Strategic elements for environmental justice.

Damage arising from putative environmental sources has been found in the DNA of the gastric and colorectal mucosae of patients presenting with gastrointestinal disorders from the South Manchester area. 06-Methylguanine (06-MeG) in the range 0.010->0.300timolesmole-' adenine was heterogeneously distributed both between and within individuals. The pattern of alkylation of tissue DNA appears to differ when comparison is made between gastric and colorectal samples. Most of the gastric tumour DNA samples were alkylated (5/6; 0.087 + 0.097), whereas the DNA of the associated mucosa was alkylated less frequently (2/7) and to a lesser extent; (0.017 ± 0.030; P = 0.07). Conversely, colorectal tumour DNA was alkylated infrequently (1/7) and to a lower extent (0.003 ± 0.007) than the DNA of the adjacent mucosa (8/10 samples alkylated with a mean of 0.083 + 0.106; P = <0.01), or indeed of any other tissue. Although increased levels of DNA damage in tissue associated with malignant disease have been indicated by independent studies of DNA damage at other cancer sites, significant differences were not observed in the present report, neither was there any suggestion of a relationship with smoking or alcohol consumption. The data provided by this report indicate that exposure to putative environmental alkylating agents occurs in the UK at levels comparable to those previously detected in areas of higher cancer risk. Although we cannot determine the extent to which this DNA damage is attributable to normal background exposures, it is evident that the alkylation of tissue DNA occurs and is not uniform. In conjunction with other reports, therefore these differences may begin to provide indications of mechanisms that could be of relevance in the aetiology of gastrointestinal cancers.

