Levels of Malondialdehyde-Deoxyguanosine in the Gastric Mucosa: Relationship with Lipid Peroxidation, Ascorbic Acid, and Helicobacter pylori

Although the incidence of gastric cancer is declining, it remains an important disease worldwide. The role of Helicobacter pylori in gastric carcinogenesis has been established in epidemiological studies and, more recently, in animal models, but the precise means through which this occurs have not yet been established (1, 2, 3, 4). One possible mechanism for H. pylori-mediated carcinogenesis is through induction of DNA damage and mutations as a result of increased activity of reactive oxygen species in the gastric mucosa (5). These compounds can damage DNA directly by causing strand breaks, apurinic sites, or DNA adducts (6). In addition, reactions with the lipid bilayer result in the accumulation of degradation products, such as MDA,² a compound that has been shown to be present in increased concentrations in H. pylori gastritis (7, 8). Vitamin C is believed to be one of the major defenses against oxidative stress in the stomach and is concentrated several-fold from plasma to the gastric mucosa. Vitamin C exists in two forms, dehydroascorbic acid and the potent antioxidant ascorbic acid. Concentrations of the latter are diminished in gastric juice of H. pylori-infected patients, a process that may serve to exacerbate oxidative damage (9, 10).

Accumulation of MDA in H. pylori-infected gastric mucosa not only provides evidence of increased oxidative stress and lipid peroxidation, but may also have a carcinogenic role. MDA can react with DNA to form adducts and cross-links and has been shown to be mutagenic in bacterial and mammalian systems (11, 12, 13) and carcinogenic to rats (14). MDA induces a diverse spectrum of mutations, such as frameshift mutations and base pair substitutions in Escherichia coli (15). In humans, increased concentrations of lipid peroxidation products have been found in the serum of gastric cancer patients (16). The main DNA adduct is M₁-dG, which has been shown in further E. coli studies to result in a mutation frequency, after correction for strand bias, of 18% (a 500-fold increase over unmodified DNA; Ref. 17). M₁-dG has been detected in human liver, WBCs, and recently, in gastric tissue (18, 19, 20, 21), and levels of M₁-dG and other MDA adducts were greater in breast tissue from patients with breast cancer than from those without breast cancer (22).

Detection of M₁-dG in human tissue has previously required gas chromatography/mass spectrometry or ³²P-postlabeling techniques (19, 21). However, these techniques require relatively large amounts of DNA, are time consuming, and are not easily applied to studies of small tissue samples from large numbers of subjects. More recently, a monoclonal antibody against M₁-dG has been developed and characterized (23), which has allowed validation of a sensitive immunoslot-blot assay for the adduct with a limit of detection of 2.5 adducts/10⁸ normal bases (18, 24).

The purpose of this study was to determine the effect of H. pylori and lipid peroxidation on levels of M₁-dG in gastric biopsies and to examine whether ascorbic acid protects against this process. We measured levels of M₁-dG using the immunoslot-blot technique alongside mucosal MDA and plasma, mucosal, and gastric juice ascorbic acid and total vitamin C. We studied patients with normal and H. pylori-infected mucosa and followed H. pylori infected patients for up to 12 months after eradication of the organism.

One hundred twenty-four patients attending for routine outpatient endoscopic investigation of dyspepsia were studied (72 males and 52 females; mean age, 49.1 years; range, 25–77 years). Patients were excluded if they had been taking antibiotics in the previous month, had had previous attempts at H. pylori eradication, or had previously had gastric surgery.

H. pylori status was determined by [¹³C]urea breath test, antral biopsy for urease testing, and histological assessment of antral and corpus biopsies. Further antral biopsies were taken from all patients as follows: two for determination of M₁-dG, one for MDA assessment, and one for ascorbic acid and vitamin C determination. Prior to endoscopy, a 10-ml sample of venous blood was withdrawn into a lithium heparin tube; immediately after intubation, a sterile Teflon catheter was passed, and 5 ml of gastric juice were collected, both for assessment of ascorbic acid and vitamin C.

