© 2003 by Oxford University Press
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of the National Cancer Institute, Vol. 95, No. 13, 948-960,
July 2, 2003
© 2003 Oxford University Press
REVIEW |
Etiology of Pancreatic Cancer, With a Hypothesis Concerning the Role of N-Nitroso Compounds and Excess Gastric Acidity
Correspondence to: Harvey A. Risch, M.D., Ph.D., Department of Epidemiology and Public Health, Yale University School of Medicine, 60 College St., P.O. Box 208034, New Haven, CT 06520–8034 (e-mail: harvey.risch{at}yale.edu).
 |
ABSTRACT |
|---|
Top
Abstract
IntroductionIn the United States, pancreatic cancer is the fourth most frequent cause of cancer death in males as well as females, after lung, prostate or breast, and colorectal cancer. Each year, approximately 30 000 Americans are diagnosed with pancreatic cancer and about the same number die of it. Germline mutations in a few genes including p16 and BRCA2 have been implicated in a small fraction of cases, as has chronic pancreatitis. The one established risk factor for pancreatic cancer is cigarette smoking: current smokers have two to three times the risk of nonsmokers. Studies of dietary factors have not been entirely consistent but do suggest associations of higher risk with consumption of smoked or processed meats or with animal foods in general and lower risk with consumption of fruits and vegetables. Colonization by Helicobacter pylori appears to increase risk, and a history of diabetes mellitus may also increase risk. The purpose of this epidemiologic review is to consider the possibility that risk of pancreatic cancer is increased by factors associated with pancreatic N-nitrosamine or N-nitrosamide exposures and with chronic excess gastric or duodenal acidity. Host genetic variation in inflammatory cytokine mechanisms may also be involved in this process. Many features of the evidence bearing on the pathophysiology of pancreatic cancer appear to support connections with N-nitroso compounds and with gastric acidity.
|
INTRODUCTION |
|---|
In the United States in 2003, it has been estimated that there will be more than 30 000 new cases of pancreatic carcinoma and that 30 000 individuals will die from it (1). It is now the fourth most frequent cause of cancer death among men and women, after lung, breast or prostate, and colorectal cancer (1), and over the lifetime, more than 1% of the population is affected (2). African Americans have appreciably higher rates than whites (2). Survival with pancreatic cancer is dismal. One-year relative survival is 20%–30%; 5-year survival is approximately 5% (1,3). The majority of patients are diagnosed with advanced disease and, in general, no standard treatment appreciably alters its rapidly progressive, fatal course (4). Heritable, germline mutations in p16 (5), BRCA2 (6), and other genes appear to be associated with a total of 5%–10% of all pancreatic cancers (7), and penetrance of these mutations for the disease may be fairly low (6,8). Chronic pancreatitis of various types is associated with another 3%–4% (7). Aside from germline mutations or pancreatitis, the one established risk factor for pancreatic cancer is cigarette smoking. Current smokers have approximately two to three times the risk of nonsmokers, and risk tends to increase according to frequency or duration of smoking (9). Dietary studies have shown a fairly consistent pattern of increased risk with meat or cholesterol intake and decreased risk with fruit or vegetable consumption, although causal inferences regarding these associations are still uncertain (10). Several mechanisms by which smoking and diet could affect risk of developing pancreatic cancer have been suggested, but adequate evidence for these hypotheses in humans (as opposed to in animal models) is fairly lacking. In this review, mechanisms—including a new one— and their supporting evidence are presented that may explain many of the findings related to the development of pancreatic cancer.
I propose that, in humans, risk of pancreatic cancer is increased by long-term conditions of excess gastric/duodenal acidity and by frequent or repeated gastrointestinal or other exposures to N-nitroso compounds or their precursors. The excess acidity, largely asymptomatic for most individuals, is typically associated with chronic basal stimulation of pancreatic bicarbonate production, which—through a trophic effect of secretin, bicarbonate’s principal hormonal secretagogue—may be associated with pancreatic ductular hyperplasia and increased DNA synthesis (11). With respiratory or gastrointestinal intake or formation of N-nitroso xenobiotics, the pancreatic ductular epithelium is exposed to carcinogens through the circulation (12). The ductular epithelium can metabolically activate the carcinogens (13), if activation has not already occurred in the liver. Metabolically activated N-nitroso carcinogens induce DNA adducts and single-strand breaks (14) and appear to stimulate DNA synthesis in the pancreatic ductular epithelium (15). Chronic secretin stimulation and N-nitroso compound exposures potentially overwhelm DNA repair capabilities, acting synergistically to induce tumor development (16). Thus, the majority of this review focuses on gastric/duodenal acidity and other factors potentially affecting chronic basal pancreatic ductular bicarbonate secretion, and on factors bearing on ductular epithelial exposures to N-nitroso carcinogens.
