A large number of studies have examined the association between gallbladder disease and gallbladder removal and pancreatic cancer, but findings have been inconsistent. As nonspecific gastrointestinal symptoms resulting from an underlying pancreatic cancer can lead to the diagnosis of cholelithiasis and even to cholecystectomy prior to a pancreatic cancer diagnosis, studies that do not remove individuals with a recent diagnosis of gallbladder disease (<1 year) may be biased. Overall, a number of studies have reported an elevated risk of pancreatic cancer among individuals with a history of gallbladder disease (cholelithiasis) or with a history of gallbladder removal (cholecystectomy). In one study, a 70% increase in risk remained after allowing 20 or more years between cholecystectomy and tumor diagnosis (Silverman et al., 1999). In contrast, a number of other studies did not observe an association between cholelithiasis or cholecystectomy and pancreatic cancer risk, or in some studies the association was no longer positive after removing those who had cholelithiasis or cholecystectomy less than a year before interview or diagnosis of pancreatic cancer.
Most studies examining cholecystectomy or cholelithiasis did not control for obesity, which is a known risk factor for gallbladder disease; consequently, it is possible that previous positive associations were confounded by obesity. Given the lack in consistency across studies, it is unlikely that gallbladder disease or cholecystectomy play an important role in pancreatic carcinogenesis. However, a modest increase in risk ( 20%) for cholecystectomy cannot be completely ruled out.
Diabetes Mellitus
Adult-onset diabetes, or type 2 diabetes, has been consistently associated with an elevated risk of pancreatic cancer in observational studies. Nonetheless, there has been substantial controversy over the role of type 2 diabetes and in determining whether it is a cause or simply a consequence of pancreatic cancer. Reverse causation could not be ruled out as an explanation for positive findings from earlier case-control studies, because diabetes is often a manifestation of pancreatic cancer. To understand better the role of diabetes, a number of observational studies collected information on duration of diabetes. In 1995, a metaanalysis estimated that individuals with at least 5 years’ duration of diabetes had a twofold higher risk of pancreatic cancer than those without diabetes (95% CI 1.2 to 3.2) (Everhart and Wright, 1995). In an updated meta-analysis based on a larger number of studies, the association between chronic diabetes (5 or more years) and pancreatic cancer was slightly weaker than the earlier meta-analysis, but still statistically significant (RR 1.5, 95% CI 1.3–1.8, compared to nondiabetics) (Huxley et al., 2005). In this meta-analysis, results from 28 cohort studies were slightly weaker than those obtained from 22 case-control studies (Huxley et al., 2005).
Additional evidence for the role of diabetes comes from studies examining the relation between prediagnostic glucose levels and pancreatic cancer. To date, elevated postload or fasting glucose levels have been associated with a higher risk of pancreatic cancer in four cohort studies with 10 to 25 years of follow-up (Gapstur et al., 2000; Batty et al., 2004; Jee et al., 2005; Stolzenberg-Solomon et al., 2005). In these studies, relative risks for biochemically defined diabetes ranged between 1.7 and 4, and dose–response relationships were observed with increasing levels of glucose. In one study, the association between glucose levels (across nondiabetic ranges) and pancreatic cancer was stronger among cases whose blood had been collected 10 or more years prior to cancer diagnosis (Stolzenberg-Solomon, 2005). Taken together, the current data strongly support a causal role for type 2 diabetes in the etiology of pancreatic cancer.
History Of Allergies
A number of observational studies have examined the relation between allergies and pancreatic cancer, usually along with other exposures. In a recent meta-analysis of 14 studies, pancreatic cancer risk was lower among individuals with a history of any allergy than those without allergies (pooled RR 0.82, 95% CI 0.68–0.99) (Gandini et al., 2005). The association was stronger among those with respiratory allergies excluding asthma (RR 0.63, 95% CI 0.52–0.76), or those with dermal allergies (RR 0.66, 95% CI 0.49–0.89), but no association was found for asthma, which is not always related to atopy (RR 1.01, 95% CI 0.77–1.31) (Gandini et al., 2005). A recent prospective study observed similar findings; a history of hayfever alone was associated with a lower risk of pancreatic cancer (RR 0.85, 95% CI 0.77–0.95; 373 pancreatic cancer cases), but no association was observed for a history of asthma alone (RR 1.02, 95% CI 0.85–1.23) (Turner et al., 2005). The biological mechanisms for this association are not known, but it has been proposed that the enhanced immune surveillance (heightened Th2 response) among individuals with allergies can decrease carcinogenesis; for example, a Th2 response stimulates production of anti-inflammatory cytokines, such as interleukin-4, that have been shown to have anticarcinogenic properties (Turner et al., 2006).
Given that having allergies is not a modifiable risk factor, it is not directly obvious how this information might be used to prevent pancreatic cancer. However, understanding the underlying biological mechanisms could provide some important clues into pancreatic carcinogenesis and may present opportunities for the development of new agents for chemoprevention or treatment of this fatal malignancy.