Of the 124 patients, 39 had all tests negative for H. pylori and histologically normal gastric mucosa, and 85 patients were deemed H. pylori positive on the basis of positive histology, urease, and [¹³C]urea breath tests. Sixty-four H. pylori-positive patients received eradication therapy and were asked to return at 6 months for repeat biopsy and breath test. Of these 64 patients, 38 received open eradication therapy and 26 were in the treatment arm of a randomized double blind study of H. pylori eradication in nonulcer dyspepsia (H2BB3009) conducted by Glaxo Wellcome. All 64 patients received identical eradication therapy, which included ranitidine bismuth citrate (400 mg), clarithromycin (500 mg), and metronidazole (400 mg), all twice daily for 1 week. The remaining 21 patients received triple therapy placebo as part of H2BB3009. In addition to the above exclusion criteria, the patients in H2BB3009 all had normal endoscopies and had not been using proton pump inhibitors or NSAIDs in the preceding month. H2BB3009 had a 12-month follow-up, so a urease test was not performed at the 6-month endoscopy in these patients to aid investigator blinding. In these patients, therefore, H. pylori status at 6 months relied on histology and breath test, but they were asked to return at 12 months for further assessment, biopsy, and breath test. In all other ways the patients in the randomized trial and those receiving open eradication therapy were treated identically. This study was approved by the local research ethics committee, and all patients gave written informed consent.

Biopsies were fixed in 10% buffered formalin. Sequential 3-μm-thick sections were cut and stained with H&E and modified Giemsa stain. Gastritis was scored according to the modified Sydney classification, such that sections were graded between 0 and 3 (from absent to severe) for activity, chronic inflammation, atrophy, intestinal metaplasia, and H. pylori density by a single histopathologist who was blinded to patient details (25).

The breath test was performed largely as described by Logan et al. (26). Subjects fasted for 4 h. Breath samples were collected after subjects drank 150 ml of a solution containing 4 g of citric acid in 200 ml of water and 30 min after subjects drank the remaining 50 ml of citric acid containing 100 mg of [¹³C]urea. Samples were analyzed on a Europa Scientific ABCA mass spectrometer to give a ratio of ¹³CO₂ to ¹²CO₂. A delta difference between pre and post ratios of >5.0 was taken as a positive result for H. pylori infection. The procedure was similar for the patients in the Glaxo Wellcome study, but the test meal contained 2.4 g of citric acid in 200 ml of orange juice and the 100 mg of [¹³C]urea were dissolved in 25 ml of water followed by a 25-ml water wash.

For each patient, one antral biopsy was snap frozen in liquid nitrogen at the time of endoscopy, stored at −70°C, and assayed within 2 weeks by the procedure of Yagi (27), as modified by Drake et al. (7). Each biopsy was immersed in 4 ml of water to which was added 1 ml of thiobarbituric acid solution (0.167 g thiobarbituric acid in 25 ml of water and 25 ml of glacial acetic acid). Mixtures were heated at 100°C for 60 min. After cooling on ice, 5 ml of butan-1-ol were added to extract the MDA equivalents. The tubes were centrifuged at 3000 rpm for 10 min to separate the aqueous and butan-1-ol phases. Fluorescence of the butan-1-ol phase at 555 nm was determined using an excitation wavelength of 515 nm. Values from the tissue specimens were compared against the freshly prepared standard solutions. This technique has an interassay coefficient of variation of 14.2% (based on paired biopsies from 142 patients).

Venous blood samples were centrifuged, and 0.5-ml aliquots of plasma were added to 1.0 ml of 2% metaphosphoric acid alone and 1.0 ml of 2% metaphosphoric acid containing 9 mg of DTT (for reduction of dehydroascorbic acid to ascorbic acid to give a value for total vitamin C). Likewise, 0.5-ml aliquots of gastric juice were added to 0.5 ml of 2% metaphosphoric acid containing 0.5% sulfamic acid, alone and with 6 mg of DTT. All samples were snap frozen in liquid nitrogen and stored at −70°C. Prior to analysis, samples were thawed and centrifuged at 1000 × g. The supernatant solution was analyzed by high-performance liquid chromatography using reversed-phase ion-pair chromatography on a C₁₈ column (28). Ascorbic acid was selectively measured using an electrochemical detector set at 350 mV. Total vitamin C was determined from the solutions initially prepared with DTT after incubation at 45°C for 120 min.

Biopsy samples were frozen immediately after endoscopy in liquid nitrogen, stored at −70°C, and assayed within 2 weeks. After thawing, biopsies were homogenized in 0.5–1.0 ml of metaphosphoric acid and divided into two parts. For total vitamin C determination, DTT was added to one part to a final concentration of 6 mg/ml and incubated at 45°C for 120 min prior to analysis. High-performance liquid chromatography was performed for both parts as above. This technique previously has been shown to be highly specific for ascorbic acid and results in 90% extraction of ascorbic acid and 75% extraction of dehydroascorbic acid from tissue (29). The interassay coefficient of variation for both ascorbic acid and total vitamin C is 5%.