|
GENERAL CONSIDERATIONS |
|---|
The pancreas is composed of three cellular compartments: acinar and ductal (i.e., the "exocrine" pancreas) and endocrine. The principal function of the acini, under hormonal (major) and neural (minor) control, is to secrete isotonic saline and digestive enzymes, such as procarboxypeptidases, trypsinogens,
-amylase, and others (17). The ductal tree is a ramified network of tubular branches providing physical support to the clusters of acini and conveying the acinic secretions to the duodenum; its main secretory function, also under hormonal and some neural control, is to produce isotonic fluid rich in bicarbonate (18). The ductal fluid helps to propel the digestive enzymes toward the duodenum, while the bicarbonate serves to create an appropriate intraluminal pH for the function of the enzymes and to neutralize the acidic gastric chyme as it leaves the duodenal bulb (19). As much as 20 g of protein enzymes and 2 L of fluid may be secreted from the 90-g pancreas per day (20). The islet cells, which form the endocrine pancreas, secrete insulin, glucagon, and other hormones into the acinar and systemic circulations. The islet cells are scattered throughout the pancreas and account for only about 2% of the organ (17). In the United States, more than 90% of malignant pancreatic neoplasms are classified as ductal adenocarcinomas (3,21), yet ductal or ductular epithelial cells constitute only about 4% of the human exocrine pancreas (17). Although acini account for 80% of the organ parenchyma (22), acinar cell carcinomas are rare (3,21). In spite of the histologic and ultrastructural resemblance of ductal carcinoma cells to the cuboidal, mucus-producing cells of the normal ductular epithelium (23), there is some controversy about the true cellular origin of these neoplasms. Ductular and acinar cells are histogenetically derived from a common precursor cell that is essentially indistinguishable from duct epithelial cells by light or electron microscopy (24). Ductal adenocarcinomas in hamster pancreata treated with N-nitroso carcinogens have been suggested to arise from acinar cells that have undergone dedifferentiation, with loss of both rough endoplasmic reticulum and zymogen granules, the characteristic features of acini (25). Development of pseudoductules derived from dedifferentiated acinar cells may occur with longer carcinogen exposures (25). However, evidence favoring a ductule epithelial cell origin of ductal adenocarcinomas appears more convincing. Under normal differentiation, the two cell types develop distinct antigens that are detected by monoclonal antibodies, AC-1 against acinar cells and HP-DU-1 against ductal epithelium (24). In cultured human pancreatic explant cells, N-nitroso carcinogen exposures (N-nitrosodimethylamine [NDMA], see "Human N-Nitroso Compound Exposures" section, below), which are known to produce ductal carcinomas in the explant model, show carcinogen uptake only in normal precursor cells labeled by HP-DU-1 (24). In addition, ductal adenocarcinomas consistently express carcinoembryonic antigen, CA 19-9, DU-PAN-2, and other antigens that are detectable in normal duct epithelium (22) and that are associated with the degree of ductal epithelial atypia (26,27). In the hamster model, a blood group A-like antigen is detectable on the luminal surface of normal pancreatic ductal and ductular cells but not on acinar epithelium. With N-nitrosamine exposures [e.g., N-nitrosobis(2-oxopropyl)amine (BOP), see "Human N-Nitroso Compound Exposures" section, below], this antigen is highly expressed on hyperplastic and neoplastic ductal and ductular cells (28). Furthermore, in N-nitrosamine-treated hamsters, secretory abnormalities involve only pancreatic ductular fluid and bicarbonate output and not acinar protein production; these abnormalities antedate tumor appearance and are not caused by tumor obstruction (29). Finally, human ductal adenocarcinomas express the same set of intermediate filaments, villin and keratins 7, 8, 18, and 19, as normal pancreatic ductal epithelium, whereas acinar cells usually express neither villin nor keratins 7 or 19 (22). Thus, the evidence seems to favor pancreatic ductal adenocarcinomas as arising from the ductular epithelial lining.
|
ANIMAL AND HUMAN EXPLANT CELL MODELS OF PANCREATIC ADENOCARCINOMA |
|---|
Various animal models have been developed for studying pancreatic adenocarcinoma. These models typically expose a given species to one of a number of xenobiotic carcinogens and have the pancreas as the primary or most prominent site of tumor development. The types of tumors produced are specific to the species and to the particular compound. Rats, mice, guinea pigs, hamsters, and dogs have been used. The animal model most applicable to human pancreatic cancer involves the application of N-nitroso compounds to Syrian golden hamsters, and ductal adenocarcinomas have been produced with the carcinogens shown in Table 1
(30,31). These N-nitroso compounds are generally metabolized or interconverted and, more importantly, -hydroxylated to their proximate reactive forms in hepatocytes and pancreatic ductal epithelium and acini (13,32–34). Regardless of the route of administration, these carcinogens reach the pancreas via the bloodstream, and there is no evidence for distribution through reflux of duodenal contents or bile (30). Interestingly, the carcinogenic effect of continuous-exposure regimens of N-nitrosamines appears to be more related to the stimulated synthesis and replication of adduct-bearing DNA, particularly in the pancreatic ductular epithelium, than to the amount or concentration of the DNA adducts formed by the N-nitrosamine (15). After single injection or continuous administration of N-nitroso(2-hydroxypropyl)(2-oxopropyl)amine (HPOP), hamster lung, liver, and kidney DNA show much higher levels of adducts than pancreas DNA, yet in contrast to the pancreas, these organs are resistant to carcinogenesis induced by this N-nitrosamine (15). HPOP administration triples DNA synthesis in pancreatic ductular cells but only slightly affects DNA synthesis in acinar and other cells (15).
