Screening for Pancreatic Cancer

Abstract

Despite improvements in the clinical and surgical management of pancreatic cancer, limited strides have been made in the early detection of this highly lethal malignancy. The majority of localized pancreatic tumors are asymptomatic, and the recognized presenting symptoms of pancreatic adenocarcinoma are often vague and heterogeneous in nature. These factors, coupled with the lack of a sensitive and noninvasive screening method, have made population-based screening for pancreatic cancer impossible. Nevertheless, at least two large institutions have performed multimodality-screening protocols for individuals with high risk of pancreatic cancer based on genetic predisposition and strong family history. Abnormalities noted during these screening protocols prompted further investigation or surgery that resulted in the discovery of benign, potentially malignant, and malignant pancreatic lesions. In addition to ductal epithelial pancreatic intraepithelial neoplasia, greater sensitivity has recently been achieved in the identification and characterization of precancerous mucinous pancreatic tumors. Advancements in proteomics and DNA microarray technology may confirm serum-based biomarkers that could be incorporated into future screening algorithms for pancreatic cancer.

There are few malignant diagnoses with a worse prognosis than pancreatic cancer. In the United States, it has been estimated that the number of new cases of pancreatic cancer diagnosed in 2007 will exceed 37,000, with a nearly equivalent number of deaths.¹ The median survival for patients with pancreatic cancer is approximately 6 months, and the percentage of patients in whom surgery is a viable curative option is in the range of 10%.² Pancreatic cancer lacks recognized early symptoms, and the common presenting symptoms, such as pain, jaundice, and weight loss, rarely occur before the cancer is locally or regionally metastatic. Even with surgery, pancreatic cancer almost always recurs, and fewer than 5% of all pancreatic cancer patients are alive 5 years past diagnosis.³

Promotion of early detection has been shown to be cost-effective and may increase survival rates for breast, colon, cervical,and prostate cancers.⁴ Although advancements have been made in the diagnosis and clinical management of pancreatic cancer, finding an accurate method with which to detect a precancerous pancreatic lesion and intervene at a curable stage has proven to be challenging. Currently, there is no consensus on the most appropriate screening regimen for early pancreatic cancer.

Does Early Detection Improve Survival?

Early disease stage at diagnosis and administration of curative intent surgery (ie, resection) provide the best opportunities of achieving long-term survival among patients with pancreatic cancer.² Sohn and associates⁵ prospectively evaluated outcomes among 616 resected pancreatic cancer patients and noted that patients with tumors measuring smaller than 3 cm, negative resection margins, and no lymph node involvement demonstrated a 5-year survival rate of 31% compared with the study's overall 5-year survival rate of 17%. Lymph node metastases indicate a poor prognosis for pancreatic cancer patients,^6,7 and even nodal microinvolvement confers a survival disad-vantage.⁸ Disease-free resection margins are an important prognostic factor, although one group evaluated survival in 360 pancreatic cancer patients who underwent pancre-aticoduodenectomy and found no statistically significant difference in median survival for those with negative margins compared to those with microscopic invasion (21.5 vs 27.8 months).⁹ Another recent study observed that even in pancreatic resections without histologic evidence of disease, patients with margins that tested positive for K-ras mutations—an indicator of potentially malignant disease—had significantly shorter survival times than those without K-ras mutations (15 vs 51 months).¹⁰

A crucial predictor of outcome in pancreatic cancer patients is tumor size. A retrospective evaluation from the Japanese literature, including 36 cases of pancreatic cancer tumors measuring smaller than 1 cm in diameter, yielded an overall 5-year survival rate of 57%.¹¹ In this study, the 5-year survival rate was 34% in 13 patients with jaundice and/or a tumor mass noted on computed tomography (CT) or ultrasound compared to 69% for 22 patients who had neither jaundice nor a mass depicted on imaging (P<.05). A study conducted in Tokyo evaluated prognostic factors among 77 resected pancreatic adenocarcinoma patients and observed a 5-year cumulative survival for the subset (10%) of patients whose tumors measured less than 1 cm.¹² As a whole, these data demonstrate that the best chance for curing pancreatic cancer is the discovery of an adenocarcinoma that is small (ideally <1 cm in diameter), limited to the ductal epithelium, and without microscopic invasion of the surrounding pancreatic parenchyma.

