Prospective study of urinary excretion of 7-methylguanine and the risk of lung cancer: Effect modification by mu class glutathione-S-transferases



Nitrosamines are mainly mutagenic through methylation of DNA. 7-Methylguanine (m7Gua) is a product of base excision repair and spontaneous depurination of such lesions in DNA and a metabolite from RNA. Associations between urinary excretion of m7Gua and risk of lung cancer were examined in a population-based cohort of 25,717 men and 27,972 women aged 50–64 years. During 3–7 years follow-up 260 cases with lung cancer were identified and a subcohort of 263 individuals matched on sex, age and smoking duration was selected for comparison. Urine collected at entry was analyzed for m7Gua by HPLC. Effect modification by glutathione-S-transferases GSTM1, GSTM3, GSTT1 and GSTP1 was investigated. We found higher excretion of m7Gua among current smokers than among former smokers. The IRR (incidence rate ratio) of lung cancer was 1.20 (95% CI: 1.00–1.43) per doubling of m7Gua excretion in unadjusted analysis and 1.12 (95% CI: 0.93–1.35) after adjustment for smoking status, intensity and duration at entry. This association was mainly present among current smokers. Comparing the highest with the lowest tertile of m7Gua excretion the IRR of lung cancer was 1.75 (95% CI: 1.04–2.95) irrespective of genotype and 2.75 (95% CI: 1.33–5.81) in subjects with GSTM1 null genotype. If not caused by residual confounding by smoking a possible association between m7Gua excretion and lung cancer supports the importance of methylation of guanine. The finding of an association between m7Gua excretion and lung cancer risk mainly among current smokers and subjects with GSTM1 null genotype supports causality in this respect. © 2007 Wiley-Liss, Inc.

The causal factors of lung cancer include active smoking, environmental tobacco smoke, various occupational exposures and probably ambient air pollution.1, 2, 3, 4 Tobacco smoke contain more than 50 known carcinogens including polycyclic aromatic hydrocarbons, aromatic amines and tobacco-specific nitrosamines (TSNA), such as 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK). TSNA are formed by nitrosation of nicotine and secondary amines during curing and smoking of tobacco.5 Upon metabolic activation TSNA can form DNA adducts of which N7-methylguanine (m7Gua) is by far the most abundant, representing 70–90% of methylation products.6, 7, 8, 9, 10 Whereas m7Gua is innocuous in DNA, the levels increase dose-dependently in DNA upon administration of methylating agents, correlating with pro-mutagenic and carcinogenic methyl-adducts, such as O6-methylguanine and methyladenine,11, 12, 13 and may thus serve as biomarker of exposure to methylating agents. Indeed, a physiologically based pharmacokinetic model for interspecies dose extrapolation of dimethyl sulphate has been based on m7Gua adducts in the nasal cavity.14

Glutathione-S-transferases (GST) are a superfamily of genetically polymorphic enzymes detoxifying carcinogens, including many of those from tobacco smoke. The mu class of GSTs includes GSTM1 and GSTM3, whereas GSTT1 and GSTP1 belong to the theta and pi classes, respectively. In the most recent meta-analysis based on almost 20,000 cases the GSTM1 null genotype was associated with a relative risk of lung cancer of 1.18, although this was reduced when restricted to 5 large studies or to subjects of European origin. The GSTT1 null genotype showed similar albeit weaker results.15 No significant associations were found for the GSTP1 and GSTM3 variant alleles in relation to lung cancer risk.15 Interestingly, however, increased sensitivity to the genotoxic effects in terms of chromosomal aberrations and sister chromatid exchange of NNK was found in cell lines with deletion of GSTM1.16 Moreover, the level of m7Gua-adducts was found to be higher in bronchial lavage cells from smokers than in such cells from nonsmokers, whereas the levels were further increased in subjects with the combined GSTM1 null, GSTT1 null and GSP1 ile/ile genotype.9 In contrast, m7G-adducts were lower in normal bladder DNA from patients with tumors induced by Schistosoma haematobium infection and with GSTM1 null genotype as compared with corresponding patients with the wild type genotype, whereas the GSTT1 and GSP1 genotypes showed no effects.17 Nevertheless, GSTM1 null and GSTM3 and GSTP1 variant genotypes may convey increased sensitivity to methylating agents, such as nitrosoureas in chemotherapy.18, 19, 20 This could be due to importance of detoxification of the methylating species by glutathione conjugation.