|
A second model of pancreatic adenocarcinoma involves in vitro culture of adult human normal pancreatic ductal epithelium obtained from cadaver donors immediately after death (24). These explant cells are exposed to various N-nitroso compounds and then evaluated for abnormal proliferation and mitotic activity or injected subcutaneously into nude mice to evaluate tumor production. As shown in Table 1
, both N-nitrosamines and N-nitrosamides have been studied with this technique (24). Ductal adenocarcinomas have been produced in nude mice with explant cell exposures to NDMA and N-nitroso-N-methylurea (MNU) (35) and to N-methyl-N'-nitroso-N-nitrosoguanidine (MNNG) and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) in similar experiments with human fetal pancreatic explant cells (24). These experiments also demonstrate that human pancreatic explant cells can activate NDMA and NNK to their carcinogenic forms. This fact is important because NNK is present in appreciable concentrations in mainstream and sidestream tobacco smoke (36) and in other tobacco products (37) and because NDMA is found in sidestream tobacco smoke (9), in certain occupational settings (38), and in appreciable amounts in foodstuffs (39,40) and may also be synthesized from dietary amines in vivo (41–43). |
HUMAN N-NITROSO COMPOUND EXPOSURES |
|---|
N-Nitroso compounds are ubiquitous in the human environment but are present at exceedingly low levels. Humans are exposed to them or to their precursors at higher levels through tobacco, occupational, and dietary sources, and these substances (NDMA, NNK, and others) specifically induce pancreatic ductal adenocarcinomas in human and relevant animal cell assays. However, the levels of N-nitroso compounds reaching pancreatic ductal epithelial cells from tobacco, occupational, or dietary exposures are likely to be orders of magnitude lower than the concentrations used in carcinogenesis experiments (24), although these common types of human exposures usually last for much longer periods of time. In any case, it is clear from more than two dozen case–control and cohort studies that cigarette smoking is related to an increased risk of pancreatic cancer (4,9). Whether this increased risk is caused by N-nitrosamines present in mainstream and sidestream tobacco smoke (9,36) is plausible but by no means certain; there are many other potentially carcinogenic substances in the smoke, including heterocyclic amines, polycyclic aromatic hydrocarbons (PAHs), and cadmium, that could be responsible (9).
It has been observed that workers employed in the manufacturing of cured rubber products are also exposed to N-nitrosamines. In some plants in the United States (38) and France (44), among various N-nitrosamines, NDMA was detected at the highest levels or was the one most frequently encountered, whereas at other plants in the United States (45) and in Germany (46), N-nitrosomorpholine (NMOR) appeared to predominate and NDMA occurred at lower levels. In a study of two industrial rubber plants where appreciable concentrations of atmospheric NDMA were detected, the frequencies of N7- and possibly O6-methyldeoxyguanine (mdG) DNA adducts in peripheral blood leukocytes appeared to be associated with exposure category levels (47). In the categories of highest exposure to NDMA, average concentrations of N7-mdG adducts were approximately 3.7 adducts per 107 guanine bases (47). This adduct concentration is approximately 200-fold lower than that required for the induction of neoplasia in the short-term HPOP hamster model (15) and is comparable with peripheral blood leukocyte adduct concentrations measured in nonsmokers (3.4 adducts per 107 guanine bases) with the same 32P-postlabeling high-pressure liquid chromatography methods (48). The similar adduct levels between normal nonsmokers and highly N-nitrosamine-exposed rubber workers may explain why the latter were apparently not found to have an increased risk of pancreatic cancer (46). Adduct concentrations in smokers are approximately double those of nonsmokers (48), suggesting that very long-term occupational exposures to appreciably greater N-nitrosamine levels would be required to detect an increased risk.
Aside from tobacco usage, the main route of human exposure to N-nitroso compounds is ingestion from dietary sources. Preformed N-nitrosamines and endogenously generated N-nitrosamines and N-nitrosamides may contribute to this exposure. N-Nitrosamines (primarily NDMA) form in protein-containing foods dried at high temperatures (beer ingredients, nonfat dry milk, cooked bacon, or dried meats) or preserved with nitrite (cured, smoked, or pickled meats and fish) (39,40,49). Estimates of the average daily dietary intake of exogenous N-nitrosamines in western diets range from 0.2 to 1.0 µg/day (39,40). N-Nitrosamides, short-lived compounds that do not persist in foods, are formed in the stomach from nitrite and ingested amides, such as creatinine, in foods of animal origin (50,51). Formation of endogenous N-nitroso compounds is thought to be proportional to the amine or amide concentration, which is in excess, and to the square of the nitrite concentration (49). Total N-nitroso compound exposures from smoked or processed meats are derived at least as much from their nitrites as from their N-nitrosamines. It has also been suggested that approximately 25% of ingested nitrate is recirculated into the saliva and that 20% of salivary nitrate is reduced to nitrite, yielding a conversion of 5% of exogenous nitrate to endogenous nitrite (49,50,52). According to this theory, most human nitrite exposure would be from the consumption of vegetables common in the diet. However, for gastric cancer, which is most likely related to nitrite exposure, empirical evidence relates exogenous nitrite intake to increased risk (53,54), whereas nitrate intake (mostly from vegetables) is associated with decreased risk (53). Large amounts of nitrate and protein are required to increase endogenous N-nitrosamine formation (42,43), suggesting that nitrate intake from vegetables is not particularly important for nitrite-related risk effects. Perhaps by the time that gastrointestinal nitrate absorption, salivary recirculation, and conversion to nitrite has occurred, protein or amides in the gastric contents have left the stomach and are no longer available for N-nitrosation. In any event, determining the intake of nitrite (and possibly exogenous N-nitrosamines) is likely the most relevant method for assessing human dietary exposures to N-nitroso compounds.
