Consistent risk factors in pancreatic cancer

Pancreatic cancer is a leading cause of cancer-related death in this country. There are no early screening tests and no effective treatment options for this deadly disease. Prevention of pancreatic cancer is difficult because little is known about the etiology. Main risk factors for pancreatic cancer include cigarette smoking, dietary factors, and maybe occupational exposure to certain carcinogens. A few molecular epidemiologic studies of pancreatic cancer have been conducted so far. Although these studies have had a very limited sample size, they have provided some clues to the etiology. DNA adducts derived from exposure to polycyclic aromatic hydrocarbon and aromatic amines have been detected in pancreatic tissues in relation to the cancer risk. Oxidative DNA damage and lipid peroxidation—induced DNA adducts are also shown to be present in the pancreas. The level of aromatic DNA adducts has been correlated with the spectrum of K-ras mutation in the tumor. The mutation profiles of p53 and K-ras genes in pancreatic cancer resemble that in bladder cancer rather than that in lung cancer. The high proportion of G to A transition in pancreatic cancer suggests the involvement of nitrosamines or alkylating agents in pancreatic carcinogenesis. This notion was supported by the observation that individuals with deficient repair of the DNA alkylation products seemed to have an increased risk for pancreatic cancer. K-ras mutation in pancreatic cancer has also been associated with alcohol consumption, organochlorines, and dietary habits. The associations between these environmental factors and K-ras mutations suggest that the mortality of pancreatic cancer can be decreased by changing of behavior and reducing exposure to environmental carcinogens.

Etiology, K-ras mutation, DNA adduct, oxidative DNA damage, polymorphism, carcinogen metabolism, genetic susceptibility

Pancreatic cancer is the fourth leading cause of cancer death in both men and women in the United States. This year, an estimated 29,200 new cases of pancreatic cancer will be diagnosed, and 28,900 patients will die of the disease. With no early screening tests and limited therapeutic options, prevention of the disease is currently the best hope for decreasing its mortality. However, prevention of pancreatic cancer is prohibited by the lack of knowledge of the etiology. Novel approaches and strategies are needed for a major impact to be made on the prevention of pancreatic cancer. So far, very few molecular epidemiologic studies of pancreatic cancer have been conducted. Most of our knowledge of the etiology of this disease comes from conventional epidemiologic studies and from studies in animal models.

The most prominent and consistent risk factor in pancreatic cancer is cigarette smoking, the relative risk of smokers being at least 1.5. The risk increases as the level of cigarette smoking increases: the highest risk ratio, 10-fold, has been seen in males who consume more than 40 cigarettes daily. The excess risk levels off 10 to 15 years after smoking cessation.

The second most important risk factor associated with pancreatic cancer seems to be diet, although the data for diet are not quite as consistent as those for smoking. Generally, increased risks have been associated with animal protein and fat consumption, and decreased risks, with intake of vegetables and fruits. Methods of food preparation evaluated in several studies showed an association of increased pancreatic cancer risk with high consumption of salt, smoked meat, dehydrated food, fried food, and refined sugar. An inverse association was found with the consumption of food containing no preservatives and additives, raw food, food prepared by high-pressure cooking, and food prepared in an electric or microwave oven.

The third suspected risk factor for pancreatic cancer is occupational exposure to some carcinogens. Excessively high rates of pancreatic cancer have been reported in workers in certain occupations, such as chemists, coal and gas exploration workers, those in metal industries, leather tanning, textiles, aluminum milling, and transportation. The available evidence is insufficient, however, to identify any common occupations as certifiable causes of pancreatic cancer. Suggestive findings exist in relation to the products of incomplete combustion, to certain pesticides, and to other chemicals and chemical processes.

Other risk factors for pancreatic cancer include some medical conditions, chronic pancreatitis, and inherited susceptibility, as reviewed in several articles. Overall, the primary cause of pancreatic cancer is poorly understood, and known risk factors suggest a role for exposure to carcinogens present in cigarette smoke, diet, and the workplace.

