African Americans

Combining all types of cancer, African Americans have the highest death rate (Ward et al, 2004) at 1.4 times higher than that of white males and 1.2 times higher than that of white females. In a 2004 study, Chang and colleagues (2004) used the California Cancer Registry of the Cancer Surveillance Section of the California Department of Health Services to determine the incidence of PDA between 1988 and 1998. African Americans had a higher age-adjusted incidence rate (8.8 per 100,000) for PDA when compared with all other races and ethnicities combined. African Americans were more likely to be seen with advanced disease and were least likely to receive surgical treatment, regardless of stage of disease (82.9% had no surgery compared with 77% of Asians, 67.8% of Hispanics, and 62.5% of non-Hispanic whites). Interestingly, serum cotinine levels, the primary metabolite of nicotine, are consistently higher in African Americans than in whites and Mexican Americans, even after adjustment for the number of cigarettes smoked per day, the number of smokers in the home, the number of hours of environmental tobacco smoke (ETS) exposure at work, the number of rooms in the home, and the region of the country where the subject lived (Caraballo et al, 1998). This suggests that genetic differences in cigarette product metabolism may influence rates of PDA.

Asians

Asian patients with PDA tend to have less aggressive tumors than do non-Asians (either black or white patients) and higher survival rates when assessed on a stage-adjusted basis (Clegg et al, 2008). To determine whether Asians develop a histologically different type of PDA than Western populations, Longnecker and colleagues (2000) conducted a population-based study using three SEER geographic areas: Hawaii, San Francisco, and Seattle. They compared PDA cases in Japanese, Chinese, Filipino, Hawaiian, black, and white patients. The study revealed that Japanese patients had the highest fraction of localized tumors with the lowest grade, and that Chinese, Filipino, and Japanese women had longer survival times than did whites, although survival time was significantly different for Japanese women only. Race-related genetic and environmental exposure factors may be involved in these survival discrepancies.

Ashkenazi Jews

Pancreatic cancer occurs in excess in Jews, particularly Jews of Ashkenazi heritage (Lynch et al, 1996). The age-standardized rates for PDA in Jews in Israel significantly exceed the rates for Israeli non-Jews (7.2 per 100,000 for all Jewish men and 5.7 per 100,000 for Jewish women vs. 4.0 per 100,000 for non-Jewish men and 2.9 per 100,000 for non-Jewish women). Approximately 5% to 10% of this discrepancy is attributed to the BRCA2 mutation, which is associated with breast, gastric, ovarian, and bile duct cancers.

Familial Pancreatic Cancer

Family history of pancreatic cancer is a strong risk factor in 5% to 10% of cases (Lowenfels et al, 2005). When a history of PDA exists in a close relative, a number of responsible genes have been identified (Wang et al, 2009). In the National Familial Pancreas Tumor Registry (NFPTR) at Johns Hopkins, prospective studies of family members of pancreas cancer kindreds found a twofold increased risk of pancreas cancer in the first-degree relatives of persons with sporadic pancreas cancer and a ninefold increased risk in first-degree relatives of those with familial pancreatic cancer (Klein et al, 2004; Wang et al, 2009). Germline mutations of BRCA2 are found in 6% to 19% of familial pancreatic cancer patients (Murphy et al, 2002). When a more stringent definition of familial pancreas cancer is used, a 57-fold increased risk of PDA was reported in kindreds with three or more family members affected with PDA (Tersmette et al, 2001). This corresponds to an incidence rate of 301 cases per 100,000 per year compared with the SEER age-adjusted rate for the entire U.S. population of 8.8 cases per 100,000 per year. A recent report identified a germline truncating mutation of the PALB2 gene in 3% of familial pancreas cancer patients (Jones et al, 2009; Blackford et al, 2009). Despite these many advances in the understanding of the mutations involved in the pathogenesis of PDA, no single “pancreas cancer gene” has been identified.