Gastrointestinal (GI) cancers are common, prognosis is poor and as yet with little sign of improvement. In this situation a fuller understanding of the aetiology of the disease could well lead to better control through improved strategies for prevention and earlier diagnosis.
Epidemiological studies suggest that GI cancers mainly result from exposure to environmental agents but the concensus remains inconclusive (Doll, 1988). GI cancers, particularly those of the upper regions of the tract have been extensively linked to exogenous and endogenous exposures to N-nitrosocompounds and although this hypothesis is supported by good animal models, the evidence so far is not compelling (Magee, 1989). Similarly, exposure to nitrates and nitrites as precursors of N-nitrosocompounds in the development of upper GI cancers finds some, but not conclusive epidemiological support (Preston-Martin & Correa, 1989). Evidence for a causal association of nitrate in the development of gastric cancer has been questioned from an analysis of the literature (Forman, 1989) as well as by models which predict intragastric rates of nitrosation (Licht & Deen, 1988).
If answers to these complex issues are to be found they will come eventually from direct measurements of exposure, rather than from estimates of potential exposure. Techniques are now available which permit the detection in target tissues of DNA damage caused by environmental agents. One such procedure is the use of radioimmunoassays (RIAs) to detect specific DNA lesions arising from exposure to alkylating agents. In the case of the simple alkylating agents, 13 products have been detected in DNA and of these, the promutagenic lesions 06-alkylguanine and 04-alkylthymine are thought to play a critical role in tumour initiation (Saffhill et al., 1985). A monoclonal antibody (McAb) specific for the detection of 06-methyl-2'-deoxyguanosine (06-MedG) (Wild et al., 1983;Myers et al., 1988;Saffhill et al., 1988a) has been used extensively to measure relatively high concentra-tions of 06-MeG in small samples of DNA extracted either from cell cultures (Boyle et al., 1986(Boyle et al., , 1987 or from mitochondrial DNA (Myers et al., 1988). The same procedures can be used to measure very small amounts of this modified base in much larger samples of DNA thus permitting the detection of 06-MeG in human DNA. This has been achieved for oesophageal and stomach DNA of patients from Lin Xian, a district of N. China (Umbenhauer et al., 1985) and from SE Asia (Saffhill et al., 1988a). In Lin Xian, oesophageal cancer risk is exceptionally high at 151 per 100,000 in males and 115 per 100,000 in females (Lu et al., 1986). In Singapore (WHO, 1987), the cancer risk in oesophagus and stomach is 14 and 37 per 100,000 in Chinese males (4 and 15, respectively in females). In these earlier studies there were few control samples. The aim of this study therefore was to investigate alkylation of DNA in a wide variety of both benign and malignant gastrointestinal conditions in a region where the incidence of GI cancers is similar or lower to that in Singapore, but much lower than that for Lin Xian. The world standardised rates for 1986 and 1987 in the NW cancer registry area for stomach, colon and rectum were approx. 18, 17 and 14 per 100,000 in males and 8, 15 and 8 per 100,000 in females, respectively (C. Hart, pers. communic) i.e. similar to the data published for these sites in the NW Registry for the period 1979period -1982period (WHO, 1987. Samples were taken from surgical specimens of different GI organs from patients living in the Manchester area. A preliminary account (Saffhill et al., 1988a) has indicated that alkylation damage can be detected in these human tissues and the following report presents the findings from this latter series in the context of the relevant clinical data.
Isolation and analysis of DNA During operations to remove diseased GI organs, tissue was dissected to isolate samples of mucosa for study. In cases where disease was non-malignant a single representative sample was taken. Where the disease was malignant (and in one case of gastric polyps), a sample was taken of the tumour itself together with an adjacent piece of uninvolved mucosa for comparison. Tissues (5-1O g) were frozen onto dry ice and stored at -80°C. Batches of DNA were prepared from the thawed tissue using a modified phenol procedure (Kirby & Cook, 1957) and digested to nucleosides using DNAase I (0.1 mg ml 1), venom phosphodiesterase (0.03 U ml-') and alkaline phosphatase (0.3 U ml-') in 50 mM Tris-HCl, 5 mM MgCl2 and 3 mM sodium azide at pH 7.5 in the presence of 1 tLM 2'-deoxycoformycin for 4 h at 37°C. The latter was added to inhibit adenosine deaminase (Fox & Kelly, 1978), which demethylates 06-MedG (O'Connor & Saffhill, 1979) and may occur as a contaminant of some DNA preparations. Hydrolysates of up to 5 mg DNA were applied to a column of Aminex-6 (25 cm x 1 cm) maintained at 50C during elution with 10 mM ammonium bicarbonate, pH 8.0. Where >5 mg DNA was available repeat separations were performed and the column fractions from both runs were pooled for analysis by RIA. This procedure separates the four major deoxynucleosides from 06-MedG, thereby enabling spectrophotometric measurement of the amount of normal purines in the DNA sample applied to the column and collection of the region of the elution profile which corresponds to 06-MedG. This system also separates ribonucleosides from deoxyribonucleosides and therefore permits analysis of 06-MedG in DNA which contains traces of RNA. The putative 06-MedG containing fractions and control fractions (i.e. a similar volume of buffer) from a blank region of the column elution profile were lyophilised after addition of 0.2-0.5 ml phosphate buffered saline containing 1% horse serum and 3 mM sodium azide. They were reconstituted in a similar volume of water for analysis by radioimmunoassay (RIA) using a monoclonal antibody to 06-MedG (Wild et al., 1983). A full account of the above procedures and the RIA have been given previously (Wild et al., 1983, Saffhill et al., 1988b, Myers et al., 1988. Results of analyses are expressed as JLmoles 06-MedG mole-' deoxyadenosine, since experience has shown that spectrophotometric measurements of deoxyadenosine concentrations can be made more accurately than those of deoxyguanosine. Clinical details A medical history was taken from each patient by the clinical member of the study group. Details of smoking habits and alcohol consumption indicated that of the 35 patients only four smoked to a significant extent (0.75-1.5 packs per day); three smoked <0.25 packs per day and one was a pipe smoker, while 21 were non-smokers. For alcohol consumption, 11 were non-users, 15 claimed to be social drinkers and only two consumed 20-40 units per week, a unit being defined as I pint of beer or single measure of spirit. Smoking and drinking habits were not obtainable for six and seven patients, respectively. Drugs used in treatment, or for premedication prior to surgery were classified as antiemetic, antihypertensive, antacids/antiulcer, tranquilliser, non-steroidal, anti-inflammatory etc., for 22 categories of treatment. Statistical evaluations using the unpaired Fischer's exact test and the Mann-Whitney universal test were made as indicated in the text. These tests were employed in view of the nonnormal distribution due to the high proportion of values which were negative according to the test employed. Tables giving details of lifestyle, medication and occupation in relation to the diagnosis and levels of DNA methylation are available on request.

Results
Gastrointestinal DNA obtained from people living in the South Manchester area contained low levels of 06-MeG in the range 0.01->0.30jmolesmole-' dA, but the pattern of alkylation was heterogeneous both within and between individuals (Tables I and II).
Out of a total of 53 DNA samples analysed, 26 (49%) had no detectable 06-MeG (i.e. <-0.010 ;moles mole-' dA), although this ratio was biased by the high proportion (6/7) of malignant colorectal tumour DNA samples which were negative (see Table II). A total of 35 individuals are included in this survey and for 13 of these (36%) there was no detectable alkylation of the tissue DNA. However, as only one sample of tissue DNA was available for analysis for eight of the 13 negative individuals, more might have been positive if more than one sample had been available for examination.