Biopsies were snap frozen in liquid nitrogen and stored at −80°C until assay. Following homogenization, the gastric tissue was incubated with RNase A and proteinase K at 45°C for 2 h. The DNA was isolated using the Qiagen genomic DNA extraction kit (Qiagen Ltd). The amount and purity of the DNA was established by determining the UV absorbance at 260 and 280 nm (Kontron Uvikon 860 spectrophotometer). Adduct recovery after extraction had been determined previously as 96%.

The method has been described previously by Leuratti et al. (18). Briefly, DNA samples were sonicated and then heat denatured; the resulting single-stranded DNA was immobilized (1 μg) in triplicate onto a nitrocellulose filter using the Minifold II blotting apparatus (Schleicher and Schuell). The filter was then heated at 80°C for 1.5 h and blocked for nonspecific binding using nonfat milk powder dissolved in PBS containing 0.1% Tween 20 for an additional 1 h. After blocking, the filter was incubated overnight with the primary antibody specific for M₁-dG (23) at 4°C and then incubated with horseradish peroxidase-conjugated secondary antibody (goat antimouse) for 2 h at room temperature. After the filter was bathed with chemiluminescent reagents (Super Signal Ultra; Pierce and Warriner), it was exposed to chemiluminescence-sensitive Hyperfilm. An image of the filter was acquired using a Fluor-S MultiImager (Bio-Rad), which was used for quantitation of adduct levels. The level of M₁-dG adduct in the samples was determined from a calibration curve generated by the dilution (with control calf thymus DNA) of standard calf thymus DNA (treated with 2 mm MDA) containing known amounts of the adduct, blotted onto the same filter. A positive control of genomic human blood DNA (from Boehringer Mannheim) was run with each filter. The M₁-dG concentration in the control DNA from 12 consecutive filters was 90.4 ± 21.2 (SD) adducts/10⁸ nucleotides, giving a coefficient of variation of 23.5%.

For calculating the required sample size, we assumed a SD of log-transformed M₁-dG of ∼0.25 adducts/10⁸ normal bases, that H. pylori increases M₁-dG levels by 40% (e.g., from 50 to 70 adducts/10⁸ normal bases), and a ratio of H. pylori-positive to -negative patients in the sample of 2:1. The number of patients required is then ∼120 patients in total to achieve statistical significance at the 5% level, with 80% power.

For skewed data, medians with interquartile ranges are presented. Subsequently, measures of MDA, ascorbic acid, total vitamin C, and M₁-dG were log-transformed to meet the normality and constant variance requirements of parametric statistical methods. Paired and unpaired t tests were performed on the transformed data. Spearman’s correlation was used for assessment of the effect of gastritis, and Fisher’s exact test was used for categorical data. To determine the relationship between M₁-dG and other continuous variables, univariate linear regression was performed. To account for the effect of H. pylori on MDA and ascorbic acid, each regression analysis was repeated with H. pylori as a second variable. Associations are presented as regression slopes with 95% CIs. The regression slopes for log-transformed data were then back-transformed and the percentage of increase in the dependent variable (either M₁-dG or MDA) was calculated for a 2-fold increase (doubling) of the independent variable. Two-tailed Ps were determined in all cases. Data were analyzed using SPSS for Windows, version 8.0.

One hundred twenty-four patients were studied, of whom 85 were H. pylori positive and 39 had normal gastric histology and negative H. pylori tests. Age, sex, smoking habits, endoscopic findings, use of NSAIDs, and acid-suppressive therapy in the month prior to endoscopy are all detailed in Table 1. H. pylori-positive patients were older and more likely to smoke than H. pylori-negative patients, but the groups were otherwise comparable.