Obesity
A number of early studies did not report any association between body mass index and pancreatic cancer. However, because weight loss is a common symptom of pancreatic cancer, it is possible that earlier studies did not obtain the patient’s weight several years prior to diagnosis. Furthermore, most of the earlier studies had low participation rates and relied heavily on next-of-kin (often called ‘proxies’) to obtain exposure data, such that measurement errors were likely to have occurred. Over the past 5 years, a large number of cohort studies (which are less prone to bias) have reported positive associations between obesity and pancreatic cancer. These data are difficult to refute, but because the evidence is very recent, a number of recent reviews on pancreatic cancer have failed to mention obesity as a potential risk factor. To date, at least 10 prospective cohort studies with a total of over 15 000 cases have reported an elevated risk of pancreatic cancer for obese individuals (BMI 30 kg/m2), compared to individuals of healthy weight (BMI <25), with relative risks typically between 1.5 and 2.0. Although a few prospective studies failed to detect an association with BMI, most of these did not include a category of exclusively obese individuals. In one study in which BMI was not associated with risk, a statistically significant 46% increase in risk was observed among individuals who had gained 12 or more kg as adults, compared to those who gained 2–5 kg. Elevated risks of pancreatic cancer have also been reported for overweight men and women in four recent case-control studies in which only direct interviews were used (i.e., the patient provided the responses, not a relative or friend).
A meta-analysis of 14 studies on obesity and pancreatic cancer risk estimated a 19% increase in risk among obese individuals compared to those with a normal body weight (RR 1.19, 95% CI 1.10–1.29, for BMI 30 kg/m2 vs. 22 kg/m2) (Berrington de Gonzalez et al., 2003). However, the results from this meta-analysis include all studies, regardless of study design, and therefore the overall effect is almost certainly underestimated. The relative risk estimates were higher when the authors excluded case-control studies with proxy data or those studies that had not adjusted for smoking in their analyses.
Obesity results in metabolic abnormalities, including hyperinsulinemia, insulin resistance, and impaired glucose tolerance, which can contribute to the development of diabetes. In vitro studies show that insulin promotes growth of hamster, rat, and numerous human pancreatic cell lines. Studies suggest that islet cell turnover, associated with insulin resistance, may play an important role in pancreatic carcinogenesis. Stimulation of islet cell proliferation enhances pancreatic ductal carcinogenesis in hamsters, and the destruction of islet cells by streptozotocin or alloxan inhibits cancer induction. In a recent study, hamster pancreatic cancer was inhibited by the drug metformin, which normalized insulin levels and the rate of islet cell turnover (Schneider et al., 2001). In a recent prospective study of Finnish male smokers, men with the highest levels of prediagnostic insulin had a twofold increase in risk of pancreatic cancer compared to men with the lowest levels (insulin was measured 5 or more years prior to cancer diagnosis; 95% CI 1.03–3.93) (Stolzenberg-Solomon et al., 2005).
Given the strength of the evidence, obesity is now an accepted risk factor for pancreatic cancer; however, the magnitude of the effect ( 20% increase in risk for every 5 kg/m2 increment in BMI) is not as strong as for other obesity-related cancers.
Physical Activity
Data on physical activity and pancreatic cancer are less consistent than for obesity. Hypothetically, because exercise is known to improve glucose tolerance, even in the absence of weight loss, this factor could be an important modifiable risk factor for pancreatic cancer. To date, five studies have reported a lower risk of pancreatic cancer with higher physical activity but an equal number of studies have reported no association. In one study, regular moderate exercise (vs. sedentary) was associated with greater than 50% reduction in risk of pancreatic cancer in both men and women (Michaud et al., 2001). The association was strongest among individuals who were obese (RR 0.59, 95% CI 0.37–0.94, for moderate/heavy exercise vs. sedentary), but was not apparent among those who were not overweight. Because lean individuals are less likely to have metabolic abnormalities, they may not benefit from exercise in the same way as overweight individuals. More studies are needed to elucidate the role of physical activity on pancreatic cancer.
Aspirin And Other Anti-Inflammatory Drugs
Data from experimental studies suggest that nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit pancreatic cancer. Furthermore, aspirin and other NSAIDs have been shown to decrease the risk of a number of gastrointestinal cancers, including colorectal, stomach, and esophagus. Inconsistent results have been found in studies examining the use of aspirin and other NSAIDs and pancreatic cancer. The Iowa Women’s Health Study is the only study to date that reported a strong inverse association between aspirin use and the risk of pancreatic cancer (RR 0.40, 95% CI 0.20–0.82, for use six or more times per week compared with no use) (Anderson et al., 2002a), but no information was available on duration of use. Despite a recent study reporting an increase in risk of pancreatic cancer among women taking five or more aspirin per week for more than 10 years (relative risk 1.78, 95% CI 1.18 to 2.60) (Schernhammer et al., 2004), several other studies have not reported any association with aspirin or NSAID use, even for long-term use of aspirin (RR 0.96, 95% CI 0.69–1.33, for 20 or more years of use compared with nonusers ( Jacobs et al., 2004)). Overall, aspirin and other NSAIDs do not appear to play a substantial role in pancreatic cancer.