Six Genetic Syndromes Associated with Pancreatic Ductal Adenocarcinoma

Six genetic familial syndromes and their respective predisposing genes have been identified and linked to the development of PDA; these are described briefly below (see also Table 58A.2). Although individuals with these syndromes have an increased risk of developing PDA, collectively these syndromes account for less than 5% of the familial aggregation of PDA. The mean age of onset of familial PDA is similar to that of nonfamilial cases: 65.8 years versus 65.2 years (Hruban et al, 1998). Familial cases have also been noted to have a somewhat increased incidence of secondary primary cancers (23.8%) when compared with their nonfamilial counterparts (18.9%).

Table 58A.2 High-Risk Genetic Disorders Associated with Familial Pancreatic Cancer

FAMMM, Familial atypical multiple mole melanoma; PDA, pancreatic ductal carcinoma

Tobacco Exposure

Cigarette smoking is a contributing factor in approximately 20% to 25% of the cases of PDA and is the most consistently reported risk factor (Iodice et al, 2008; Jemal et al, 2009). Smoking has been associated with increased risk of PDA in at least 29 epidemiologic studies (Silverman et al, 1994), and smokers have a 70% increased risk of PDA compared with nonsmokers. Smoking habits in the 15 years preceding the diagnosis of PDA appear to be more relevant to increased risk, whereas former smokers who have quit for more than 13 years decrease their PDA risk to that of a lifetime nonsmoker (Howe et al, 1991).

Prospective and retrospective studies have found that the risk of PDA increases consistently with cigarette smoking but only inconsistently with cigar or pipe smoking (Wynder, 1975), suggesting perhaps that smoke must be deeply inhaled to exert its carcinogenic effects. A dose-response relationship between PDA and the number of cigarettes consumed has been documented in several investigations (Ahlgren, 1996; Howe et al, 1991). For current smokers who also have a family history of PDA, the relative risk of PDA has been reported to be as high as 8.23 (Schenk, 2001). Lowenfels and colleagues (1997) found that individuals with familial PDA tend to smoke more than those with sporadic PDA. These observations raise the possibility that smoking is interactive, perhaps multiplicatively, with genetic mutations known to be present in persons with familial PDA. Postmortem examinations of the pancreatic ducts of smokers have found widespread ductal hyperplasia, now called pancreatic intraepithelial neoplasia (PanIN), considered to be a premalignant lesion. Neugut and associates (1995) found a relationship between cigarette smoking and the development of PDA as a second malignancy in patients with a smoking-related first malignancy, such as lung, head and neck, or bladder cancer. This relationship is likely due to overlapping smoking-related genetic mutations among these different malignancies.

The socioeconomic status (SES) gap among smokers is widening in the United States and globally. Decreases in smoking prevalence in the U.S. population as a whole have not been realized in those from a lower SES background (Gilman et al, 2003). In a 2003 study of 657 adults aged 30 to 39 years that used parental occupation, adult educational attainment, and household income as indicators of SES, those of a lower SES were more likely to start smoking and to become regular smokers and were less likely to quit smoking than their higher SES counterparts.

The carcinogenic components of cigarette smoke are usually cleared from the bloodstream and excreted into bile; they reach the pancreas from the bile duct through biliary-pancreatic reflux, or they may be carried into the pancreatic parenchyma directly by the bloodstream. Cigarette smoke contains more than 60 known carcinogens, including polycyclic aromatic hydrocarbons (PAHs), nitrosamines, benzo[a]pyrene, β-napthylamine, methylfluoranthenes, and arylamines (Hecht, 2003). These carcinogens may bind to DNA and form adducts, which increase the risk of somatic mutations and cancer if unrepaired. Nitrosamines are known to be organ specific, causing pancreatic carcinomas in hamsters that are histologically similar to the type found in humans (Wang et al, 2009). Little research has been conducted on the other carcinogens in cigarette smoke and their relationship to PDA. Another possible mechanism by which cigarette smoke leads to PDA is unrelated to the carcinogens in the smoke. Rather, higher cholesterol (lipid) levels are measured in smokers than in nonsmokers. The pathophysiologic mechanism of hypercholesterolemia may be partially due to nicotinic stimulation of circulating catecholamines, which increase serum cholesterol levels. Elevated lipid levels may be causative in the induction of PDA (Wang et al, 2009).