Until recently, most cancer epidemiology studies were limited to assessing possible causative associations between exposure to potential causative environmental agents and disease outcome. The modulation of environmental factors by host susceptibility was rarely evaluated. However, in the past few years, there have been some major advances in the understanding of environment-susceptibility interactions in human cancers. One typical example is lung cancer, a major smoking-related human cancer. By use of DNA adducts as biomarkers, some specific carcinogens in cigarette smoke, such as polycyclic aromatic hydrocarbons and tobacco-specific nitrosamines, have been linked to lung cancer. Individual variations in metabolism of these carcinogens and in repair of DNA damages caused by these carcinogens have also been shown to be predisposing risk factors in lung cancer. Moreover, the characteristic mutation patterns in the p53 gene have been linked to smoking and to a particular type of carcinogen. Understanding of the mechanism of carcinogenesis and the molecular events involved in the process has provided valuable information for developing novel strategies for the prevention of this cancer. Unfortunately, very few molecular epidemiologic studies of pancreatic cancer have been conducted so far, perhaps because of the difficulties in recruiting sufficient numbers of patients in any single institute. This review focuses on some limited information from the studies of DNA damage, genetic polymorphisms, and K-ras mutations in pancreatic cancer.
EXPOSURE MARKERS

If the hypothesis that exposure to carcinogens contributes to human pancreatic cancers is true, one would expect to see a greater burden of carcinogen-induced DNA damage in patients with pancreatic cancer. To test this hypothesis, numerous studies have examined DNA adducts in human pancreatic tissues. In the first, 37 autopsy samples demonstrated a significantly lower level of adducts in the pancreas compared with those in the lung. No significant correlation was seen between smoking history and DNA adduct levels in the pancreas, with average levels being 3.35, 2.45, and 2.0 per 108 nucleotides in the 14 persons who had smoked, the seven who had quit, and the 16 who had never smoked, respectively. The second study involved pancreatic tissues from 29 organ donors. In five of 13 smokers and three of 16 nonsmokers, a major DNA adduct was observed that was chromatographically identical to the predominant 4-aminobiphenyl (ABP)-DNA adduct, N(deoxyguanosine-8-yl)-ABP. 4-ABP is a representative aromatic amine in cigarette smoke. Although overall adduct levels were relatively low, ranging from 0.2 to 1.1 adducts per 108 nucleotides, detection of the putative ABP adduct in the pancreatic tissues suggests that the human pancreas may be a target organ for carcinogenic aromatic amines. In our own study, DNA adducts were measured by 32P-postlabeling in 13 normal tissues adjacent to tumors and 20 tumors from patients with pancreatic cancer. Normal pancreatic tissues from five nonpancreatic cancer patients and 19 healthy organ donors served as controls. To correlate the DNA adduct level with the patients' characteristics, information on age, sex, body mass index, and smoking status was collected from medical records. A significantly higher level of total DNA adducts was detected in pancreatic cancer patients than in the controls. The mean level of total adducts per 108 nucleotides in adjacent tissues was 102 ± 21, compared with 39 ± 6 and 13 ± 1 in tumor and control tissues, respectively. The apparently higher level of DNA adducts observed in our study than in the previous reports can be partially explained by the different versions of the 32P-postlabeling assay and various chromatographic methods used. In our study, some specific adducts with unknown identity were correlated positively with smoking status. Seven of 10 smokers and three of 10 nonsmokers displayed these adducts, compared with seven of 10 nonsmokers who did not have these adducts (P < 0.05 by ?² test). The level of total DNA adducts in smokers tended to be higher than that in nonsmokers (P = 0.09, t test). These observations suggested that some tobacco carcinogens, such as aromatic amine and polycyclic aromatic hydrocarbons, may contribute to pancreatic carcinogenesis through the induction of DNA adduct formation. In fact, we have correlated the K-ras mutation spectrum to the level of aromatic DNA adducts in a small set of human pancreatic tumor samples. Mutation of the K-ras gene was found in 25 (81%) of 31 pancreatic tumors and included three G to A transitions and 22 G to T transversions at codon 12. The average aromatic DNA adduct level (relative adduct labeling × 109), as measured by 32P-postlabeling, in patients with a wild-type K-ras was 4.0 ± 5.0, those with G to A transitions was 6.8 ± 5.8, and those with G to T transversions was 34.9 ± 26.9 (F = 4.62, P = 0.02). This observation supports the hypothesis that benzo[a]pyrene-type of compounds contributed to the G to T transversion of K-ras in human pancreatic cancers.

