Risk Factors for Pancreatic Adenocarcinoma and Prospects for Screening

Abstract

Pancreatic cancer has one of the worst survival rates of any cancer and is the fourth leading cause of cancer mortality. Early detection and surgery are the patient's best chance for cure. However, symptoms are typically vague and occur when the cancer is unresectable. Population-based mass screening is not practical for this rare disease, though screening and early detection in asymptomatic high-risk patient populations may be indicated.

Pancreatic cancer has one of the worst prognoses of any type of cancer, with a 5-year survival rate of only 4.6% and a median survival rate of less than 6 months. Although pancreatic cancer accounts for less than 2% of new cancers, it is the fourth leading cause of cancer mortality and accounts for 6% of all cancer deaths.¹ The best chance for curing the disease is early detection and surgery, which has a 5-year survival rate of approximately 21% when combined with chemotherapy.² The frequent vagueness of symptoms and typically unresectable nature of the disease at the time of initial diagnosis are the Achilles heel of early detection. Population-based mass screening is not practical in a disease as rare as pancreatic cancer (which has an incidence of 8-12 per 100,000).¹ However, screening and early detection in asymptomatic high-risk groups may be indicated, and will be examined in further detail in this paper.

Defining Early Pancreatic Cancer

Fewer than 20% of patients with pancreatic cancer present with potentially curable disease. The 5-year survival rate remains only approximately 20% in patients undergoing operation with curable intent.^1,3 The major determinants of survival are tumor size, lymph node status, and extrapancreatic perineural invasion.^4-6 Tumors less than 1 cm in size with no nodal involvement carry a significantly better prognosis; in fact, Japanese researchers have had excellent results with early-stage disease. Ariyama and associates reported a 100% 5-year survival rate in patients with tumors less than 1 cm in size.⁷ In another study from Japan, at the time of resection, 26% of patients had T1 lesions (<2 cm), which conferred a 5-year survival rate of 48.1%, whereas T3 lesions (tumors extending beyond the pancreas but without involvement of the celiac axis or superior mesenteric artery) had only a 5-year survival rate of 27.9%.⁸ Studies in the United States have reported a 5-year survival rate of approximately 21% for resected pancreatic cancers and a higher rate (28%) in the subgroup of tumors less than 3 cm in size.⁵

Pancreatic cancer must be detected at an early stage to have a chance of being cured, and therein lies the problem: patients typically become symptomatic when the disease is incurable, and furthermore, the diagnostic tools at our disposal are poor at detecting small lesions in the pancreas.

Tools Currently Available for Early Detection of Pancreatic Cancer

Serologic Tumor Markers

Carbohydrate antigen 19-9 (CA 19-9) exists in tissue as an epitope of sialylated Lewis^a blood group antigen. Most clinicians are aware of the use of CA 19-9 as a marker of pancreatic cancer but are not necessarily aware of its sensitivity, specificity, and positive predictive value. Approximately 5% of individuals, those who are genotypically Lewis^a-b, lack the enzyme necessary to synthesize CA 19-9, further limiting its sensitivity.⁹ CA 19-9 lacks the sensitivity for detecting early pancreatic cancer and is elevated in only 50% of pancreatic adenocarcinomas less than 3 cm in size.¹⁰ In addition, data indicate that poorly differentiated pancreatic cancers may produce lower levels of CA 19-9 compared to well-differentiated tumors.¹⁰

CA 19-9 is also not specific for pancreatic cancer. It may be elevated in gastric cancer, colorectal cancer, chol-angiocarcinoma, and urothelial malignancy, as well as in benign conditions such as biliary obstruction, hepatitis, acute and chronic pancreatitis, and thyroiditis.^11-15

Among symptomatic patients, CA 19-9 has a sensitivity of 81-85% and a specificity of 81-90% when a cut-off of 37 U/mL is used.^10,16 As a screening test, CA 19-9 would be more useful in the asymptomatic population, but, unfortunately, it has a low positive predictive value. In a South Korean asymptomatic population, Kim and colleagues found its sensitivity and specificity to be excellent (100% and 98.5%, respectively) but noted a positive predictive value of only 0.9%, making CA 19-9 a very poor screening tool in asymptomatic patients.¹⁷ The American Society of Clinical Oncology does not recommend the use of CA 19-9 as a screening test for pancreatic cancer.¹⁸ Several other tumor markers, including K-ras, p53, and mucin, have been investigated; however, none of these markers is sensitive enough to be recommended for clinical use.