Finally, N-nitrosamines (mainly, N-nitrosodiethylamine [NDEA] and NDMA) or N-nitroso compounds in general are present and measurable in the gastric juice of subjects after overnight fasting (41,55). This basal endogenous formation occurs through acid-catalyzed N-nitrosation at gastric conditions below pH 2.5 and through bacterially catalyzed N-nitrosation, to an increasing degree, as gastric pH increases from 5 to 8 (41,55). Exposure levels from basal endogenous N-nitroso compounds appear to be an order of magnitude lower than those from typical dietary sources (39–41,55).
|
PANCREATIC CANCER AND EXPOSURE TO DIETARY N-NITROSO COMPOUNDS |
|---|
Many studies have analyzed N-nitroso compound-related exposures and the risk of pancreatic cancer. A few studies have observed no particular associations with smoked or processed meat intake (56,57) or with calculated average daily nitrite intake (58,59). Most published studies, however, have found increased risks (60–62) or statistically significant trends in increasing risk (63–67) associated with consumption of smoked or processed meats. One study (59) that calculated an index of exogenous N-nitrosamine intake also found a positive association with risk. Intake of vitamin C, which inhibits N-nitrosation, has been associated with a statistically significantly decreased risk in most studies (10,57,59,64,67–69), although not uniformly (70). Because most of these studies have found positive risk associations with meat or cholesterol consumption and decreased risks with fruit or vegetable consumption, it is not yet clear whether the associations with N-nitroso compound-related exposures are actually attributable to the exogenous nitrite, N-nitrosamine, or vitamin C.
|
CONTROL OF GASTRIC ACIDITY |
|---|
In addition to associations with N-nitroso compound exposures, the proposed hypothesis of this review specifies that risk of pancreatic cancer is increased by long-term conditions of excess gastric/duodenal acidity. To develop this part of the hypothesis, I first consider normal mechanisms of gastric acid physiology.
Upper gastrointestinal and pancreatic function is controlled by many interacting hormones and modulated by neuroendocrine processes. The simplifications provided below do some injustice to the complexity of gastrointestinal physiology but help to identify physiologic mechanisms through which risk factors for pancreatic cancer could operate. To start with, three principal hormones are involved in gastric acid (i.e., HCl) regulation: gastrin, somatostatin, and secretin (Table 2). These hormones exist in a number of precursor and mature polypeptide forms, variants that are beyond the scope of this review.
|
The stomach is divided anatomically into the cardia (the upper entry), the fundus (the upper dome), the body or corpus (the large main section), the antrum (the narrowed section preceding the pylorus), and the pylorus (the outflow valve region). HCl is secreted into the gastric lumen from parietal cells in glands of the oxyntic mucosa, occupying most of the body of the stomach (71,72). The major physiologic mechanism responsible for the acid release is gastrin secretion into the bloodstream by the antral G cells (73). This circulating gastrin, released from the gastric mucosal capillary bed, stimulates enterochromaffin-like (ECL) cells lying in the basal halves of the oxyntic glands to release histamine, which then causes nearby parietal cells of the glands to secrete acid into the lumen (72). Gastrin also has a lesser, direct stimulatory effect on the parietal cells (73). A basal level of acid secretion (10% of maximal) is always present in humans (74). During the cephalic (initial) phase of the meal response, antral G cells release gastrin through central neurologic integration of positive and negative stimuli (e.g., sight, smell, taste, physical eating sensations, emotional arousal), mediated by the vagus nerve and its gastrin-releasing peptide (19). Another vagal reflex mechanism appears to be involved in stimulation of the pancreatic acini to secrete digestive enzymes (75). This reflex may be initiated by gastric distention during the gastric phase of meal digestion (19). Pancreatic digestive enzymes are also secreted during the intestinal phase of digestion because of cholecystokinin (CCK) released into the circulation from endocrine I cells scattered throughout the proximal two-thirds of the small intestine, in response to the presence of partially digested proteins and fats (76).