In DNA m7Gua is repaired by excision by methylpurine glycosylase or spontaneously, liberating free m7Gua for urinary excretion. Excretion of m7Gua was increased in laboratory animals after exposure to methylating agents21, 22, 23, 24 and the urinary excretion of m7Gua has been shown to be higher among smokers than among nonsmokers.25 Moreover, the excretion decreased by 54% upon smoking cessation.26 The excretion has also been reported elevated in a small number of patients with colon cancer,27 although not in patients with gastric cancer.28 m7Gua is also a metabolite from RNA,28, 29, 30 which affects interpretation as biomarker of exposure to methylating agents. On the other hand, the validity of a biomarker also includes the ability to predict outcome, e.g. cancer. Moreover, effect modification by relevant detoxifying enzymes may support causal relationships.31 So far, the level of bulky adducts to leukocyte DNA is the only biomarker of exposure validated in terms of prediction of cancer risk in several studies,32, 33, 34 although effect modification by GST enzymes in this respect has not yet been found.35

In this study, we examined the association between the urinary excretion of m7Gua and risk for lung cancer in a large population-based Danish cohort for which detailed information on smoking patterns and other lifestyle factors at enrolment is available. Moreover, we assessed potential effect modification by GSTM1, GSTM3, GSTT1 and GSTP1.

Material and methods

Study population

Diet, Cancer and Health is a Danish prospective follow-up study approved by the scientific ethics committees. Participants invited were 160,725 individuals aged 50–64 years, of which 57,053 individuals with no previous cancer diagnosis were recruited.36 All participants were born in Denmark and virtually all were Caucasians (race was not registered). At enrolment (1993–1997), detailed information on diet, smoking habits, lifestyle, reproduction, medical treatment and other socioeconomic characteristics and environmental exposures, including second hand smoke, were collected. Body weight and height were measured and spot urine samples were voided at the clinic and stored at −150°C.

Among the 57,053 persons recruited to the Diet Cancer and Health study, 542 were later registered in the Danish Cancer Register as having cancer diagnosed before the date of enrolment and were therefore excluded. Among the 56,511 individuals with no previous cancer diagnosis, we excluded 2,291 individuals for whom information regarding smoking habits was incomplete, inconsistent or missing. Among the remaining 54,220 cohort members included in this study, 265 cases of lung cancer, diagnosed between 1994 and 2001, were identified in the files of the nationwide Danish Cancer Registry.37 The three most common histological diagnoses were adenocarcinoma (81 cases), small cell carcinoma (55 cases) and squamous cell carcinoma (51 cases). Also from among the cohort members, a subcohort was selected. For this procedure, we divided the entire cohort (inclusive cases) in strata defined by sex, year of birth (5-year intervals) and duration of smoking (10-year intervals). For the subcohort, we sampled 272 persons (including 4 of the lung cancer cases) such that the number of subcohort members in the different strata approximated the number of cases. We used a random procedure to select subcohort members within each stratum. Urine was available for 523 of the 533 selected individuals, and laboratory analyses of urine failed for one sample, leaving 522 persons (251 cases and 261 subcohort members) for analyses of lung cancer risk in association with urinary m7Gua excretion.

Determination of m7Gua and GST genotypes

The urinary concentration of m7Gua was determined by high-performance liquid chromatography using anion exchange separation and detection at 305 nm as described elsewhere.38 The reliability of the assay was confirmed by collection of the peak corresponding to m7Gua and reinjection on a HPLC system connected with a photo diode array UV detector. This confirmed the absorbance spectrum of the collected fraction to be identical with that of a pure m7Gua standard. During the assay of the current samples, typically a series of around 100 samples from a random and blinded mix of case and subcohort were analyzed in one sequence. During each analysis series, 6 samples of an m7Gua control (10 mg/L) were also analyzed. The coefficient of variation of 6 standard samples varied between 1.3% and 3.1% in separate analysis runs. The interday coefficient of variation for 60 urine samples was 7%. All samples have been subjected to 2 freeze-thaw cycles. The urinary concentration of creatinine was determined by a standard colorimetric method.