Pancreatic Imaging

Computed Tomography Although computed tomography (CT) is important as a screening tool, almost all of its studies have been performed in symptomatic patients. The sensitivity of CT ranges from 20% to 89%,^19,20 depending upon the tumor size. From a screening perspective, CT is poor at detecting lesions less than 1—2 cm in size²¹ and pancreatic adenocarcinoma in the background of chronic pancreatitis.

The quality and speed of CT imaging has significantly improved with the introduction of multidetector row spiral CT (MDCT). Furthermore, the use of a higher injection rate of iodinated contrast material (at 8 mL/sec) optimizes the bolus of intravenous dye to the faster acquisition speed of MDCT scanners. Compared to the standard flow rate of 4 mL/sec, this change results in greater pancreatic enhancement and greater conspicuity of the tumor relative to the background pancreas.²² MDCT enables high-resolution three-dimensional imaging, and coronal multiplanar images have become a routine component of the CT staging of pancreatic neoplasm.^23-25 Curved coronal images are ideal for displaying the full length of the pancreatic duct that leads to an obstructing mass or for unfolding a tortuous vessel to show a perivascular tumor abutting the walls of the superior mesenteric vein, superior mesenteric artery, or celiac artery branches.^26,27 Coronal maximum-intensity projections are useful to show narrowing of the portal-splenic-mesenteric venous confluence and associated venous collaterals that engorge the branches of the gastrocolic venous trunk.^28-30 Helical CT is superior to magnetic resonance imaging (MRI) for preoperative staging of pancreatic carcinoma. To determine surgical resectability, CT imaging shows invasion of peripancreatic tissue, portal and superior mesenteric veins, and peripancreatic arteries more reliably than MRI.³¹

The advantage of CT over MRI applies to patients with definite pancreatic carcinoma. CT imaging in a screening program of asymptomatic patients raises concerns regarding radiation dose, particularly the cumulative radiation dose of annual CT screening.³² Therefore, MRI of the pancreas, without any ionizing radiation, offers advantages for imaging the pancreas in a screening population.

Magnetic Resonance Imaging The role of MRI for evaluation of pancreatic malignancy continues to evolve. According to a meta-analysis of 68 studies, Bipat and coworkers found that for the detection of pancreatic adenocarcinoma, MRI had a sensitivity of 84% compared to 91% for helical CT, whereas the sensitivity for tumor resectability was essentially equal for both (MRI: 82%; CT: 81%).³³ However, the publication years for the cited articles ranged from 1990 to 2003. With the significant technological advancements with MRI and CT that have occurred since that time period, it is difficult to draw conclusions from this study. For MRI, these advances include the development of high-performance multichannel phased-array coils and parallel imaging acquisitions, which allow for more rapid imaging sequences. In addition, the advent of the three-dimensional gradient-echo pulse sequence for dynamic postcontrast imaging as well as three-dimensional magnetic resonance cholangiopancreatography have allowed for increased spatial and contrast resolution, as well as decreased artifacts, with resultant improved visualization of the pancreas.³⁴ A more recent study by Park and associates concluded that MRI was superior to CT in the detection of pancreatic carcinoma and that both modalities performed similarly with respect to the evaluation of tumor resectability.³⁵ Moreover, with its superior contrast resolution, MRI has been shown to more reliably detect smaller, noncontour-deforming tumors compared to CT^21,36 and can more accurately detect and characterize smaller hepatic metastases.^37,38 With respect to screening, MRI is limited by its relatively higher cost, reduced availability, requirement for intravenous contrast administration, and increased examination time. However, the possibility of customized, unenhanced examinations at reduced cost and acquisition time are possible with advanced sequences such as diffusion-weighted imaging.³⁹ Therefore, supplementary studies, and perhaps additional innovation, are required for understanding the efficacy of MRI as a screening modality.