Besides aromatic DNA adducts, biomarkers of exposure to other carcinogens have also been detected in pancreatic cancer. For example, organochlorine compounds are long-lived chemicals that are present in both ambient human environment and workplace. An increased risk for pancreatic cancer has been associated with exposure to organochlorine compounds in both occupational and population-based studies. In a recent case-control study, a significant dose-response relationship between the level of serum organochlorine and pancreatic cancer was found. The median concentrations of 1,1-dichloro-2,2-bis(p-chlorophenyl) ethylene (DDE) (1290 vs 1030 ng/g lipid), polychlorinated biphenyls (330 vs 220 ng/g lipids), and transnonachlor (54 vs 28 ng/g lipid) were all significantly higher among cases (N = 108) than controls (N = 82). Subjects in the highest tertile of polychlorinated biphenyl had an odds ratio of 4.2 (95% confidence interval = 1.8–9.4).

Cadmium is another carcinogen that has been proposed to play a role in human pancreatic carcinogenesis.[ 33] So far, however, no molecular epidemiologic study has been performed to test this hypothesis.
OXIDATIVE STRESS AND LIPID PEROXIDATION

DNA damages can be also derived from endogenous processes such as oxidative stress. Oxygen radicals are well known to play an instrumental role in the development of inflammatory tissue damage, degenerative diseases, and cancer. Oxygen radicals can be derived not only from cigarette smoking or other carcinogen exposure but also from normal cellular functions, such as metabolism of nutrients and hormones. Free radicals react most readily with polyunsaturated fatty acid, resulting in a chain reaction of lipid peroxidation. In experimental models of acute pancreatitis (cerulein-, dietand sodium taurocholate-induced pancreatitis), lipid peroxidation products increased in the pancreatic tissues before morphologic changes could be detected. In patients with acute or chronic pancreatitis, increased serum and tissue levels of lipid peroxidation products and decreased levels of glutathione have also been reported, suggesting ongoing peroxidation of lipids caused by enhanced generation of oxygen radicals. Moreover, detection of lipid peroxidation product—induced DNA adducts, such as malondialdehyde adduct and etheno adducts, in human pancreatic tissues has been reported in several studies.

When we examined the DNA damage in pancreatic tissues that was induced by oxygen radicals and lipid peroxidation products, we found levels of oxidative DNA damage— and lipid peroxidation—related DNA adducts to be significantly higher in patients with pancreatic cancer than in controls. Using the high-performance liquid chromatography electrochemical detection method,[ 43] we detected 8-hydroxyguanosine, a mutagenic oxidized DNA base, at levels of 13.0 ± 2.3 (mean ± SE), 11.9 ± 3.0, and 7.1 ± 1.06 per 105 guanine nucleotides in tumors (N = 23), adjacent tissues (N = 11), and controls (N = 23), respectively. The difference between controls and adjacent tissues was at borderline significance (P = 0.08); between control samples and tumors, it was statistically significant (P = 0.03). In addition, putative DNA adducts derived from malondialdehyde, an end-product of lipid peroxidation, was detected in all tissue samples examined, and the level was significantly higher in the adjacent tissues than in the tumors and control samples. These findings support the notion that the human pancreas is susceptible to oxidative stress— and lipid peroxidation—related damages, which may play an important role in pancreatic carcinogenesis. However, the questions remain unanswered as to what factors contribute to the oxidative stress in the pancreas and who is more susceptible to such stress. In some of the studies, the level of these endogenous DNA damages were examined in relation to smoking, age, sex, body mass index, and genetic polymorphisms of some antioxidant enzymes, but no significant association was detected. In our study of oxidative DNA damage, we did not find any significant correlation between the level of 8-hydroxydeoxyguanosine and polymorphisms of manganese superoxide dismutase and 8-hydroxy-2-deoxyguanosine DNA glycosylase/apurinic lyase. Search for more reliable biomarkers of oxidative stress—induced damages and biomarkers of individual susceptibility to oxidative stress are being carried out in our laboratory.
GENETIC POLYMORPHISMS