During the gastric phase, food in the stomach (primarily proteins) strongly buffers the released acid, which is secreted at near-maximal rates (19). As gastric emptying progresses, the moderately acidic chyme in the antrum and duodenum stimulates the secretion of somatostatin from D cells of the antrum and secretin from S cells of the duodenum and proximal jejunum (73). The somatostatin turns off the release of gastrin from nearby antral G cells and inhibits the histamine secretion of the ECL cells by an endocrine route (73). The net effect is to return parietal cell acid secretion to basal levels. The secretin released by duodenal and jejunal S cells travels through the circulation to the pancreatic ductal epithelium, where it stimulates the production of fluid and bicarbonate. Postprandial plasma levels of secretin in humans tend to be low but still functionally important for ductal bicarbonate production (77–79), although CCK and neuropeptide factors have been suggested as potentiators of the secretin effect (19). Nevertheless, there is no doubt that gastric acid drives pancreatic bicarbonate secretion (19,78). The bicarbonate, entering the gut through the papilla of Vater, progressively neutralizes the acidic chyme as it travels distally through the duodenum, with a steep pH gradient, typically from pH 2.0–3.5 in the first few centimeters to pH 5.0–6.0 in the mid-duodenum (19,80). Hepatic bicarbonate may also participate in the neutralization (81).
|
EXCESS GASTRIC/DUODENAL ACIDITY AND PANCREATIC CANCER |
|---|
Evidence exists supporting an association between conditions of excess gastric acidity and increased risk of pancreatic cancer. To start with, duodenal ulcers occur in individuals chronically suffering from greater-than-normal levels of basal gastric acid output (82–86). Patients with duodenal ulcers experience longer durations of basal hyperacidity (pH <2.5) in the proximal duodenum (80,87) and can have higher levels of stimulated acid output (88) than normal individuals. Virtually all (98%) chronic duodenal ulcers arise in the first few centimeters of the duodenum (89), an area with a strong pH gradient, consistent with neutralization of the excess acid by chronic pancreatic or hepatic bicarbonate secretion below the papilla of Vater. Patients with gastric ulcers in the prepyloric region may also have higher basal acid output; however, patients with gastric ulcers in other regions of the stomach (the great majority of those with gastric ulcers) have normal or low levels of gastric acid secretion (82,83). The risk of gastric cancer is elevated among patients with non-prepyloric gastric ulcers (90) or with gastric ulcers in general (91,92), but the risk of pancreatic cancer apparently is not (93). In contrast, elevated risk of pancreatic cancer is generally found among patients with duodenal ulcers (91,93), individuals reporting a history of ulcers (of unspecified type) (94,95), or those who have had a gastrectomy procedure for ulcer treatment (60,94–100). Still, a few studies have not shown this association (56,101–103). With minor exceptions, most studies find that duodenal ulcers are not associated with an increased risk of gastric cancer [for review, see (102)]. Because duodenal ulcers are frequently associated with a history of cigarette smoking, it is also possible that some of the observed association with pancreatic cancer could be attributable to the smoking; other smoking-related diseases occur at greater-than-expected frequency among individuals with duodenal ulcers. However, one report (104) that adjusted for smoking showed an elevated risk of pancreatic cancer among peptic (mostly duodenal) ulcer patients, and a statistically significant positive association was seen after adjustment for smoking in the study of Adventists (94). Finally, there is some evidence that duodenal ulcer patients treated with vagotomy and gastrojejunostomy or pyloroplasty may have elevated gastric concentrations of nitrites and N-nitroso compounds (41,105–107), which could be involved in the pancreatic neoplasia.
|
HELICOBACTER PYLORI AND GASTRIC/DUODENAL ACIDITY |
|---|
H. pylori colonization of the gut is among the most common infections of humans, especially in developing countries, where colonization is almost universal (108). Approximately one-third of adults in the United States are colonized, with greater prevalence of infection among non-whites, individuals of lower socioeconomic status, and older persons (108). Colonization appears to last for decades or perhaps indefinitely unless adequately treated (109,110). The majority of individuals colonized with H. pylori are asymptomatic and ostensibly disease-free, although the infection does generally affect gastric function (111). The occurrence of both gastric and duodenal ulcers is strongly associated with H. pylori infection; statistically significant odds ratios of three- to fourfold for H. pylori seropositivity are typically obtained (112,113). Studies outside the United States find more than 80% of patients with gastric or duodenal ulcers or endoscopically confirmed non-ulcer duodenitis to have evidence of H. pylori infection (112,114–116); in the United States, 50%–65% of duodenal ulcer cases, and a similar fraction of gastric ulcer cases not related to nonsteroidal anti-inflammatory drug usage, are attributable to H. pylori (108). Different strains of H. pylori may be involved in gastric versus duodenal ulcers (117), age at colonization may define the risks for the two types (118,119), or dietary factors may be involved (118,119). However, host acid-secretory capacity and inflammatory cytokine genetic factors most likely underlie the divergent physiologic and colonization behaviors of the organism (120).