Endoscopic Ultrasound Endoscopic ultrasound (EUS) is the most sensitive test for detecting pancreatic cancer. This conclusion is supported by multiple studies,^19,40,41 including one by Müller and colleagues that noted a sensitivity of 94% for all cases and 93% for lesions less than 3 cm in size.¹⁹ De Witt and coworkers reported similar findings in a study comparing EUS to CT in 80 patients with pancreatic cancer. EUS had a sensitivity of 98% compared to 86% for CT.⁴⁰ However, EUS is usually performed after CT or MRI has already detected a pancreatic mass; without this advantage, the reliability of EUS in a widespread screening program is less certain. EUS occasionally suffers from a limited field of view with relative blind spots in the uncinate process or tail of the pancreas.

Endoscopic Retrograde Cholangiopancreatography Endoscopic retrograde cholangiopancreatography (ERCP) is an excellent method for visualizing the pancreatic duct, with a sensitivity of 96% and specificity of 92% for diagnosing pancreatic cancer.¹⁶ ERCP also has the potential for tissue sampling from the pancreatic duct. The shortcomings of ERCP as a screening tool include its invasive nature and significant morbidity.⁴²

Positron Emission Tomography Fluorodeoxyglucose positron emission tomography (FDG-PET) relies on the increased uptake and metabolism of the radionuclide ¹⁸F-fluorodeoxyglucose by neoplastic cells. FDG-PET has a sensitivity of 88-92% and a specificity of 83-85%^43,44 for pancreatic cancer, but it lacks anatomic detail. The combination of FDG-PET and CT adds the anatomic detail and localization that PET by itself lacks.^45,46 PET may miss lesions when they are desmoplastic or when there are not enough hypermetabolic tumor cells to accumulate glucose. The disturbance of glucose metabolism in diabetics also reduces the sensitivity of PET. The future role of PET in screening for pancreatic cancer is doubtful. PET will likely continue to have a limited role in pancreatic cancer staging, assessment of response to chemotherapy, and diagnosis of recurrent disease.

Molecular Analysis of Pancreatic Juice Pancreatic cancer biomarkers may be detected in pancreatic juice. Sampling of pancreatic juice is performed at the time of ERCP and carries the risk of inducing pancreatitis. In a study by Yan and associates, pancreatic juice was extracted in 146 patients with either pancreatic cancer, chronic pancreatitis, or biliary tract stones.⁴⁷ The pancreatic juice was analyzed for the p53 and k-ras mutations as well as the proportion of pl6(INK4a) promoter methylation. Combination molecular analysis increased the discrimination between patients with malignant disease and those with benign disease and allowed patients in high-risk groups to be stratified from negligible risk to an over 50% probability of early cancer.⁴⁷ Nakashima and colleagues measured human telomerase reverse transcriptase (hTERT) expression in the pancreatic juice of 115 patients with pancreatic adenocarcinoma preop-eratively.⁴⁸ hTERT expression was detected in 84% of pancreatic adenocarcinoma patients, whereas only 62% were positive based upon cytology. The European Registry of Hereditary Pancreatic and Familial Pancreatic Cancer is currently studying the molecular analysis of pancreatic juice with CT and EUS as part of a secondary screening program. The limiting factor in the application of biomarkers from pancreatic juice is the invasive test required for sample collection.

Considerations for a Screening Strategy

Screening for a disease in an asymptomatic patient involves many considerations. The disease for which patients are being screened should have a substantial morbidity and mortality. Pancreatic cancer certainly meets these criteria.¹ Early detection should improve prognosis. There is evidence that this statement is true in pancreatic cancer and that it is not merely due to lead-time bias.^1,5-8 Next, the pretest probability of a disease and the target population should be defined. The annual incidence of pancreatic cancer in the general population is only 0.01%.¹ According to Bayes' theorem, a disease with such a low prevalence dramatically reduces the post-test probability of having the disease, even when using a test with a high sensitivity and specificity. The cost, acceptability, and efficiency of a screening strategy should be considered. An affordable serologic test with a high sensitivity and specificity would be acceptable and efficient. A positive serologic test could lead to combinations of other tests, each with their own potential for harm that must be considered when a screening strategy is designed. Even if one considers the devastating prognosis of pancreatic cancer, it is difficult to conceive that mass population screening for such a rare disease would be feasible. Certain populations are at a much higher risk for pancreatic cancer, to the extent that screening may become feasible.