Metabolic activation is a prerequisite for the carcinogenic effect of many carcinogens, and considerable inter-individual variation exists in the metabolic capacity of carcinogen activation. Many studies of smoking-related human cancers have shown that genetic variations in carcinogen metabolism affect individual susceptibility to carcinogen exposure and, in turn, cancer risk. Unfortunately, so far, only three published studies have explored the association between genetic polymorphisms of drug-metabolizing enzymes and the risk of pancreatic cancer. In a study of 45 cases and 53 controls, no association was found between susceptibility to pancreatic cancer and genetic polymorphisms of cytochrome P450 1A1, 2D6, and 2E1, enzymes that activate chemical carcinogens. In another study of 149 cases and 146 controls, no association was found between the risk of pancreatic cancer and polymorphisms of GSTM1, GSTT1, and CYP1A1.[ 46] The third study of 81 pancreatic cancer cases, 78 controls, and 119 patients with chronic pancreatitis, a nonsignificant excess of NAT1 slow acetylator was found in cancer cases versus controls. A significant overexpression of GSTM1 AB or B genotype was found in all pancreatic disease cases. These observations suggest that the polymorphism of GSTM1 and NAT1 enzymes may be associated with a modest increase in susceptibility to pancreatic diseases. Our preliminary data from a pilot study with very limited sample size also show a similar trend in NAT1 (data not shown). In addition, a polymorphism of cytochrome P450 1A1 gene that confers a higher enzyme activity[ 48] seemed to have a significant effect on the level of aromatic DNA adducts in pancreatic tissues.[ 29] The average level of aromatic DNA adducts (in 109 nucleotides) was 23.4 ± 36.9, 32.3 ± 36.2, 117.8 ± 163.6 among patients with wild-type (N = 50), heterozygous mutant (N = 11), and homozygous mutant (N = 3) CYP1A1, respectively (P = 0.005 by analysis of variance). This observation is consistent with previously reported similar findings in other smoking-related human cancers.

In addition to carcinogen-metabolizing enzymes, we have also explored the role of DNA repair in pancreatic cancer. O6-Alkylguanine DNA transferase (AGT) is a repair enzyme that transfers the alkyl group from O6-alkylguanine adduct to the cysteine residual at position 145 of the protein. Because nitrosamines and alkylating agents are suspected to be responsible for the G to A transitions of the K-ras gene frequently seen in pancreatic cancer through induced formation of O6-alkylguanine, which is a highly mutagenic DNA adduct,[ 54] repair efficiency of this adduct may affect individual risk for pancreatic cancer. One amino acid-altering polymorphism in close proximity to the cysteine acceptor site (codon 143 in exon 5) has been identified in the AGT gene. This polymorphism may contribute to the large interindividual variation found in human AGT activity and thus may determine the individual's susceptibility to environmental alkylating carcinogens. A marginally significant association between the AGT codon 143(Ile/Val) genotype and increased risk for lung cancer has been reported. We have studied this polymorphism in a pilot study of 94 pancreatic cancer cases and 51 frequency-matched controls. It was found that 22.3% of the cases and 9.8% of the controls carried this polymorphism, and the odds ratio was 2.65 (95% confidence interval: 0.96-7.24, P = 0.06). This preliminary observation is in support of the hypothesis that alkylating agents contributes to the development of pancreatic cancer and that individuals with deficient DNA repair mechanism are at increased risk. Confirmation of this observation is underway in a larger scale study.
GENE MUTATION AND ETIOLOGY

Studies of mutation spectra of the tumor suppressor gene p53 have shown that specific endogenous or exogenous mutagens may induce characteristic patterns of DNA alteration in the tumor as a fingerprint of exposure. For example, 24% of p53 mutations in lung cancers were G to A transition, and 40% were G to T transversions. The frequency of p53 mutation and of G to T transversion on the nontranscribed strand has been positively correlated with lifetime cigarette consumption or history of tobacco and alcohol use. G to T transversion has been associated with benzo[a]pyrene, a classical tobacco carcinogen, in several experimental systems. The mutation frequency of p53 gene in pancreatic cancer is about 44%. The mutational pattern in pancreatic cancer is similar to that in bladder cancer, another smoking-related cancer, but not to lung cancer. Forty-one percent of the p53 mutations in pancreatic cancer were G to A transitions, and 13% were G to T transversions. Interestingly, when mutations were found in both p53 and K-ras genes in a tumor, the same type of mutation (either transition or transversion) was detected in the same tumor, suggesting a common mechanism.