Mulder and colleagues (1999) developed multicountry computer-based models to estimate the impact of smoking cessation on the incidence of PDA in the European Union (EU). According to these models, the EU would reach a projected 600,000 new cases of PDA per year by 2020 if the present smoking levels were to continue. If all current smokers quit smoking, the predicted number of incident cases of PDA per year in the EU would be 175,000 by 2020. If 45% to 50% of smokers were able to successfully quit smoking, the reduction of new cases would still be a substantial 68,000 cases per year.

High-Risk Occupations

The evidence linking occupational exposures to PDA is inconsistent, reflecting the difficulty of quantifying workplace exposure to carcinogens and of differentiating these exposures from other risk factors. A number of epidemiologic investigations have suggested excess risk of pancreatic cancer in certain occupations. Definitively establishing certain occupations as high risk for the development of PDA is difficult because of the problems with self-reported exposures, lack of objective quantitative monitoring data, personal comorbidities, and the presence of other confounding and modifying risk factors. The available studies are summarized in Table 58A.3.

Occupational Exposures

Studies examining occupational exposures that increase a worker’s risk of developing PDA have been conducted both in the United States and in Europe, particularly in the Scandinavian countries. Occupational exposures linked to PDA are summarized in Table 58A.4. Overall the occupational etiologic fraction for PDA was estimated at 12% in a meta-analysis of 20 occupational studies conducted between 1969 and 1998 (Ojajarvi et al, 2000). A retrospective analysis of more than 22 occupational and environmental studies reported by people with PDA found that exposure to asbestos, pesticides and herbicides, residential radon, coal products, welding products, and radiation were the most commonly reported exposures (Yeo et al, 2009).

Table 58A.4 Occupational Exposures Linked to Pancreatic Cancer

DDD, Dichlorodiphenyl-dichlorethane; OR, odds ratio; RR, relative risk

Data from references cited in Table 58A.3.

Calvert and colleagues (1989) conducted a review supported by the National Institute for Occupational Safety and Health (NIOSH) of five cohort studies reporting an association between the use of metalworking fluids (MWFs) used in industrial machining and grinding operations and the development of all types of cancer, including PDA (Acquavella et al, 1993; Hoppin et al, 2000; Porta et al, 1999; Rotimi et al, 1993; Tolbert et al, 1992). More than 1 million workers are exposed to MWFs according to NIOSH estimates (Calvert et al, 1989). Substantial evidence was found for an increased risk of cancer at several sites, including the pancreas, larynx, rectum, skin, scrotum, and bladder. MWFs contain a number of compounds suspected to be cancer initiators or promoters including long-chain aliphatics, polyaromatic hydrocarbons, nitrosamines, sulfur-containing compounds, formaldehyde-releasing biocides, and heavy metals (Tolbert et al, 1992).

Diabetes Mellitus

Two theories regarding the relationship between type 2 diabetes mellitus (DM) and PDA have been postulated; in one theory, DM is viewed as a predisposing risk factor for PDA; in the other theory, type 2 DM and its attendant altered glucose homeostasis are considered a consequence of PDA (Yalniz et al, 2005). The majority of PDA patients (80%) have either subclinical impaired glucose tolerance or type 2 DM. Many investigators regard DM as a clinical manifestation of PDA rather than a risk factor, a result of the alteration of islet cell function—specifically, the loss of β-cell mass from tumor growth—or a result of disruptions in acinar–islet cell interactions.