Populations at Risk for Pancreatic Cancer

Inherited Risk Factors (Table 1)

Table 1.

Inherited Risk Factors for Pancreatic Cancer

	Risk for pancreatic cancer
Hereditary pancreatitis	5251
Cystic fibrosis	31.557
Peutz-Jeghers syndrome	13260
Hereditary nonpolyposis colorectal cancer with MLH1 mutation	<664,65
Familial atypical multiple mole melanoma syndrome	13.168
Hereditary breast and ovarian cancer	23.1 in BRCA1 carriers70,102 5.9-6.6 in BRCA2 carriers70,71
Familial pancreatic cancer	6.4-3254

Hereditary Pancreatitis Hereditary pancreatitis is an autosomal dominant condition that was first described by Comfort and Steinberg in 1952.⁴⁹ These patients usually present with recurrent attacks of acute pancreatitis starting in childhood and progress to chronic pancreatitis at a relatively young age. In 1996, Whit-comb and coworkers reported that mutations in the cationic trypsinogen gene (PRSS1) were associated with hereditary pancreatitis.⁵⁰ Mutations in this gene prevent inactivation of trypsin, resulting in autodigestion of the pancreas. More than 25 mutations in the PRSS1 gene have been described to date. A less common mutation involves the pancreatic secretory trypsin inhibitor gene (PST1), also known as the serine protease inhibitor Kazal type 1 (SPINK1). Mutations in this gene have been associated with chronic pancreatitis in children, tropical pancreatitis, and alcoholic chronic pancreatitis. Patients with hereditary pancreatitis have an estimated lifetime risk of 40% for developing pancreatic cancer.⁵¹

Familial Pancreatic Cancer In 1989, Lynch and coworkers described 12 families in which pancreatic cancer occurred in 2 or more first-degree relatives.⁵² Subsequently, tumor registries were established to collect data on families with pancreatic cancer. Familial pancreatic cancer is defined as ductal adenocarcinoma of the pancreas affecting at least 2 first-degree relatives who do not fulfill the criteria for another inherited tumor syndrome.⁵³ The ongoing National Familial Pancreas Tumor Registry estimates the risk of developing pancreatic cancer in an individual with 2 affected family members to be 6.4-fold (95% confidence interval [CI], 1.8-16.4). With 3 or more affected family members, the risk may be as high as 32-fold (95% CI, 10.2-74.7).⁵⁴ Kindreds of families with sporadic pancreatic cancer appear to not have an increased risk of pancreatic cancer. There is evidence for anticipation of familial pancreatic cancer that does not appear to be the result of biases and is independent of smoking.⁵⁵ As of yet, there has been no identification of a gene that plays a major role in the development of familial pancreatic cancer, except for BRCA2, which is present in a minority of cases (17%) of familial pancreatic cancer.⁵⁶

Cystic Fibrosis Cystic fibrosis is one of the most commonly inherited diseases of the exocrine pancreas. These patients are at a higher risk for digestive tract cancers (odds ratio [OR], 6.5; 95% CI, 3.5-11.1).⁵⁷ The frequency of pancreatic cancer in cystic fibrosis families is increased compared to that of the normal population; however, it is still very low.⁵⁸ Neglia and coworkers found only 2 cases of pancreatic cancer in 28,000 individuals with cystic fibrosis.⁵⁷

Hereditary Syndromes Associated With Pancreatic Cancer

Peutz-Jeghers Syndrome Peutz-Jeghers syndrome is a rare autosomal dominant condition characterized by mucocutaneous pigmentation and hamartomatous gastrointestinal polyps. These patients are also at risk for developing gastrointestinal and extragastrointestinal cancers. The most common cause of Peutz-Jeghers syndrome is a germline mutation in the STK11/LKB1 gene, which is involved in regulation of cell proliferation and polarity.⁵⁹ The increased risk of these patients for developing pancreatic cancer is as high as 132-fold (95% CI, 44-261), with a lifetime risk of 36%.⁶⁰

