intervals (CI) are reported. The UNISCALE scores and the FACT-L summated scores were transformed to a 0 to 100 scale, with higher scores representing better status. A clinically significant decline (CSD) in QOL was defined as a 10% (i.e., 10 point) decrease from baseline to week-8. Multivariate logistic regression models (adjusting for the same factors as in the Cox models) were used to evaluate the association of the genotypes at each locus on a CSD in QOL, and the incidence of a severe (grade 3 or higher) AE.11 Odds ratios (OR) and 95% CI are reported. All analyses were carried out using SAS version 9.1.3, and p-values <0.05 were considered significant. No adjustments for multiple comparisons were made as these analyses are considered exploratory and hypothesis generating due to the limited sample size.


There were no significant differences in the distribution of baseline characteristics of the different cohorts (Table 1). The median follow-up of alive patients in cohorts I and II were similar to the overall cohort (63.2, 60.5, and 60.4 months, respectively). In cohorts I and II, 94 and 93% of deaths were due to lung cancer, respectively. The median OS (in months) of patients in the 3 cohorts were: 10.9 (overall), 10.6 (cohort I), and 12.4 (cohort II) months, respectively (log-rank p = 0.36).

Baseline Characteristics

Genotypes and Clinical Outcomes

Table 2 provides genotyping distribution of the six DNA markers representing GSH-pathway genetic variation. No statistically significant differences in the incidence of severe AEs were observed for any of the genotypes (data not shown). Figure 2 depicts the Kaplan-Meier survival curves by GCLC-77 carrier status of the patients, with a median survival time of 8 months for patients with the 77 genotype compared with 12 months for patients with other genotypes (7* and **). The results from the multivariate Cox proportional hazards models adjusted for baseline factors (see Methods section) show that, although no significant differences in TTP were observed for any of the genotypes, patients carrying a GCLC-77 repeat genotype (homozygous) had a worse OS (HR: 1.6, p = 0.05 for 77 versus 7* and **). In addition, gender was a significant predictor of OS in all genotype models (with males having a worse prognosis; HR = 1.5–1.7; p = 0.02–0.06). Gender and age were significant predictors of TTP in all genotype models (with males [HR = 1.4–1.5; p = 0.04–0.08] and patients younger than 65 years of age [HR 0.5–0.6; p = 0.01] having a worse outcome).

Kaplan-Meier survival curves for overall survival for glutamate-cysteine ligase (GCLC) variants. a7*” denotes the GCLC repeat variants other than a “7.” b Multivariate Cox Model results.
Genotype Distribution

We also tested combined effects at two loci on OS and TTP outcomes based on our previous findings where GCLC-77 and GSTT1-present together was a significant predictor for OS of lung cancer. Again, although none of the combined genotypes was significant for TTP, we found GCLC-77 + GSTT1-present and GCLC-77 + GPX1-CC/CT were associated with a worse OS. However, the estimated HR, 95% CI, and p value for combined genotypes were all similar to that of GCLC-77 alone, suggesting it is unnecessary to use other SNP markers.

Genotypes and Quality of Life (QOL)

In this exploratory analysis, we systematically assessed the association of each candidate genetic variant and CSD of multiple QOL measures. Table 3 shows the CSD percentages by the genotypes at each locus for UNISCALE and FACT-L (both total and for the listed constructs); except for GCLC, variants in each SNP showed differing association of CSD percentage with varied QOL measures. For example, 48% of patients with GSTM1-present genotype reported a CSD in the UNISCALE versus 19% of those with GSTM1-null; 24% with GSTT1-present versus 11% with GSTT1-null in the total FACT-L; 46% with GSTP1(I105V)-AA versus 20% with GSTP1 (I105V)-other for the functional component; 42% with GPX1-CC versus 9% with GPX1-other for the physical component; and 28% with GSTP1(A114V)-CC versus 14% with GSTP1(A114V)-other for the emotional component.

Clinically Significant Decline (CSD) Percentages for the Genotypes at each Locus for UNISCALE, Total FACT-L, and FACT-L Constructs

When modeling each locus adjusting for the baseline characteristics in a multivariate logistic regression model, only the GPX1 gene showed a consistent association with multiple QOL measures (Table 4). Specifically, patients carrying the GPX1-CC genotype had a CSD in the UNISCALE (OR: 7.5; p = 0.04), total FACT-L score (OR: 11.0; p = 0.04), FACT-L functional well-being construct (OR: 5.2; p = 0.04), FACT-L physical well-being construct (OR: 7.1; p = 0.03), and FACT-L emotional well-being construct (OR: 23.8; p = 0.01). The CI for these associations was relatively wide (possibly due to sparse data). We also systematically evaluated the association of combined SNP loci and a CSD in QOL measures based on the distribution of the CSD percentages by the genotypes at each locus in Table 3 using multivariate logistic regression models. Results of two combined genotypes between GPX1 and GSTP1 are presented in Table 5, confirming the same association as GPX1 alone.