Gastric samples
Only 3/6 of the non-malignant gastric samples were positive and the highest value in the entire series was found in the adjacent mucosal DNA of a patient with benign gastric polyps (patient No. 1); (Table I). In patients with malignant disease only 2/7 tumour adjacent mucosal samples were positive while 5/6 tumour DNA samples contained 06-MeG. The mean values for the DNA of non-malignant mucosa, tumour adjacent mucosa and malignant tumour were, 0.441 ± 1.022 (0.024 ± 0.038 if the atypically high value for Patient No. 1 is omitted), 0.017 ± 0.030 and 0.087 ± 0.097, respectively. The combined value for all patients with malignant disease is 0.049 ± 0.076 (Table III).

Colorectal samples
In DNA from mucosae of patients with non-malignant disease, 5/15 samples (including the ileum) and 5/11 indi-  . ) enm %00eI0-0r.I NC s 1 viduals were positive. In patients with malignant disease, tumour adjacent mucosal DNA was positive in 8/10 individuals, whereas tumour DNA was positive in only 1/7 individuals (Table II). The mean O6-MeG level for the non-malignant mucosal samples was 0.041 ± 0.060 (5/11 positive individuals). Of the two ileal samples, one was negative and the other just positive (0.014pmolesmole-' dA). In patients with malignant disease the mean value for the tumour adjacent tissue was 0.083 ± 0.106 vs 0.003 ± 0.007 for the malignant tumour DNA itself (see Table III). Alkylation was approximately 2-fold greater in the DNA of mucosa adjacent to malignant colorectal tumours than in the DNA of the mucosa of nonmalignant colorectal disease, 0.083 ± 0.106 vs 0.041 ± 0.060, respectively (Table III).

Effects of life style and medication
No evidence of a relationship was observed for the effects of either smoking or alcohol consumption. Alcohol and tobacco consumption were not excessive in the patients studied (see Methods) and these factors did not appear to influence the levels of DNA alkylation.
Analysis was also performed for occupation and drug treatment but not surprisingly in this small sample no relationship was found.