Hereditary Nonpolyposis Colorectal Cancer (Lynch Syndrome) Hereditary nonpolyposis colorectal cancer (HNPCC) is an autosomal dominant condition associated with predominantly right-sided colorectal cancers at a relatively young age.⁶¹ Patients are also at risk for noncolorectal cancer, including pancreatic,⁶² ovarian, gastric, endometrial, and small-bowel cancer.⁶³ HNPCC is caused by a mutation in one of the DNA mismatch repair genes, the most common being MLH1 and MSH2. The relative risk (RR) for developing pancreatic cancer in patients with HNPCC is less than 6.^64,65 In a study of Korean and Dutch patients with HNPCC, pancreatic cancer was only found in patients with the MLH1 mutation.⁶⁶

Familial Atypical Multiple Mole Melanoma Familial atypical multiple mole melanoma (FAMMM), also known as dysplastic nevus syndrome, is an autosomal syndrome presenting with numerous dysplastic nevi or melanomas. Approximately 50% of patients with this syndrome have germline mutations in the cyclin-dependent kinase 2A (CDKN2A) gene, which appears to be required for the development of pancreatic cancer.⁶⁷ These patients have a 13.1-fold (95% CI, 1.5-47.4) increased risk for pancreatic cancer.⁶⁸

Melanoma-Pancreatic Cancer Syndrome A family was identified in 1995 who had pancreatic cancer and melanoma associated with the CDKN2A gene, but without the phenotype of FAMMM. This situation is now regarded as a cancer syndrome distinct from FAMMM.⁶⁹

Hereditary Breast and Ovarian Cancer Carriers of the BRCA1 or BRCA2 germline mutations have an increased risk of pancreatic cancer, though the risk is greater for carriers of the BRCA2 mutation. According to Risch and associates, the RR of pancreatic cancer is 3.1 (95% CI, 0.45-21) and 6.6 (95% CI, 1.9-23) in carriers of the BRCA1 and BRCA2 mutations, respectively.⁷⁰ Van Asperen and colleagues calculated a RR of 5.9 (95% CI, 3.2-10) in carriers of the BRCA2 mutation.⁷¹

Noninherited Risk Factors (Table 2)

Table 2.

Noninherited Risk Factors for Pancreatic Cancer

	Risk for pancreatic cancer
Smoking	1.6-2.572-74
Diabetes	2.0 for type 1 diabetes76 1.8 for type 2 diabetes77
Chronic pancreatitis	18.581-26.383
Obesity	1.7286

Smoking Tobacco smoking increases the risk of pancreatic cancer up to 2.5-fold compared to that of nonsmokers, and has been noted in up to one third of pancreatic cancer patients.^72,73 The risk appears to be dose-related but does drop off significantly after cessation of smoking.^72,74 It has been suggested that smoking cessation could prevent 25% of the deaths attributed to pancreatic cancer in the United States.⁷²

Diabetes Mellitus Pancreatic cancer occurs with increased frequency among individuals with diabetes.^75-77 A metaanalysis of 36 studies found that type 2 diabetes increases the risk of pancreatic cancer by 82% (OR, 1.82; 95% CI, 1.66-1.89).⁷⁷ The risk of pancreatic cancer was 50% higher if diabetes was diagnosed within the preceding 5 years.⁷⁷ Although earlier studies have focused on type 2 diabetics,⁷⁸ a recent meta-analysis by Stevens and coworkers found that type 1 diabetes increased the risk of pancreatic cancer 2-fold (RR, 2.0; 95% CI, 1.37-3.01).⁷⁶ Chari and associates estimated that 1% of diabetic subjects over 50 years of age will be diagnosed with pancreatic cancer within 3 years of meeting diagnostic criteria for diabetes.⁷⁹ The high prevalence of diabetes in the general population excludes hyperglycemia as a practical screening tool for pancreatic cancer. Hyperglycemia may become more relevant as a screening test if it becomes possible to differentiate pancreatic cancer—induced diabetes from other causes of hyperglycemia by means of better clinical characteristics or the use of serologic markers.