Multivariate Logistic Regression Models of Genotypes at 6 Loci and a CSD in QOL for the UNISCALE, FACT-L Total Score, and FACT-L Constructsa
Multivariate Logistic Regression Model for Selected Genotype Combinations and a CSD in QOL for the UNISCALE, FACT-L Total Score, and FACT-L Constructsa

Other constructs within the FACT-L, i.e., social and family, and additional concerns were not significantly associated with any of the genotypes (data not shown).


Two major categories of targeted therapy are tumor-specific and patient-specific treatment modalities.12 Tumor-specific targets are usually developed based on changes in angiogenesis pathways, apoptosis, proteosome regulation, and cell cycle control; particularly in the recent literature are alterations in the tyrosine kinase receptors and vascular endothelial growth factor.13 Results from our current study support previous reports that genomic variations of patients, representing patient-specific genomic markers, affect OS. Specifically, we have validated our previous findings that variants in the GCLC gene that encodes the rate-limiting enzyme γGCS in the GSH pathway could make a difference in OS of patients who received platinum-based chemotherapy.

As an ancillary correlative study based on NCCTG-97-24-51, which was a negative and early terminated phase III trial,4 we took advantage of the design where most subjects were previously treated with platinum drugs and had well-documented responses, AEs, times to progression, and survival. An a priori hypothesis was that genomic variations representing platinum drug metabolizing functions may predict clinical outcome independent of other known factors. Among the six carefully selected GSH-related polymorphic markers, either SNPs or short-tandem repeats, we confirmed that GCLC repeats variation was significantly associated with OS in this study; specifically, patients with GCLC-77 genotype had a worse OS outcome compared with those with other genotypes. This result is biologically plausible because GCLC encodes the catalytic function of a rate-limiting enzyme in GSH synthesis; the mechanisms where how GSH level and activity affect platinum drug efficacy has been previously reviewed.1

It is not a surprise that we did not find TTP and AEs to be associated with the candidate genomic variations. The NCCTG-97-24-51 trial selection entry criteria excluded those who had progressed on platinum drugs (only patients who were stable or responding were randomized to CAI or a placebo). The AEs (primarily grade 1 and 2) were experienced more often by patients in the CAI arm,4 and it is unlikely that AEs related to CAI would bear a confounding effect on the association of OS and GCLC-77 genotype.

In parallel to the ongoing efforts in effectively treating advanced lung cancer as a disease, QOL of survivors has always been an important concern of patients and care providers, and has been an increasing focus in the medical community. It is inevitable that both the quantity and the QOL of lung cancer patients needs to be carefully evaluated and balanced when declaring a new treatment superior to the existing ones or receiving any chemotherapy. If a feasible test of genomic markers could predict the impact of chemotherapy on major domains of QOL, it could assist patients and physicians in decisions and preparation before choosing treatment plans. In the current study, we conducted exploratory analyses of the potential predictive value using ORs for the same candidate markers under evaluation for clinical outcomes. From multivariate logistic models under each QOL construct, total or subscales, variations at the GPX1 locus showed the most significant association with CSD percentage, with improved OR measures when combined with variations at the GSTP1 gene. Five QOL measures associated with genetic variations are UNISCALE, total FACT-L, physical well-being, functional well-being, and emotional well-being constructs. The Physical construct deals with the aspects like energy, pain, and side effects from treatment; whereas the functional construct deals with mobility and general outlook on life.

Our findings are among the pioneer work in linking genetic variations to QOL measures. To note, NCCTG investigators previously correlated allelic variants in 5-FU metabolizing genes with QOL in colorectal cancer patients.14 Unlike GCLC locus variations and OS, plausible mechanisms and interpretations of the association between genomic variations and CSD percentage are underdeveloped. Although direct causal relationship is far reaching, several indirect “pathways” are postulated. First, our results are based on relatively small samples with wide confidence intervals; chance findings are not ruled out until robust replicate results become available. Second, although the CSD percentage was derived based on patient-reported QOL scores, the CSD percentage in only two specific constructs (physical and functional) along with total FACT-L and UNISCALE are significantly associated with GPX1/GSTP1 SNPs, indicating the underlying physical and functional conditions that drive the decline in these constructs may be associated with these genes. Third, GPX1 and GSTP1 genes may be surrogates for yet to be discovered “QOL genes.” Beyond cautious interpretation of results from a small study and based on limited evidence in experimental models, we hypothesize that there are likely downstream immune response effects modulated by GPX1 expression and activity levels. In transgenic mice that overexpress human antioxidant enzymes (particularly intracellular GSH peroxidase), the mechanism of protection involves the modulation of inflammatory response as well as reduced sensitivity of neurons to cytotoxic cytokines.15 These animals show significant reduction of expression of chemokines, IL-6, and cell death-inducing ligands as well as corresponding receptors. One well known inflammation pathway includes nuclear factor-kappaB (NFkappaB), which can be inhibited by GPX1 expression.16 Nonetheless, the applications of reliable and accurate predictive genetic markers may not require fully understanding their biologic mechanisms.