Many mechanisms have been proposed for how H. pylori induces duodenal or gastric ulcers. Mechanisms for duodenal ulcers generally involve H. pylori-related hypergastrinemia or increased sensitivity to gastrin and consequent hyperacidity (121). The hyperacidity leads to duodenal gastric metaplasia and colonization of the duodenum by H. pylori, mucosal inflammation, loss of duodenal mucosal bicarbonate, and ulceration (120). Interleukin 1
(IL-1) and other inflammatory cytokines are involved in the inflammatory process (120). In general, H. pylori flourishes on the surface of non-acid-secreting gut epithelium, particularly the gastric antrum, colonizes the corpus only when acid production there is inadequate or suppressed (122,123), and is lost from the corpus with progressive atrophy and other preneoplastic changes (124). Colonization of the antrum by H. pylori has little effect on the number of G cells but involves specific reductions in the number and function of antral D cells (125–136), possibly mediated by luminal ammonia produced by the organism (125). Somatostatin levels in the antral mucosa are decreased because of a reduction in the number of D cells, but synthesis of somatostatin mRNA per D cell is also reduced (126,130). The suppression of somatostatin may also disinhibit ECL cell histamine secretion (137,138). Thus, the net effects of antral H. pylori colonization are paracrine disinhibition of antral G-cell function, hypergastrinemia, and hyperacidity. The suppression of antral D cells is observed in normal, asymptomatic H. pylori carriers (127,130,131), not just in carriers with duodenal ulcers or non-ulcer dyspepsia. Asymptomatic H. pylori carriers have higher basal and postprandial plasma gastrin levels than normal individuals (130,139).
|
H. PYLORI, DUODENAL ULCERS, SECRETIN, AND BICARBONATE |
|---|
A second function of somatostatin is to inhibit the secretion of secretin by S cells of the duodenum and proximal jejunum (73). In H. pylori carriers, a consequence of the hyperchlorhydria and suppressed somatostatin is a general increase in secretin and, perhaps more importantly, pancreatic bicarbonate output. Basal pancreatic bicarbonate may be particularly increased, as observed in duodenal ulcer patients compared with normal subjects (140–142). Whether stimulated (postprandial) pancreatic bicarbonate is increased in duodenal ulcer patients or in asymptomatic H. pylori carriers is uncertain, because these individuals tend to have a different time course of pancreatic response to the secretin than normal subjects; thus, studies testing all subjects at a fixed latency after stimulation may not detect a relevant difference (143). In any event, the basal acid hypersecretion results in increased basal secretin release (142). Over the day, the highest plasma secretin levels occur in a basal context during the late nighttime and early morning hours (144).
|
SECRETIN (OR BICARBONATE) AND PANCREATIC CANCER |
|---|
Although somewhat weaker than that of CCK, secretin appears to have some trophic effect on murine pancreatic growth and DNA synthesis (11,145–149). The magnitude of this effect appears insufficient to increase the risk of pancreatic neoplasia (16). Studies have generally examined trophic effects on the whole pancreas, and if the effect is specific to the ductular epithelium, then the small amount of ductular epithelium in the pancreas would limit how much trophic effect could be detected. In the hamster-BOP N-nitrosamine model, however, adding a very low dose of secretin to the carcinogen regimen statistically significantly accelerated the development and frequency of pancreatic ductular cell dysplasia, ductal carcinoma in situ, panlobular ductular proliferations, and frank ductular adenocarcinomas (16). Thus, it is plausible that low-level, prolonged increases in secretin or pancreatic bicarbonate production increase the rate of ductular epithelial cell activity and turnover sufficiently to enhance the carcinogenic effect of applied (and by inference, environmental or endogenous) N-nitroso carcinogens. In hamsters, short continuous treatment with HPOP yields a high frequency of pancreatic ductular tumors. This carcinogen produces appreciable numbers of ductular epithelial cell DNA adducts and strongly increases the mitogenic behavior of these cells (15). A majority of the ductular cells incorporating the carcinogen divide for one or two cycles and then die, suggesting that the amount of adduct damage in this assay is rather high and that ductular cells remaining viable still may have appreciable genetic damage from the adducts and thus be at increased risk of neoplastic transformation. Pancreatic acini also undergo a similar degree of adduct damage, but the relative increase in mitotic activity in acini is much smaller than that in ductular epithelium, a fact that may account for the lack of acinar cell tumors observed in this model (15). This observation is consistent with the reasoning that factors increasing ductular epithelial cell proliferation (such as secretin or the stimulation of ductal bicarbonate production in general) in the presence of pancreas-specific carcinogens should increase the risk of neoplastic transformation. Individuals with increased basal secretin or pancreatic bicarbonate activity, such as those having asymptomatic H. pylori colonization, non-ulcer dyspepsia, or duodenal ulcers, with dietary or other exposure to N-nitroso carcinogens, therefore, would be at increased risk for developing pancreatic cancer.