Chronic Pancreatitis Chronic pancreatitis is a progressive inflammatory disease characterized by irreversible histologic transformation that clinically presents with pain or loss of function.⁸⁰ Most cases of chronic pancreatitis are secondary to excessive alcohol consumption, yet only 5—10% of chronic alcoholics develop chronic pancreatitis. Therefore, the causal relationship is not straightforward. Other dietary, environmental, and genetic factors likely play a role. Much less frequent causes include any obstructions of the pancreatic duct such as a traumatic disruption, pancreas divisum, stones, or tumors. Tropical pancreatitis and autoimmune pancreatitis are recognized causes of chronic pancreatitis, as are certain systemic conditions such as hypertriglyceridemia and hyperparathyroidism. Chronic pancreatitis may be part of a familial syndrome.

Chronic pancreatitis elevates the risk of pancreatic cancer, which increases over time.^81-83 Quantifying the risk is difficult because of confounding factors such as smoking, alcohol, and diet. The standard incidence ratio has been estimated to be between 18.5⁸¹ and 26.3.⁸³ Stated in more meaningful terms, Lowenfels and coworkers found the cumulative risk to be 1.8% after 10 years and 4% after 20 years.⁸³ Tropical pancreatitis is an even greater risk for pancreatic cancer, with a RR of 100 (95% CI, 37-218).⁸⁴ In hereditary pancreatitis, the risk is also very high, with a cumulative lifetime risk of 40%.⁵¹

Obesity and Physical Activity Excessive body weight appears to increase the risk of pancreatic cancer.^85-87 The Health Professionals Follow-up Study and the Nurses Health Study found that men and women with a body mass index (BMI) of at least 30 kg/m² had a 72% increased risk of pancreatic cancer when compared to those with a BMI of less than 23 kg/m² (RR, 1.72; 95% CI, 1.19-2.48).⁸⁶ Physical activity appeared to reduce the risk of pancreatic cancer by 55% (RR, 0.45; 95% CI, 0.29-0.70). In Cancer Prevention Study II, the risk of pancreatic cancer was 41% higher in subjects with a BMI between 30.0 kg/m² and 34.9 kg/m² (RR, 1.41; 95% CI, 1.19-1.66) when compared to those with a normal BMI (<25 kg/m²).

Cystic Lesions of the Pancreas Cystic neoplasms of the pancreas are a frequent incidental finding due to the increased use of high-resolution cross-sectional abdominal imaging. The most common cystic neoplasms of the pancreas are serous or mucinous in nature. Serous cystic neoplasms of the pancreas have a characteristic honeycomb appearance on CT scan, MRI, and ultrasound, and they do not have malignant potential. In 1996, mucinous cystic neoplasms were classified by the World Health Organization as either mucinous cystic neoplasms (MCNs) or intraductal papillary mucinous neoplasms (IPMNs), both of which have the potential for malignant transformation.⁸⁸ MCNs are usually solitary neoplasms, occur almost exclusively in women, and are usually found in the pancreatic body or tail. MCNs contain ovarian type stroma and, therefore, require surgical resection and histologic examination to make a definitive diagnosis.⁸⁹ IPMNs affect men and women equally, are usually located in the head of the pancreas, and may be multiple. IPMNs arise from the pancreatic duct and can be classified as main-duct IPMNs, side-branch IPMNs, or a mixed phenotype. Main-duct IPMNs are characterized by dilation of the main duct (> 1 cm), whereas side-branch IPMNs consist of side-branch mucinous cysts without main duct involvement. A connecting side branch between the cyst and the main pancreatic duct is frequently identified on pancreatic imaging. Mixed IPMNs have features of both main-duct and side-branch IPMNs. The distinctions between the three types of IPMNs are important, as main duct involvement carries significant malignant potential and the need for resection, whereas branch-duct IPMNs of less than 3 cm and without mural nodules can be managed conservatively.⁸⁹ A more detailed discussion of the management of pancreatic cystic neoplasms is beyond the scope of this review and has been published elsewhere.^90-92