Intra-individual variation in excretion of m7Gua has been assessed in 10 individuals with 2–5 repeated spot urine samples collected within one month. The average intraindividual coefficient of variation in m7Gua per creatinine was 16% (unpublished data).

The genotypes of GSTM1 and GSTT1 were determined as the homozygous deleted null genotype or the nonnull genotype,39 whereas the GSTM3 3-base deletion in intron 6 (*B allele) and the GSTP1-105 single nucleotide polymorphism (val-ile) were assessed by PCR-based reactions as described elsewhere.39

Statistical methods

The data were sampled according to the case-cohort design within the sampling strata.40 According to the sampling strategy, the unweighted case-cohort approach41 was used to estimate incidence rate ratios (IRR) for lung cancer using a Cox proportional hazards model stratified according to the sampling strata. Age was the time axis, which ensured that the estimation procedure was based on comparisons of individuals of the same age. The analyses were corrected for delayed entry, such that persons were considered at risk only from the time of enrolment in the cohort. We calculated 95% confidence intervals (CI) and p-values based on Wald's test of the Cox regression parameter on the log rate ratio scale using robust estimates of the variance–covariance matrix.42

Interactions between the genotypes and the urinary excretion of m7Gua per creatinine were analyzed by estimating linear effects of m7Gua for each genotype and then test of the null-hypothesis that the effect of m7Gua was the same for the 2/3 genotypes. The hypothesis of a linear association was evaluated using a linear spline with 3 boundaries, placed at the quartiles among cases, as covariates in the Cox model.43 The linearity was evaluated graphically and by a numerical test using the likelihood ratio test statistic to compare the model assuming linearity with the linear spline model. We found that there was no statistically significant deviation from linearity for the log transformed urinary excretion of m7Gua per creatinine.

IRRs were also calculated for tertiles based on the concentration of m7Gua per creatinine. All analyses were also performed excluding cases diagnosed within the first year of enrolment and with adjustment for storage time of the urine samples.

Differences in urinary excretion of m7Gua between subgroups defined by GST genotype, sex, age and smoking were tested by univariate and multiple linear regression analyses. Associations between excretion of m7Gua and daily consumption of rye bread, fish and the energy percents of carbohydrate, protein and fat in the diet were tested by linear regression. SAS version 8.2 was used for all analyses.


Only 8 never-smokers of whom 6 were women developed lung cancer during the follow up period. There was no difference with respect to exposure to second hand smoke between never-smoking cases and subcohort members. Among the cases and the subcohort 55% and 56% were men, respectively, as compared with 48% in the total cohort.

Potential determinants of the excretion of m7Gua are summarized in Table I. Current smokers excreted more m7Gua than former smokers but no clear differences were evident between high and low smoking intensity among the current smokers, whereas never-smokers had the highest excretion. Women had higher levels of m7Gua corrected for creatinine than men did, whereas the oldest age group (60–64 years) had higher levels than the youngest group. In multiple regression analysis the m7Gua per creatinine in urine was significantly associated with sex (p < 0.0001) and smoking status (p = 0.001) as categorical variables and with age as a continuous variable (p = 0.04). There was no dose-response relationship between smoking and m7Gua excretion. None of the GST genotypes had significant influence on the m7Gua excretion. The excretion of m7Gua was inversely associated with the daily consumption of rye bread (Rs: −0.2, p < 0.0001). This association was independent of consumption of fruits and vegetables and smoking, whereas sex explained a part of it. There were no significant associations with consumption of fish and the energy percents of carbohydrate, protein and fat.