Parenthetically, a similar mechanism of pancreatic neoplasia, involving N-nitrosamine exposures along with excessive CCK stimulation, has been proposed (150). CCK is a stimulator of pancreatic growth and protein and DNA synthesis (147,149, 151), with a major trophic effect in the acini (145,152,153). Dietary proteases, found in raw soy flour and fermented raw soybeans (but not heat-treated flour or beans), strongly increase CCK secretion and its pancreatic stimulation (154) and, in large doses, induce murine pancreatic adenomas and invasive cancers (155). However, there is little empirical evidence from humans to support this hypothesis. Raw soy products or other foods containing dietary proteases constitute a minuscule fraction of the normal western diet. One study that examined fasting serum CCK levels failed to detect a difference between case patients with pancreatic cancer and control subjects (156). Finally, it is unclear whether factors that exert their primary physiologic effects on pancreatic acini are relevant for adenocarcinomas arising from the ductular epithelium.
|
H. PYLORI AND PANCREATIC CANCER |
|---|
Two recent reports have demonstrated positive associations between H. pylori seropositivity and risk of pancreatic cancer. The first study, performed in Austria (157), found an odds ratio of 2.1 (95% confidence interval = 1.1 to 4.1) for pancreatic cancer associated with colonization. The second study, performed in Finland, was a matched case–control analysis (158) within the Alpha-Tocopherol, Beta-Carotene (ATBC) Cancer Prevention Study, a randomized, controlled, prospective study of male smokers only. This study, adjusted for years of smoking, found a positive association between H. pylori colonization and risk of pancreatic cancer (odds ratio = 1.87, 95% confidence interval = 1.05 to 3.34), with little difference between CagA-positive and -negative subjects. Study subjects harboring H. pylori were almost twice as likely as those without H. pylori to report a history of peptic or duodenal ulcer (158). These two studies were fairly small, with 92 and 121 patients with pancreatic cancer, respectively, yet still showed statistically significant positive associations.
Although H. pylori does appear to colonize the bile ducts and may be responsible for a proportion of malignant biliary tract disease (159,160), there is no evidence for direct pancreatic colonization by the organism. The study in Austria (157) obtained tumor and surrounding normal tissue specimens from 20 case patients and observed no bacteria in any of the malignant tissue or in any adjacent normal pancreatic ducts. Researchers in a study in Sweden identified Helicobacter genus-specific DNA in five of six pancreatic ductal adenocarcinoma biopsy specimens but, with H. pylori-specific primers in the polymerase chain reaction assays, detected no H. pylori (160). Two normal pancreatic tissue biopsy specimens were negative for Helicobacter by genus-specific polymerase chain reaction primers (160). Pancreatic juice exerts some antibacterial activity (161) and may limit the ability of H. pylori to colonize the pancreas.
|
MODULATION OF GASTRIC ACIDITY BY INFLAMMATORY CYTOKINES |
|---|
Host inflammatory cytokine mechanisms play a major role in initiating and amplifying the inflammatory response to H. pylori (162–164). These cytokines, especially IL-1
but also tumor necrosis factor- and IL-10, modulate gastric acid production (165–167) and, in particular, have strong suppressive effects (120). For example, carriage of the -31T allele of the IL-1B gene is strongly related to hypochlorhydria in H. pylori-colonized normal individuals (168) and gastritis patients (169). Thus, in addition to host variation in acid-secretory capacity, host genetic variation in inflammatory cytokine mechanisms may be involved in the ability of H. pylori to colonize the gastric corpus from the antrum (168,170). Recent studies show that risk of non-cardia gastric cancer is strongly associated with polymorphisms in the cytokine genes IL-1B, tumor necrosis factor-A, and IL-10 (168,171,172): threefold, fivefold, and more than 20-fold increased risks were associated with the presence of one, two, or three or more variants (171). Risks for cardia and esophageal adenocarcinomas, however, were not associated with these polymorphisms (171). It is apparent that, given colonization by H. pylori, host polymorphisms in these genes are related to the climate of gastric acidity and may determine, at least in part, the divergent behavior of the organism, leading typically either to hypoacidity, corpus atrophy, and gastric cancer or to hyperacidity, possibly duodenal ulcer, and (as suggested herein) pancreatic cancer. |
OTHER RECENT HYPOTHESES ON THE ETIOLOGY OF PANCREATIC CANCER |
|---|
Cadmium
An interesting review (173) proposed that various aspects of the epidemiology of pancreatic cancer might be explained by long-term exposure to cadmium, which accumulates in the pancreas. Risk factors for pancreatic cancer, related to cadmium exposure, include cigarette smoking and work in certain occupations or industries involving metal welding or soldering, pesticides, paints, or batteries (173). Schwartz and Reis (173) suggested that molecular substitution of cadmium for zinc might underlie its carcinogenic mechanism. In rodent models, cadmium is both mitogenic and carcinogenic to the pancreas but apparently not to the ductular epithelium per se (174,175). Cadmium accumulation in the human renal cortex is approximately 30 times that in the human pancreas (176,177), although in animal models, administration of cadmium does not cause kidney tumors (175). It is perhaps most relevant for this review that cadmium appears to potentiate the effect of secretin on the volume of pancreatic fluid output (178), without involving CCK (178). Although dietary sources of cadmium such as shellfish, cereals, roots, and tubers are common (179), cigarette smoking may be a more important source, even in areas with high cadmium contamination (180). Absorption and metabolism of dietary cadmium may be strongly affected by intake of zinc (181) or fiber (182). Aside from cigarette smoking, occupational sources may provide nontrivial exposures to cadmium [for review, see (173)]. However, a meta-analysis of occupational studies examining cadmium exposures did not find an increased risk of pancreatic cancer (183).