In conclusion, although the addition of CAI after chemotherapy does not provide clinical benefit or improvement in QOL over a placebo in advanced NSCLC, genotypes of GSH-related enzymes, especially GCLC, may be used as host factors in predicting patients' survival after platinum-based chemotherapy. Meanwhile, GPX1 may be an inherited factor in predicting patients' QOL after platinum-based chemotherapy. Furthermore, investigations to define and measure the effects of these genes in chemotherapeutic regimens, drug toxicities, disease recurrence or progression, and measure the direct or indirect effects of these genes on QOL are critical.


Supported, in part, by NIH research grants CA77118, CA80127, CA25224, and CA84354.

We would like to thank Susan M. Ernst, MA, for her technical assistance with this manuscript.


Disclosure: The authors declare no conflicts of interest.

Presented as abstracts at the 2006 and 2007 Annual Meeting of the American Society of Clinical Oncology.


1. Yang P, Ebbert JO, Sun Z, et al. A role of the glutathione metabolic pathway in lung cancer treatment and prognosis: a review. J Clin Oncol. 2005;24:1761–1769. [PubMed]
2. Yang P, Yokomizo A, Tazelaar HD, et al. Genetic determinants of lung cancer short-term survival: the role of glutathione-related genes. Lung Cancer. 2002;35:221–229. [PubMed]
3. Yang P, Williams B, Adjei AA, et al. Platinum-based chemotherapy, glutathione metabolic pathway genes and lung cancer survival [Abstract of IASLC, 11th World Conference on Lung Cancer, Barcelona, Spain (July 3–6, 2005)] Lung Cancer. 2005;49(S2):S4.
4. Johnson EA, Marks RS, Mandrekar SJ, et al. Phase III randomized, double-blind study of maintenance CAI or placebo in patients with advanced non-small cell lung cancer (NSCLC) after completion of initial therapy (NCCTG-97-24-51) Lung Cancer. 2008;60:200–207. [PubMed]
5. Cancer Therapy Evaluation Program, CTC V2.0. [April 30, 1999]. Available at
6. Spitzer WO, Dobson AJ, Hall J, et al. Measuring the quality of life of cancer patients: a concise QL-index for use by physicians. J Chron Dis. 2001;34:585–597. [PubMed]
7. Cella DF, Bonomi AE, Lloyd SR, et al. Reliability and validity of the Functional Assessment of Cancer Therapy-Lung (FACT-L) quality of life instrument. Lung Cancer. 1995;12:199–200. [PubMed]
8. Yang P, Bamlet WR, Ebbert JO, et al. Glutathione pathway genes and lung cancer risk in young and old age populations. Carcinogenesis. 2004;25:1935–1944. [PubMed]
9. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. JASA. 1958;53:457–481.
10. Cox DR. Regression Models and Life-Tables. J R Stat Soc Series B (Methodol) 1972;34:187–220.
11. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd. New York: John Wiley & Sons; 2000.
12. Molina J, Yang P, Cassivi SD, Schild SE, Adjei AA. Non-small cell lung cancer: epidemiology, risk factors, treatment, and survivorship. Mayo Clin Proc. 2008;83:584–594. [PMC free article] [PubMed]
13. Bunn PA, Thatcher N. Systemic treatment for advanced (stage IIIb/IV) non-small cell lung cancer: more treatment options; more things to consider. Conclusion Oncologist. 2008;13(Suppl 1):37–46. [PubMed]
14. Sloan JA, McLeod H, Sargent D, et al. Preliminary evidence of relationship between genetic markers and oncology patient quality of life (QOL) (Abstract 5) ASCO Proceedings J Clin Oncol. 2004;22:2S.
15. Ishibashi N, Prokopenko O, Reuhl KR, et al. Inflammatory response and glutathione peroxidase in a model of stroke. J Immunol. 2002;168:1926–1933. [PubMed]
16. Li Q, Sanlioglu S, Li SS, et al. GPx-1 Gene delivery modulates NFkB activation following diverse environmental injuries through a specific subunit of IKK complex. Antioxid Redox Signal. 2001;3:415–432. [PubMed]