Table I. Determinants of Urinary Excretion of m7Gua Per Creatinine at Enrolment Among Cases of Lung Cancer and Subcohort from the Danish Diet Cancer and Health Cohort
 NUrinary m7Gua per creatinine (μmol/mmol) Median (5–95% percentiles)p
Disease status  0.18
 Cases25118.6 (7.6–98.3)
 Subcohort26117.8 (8.1–105)
Gender  <0.0001
 Male28414.8 (7.2–56.5)
 Female22425.0 (9.1–158)
Age  0.19
 50–54 years12317.8 (7.2–155)
 55–59 years14816.5 (8.4–74.2)
 60–64 years23719.8 (8.0–112)
Smoking at enrolment  <0.0001
 Never smokers1521.5 (8.2–158)
 Former smokers9414.1 (6.8–33.3)
 Current smokers39919.2 (8.8–125)
  Of ≤18 g tobacco per day18219.1 (7.5–160)0.78
  Of >18 g tobacco per day21719.2 (9.6–105)
GSTT1 genotype  0.06
 Present45018.6 (8.1–112)
 Null5216.1 (6.7–79.1)
GSTM1 genotype  0.63
 Present21218.4 (7.6–109)
 Null29018.3 (8.2–105)
GSTM3 genotype  0.51
 *A/*A36818.6 (8.1–112)
 *A/*B11917.3 (7.4–90.2)
 *B/*B1114.0 (7.8–155)
GSTP1-105 genotype  0.34
 val/val22716.9 (7.5–98.3)
 val/ile22419.2 (8.2–125)
 ile/ile5419.8 (10.0–112)

Overall, the unadjusted analysis showed that the IRR for lung cancer increased by 20% per doubling of the m7Gua excretion, which was of borderline statistical significance and mainly present among current smokers at entry (Table II). This association was weakened and not statistically significant after adjustment for smoking status, intensity and duration at entry. However, the risk of lung cancer was significantly higher in the upper m7Gua excretion tertile with a dose-response relationship even after adjustment for smoking intensity and duration (Table III). In analysis stratified according to tumor histology the excretion of m7Gua appeared to be particularly associated with the risk of small cell carcinoma (Table II).

Table II. Incidence Rate Ratios (IRR) and 95% Confidence Intervals (95% CI) for Lung Cancer Per Doubling of Urinary Excretion of m7Gua/Creatinine (μmol/mmol) Stratified for Smoking Status at Enrolment, Sex and GSTM1, GSTM3, GSTT1 and GSTP1-105 Genotype
IRR95% CIp interactionIRR95% CIp interaction
  • *p-value for interaction.

  • 1

    Adjusted for smoking status, intensity and duration at entry.

All251/2611.201.00–1.42 1.120.93–1.35 
Smoking stratified       
 Never smokers8/80.940.33–2.69  
 Former smokers36/581.030.65–1.62 0.970.61–1.56 
 Current smokers217/1981.140.94–1.39 1.110.90–1.36 
Histology stratified       
 Adenocarcinomas55/2571.080.77–1.51 1.030.71–1.49 
 Squamous cell carcinomas81/2571.150.89–1.49 1.090.82–1.45 
 Small cell carcinomas51/2581.421.09–1.85 1.310.96–1.80 
GSTT1 genotype stratified   0.90  0.70
 Present212/2421.221.01–1.47 1.150.94–1.40 
 Null37/151.260.75–2.13 1.280.75–2.18 
GSTM1 genotype stratified   0.20  0.18
 Present112/1011.050.81–1.36 0.980.74–1.29 
 Null137/1561.301.04–1.62 1.230.98–1.56 
GSTM3 genotype stratified   0.71  0.73
 *A/*A178/1921.150.94–1.41 1.080.88–1.34 
 *A/*B65/561.320.88–1.99 1.220.80–1.87 
 *B/*B3/81.390.76–2.54 1.340.72–2.52 
GSTM1/GSTM3   0.04  0.04
 Present/*A/*A66/570.910.67–1.25 0.830.59–1.16 
 All others180/1971.321.08–1.62 1.261.01–1.56 
GSTP105 genotype stratified   0.59  0.51
 val/val116/1151.341.01–1.78 1.260.94–1.71 
 val/ile102/1221.120.88–1.42 1.030.79–1.34 
 ile/ile31/231.200.75–1.92 1.200.75–1.94 
Table III. Incidence Rate Ratios (IRR) and 95% Confidence Intervals (95% CI) for Lung Cancer According to Tertiles of Urinary Excretion of m7Gua and Stratified for GSTM1 Genotype
IRR95% CIpIRR95% CIp
  • 1

    Adjusted for smoking status, intensity and duration at entry.