As expected, and reported previously, ascorbic acid and vitamin C levels in gastric juice and plasma were significantly lower in H. pylori-positive patients than negative controls, whereas MDA concentrations were significantly higher (7, 10, 30). Concentrations of M₁-dG were, however, similar in H. pylori-positive and -negative patients (Table 2; Fig. 1). In addition, levels of M₁-dG were unaffected by age, smoking history, sex, the presence of peptic ulcer disease, and the use of histamine receptor antagonists, proton pump inhibitors, or NSAIDs. Sydney gastritis scores were available for 43 H. pylori-positive patients (all in the randomized trial), and no significant associations existed between M₁-dG and level of antral polymorph infiltrate, chronic inflammation, atrophy, intestinal metaplasia, or H. pylori infection density. The MDA concentration was positively related to antral polymorph infiltrate with borderline significance (Spearman’s ρ = 0.32; P = 0.05), but did not correlate with any other of the parameters of gastritis. Total gastric juice vitamin C was significantly positively associated with antral chronic inflammatory infiltrate (ρ = 0.42; P = 0.01), but there were no other significant correlations between plasma, antral, or gastric juice ascorbic acid or total vitamin C and any of the gastritis scores.

After logarithmic conversion, univariate linear regression revealed a statistically significant positive relationship between MDA and total antral vitamin C. In addition, there were significant inverse relationships between MDA and total gastric juice vitamin C, plasma ascorbic acid, and total plasma vitamin C (Table 3). However, after adjustment for the presence or absence of H. pylori, the associations of total gastric juice vitamin C, plasma ascorbic acid, and total plasma vitamin C with MDA were all nonsignificant, whereas there was a consistently significant positive relationship between presence of H. pylori and increasing MDA concentration (P < 0.001), suggesting that the aforementioned inverse relationships were the result of confounding by H. pylori infection. On the other hand, the positive relationship between total antral vitamin C and MDA remained significant (P = 0.02), and that between antral ascorbic acid and MDA became conventionally significant (P = 0.04). Thus, doubling of antral ascorbic acid and total antral vitamin C was reflected in 6% (95% CI, 0.2–11%) and 8% (95% CI, 1–15%) increases in MDA, respectively.

Application of the same analysis with M₁-dG as the dependent variable revealed a statistically significant positive relationship between M₁-dG and both antral and gastric juice ascorbic acid and total vitamin C (Table 4). The relationships between plasma ascorbic acid and total plasma vitamin C with M₁-dG were nonsignificant. After adjustment for the presence of H. pylori, gastric juice ascorbic acid and total gastric juice vitamin C ceased to be significantly associated, whereas antral ascorbic acid and total antral vitamin C remained significantly positively related to M₁-dG. Thus, doubling of antral ascorbic acid and total antral vitamin C was reflected in 14% (95% CI, 4–27%) and 18% (95% CI, 4–34%) increases in M₁-dG, respectively.

Likewise, univariate analysis demonstrated a weak positive relationship between MDA and M₁-dG, which became statistically significant after adjustment for the presence of H. pylori [a doubling of MDA reflected in a 36% (95% CI, 6–75%) increase in M₁-dG; P = 0.02]. This suggests an effect modification by H. pylori infection. Indeed, the relationship between MDA and M₁-dG was greater for H. pylori-positive cases [a doubling of MDA reflected in a 40% (95% CI, 3–93%) increase in M₁-dG; P = 0.04] than for H. pylori-negative controls [a doubling of MDA reflected in a 23% (95% CI, −23 to 87%) increase in M₁-dG; P = 0.3], although the difference in regression slopes was not statistically significant (P = 0.6; Fig. 2).

Forty-five of 64 H. pylori-positive patients who received eradication therapy returned at 6 months. Forty-two of these were negative, 1 was positive, and 2 were indeterminate (breath test positive and histology negative) for H. pylori. For patients with successful eradication, mean antral and corpus chronic inflammation scores fell significantly between 0 and 6 months [from 1.9 to 1.2 (95% CI of difference, 0.5–1.0; P < 0.001), and from 1.4 to 0.8 (95% CI, 0.1–1.2; P = 0.03), respectively], as did antral and corpus polymorph infiltration [from 1.25 to 0.33 (95% CI of difference, 0.2–1.7; P = 0.02), and from 0.8 to 0.1 (95% CI, 0.3–1.1; P = 0.005), respectively], but no significant changes were seen in atrophy or intestinal metaplasia scores. As expected, MDA concentrations fell (from 134.9 to 106.2 nmol/g; P = 0.007), and gastric juice ascorbic acid and total vitamin C both increased significantly 6 months after successful eradication of H. pylori (from 6.7 to 13.0 μg/ml for ascorbic acid; from 9.4 to 17.0 μg/ml for total vitamin C; P < 0.001 for both). No changes were seen in plasma or antral ascorbic acid or total vitamin C. Despite these potentially favorable changes, there was no change in M₁-dG concentration 6 months after successful eradication of H. pylori (Table 5).