A meta-analysis of epidemiologic studies focused on DM and the risk of PDA found that the pooled relative risk (RR) of PDA in people diagnosed with DM 5 years or more prior to the PDA diagnosis was twice the risk (RR, 2.0; 95% CI, 1.3 to 2.2) of those without DM (Everhardt et al, 1995). In a population-based case-control study of 484 PDA cases and 2099 controls in three geographic areas of the United States, those diagnosed with DM at least 10 years prior to the development of PDA had a 50% increased risk of PDA compared with the control group (Silverman et al, 1998). Insulin treatment did not appear to alter the risk of development of PDA (OR, 1.6 with insulin and 1.5 without).

During an 18-year follow-up study of 88,802 women in the Nurses’ Health Study (NHS), 180 incident cases of PDA were reported (Michaud et al, 2001). A positive association was observed between fructose intake and impaired glucose metabolism and increased PDA risk. The risk was most apparent in women with an elevated body mass index (BMI) above 30 kg/m² and low physical activity. A diet high in glycemic load (GL), the glucose response to each unit of carbohydrate-containing foods, may increase the risk of PDA in women with underlying insulin resistance as a result of obesity. Chronically elevated plasma glucose levels and increased PDA risk were also observed by Gapstur and colleagues (2000), further suggesting that impaired glucose tolerance, insulin resistance, and hyperinsulinemia may be involved in the etiology of PDA.

Chronic Pancreatitis

Chronic pancreatitis has been linked to the development of PDA. It is not clear whether chronic pancreatitis is a risk factor or if it represents an indolent presentation of PDA. The standardized incidence ratio of PDA in individuals with a diagnosis of chronic pancreatitis was 16.5 compared with the 1.76 expected number of cases (Lowenfels et al, 1993). Because PDA may mimic chronic pancreatitis, misdiagnosis is a possibility. A review of the records of patients with acute, chronic, or unspecified pancreatitis was conducted on all inpatient medical institutions in Sweden from 1965 to 1983. In 7956 patients with pancreatitis, 46 pancreas cancers were diagnosed compared with 21 expected cases—a standardized incidence ratio of 2.2 (CI, 1.6 to 2.9) (Ekbom et al, 1994). Although it seems reasonable to conclude that the cellular destruction and glandular dysfunction caused by pancreatitis may yield an environment favoring the initiation of tumor growth, it remains problematic that some premalignant lesions may be initially incorrectly diagnosed as chronic pancreatitis.

Cholecystectomy

A history of cholelithiasis and/or cholecystectomy may precede the diagnosis of PDA, yet cholecystectomy per se is not believed to promote its development. Individuals who underwent cholecystectomy in the year preceding the diagnosis of PDA had an exceedingly high risk of developing PDA (OR, 57.9) in one investigation (Everhardt et al, 1995). This finding is most likely explained by the misdiagnosis of gallbladder disease, either cholelithiasis or cholecystitis, in the setting of cryptic PDA, often prompting gallbladder removal with an undetected tumor being left in place. However, when data subset analyses were performed in patients with a history of cholecystectomy 20 years or more prior to the diagnosis of PDA, a 70% excess risk of PDA was still evident.

Hormonal Factors

In recent years, a link between female reproductive factors and PDA has been hypothesized. A case-control study of 52 postmenopausal women with PDA and 233 matched controls was conducted as a component of the Canadian Enhanced Cancer Surveillance Project (Kreiger et al, 2001). A questionnaire focused on reproductive history, cigarette smoking, physical activity, diet, occupation, residential history, and sociodemographic information was mailed to eligible patients diagnosed with PDA. Multiparity, defined as three or more children, and use of oral contraceptives were associated with a decreased risk of PDA (OR, 0.22 and 0.36, respectively). Late age at birth of first child significantly increased the risk of PDA (OR, 4.05; CI, 1.50 to 10.92 for ages 25 to 29 years; for women 30 or older, OR, 3.78; CI, 1.02 to 14.06). No relationships were found between age of menarche or menopause and PDA. The inverse relationship between parity and PDA suggests that PDA may be partially an estrogen-dependent transformation, or that estrogen may act in an inhibitory manner upon pancreatic carcinogenesis.