  • 2

    T1 = 2.46–15.4 m7Gua/creatinine (μmol/mmol).

  • 3

    T2 = 15.5–24.9 m7Gua/creatinine (μmol/mmol).

  • 4

    T3 = 25.0–355 m7Gua/creatinine (μmol/mmol).

 T1284/1111  1  
Current smokers
 T1259/751  1  
GSTM1 present
 T1242/481  1  
GSTM1 null
 T1240/611  1  

The association between m7Gua excretion and lung cancer risk appeared mainly to be present among subjects with the GSTM1 null genotype, although the interaction analysis did not reach significance when considering the excretion a continuous variable (Tables II and III). In the highest tertile of m7Gua excretion the IRR for lung cancer was 2.75 (95% CI: 1.33–5.71) in subjects with the GSTM1 null genotype and 1.59 (95% CI: 0.74–3.40) among subjects with the non-null genotype (Table III). The presence of either the GSTM1 null genotype or a GSTM3 variant allele or both had significant modifying influence on the association between m7Gua and lung cancer risk. The interaction analysis was significant and the IRR was 1.26 (95% CI: 1.01–1.56) per doubling of m7Gua excretion after adjustment for smoking among subjects with this combined genotype (Table II).

As previously reported, the GSTT1 null genotype was associated with an increased risk of lung cancer in the present material.39 However, the genotype had no influence on the association between m7Gua and lung cancer risk (Table II). Similarly, there was no obvious association between the m7Gua excretion and the GSTP1-105 genotype in relation to lung cancer.

All associations were also analyzed with adjustment for time of frozen storage before analysis and with exclusion of all cases diagnosed within one year after enrolment in the cohort. This had no effect on the results (data not shown).


In this study, we examined the association between urinary excretion of m7Gua and subsequent risk of lung cancer in a large population-based cohort study. There was a high risk of lung cancer among subjects with the highest urinary m7Gua level, although this association was only robust to adjustment for smoking for categorized excretion. This association was mainly present among current smokers in linear analysis and in subjects with GSTM1 null and/or GSTM3 variant genotype.

The urinary excretion of m7Gua was higher in current smokers than in former smokers among both cases and subcohort members, although there was no clear dose-response relationship with number of cigarettes smoked. The high urinary ratio of m7Gua to creatinine among never-smokers may be due to the fact that 6 out of 8 were women, who had higher ratios than men did. The results confirm a previous study showing increased m7Gua excretion in smokers and decreased excretion upon cessation of smoking.25, 26 However, the effect of smoking may also complicate the interpretation of associations between the urinary excretion of m7Gua and the risk of cancer and this relies on the understanding of possible causal pathways.44

There was a borderline significant association between cancer risk and m7Gua in the unadjusted linear analysis. The association was reduced by adjustment for current smoking at entry. In linear and categorized analysis there was a dose-response relationship and the association with cancer risk was strong and robust to adjustment for smoking in the tertile with the highest m7Gua excretion. Even though we matched the subcohort to the cases for smoking duration and adjusted the analyses for smoking status, intensity and duration we cannot eliminate the possibility that the association between risk of lung cancer and urinary excretion of m7Gua was related to residual confounding by smoking, although the finding of association mainly among current smokers at entry into the cohort is in support of causality. Moreover, the possible effect modification by GSTM1 with a significant interaction when combined with GSTM3 supports causality in this respect. This interaction is not likely to be caused by confounding because the GSTM1 genotype has not been found to modify the lung cancer risk associated with smoking to a significant extent in a large pooled analysis.45