Evidence From Published Screening Studies in Asymptomatic High-risk Populations

Several models have evaluated the possible benefits of screening in high-risk populations. Rulyak and associates used a decision analysis protocol to compare a one-time EUS screening for pancreatic dysplasia with no screening in a hypothetical cohort of 100 familial pancreatic cancer patients.⁹³ They concluded that EUS was cost-effective and increased life expectancy, provided that the population was at high risk, with a prevalence of dysplasia over 16% and a sensitivity of EUS over 84%. On the other hand, a similar model devised by Rubenstein and colleagues looked at 4 management strategies: no surveillance, total pancreatectomy, EUS surveillance, and EUS with biopsy, based upon a 45-year-old man with a first-degree relative with pancreatic cancer and a history of chronic pancreatitis.⁹⁴ Although the lifetime risk of cancer was 20%, no surveillance provided the lowest cost and the greatest remaining quality and years of life. Total pancreatectomy was only beneficial if the lifetime risk of cancer was 46%. This study demonstrated the weakness of any screening approach in pancreatic cancer: even with early diagnosis, the mortality is high; thus, early diagnosis does not have an impact on life expectancy.⁹⁵

Only a few series have looked at screening in asymptomatic high-risk populations. In 14 patients with a family history of pancreatic cancer in first-degree relatives, Brentnall and coworkers diagnosed 7 patients with pancreatic dysplasia based upon a combination of patient history, EUS, and ERCP, which was confirmed pathologically at the time of pancreatectomy.⁹⁶ At the University of Washington, 46 patients in familial pancreatic cancer kindreds have been screened by EUS.⁹⁷ Changes of chronic pancreatitis were found in approximately half of the patients. ERCP was performed on those with abnormal EUS findings. If a pancreatography was found to be abnormal, pancreatectomy was recommended. In all pathologic specimens, widespread dysplasia was found, but no invasive cancer.

Canto and coworkers studied 2 groups with a high risk of pancreatic cancer (patients with Peutz-Jeghers syndrome and relatives of patients in familial pancreatic cancer kindreds).⁹⁸ A total of 78 patients and 149 control subjects were evaluated with a baseline and 12-month EUS and CT scan. If the EUS was abnormal, EUS-fine-needle aspiration and ERCP were performed. Abnormalities suggestive of chronic pancreatitis were common in the high-risk groups (78% by EUS and 73% by ERCP). Seventeen high-risk patients were diagnosed with neo-plastic-type lesions on EUS. Eight of these patients had pancreatic neoplasms diagnosed by fine-needle aspiration or surgery. In 7 patients who underwent surgery, branch-duct—type IPMNs were diagnosed pathologically. The authors commented that IPMNs should be considered part of the phenotype of familial pancreatic cancer. In this study, ERCP and CT were inferior to EUS for the detection of small precursor lesions.

A recent series of 44 asymptomatic patients with a combination of hereditary risk factors for pancreatic cancer were screened using EUS. Three (6.8%) patients were found to have mass lesions that were resected and identified as adenocarcinoma. Seven patients (16%) were diagnosed with premalignant lesions. In this series, EUS screening was feasible and safe; however, the impact on long-term survival was unclear.⁹⁹

Conclusions

Although there are no standard screening guidelines for patients at high risk of pancreatic cancer, astute physicians can potentially diagnose early cases if they are aware of the risk factors for pancreatic cancer and collaborate with a radiologist and endoscopist experienced in pancreatic diseases.

Familial pancreatic cancer and Peutz-Jeghers syndrome are perhaps closest to having established screening guidelines. It has been suggested that relatives from familial pancreatic cancer kindreds start screening at 40 years of age or 10 years before the youngest family member with pancreatic cancer.^97,98 Because patients with Peutz-Jeghers syndrome develop pancreatic cancer at an even earlier age, screening should begin at ages 25—30.¹⁰⁰ The optimal frequency of screening has not been critically assessed, though several institutions recommend annual screening.⁹⁸

Of all the candidate tests for detecting early pancreatic cancer in high-risk individuals, EUS is the most sensitive. However, EUS is expensive, invasive, operator-dependent, and impractical for mass screening on a large scale. The ideal screening tool would be a molecular marker detected in the blood of patients with early stage, or surgically resectable disease, such as the plasma marker recently reported.¹⁰¹ However, any such marker has yet to be validated. The improved sensitivity of radiologic imaging such as CT and MRI are exciting and, when combined with 1 or more blood markers, may allow for earlier detection.

Although mass screening for pancreatic cancer is not feasible at this point in time, it is important that we continue to strive for earlier diagnosis of pancreatic cancer. Clinicians should at least be aware of the noninherited and inherited risk factors for pancreatic cancer, as this knowledge may allow more targeted screening on an individual basis.