Pancreatic cancer has the highest frequency (>85%) of K-ras mutation among all human cancers. The frequency of K-ras mutation in pancreatic cancer has been associated with the patients' smoking or drinking status.[ 60, 61] In a study of 82 surgically resected or biopsied primary adenocarcinomas of the pancreas from patients in the United States, K-ras mutations were found in 89% of ex-smokers and 86% of current smokers, compared with 68% of never-smokers. The same report also summarized 13 studies of pancreatic cancer, including 349 mutants of the K-ras gene. It was found that 62% of the mutations were G to A transitions, and 35% were G to T transversions. The high frequency of G to A transition in K-ras gene was consistent with that in the p53 gene. G to A transition was also exclusively seen in nitrosamine-induced pancreatic carcinoma in hamsters. In addition, G to A transition at codon 12 of the ras gene is a common mutation found in mammary and lung carcinomas in animals treated with nitrosamines and other alkylating agents. G to T transversion at codon 12, on the other hand, has been seen in murine lung carcinoma induced by benzo[a]pyrene. Thus, it is conceivable that the high frequency of G to A transition of the p53 and K-ras genes in pancreatic cancer suggests a role of nitrosamines and alkylating agents in pancreatic carcinogenesis.

Other environmental factors that have been associated with K-ras mutations in pancreatic cancer include alcohol, organochlorines, and diet. The association between K-ras mutation and alcohol consumption was examined in 51 pancreatic cancer cases, and the risk of mutation was three times higher in alcohol drinkers than in nondrinkers, with a linear trend. Porta et al[ 65] have reported an association between serum concentrations of organochlorine compounds and K-ras mutations in 51 patients with exocrine pancreatic cancer. In this study, they found that serum levels of 1,1,1-trichloro-2,2-bis(p-chlorophenyl)-ethane (DDT) and DDE were significantly higher in pancreatic cancer cases with a K-ras mutation than in cases without a mutation. Furthermore, a specific association was found between serum concentrations of DDT and DDE and a G to T transversion at codon 12 of the K-ras gene. Because DDT and DDE do not directly induce gene mutation and are considered tumor promoters, other direct carcinogens, such as benzo[a]pyrene, could have contributed to the G to T transversions in these tumors. DDT and DDE, alternatively, could have promoted the growth of the cells carrying these mutations. The same group has also observed an association between diet and K-ras mutation.[ 66] Patients with mutant K-ras (N = 83) were more likely to use milk and milk products daily than were wild-type cases (odds ratio 4.6, 95% confidence interval 1.2-17.6, P for trend = 0.02). When this food group was combined with butter, the odds ratio was 6.4 (CI, 1.4–29.2, P for trend = 0.01). Cases with mutant K-ras also had significantly lower intakes of vitamin C and vitamin E after adjustments were made for calorie intake. These results support the hypothesis that K-ras mutation was related to lifestyle environmental factors in pancreatic cancer. Alcohol consumption, organochlorines, and dietary fat intake have all been suspected to contribute to pancreatic cancer by epidemiologic studies. Even though the mechanisms underlying the association between K-ras mutation and these environmental factors are not fully understood, these findings are optimistic because they mean that pancreatic cancer can be affected by dietary change and reduced exposure to carcinogens.

The severity of the pancreatic cancer problem and the scarcity of information on its etiology call for more intensive research effort. Both genetic and environmental factors may play significant roles in the etiology of pancreatic cancer. Exposure to carcinogens through cigarette smoking, diet, and occupational contact may increase the risk of pancreatic cancer. The frequency and the spectrum of K-ras mutation in pancreatic tumors have been associated with smoking, serum organochlorine level, dietary habits, and aromatic DNA adducts in the target tissue. Individuals with deficient DNA repair and unfavorable genetic make-up in carcinogen metabolism may at increased risk for pancreatic cancer. We hope, with advances in molecular biology and new methods of molecular epidemiologic study, to develop the tools for identifying high-risk individuals in whom this deadly disease can be prevented. Although in general, the sample sizes were small in all published molecular epidemiologic studies of pancreatic cancer, some useful information has been generated to guide the future direction of research in this field. An understanding of the etiologic and molecular events leading to the development of pancreatic carcinoma may provide a basis for the development of effective strategies for the prevention, early diagnosis, and treatment of this disease.

Last updated Jan 2/07