Thirteen evaluable patients who had received eradication therapy returned at 12 months, all of whom were H. pylori negative on urease, histology, and breath test. On paired analysis, there was a borderline increase in M₁-dG from 6 to 12 months, (from a median of 60.8 to 77.5 adducts/10⁸ bases; P = 0.05), but no difference between pretreatment and 12-month levels of M₁-dG (66.7 and 77.5 adducts/10⁸ bases, respectively; P = 0.8). No significant changes were seen between 6 and 12 months for gastric juice, antral and plasma ascorbic acid, and total vitamin C or for MDA.

Twenty-one patients received placebo eradication therapy. Data were available for 12 of these patients at 6 months and for 8 patients at 12 months, all of whom remained H. pylori positive. MDA levels did not change significantly between pretreatment and 6 months (from 95.9 to 103.6 nmol/g; P = 0.5) but, surprisingly, increased significantly by 12 months (from 95.9 nmol/g at pretreatment to 178.7 nmol/g at 12 months; P < 0.001). In addition, M₁-dG did not change significantly between pretreatment and 6 months (from 38.9 to 70.6 adducts/10⁸ bases; P = 0.3) but increased significantly between pretreatment and 12 months (from 38.9 to 74.5 adducts/10⁸ bases; P = 0.02). No significant changes from pretreatment to 6 and 12 months were seen for plasma, antral, and gastric juice ascorbic acid and total vitamin C. Caution should be used, however, when interpreting these secondary analyses because of multiple testing and low power for these comparisons.

H. pylori infection of the gastric mucosa stimulates influx of polymorphonuclear leukocytes, leading to the generation of reactive oxygen and nitrogen species. Cell membranes, which are rich in polyunsaturated fatty acids, are readily attacked by these compounds, producing fatty acid radicals and lipid hydroperoxides, which can decompose in complex ways, yielding more radical species and a wide range of compounds, notably aldehydes. Of these, MDA and 4-hydroxynonenal are the most common (31). MDA, which is also formed by the breakdown of prostaglandin endoperoxides (32), is a strongly genotoxic carbonyl compound that can react directly with DNA to produce a variety of adducts. The most common, formed after reaction of MDA with deoxyguanosine, is the mutagenic pyrimidopurinone compound M₁-dG (33). Recently, it has been demonstrated that M₁-dG can also be formed independently of lipid peroxidation through DNA oxidation and base propenal intermediates (34).

We sought to identify the relationship between M₁-dG and ascorbic acid, MDA, and H. pylori in the human stomach and studied a large number of patients before and after eradication of H. pylori. The immunoslot-blot technique used in this study has been carefully validated and has a detection limit of 2.5 adducts/10⁸ normal bases, a level well below those detected in this study (18, 24). Our results are in keeping with the generation of M₁-dG in gastric mucosa at levels that are similar to those seen in healthy liver and are significantly higher than those recorded in leukocytes, pancreas, and breast (33). We found that H. pylori leads to increased concentrations of MDA and that, to a lesser extent, increased MDA is associated with an increase in M₁-dG. This is consistent with the anticipated role of lipid peroxidation and MDA as the major precursors of this adduct. The fact that the relationship between MDA and M₁-dG is relatively weak may be because MDA is generated predominantly at the cell membrane, some distance from cellular DNA.

Both ascorbic acid and total vitamin C are highly concentrated from plasma into gastric mucosa, but concentrations are reduced in the gastric juice of patients with H. pylori gastritis, a finding that was once again confirmed in our study. Ascorbic acid (the reduced form of vitamin C) is a potent antioxidant, able to scavenge reactive oxygen species in the gastric mucosa (35). However, we have shown that higher mucosal ascorbic acid and total vitamin C levels are associated with increases in both MDA and M₁-dG concentrations. This is consistent with the finding of a positive correlation between total antral vitamin C and MDA levels in a separate population of patients from our department and suggests that ascorbic acid and vitamin C may be promoting lipid peroxidation in the gastric mucosa (7). This may result from the ability of ascorbic acid to reduce iron to the ferrous form, which is a more potent catalyst of lipid peroxidation than ferric iron (36). The ability of ascorbic acid to promote formation of M₁-dG has also been demonstrated in liver homogenates, in which levels of M₁-dG were doubled after addition of ascorbic acid (33).