Pyrolysis Carcinogens
Cooking at high temperature or incomplete burning of organic materials produces low but measurable levels of mutagenic heterocyclic amines and PAHs that may be carcinogenic to the pancreas and other organs. Heterocyclic amines such as 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine are typically found in pan-fried, grilled, or barbecued meats and fish, particularly when cooked to "very well done," and in gravies made with pan drippings from such foods (184). Various heterocyclic amines are also found in mainstream and sidestream tobacco smoke and in diesel exhaust particulates (185). PAHs are typically found in tobacco smoke, automotive exhausts, and smoke from the burning of coal tar (186). Although heterocyclic amines and PAHs are generally mutagenic and can produce DNA adducts in the pancreas, they do not produce ductal tumors by themselves and thus are not pancreatic carcinogens per se (187). Two heterocyclic amines (3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole [Trp-P-1] and 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline [4,8-DiMeIQx]) are promoters for hamster pancreatic ductal adenocarcinoma after BOP administration (see Table 1); however, 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine and various other heterocyclic amines are not promoters with BOP (188). There is little evidence that exposure to these substances increases risk of pancreatic cancer in humans. Trp-P-1 and related heterocyclic amines are inactivated by nitrite, under mildly acidic conditions and nitrite concentrations that are generally seen in saliva and in the stomach (189). In a meta-analysis (183), occupational exposures to diesel exhausts and to PAHs were not associated with statistically significant increased risks. A recent case–control study of meat intake, cooking methods, and pancreatic cancer suggested that risk was elevated for frequent consumption of grilled or barbecued red meat and possibly for fried red meat (190). This study, however, lumped together consumption of beef steak and ground beef; ground beef appears to yield almost no heterocyclic amines when grilled, barbecued, or fried to rare, medium, or well done (184). Finally, in another case–control study, no statistically significant risk associations were observed for functional germline polymorphisms in phase I enzymes that metabolize heterocyclic amines and PAHs to their active forms (191).
Folate
Folate deficiency may affect the risk of developing pancreatic cancer. Two studies have examined dietary intake (192) and serum levels (193) of folate in male smokers within the ATBC prospective cohort study. Both dietary and serum folate, measured at study baseline, were associated with statistically significant, decreasing trends in risk of subsequent pancreatic cancer. Use of folate vitamin supplements was not associated with risk, however (192). Two previous case–control studies have examined folate intake and the risk of pancreatic cancer, with conflicting results: a study in Australia did show a statistically significant trend of decreasing risk with increasing folate intake (59), but a larger study in the United States found no association (57). Low-folate diets may result in impaired acinar cellular DNA synthesis, although amylase secretion is not usually reduced (194). Because fruits and vegetables are the major dietary source of folate, it is not clear whether the reduced risks observed for dietary or serum folate result from the folate itself or from some other component of these foods. Most studies of diet and pancreatic cancer have detected a decreased risk with the consumption of fruits and vegetables (56,57,59,61,63–67,70, 94,195).
Lycopene
Higher lycopene intake may be associated with a reduced risk of pancreatic cancer. A recent case–control study found that plasma levels of lycopene were statistically significantly lower in case patients with pancreatic cancer than in matched control subjects (196). Stronger evidence comes from the Washington County, MD, Cohort Study that found, in subjects followed for almost 15 years, statistically significantly lower baseline serum levels of lycopene in case patients with pancreatic cancer than in matched cohort control subjects (197,198). Dietary lycopene is also found in fruits and vegetables, particularly tomatoes and tomato products. Lower serum lycopene levels have been seen in individuals with certain chronic inflammatory conditions, suggesting that lycopene intake reduces the presence, extent, or degree of the inflammation or that lycopene is depleted by the inflammatory process (199). No studies to date have analyzed lycopene intake and risk of pancreatic cancer.
Glucose Intolerance and Diabetes Mellitus
The association between adult-onset diabetes mellitus (type 2 or non-insulin dependent) and risk of pancreatic cancer has been known but controversial for some time [(102); for review, see (200)]. Many of the studies have involved self-reporting of diabetes history by case and control subjects. The controversy arises because of difficulty in determining whether pancreatic cancer in its prediagnostic phase could suppress or damage pancreatic islet cells or whether the diabetes could induce or promote development of the cancer. Most likely, both mechanisms are operative. Two cohort studies have observed statistically significantly increased risks of pancreatic cancer associated with a history of diabetes at baseline of follow-up (201,202). A third study screened 35 658 male and female employees at baseline for postload plasma glucose level and then followed them for more than 25 years for mortality from pancreatic cancer (203). A statistically significant, direct trend in risk according to postload glucose level was observed; in addition, subjects self-reporting at baseline a history of diabetes were also found to be at statistically significantly higher risk. Excluding cases diagnosed within 5 years of baseline did not alter the findings. Obesity and lack of regular physical activity, which are involved in the development of diabetes, have also been associated with an increased risk of pancreatic cancer in cohort studies (201,204). Factors associated with abnormal glucose metabolism may thus play a role in the development of pancreatic cancer (203). A commentary on Gapstur et al. (203) suggested that elevated insulin levels might be acting as a growth promoter for pancreatic cancer (205). This idea could be substantiated by studies evaluating dietary glycemic index (