Lifestyle Factors and Height

Michaud and associates (2001) reported on the relationship between BMI, height, physical activity, and smoking and the risk of PDA in two large, prospective cohort studies: the Health Professional Follow-up Study (HPFS) and the NHS. Activity levels and body weight were ascertained prospectively. A higher risk of PDA was found among obese men and women, as was a direct association between above-average height and risk of PDA. An additional 2.54 centimeters of height above average increased the risk of PDA by 6% in the HPFS and by 10% in the NHS. The association between height and cancer risk in general has been identified in other studies (Giovannucci et al, 1995; Smith et al, 1998). Height may serve as a proxy for calorie intake or exposure to growth factors, such as insulin or insulinlike growth factor-1, in childhood.

Furthermore, an inverse relationship between moderate physical activity and PDA was observed. Walking or hiking 1.5 hours or more per week in the overweight cohort was associated with a 50% reduction in PDA risk, but physical activity had no effect on PDA risk for participants who were not overweight. Likewise, BMI had no effect on risk if the participant was a moderate exerciser. However, individuals with a BMI of at least 30 kg/m² had an elevated risk of PDA (RR, 1.72; CI, 1.19 to 2.52) compared with those with a BMI of less than 23 kg/m². The authors speculated that the PDA risk associated with obesity might be linked to glucose intolerance, peripheral insulin resistance, and/or hyperinsulinism.

Diet

It is estimated that 30% to 50% of pancreatic cancers may be attributable to dietary factors. Both butter consumption and saturated fat intake were positively associated with PDA (hazard ratio [HR], 1.40; CI, 0.87 to 2.25; and HR, 1.60; CI, 0.96 to 2.64, respectively) in the Finnish Cancer Registry (Stolzenberg-Solomon et al, 2005). Fat entering the duodenum stimulates cholecystokinin secretion, and it is possible that chronic hypercholecystokininemia may be associated with an increased susceptibility of the pancreas to carcinogens. Other possible explanations include that the increased intake of saturated fat may lead to insulin resistance and the development of DM, or foods with a high soluble fat content may be contaminated with carcinogenic organochloride compounds from the environment.

Processed meats have been linked to the development of PDA. In a prospective study of more than 200,000 people, 482 individuals developed PDA during 7 years of follow-up (Nöthlings et al, 2007). Those who had consumed the greatest amount of processed meats had a 67% increased risk of PDA. Diets laden with pork and red meat intake were associated with a 50% increase in PDA risk, but poultry, fish, dairy product, and egg consumption conferred no additional risk. Heterocyclic amines and polycyclic aromatic hydrocarbons that form during high-temperature cooking are known to be carcinogenic and may be responsible for the increased risk of PDA observed with processed and barbecued meats. These compounds have also been linked to the development of other cancers.

Several recent investigations have focused on the relationship between glycemic load and glycemic index. Consumption of fruit juices and soda has been evaluated in six studies to date with conflicting results (Michaud et al, 2002; Mueller et al, 2010; Nöthlings et al, 2007; Silvera et al, 2005; Stolzenberg-Solomon et al, 2005). Two studies found no relationship between PDA and glycemic load, carbohydrate intake, or sugar, sucrose, or fructose intake (Johnson et al, 2005; Silvera et al, 2005). Three studies reported mixed findings (Michaud et al, 2002; Nöthlings et al, 2007; Stolzenberg-Solomon et al, 2005). In the Multiethnic Cohort Study, BMI modified the effect of high sucrose intake and the occurrence of PDA, with elevated BMI increasing the risk and normal BMI decreasing the risk (Nöthlings et al, 2007). A positive association with PDA was also observed with high fructose intake, but no association was observed with consumption of soft drinks (diet or regular).