We hypothesized that increased oxidative stress and MDA concentrations in H. pylori-infected mucosa would lead to a concomitant increase in concentrations of M₁-dG. However, levels of M₁-dG were similar in H. pylori-infected and normal gastric mucosa and were not affected by any of the parameters of gastritis. Eradication of H. pylori resulted in expected reductions in mucosal inflammatory scores and MDA levels, but no significant changes were seen in M₁-dG. Although there was a small increase in M₁-dG levels from 6 to 12 months after eradication therapy, the level at 12 months was not different to pretreatment values, so this is unlikely to be a consequence of H. pylori eradication and, given the small numbers analyzed, could be the result of chance.

The lack of a direct relationship between H. pylori and M₁-dG may be explained by other effects of infection that compensate for increased MDA levels, such as induction of defenses against reactive oxygen species, lipid peroxidation, and DNA damage. Glutathione peroxidase is an important antioxidant that protects against lipid peroxidation by reducing lipid hydroperoxides to fatty acids. Concentrations are increased in Mongolian gerbils infected with H. pylori and in humans with gastritis and H. pylori infection (8, 37, 38). The other major protection against lipid peroxidation is the lipid-soluble chain-breaking antioxidant α-tocopherol. Although this compound is not concentrated in the gastric mucosa from plasma (39), levels are sustained by redox regeneration of tocopherol by ascorbic acid (which may indeed be one of its major functions in the gastric mucosa). Because proliferating cells have been shown to be resistant to the effects of lipid peroxidation, it may also be that the increased proliferative rate in H. pylori gastritis protects against formation of M₁-dG (36). Finally, repair of DNA adducts (which in this case is by nucleotide excision repair) is likely to be induced in inflamed, rapidly proliferating tissue (40). These factors may combine to counteract the elevated levels of MDA in H. pylori gastritis.

One difficulty with biopsy-based studies is that it is not possible to differentiate between the important cells in carcinogenesis (proliferating epithelial cells) and all other cells from the biopsy, including inflammatory and connective tissue cells. The concentration of these cells will differ between inflamed and normal biopsies, diluting any genuine differences in epithelial cell adduct concentrations. As a consequence, these data do not completely exclude an effect of H. pylori on levels of specific epithelial cell M₁-dG in the gastric mucosa.

Other studies of the effect of H. pylori on DNA damage are conflicting. Two studies found only small (but significant) increases in the oxidative DNA adduct 8-hydroxy-2-deoxyguanosine in H. pylori infection (41, 42). In a study of 281 subjects, urinary levels of 8-hydroxy-2-deoxyguanosine were found to be higher in persons without H. pylori infection than infected persons (43). O⁶-Methylguanine, an alkyl adduct formed after reactions of DNA with N-nitroso compounds, has been detected in variable amounts in gastric mucosa (44, 45). The Eurogast study group found this adduct more frequently in leukocytes of individuals from areas with a high incidence of gastric carcinoma, but there was no link with H. pylori (46). In an earlier work, levels of DNA adducts measured by ³²P-postlabeling were similar in normal mucosa and atrophic gastritis or intestinal metaplasia (47), whereas others have shown abnormal DNA content in H. pylori-associated atrophic gastritis associated with abnormal c-myc and p53 expression (48). Finally, in our own studies of epithelial cells isolated from gastric biopsies, we found lower levels of DNA strand breaks in H. pylori-infected than normal mucosa (49).

At present, therefore, the relationship between H. pylori and DNA damage in the gastric mucosa remains elusive. What is perhaps of greater importance are the location of DNA damage and the rate at which damage is converted into significant mutations. The increase in size of the proliferative zone in H. pylori gastritis makes any damage more likely to occur in a proliferating cell, and the more rapid turnover means that DNA damage occurring in replicating cells is more likely to be incorrectly repaired (50, 51). Both of these factors could result in an increased mutational rate in H. pylori gastritis, such that even a very small increase in DNA damage may be amplified many times in terms of cancer risk.

In conclusion, we have demonstrated significant background levels of M₁-dG in gastric biopsies, raising the possibility that, despite a lack of association with H. pylori infection in this study, it could be an important carcinogen in the human stomach. However, future work on the mechanisms of gastric carcinogenesis should be aimed at characterizing DNA damage in proliferating epithelial cells.

We thank Glaxo Wellcome for funding the M₁-dG analyses.