Discussion
These data demonstrate that GI tissue DNAs obtained from a Manchester population contain O6-MeG which is most probably derived via an environmental source. While it is not yet possible to identify the exposures involved, it is evident that they can result in levels of DNA alkylation as high as those observed in Lin Xian where there is a very high cancer risk (Umbenhauer et al., 1985).
As might be anticipated for environmental (low dose) exposures, the alkylation damage is heterogeneously distributed, both between and within individuals (Tables I and  II). Immunohistochemical observations of DNA adduct formation for several different carcinogens have all indicated a heterogeneous distribution within a given tissue (e.g. NDMA, Fan et al., 1989;Aflatoxin B,, Wild et al., 1990 and Nnitrosobis (2-oxopropyl)amine, Bax et al., 1990). In these cases metabolism is required for activation of the carcinogen to a chemically reactive form. The distribution of competent, activating enzyme systems in different cell types, therefore, is most probably responsible for this heterogeneity. After treatment, nuclear reactions can repair alkylation injury to DNA to a variable extent, thereby adding further to this heterogeneity (O'Connor et al., 1991). In human tissue, similar factors would be expected to operate since in man and the rat, similar systems are responsible for the metabolism of environmental nitrosamines such as NDMA (Yoo et al., 1988) and for the repair of O6-MeG (Pegg, 1983;Gerson et al., 1986).
Whilst the data reported here are consistent with the exposure to an environmental alkylating agent that requires metabolism for activation, there is no immediately obvious link with occupation, drug therapy or life style factors that might suggest such an exposure. It is worthy of note, however, that in rat liver, an organ which is competent for the metabolism of NDMA, tissue average DNA alkylation levels of 0.01-0.30 tmoles 06-MeG mole-' would arise from orally administered doses of 2-20 iLg NDMA Kg-' (Pegg & Perry, 1981). Although exposure levels of -1 pg or more of NDMA per day have been reported for a variety of sources (e.g. see Scanlan, 1983), no estimates of exposure arising from endogenous sources are available. Given that some cells may be DNA repair deficient (see above) these levels of DNA alkylation might accrue from repeated exposures to lower doses of environmental agents such as NDMA.
Comment on the overall levels of DNA alkylation observed must be made with reservation in view of the sample  size. In the case of gastric DNA (Tables I and III), the O6-MeG levels for non-malignant mucosa were lower than those for tumour DNA and similar to those for mucosa adjacent to malignant disease (0.024 ± 0.038; two of five samples vs 0.087 + 0.097; five of six samples and 0.017 ± 0.030; two of seven samples, respectively) if the atypically high values for Patient No. 1 are excluded. However, only the higher level of alkylation in the tumour DNA vs that of the adjacent mucosa approaches statistical significance (P = 0.07, Mann-Whitney universal test). In the colorectum (Tables II and III), the levels of DNA alkylation were higher in the uninvolved mucosa adjacent to the malignant tumours than in the mucosa of patients with non-malignant disease and were lowest of all in the malignant tumours themselves (0.083 ± 0.106, 8/10 samples; 0.041 ± 0.060, 5/11 samples and 0.003 ± 0.007, 1/7 samples, respectively). In this case the higher level of DNA alkylation in the adjacent mucosa compared with that of the tumour DNA was significant (P<0.010, Mann-Whitney universal test).
A similar observation to the trend seen for the DNA in the colorectum has been made for the presence of 04-ethylthymine in the hepatic DNA of 33 patients with either cancer or non-malignant disease. In this study the mean values for hepatic 04-ethylthymine levels were significantly higher (4-5fold) in patients with malignant disease (Huh et al., 1989). This may be a particularly appropriate lesion to follow in liver since animal tissues have shown that although 04-ethylthymine occurs initially at levels which are many fold lower than 06-MeG, it is repaired slowly, if at all, and so tends to accumulate (Swenberg et al., 1984). In the original study of alkylation damage in human DNA reported from Lin Xian, a district with high risk for oesophageal cancer in N. China (Umbenhauer et al., 1985), the DNA from subjects with oesophageal and gastric cancer which had detectable levels of O6-MeG were, on average, 3 x and 1+ times higher respectively, than those of the controls. The controls were not truly representative, however, in that they came from a European source. In this latter study, as in the present report, there were many samples in which 06-MeG was not detected. 06-MeG has also been detected in one sample of stomach mucosa from 20 individuals in a group of Athens patients with either normal or atrophic gastric mucosa (Kyrtopoulos et al., 1990), in placental DNA in 2/10 smokers and 3/10 non-smokers from the USA (Foiles et al., 1988) and in the DNA of 16/17 peripheral lung samples from smokers and non-smokers (Wilson et al., 1989). In the two American studies, as in the present report, there was no evidence of a correlation with smoking levels, among these small numbers of samples. On the other hand, when other unidentified adducts have been examined by 32P-post labelling procedures, correlations have been observed for adduct concentration in lung DNA with the severity of smoking (Phillips et al., 1988;Randerath et al., 1989, Cuzick et al., 1990. Amounts of post-labelled adducts in other tissues (e.g. bladder, aorta, heart, liver, pancreas, oesophagus and kidney) were also raised suggesting a causal association in other target tissues (Randerath et al., 1990;Cuzick et al., 1990).
The fact that the DNA alkylation level was low or nondetectable in the malignant tumours of the colorectum is worthy of note and is in contrast to the relatively consistent and higher levels of alkylation in the DNA of gastric carcinoma. The low level in colorectal tumour DNA could be due to a variety of factors. These tumours were exposed to the GI lumen so that physical impedance to absorption of an exogeneous environmental agent seems unlikely, although there could be a change in membrane permeability. Loss of the capacity to metabolise carcinogens has been observed in animal tumours Cameron et al., 1976) and an increased capacity for DNA turnover or DNA repair could be responsible. There are, however, no strong precedents for the latter interpretation and on the contrary, reduced repair of 06-MeG has been observed in about 20% of established tumour cell lines (Yarosh, 1985). In vivo, a wide range of DNA-alkyltransferase activities was found in human neural tumours (Wiestler et al., 1984). Furthermore, in human colon, the effect of malignant change on this repair activity is as yet unclear; DNA-alkyltransferase activity may be increased, decreased, or remain unchanged when tumour tissue activity is compared with that of the adjacent mucosa (Margison et al., 1990).
In conclusion, the levels of DNA alkylation reported in this study are similar to those observed in N. China (Umbenhauer et al., 1985) and in Singapore (Saffhill et al., 1988a). The detection of promutagenic lesions in GI tissue DNA samples in several regions of the world now provides a possible explanation for the presence in some GI tumours of activated ras genes (Bos, 1989) which could arise as a result of miscoding events due to the presence of alkylated bases during DNA synthesis (Saffhill et al., 1985). Although from these limited studies there is no direct indication of a relationship between cancer incidence and the extent of DNA alkylation it is anticipated that as data of this kind accumulate, it will not only be possible to define exposures to environmental sources of DNA damaging agents, but eventually to determine their importance and possibly also to begin to predict risk factors.
We gratefully acknowledge financial support from the Cancer Research Campaign for the execution of this work and to the British Council for sponsoring one of the authors (A.F.B.). We wish to thank Miss M.D. Boulter for the careful preparation of this manuscript, Mr B. Waszkowycz for advice on the classification of the medications, Dr S. Roberts for statistical analysis and advice and Ms C. Hart for access to recent data from the NW Cancer Registry.