Dietary Factors
Numerous case-control studies have examined the relation between diet and the risk of pancreatic cancer. For many dietary factors, results have been mixed. Inconsistencies in these findings may be due to a number of inherent problems with these studies, which include use of next-of-kin (proxies) to obtain dietary information, recall bias (when patients remember their past exposures, including diet, differently than healthy controls because of their disease status), low response rates among cases because of high fatality rates, and poor dietary assessment tools. For alcohol and coffee specifically, findings may be biased in population-based case-control studies because heavy alcohol and coffee drinkers may have lower participation rates than nondrinkers. Alternatively, findings may be biased in hospital-based case-control studies because alcohol and coffee may be related to the health conditions afflicting controls that are included for study. Therefore, it is important to look at prospective cohort studies when examining dietary factors in relation to pancreatic cancer risk, as dietary intakes are measured prior to disease diagnosis.
Coffee
At least 13 prospective cohort studies have examined the relation between coffee intake and pancreatic cancer risk; with three exceptions, results from these studies have been null. The International Agency for Research on Cancer (IARC) concluded, based on the existing literature in 1991, that there was little evidence to support a causal relation between coffee and risk of pancreatic cancer (IARC, 1991). The majority of cohort studies conducted since 1991 have been consistent with IARC’s conclusion of no association (World Cancer Research Fund, 1997).
Alcohol
Ten prospective studies have examined the influence of alcohol intake in nonalcoholic populations. Four studies reported an increase in pancreatic cancer with alcohol intake, but the remaining studies observed no associations between alcohol and pancreatic cancer. Given that smoking and alcohol are highly correlated, some of the elevated risks observed may have been due to residual confounding by smoking. However, it is also possible that heavy alcohol drinking does increase the risk of pancreatic cancer (Silverman, 2001); this observation may be mediated through chronic pancreatitis (which is often caused by excessive alcohol consumption). Overall, and given additional data from case-control studies (summarized in the World Cancer Research Fund report (World Cancer Research Fund, 1997)), it appears unlikely that moderate alcohol consumption plays a major role in pancreatic cancer.
Fruit And Vegetables
In 1997, a panel of experts concluded that consumption of fruit and vegetables was ‘probably’ associated with a lower risk of pancreatic cancer; this decision was largely based on case-control data, because few prospective studies had been published at that time (World Cancer Research Fund, 1997). Prospective studies, however, have not confirmed these studies; to date, no association has been observed in five prospective studies. It is likely, therefore, that findings from case-control studies were biased by differential recall of dietary intake or other selection factors, and that overall, consumption of fruit and vegetables is not associated with pancreatic cancer. It remains possible, nevertheless, that certain types of fruit and vegetables, or selected nutrients high in fruit and vegetables, do play a role in pancreatic cancer. For example, cruciferous vegetable consumption (e.g., broccoli, cauliflower, brussels sprouts, cabbage) may decrease the risk of pancreatic cancer (RR 0.70, 95% CI 0.43–1.13, for three or more servings per week compared with less than one serving per week (Larsson et al., 2006b); RR 0.5, 95% CI 0.4–0.8, for the highest compared with lowest quartile of cruciferous vegetable intake (Silverman et al., 1998)). Cruciferous vegetables contain a large variety of compounds with potential anticarcinogenic properties. Another example is folate, which is found in a variety of fruit and vegetables.
Folate
There is substantial evidence to suggest that folate plays a role in colon and breast carcinogenesis; it is involved in DNA repair and synthesis as well as DNA methylation and may thus play a role in a number of cancers. Three prospective cohort studies have reported similar findings for folate. A high dietary intake of folate (i.e., not including supplements) was associated with a 48% lower risk of pancreatic cancer in a Finnish cohort study of smokers (RR 0.52, 95% CI 0.31–0.87) (Stolzenberg-Solomon et al., 2001), and in the same cohort, pre-diagnostic serum folate levels were inversely associated with pancreatic cancer (RR 0.45, 95% CI 0.26–0.88, for high vs. low serum folate levels) (Stolzenberg-Solomon et al., 1999). In a Swedish prospective study, a strong association was reported for dietary folate intake (RR 0.25, 95% CI 0.11–0.59, for 350 vs. <200 mg of folate per day) (Larsson et al., 2006a). Finally, in a combined analysis of two U.S. cohorts, folate intake from diet alone was associated with a lower risk of pancreatic cancer (RR 0.66, 95% CI 0.42–1.03, for the highest vs. lowest quintile of folate intake), but the reduction in risk was not statistically significant (Skinner et al., 2004).
In contrast, folate from supplements (typically from multivitamins) has shown different results. Supplemental folate was associated with an elevated risk of pancreatic cancer in the study of Finnish smokers (RR 1.56, 95% CI 0.90–2.70, for reported users vs. nonusers) (Stolzenberg-Solomon et al., 2001). In the other cohorts, folic acid from supplements was not associated with risk (Skinner et al., 2004; Larsson et al., 2006a).