The increasing incidence in adenocarcinoma and decreasing incidence of squamous cell carcinoma have been suggested to be related to increasing depth of inhalation and TSNA levels as well as decreasing particle size and polycyclic aromatic hydrocarbon levels of the smoke of modern cigarettes.46, 47 However, time trends are affected by the longer persistence of risk of adenocarcinoma than of squamous cell carcinoma after smoking cessation, whereas among Asian women adenocarcinoma is the predominant type of lung cancer and usually unrelated to smoking. The incidence of small cell carcinoma shows a less clear time trend. Studies of genetic polymorphisms in enzymes involved in metabolic activation of TSNA or repair of DNA methylation suggest that both adenocarcinoma and small cell carcinoma may be relevant. Thus, excess risk of both adenocarcinoma and small cell carcinoma have been found in non- or light-smoking subjects with variant alleles of CYP2A6, CYP2E1, CYP2A13 and O6-alkylguanine alkyltransferase.48, 49, 50, 51, 52 In contrast these polymorphisms, except for CYP2A6, which may also affects smoking behavior, had less association with the risk of squamous cell carcinoma. Accordingly, it is difficult to interpret our finding of the strongest association between m7G excretion and risk for small cell carcinoma.

The m7G-adduct is innocuous and the level in DNA or the urinary excretion should mainly be regarded as a biomarker of exposure of DNA to methylating agents, including TSNA. We found that the GSTM1 and GSTM3 genotypes had no influence on the m7Gua excretion and should thus not be expected to influence the rate of formation of this adduct. Accordingly, the interpretation of our data may be that a high load of methylating agents at a given level of smoking is associated with an increased risk of lung cancer in particular in subjects with the GSTM1 null genotype. This is in keeping with the increased sensitivity to formation of chromosomal aberrations and sister chromatid exchange by NNK in cells lines with deletion of GSTM1.16 Moreover, GSTM1 and GSTM3 appear to convey resistance to methylating agents.18, 19, 20 A particularly high level of m7Gua-adducts in bronchial lavage cells from smokers has only been found in the combined GSTM1 null, GSTT1 null and GSTP1 ile/ile genotype, although pointing to relevance for this target organ.9 Similarly, the increased risk of gastric cancer in subjects with GSTM1 null genotype may be limited to exposure related to smoking subjects as indicated by a recent meta-analysis.53

The urinary excretion of m7Gua has been very little investigated as a biomarker. Except for the effect of smoking25, 26 also seen in the present study, no specific external factor are known to affect it. If not due to residual confounding by smoking, the present association between m7Gua excretion and risk of lung cancer could be interesting because TSNA methylates guanine in DNA.8 This may warrant further study of m7Gua as a product of repair of methylation damage to DNA, although m7Gua is also a metabolite from RNA.28, 29, 30 The contribution from each of these pathways is unknown and more knowledge is required for the use of m7Gua as biomarker. In this context associations with well established biomarkers of exposure to TSNA, such as NNK, would be interesting.7 Moreover, m7Gua like 3-methyladenine may be formed in grain during fumigation with methylbromide and thus be present in food.54 Diet has been shown to influence the excretion of 3-methyladenine.55 During sample collection from 1993 to 1997 the main dietary source with potential methylbromide treatment was rye and the staple food rye bread in Denmark. However, there was a negative association between consumption of rye bread and excretion of m7G and no association with other sources of carbohydrates or fish, indicating that this is not a problem in our cohort.

The present study was prospective and effects of cancer on the biomarker are thus unlikely. Only one spot sample of urine was collected at enrolment in the study and this may not be representative for the long period at risk. Urine was collected as spot samples in the present study and the concentrations of m7Gua had to be adjusted by the creatinine concentration. This is not ideal because creatinine excretion is affected by muscle mass and that may explain the higher levels of m7Gua per creatinine in women as compared with men and the higher levels with increasing age. However, age and sex associated differences in the excretion of m7Gua are not likely to affect our results concerning risk of lung cancer because the subcohort for comparison with the cases was selected with match on sex and because age was used as the time axis in the statistical analyses.

In conclusion, this population-based cohort study suggests that the urinary excretion of m7Gua is a predictor of risk of lung cancer. Although residual confounding by smoking cannot be excluded, a high risk of lung cancer among subjects with the highest excretion m7Gua in particularly among those with GTSM1 null genotype, supports an importance of methylation of guanine, e.g. by TSNA in this group.