Mueller and colleagues (2010) conducted a retrospective analysis of the Singapore Chinese Health Study regarding soft drink and fruit consumption and the risk of PDA. Among 63,257 participants, 142 incident PDA cases were identified. However, only 56.4% of the PDA cases were histologically confirmed, 5% were reported from death certificate data, and 38.8% were identified by signs and symptoms possibly related to PDA but not proven. The finding of an elevated HR for PDA in those who drink two to three soft drinks per week is suspect in this setting and requires further evaluation.

Coffee and Alcohol Consumption

Cohort studies from the 1970s and 1980s indicating that heavy coffee and alcohol consumption led to an excess risk of PDA were often confounded by excessive smoking among the heavy coffee and alcohol drinkers (Michaud et al, 2001). Three subsequent investigations have attempted to clarify this issue while controlling for smoking (Ghadirian et al, 1991; Lagergren et al, 2002; Michaud et al, 2001). Ghadirian and colleagues (1991) conducted a case-control study of residents of greater Montreal, obtaining all information through questionnaires administered to either the patient or a proxy. The results indicated that those who consumed alcohol were, in general, at lower risk for PDA (OR, 0.65; CI, 0.30 to 1.44). This was true for beer, white wine, and hard liquor. The odds ratio for red wine drinkers was elevated at 1.57, but it did not achieve statistical significance. Likewise, coffee drinkers had a lower overall risk of developing PDA (OR, 0.55; CI, 0.19 to 1.62), regardless of whether they drank caffeinated or decaffeinated coffee. The increased risk when coffee was taken on an empty stomach or between breakfast and lunch (OR, 1.30 and 1.15, respectively) was not statistically significant.

Lagergren and colleagues (2002) hypothesized that because heavy alcohol intake often causes chronic pancreatitis, which may be a risk factor for PDA, the ideal group to examine would be those with a diagnosis of alcoholism, alcoholic chronic pancreatitis, and alcoholic liver cirrhosis. Using the Swedish National Board of Health and Welfare, data were collected on 178,688 patients over a 30-year period. A total of 305 incident cases of PDA were identified in the group, representing a 40% excess risk in observed compared with expected cases. Alcoholics with chronic pancreatitis or cirrhosis had a twofold increased risk of PDA. A major limitation of the study was the lack of information on smoking. The authors estimated that the observed excess risk of PDA in the alcoholic group “may be almost totally attributable to the confounding effect of smoking” (Lagergren et al, 2002).

Michaud and colleagues (2001) examined data on coffee and alcohol consumption and other dietary factors obtained at baseline in two large, national cohort studies. The HPFS, initiated in 1986, includes 51,529 men aged 40 to 75 years who responded to a mailed questionnaire; the NHS began in 1976 and includes 121,700 registered nurses. Follow-up information was gathered via a mailed questionnaire from both groups on age, marital status, weight, height, medical history, medication use, smoking status, physical activity, and intake of coffee and alcohol. The results regarding coffee and alcohol are compelling: during 1,907,222 person-years of follow-up, 288 incident cases of PDA were diagnosed. Data were analyzed separately for each of the cohorts and then pooled to compute overall RR estimates. The results revealed that neither coffee consumption nor alcohol consumption conferred an excess risk of PDA; a pooled RR of 0.62 was reported (95% CI, 0.27 to 1.43) for more than 3 cups of coffee per day versus no coffee, and a pooled RR of 1.00 (95% CI, 0.57 to 1.76) for more than 30 g alcohol per day versus no alcohol were found.

Most recently, Brand and colleagues examined data on 29,239 histologically confirmed cases of pancreas cancer from 350 U.S. hospitals between the years 1993 and 2003 by cigarette smoking and alcohol consumption self-reported history (Brand et al, 2009). Current smokers were diagnosed with PDA at significantly younger ages (6.3 to 8.6 years) than nonsmoking counterparts. Among nonsmokers, age at diagnosis of PDA progressed linearly according to the amount of alcohol consumed, from 6 years younger for minimal drinkers to 8.7 years younger for self-described heavy alcohol consumers.