Risk Factors for Pancreatic Cancer

Pancreatic cancer has a greater incidence in men than in women, increases with advancing age, and predominates in certain ethnic/racial groups, such as African Americans, Ashkenazi Jews, Pacific Islanders, and New Zealand Maori.^13,14 The increased risk among certain populations appears to be multifactorial in nature and is likely due to a combination of environmental and inherited factors.

Factors With Significantly Elevated Risk

Environmental and host factors that may influence the development of pancreatic cancer are detailed in Table 1. Individuals who smoke cigarettes have consistently demonstrated an approximately 2-fold increased risk for developing pancreatic cancer.^15,16 Passive exposure to cigarette smoke has not been implicated as causative.¹⁷ The magnitude of risk appears to be dose-responsive, with studies of individuals who regularly smoke over 40 cigarettes a day¹⁸ or have at least a 40 pack–year smoking history¹⁹ displaying a 3- to 5-fold increased risk of pancreatic cancer. Additionally, cigarette smoking multiplicatively increases the risk of pancreatic cancer in high-risk families and decreases the age of onset by approximately 10 years.^20,21

Table 1.

Environmental and Host Risk Factors for Pancreatic Cancer

Condition or behavior	RR or OR	Comments
Chronic pancreatitis27–31	2.31–26	Higher risk in smokers
Cigarette smoking15–21	2–5	Dose response relates to degree of risk
Diabetes mellitus22–26	2.0–2.2	Direction of causality unclear
Obesity or overweight32	∼2	Exercise benefits overweight individuals
Diet high in red, fried, processed, or grilled meat33	Unknown	Relationship not evident in all studies
Occupational exposures to dyes, pigments, or solvents35,36	Unknown	Limited evidence

OR

odds ratio

RR

relative risk.

Long-standing type II diabetes mellitus is a recognized risk factor for pancreatic cancer,²² and as many as 80% of pancreatic cancer patients may have glucose intolerance or frank diabetes at the time of diagnosis.²³

In some patients, diabetes is the only clinical sign of pancreatic cancer.²⁴ The apparent route of causality is controversial, with evidence supporting both the theory that diabetes mellitus predisposes a patient to pancreatic cancer and the theory that the state of the cancer itself incurs glucose intolerance.^22,25 However, a recent population-based cohort study found that approximately 1% of diabetes subjects 50 years of age or older were diagnosed with pancreatic cancer within 3 years of first meeting the criteria for diabetes, the majority (56%) of which were diagnosed with pancreatic cancer within 6 months of their diabetes diagnosis.²⁶ The observation that pancreatic tumor resection can “cure” diabetes²³ supports the theory that the tumor itself causes the diabetes. Thus, new-onset type II diabetes in an older individual may allude to an impending diagnosis of pancreatic cancer; further research is warranted to determine whether such patients or other subsets of diabetics may be candidates for surveillance protocols.

Chronic pancreatitis of any etiology has been shown to greatly increase the lifetime risk of pancreatic cancer development.²⁷ The calculated risk of cancer in chronic pancreatitis patients, however, has varied from study to study, likely due to study methodology or etiologic varia-tions of chronic pancreatitis in different populations. For instance, a multicenter historical cohort study conducted by the International Pancreatitis Study Group²⁸ evaluated 2,015 subjects with chronic pancreatitis and demonstrated a standardized incidence ratio for pancreatic cancer of 26.3 (95% confidence interval, 19.9–34.2). A hospital-based case-control study of pancreatic cancer conducted in northern Italy between 1983 and 1992 calculated an age- and sex-adjusted relative risk of pancreatic cancer of 5.7 among patients with a history of pancreatitis,²⁹ whereas a single-center, prospective study in France found a risk of pancreatic cancer of 19.0 among pancreatitis patients.³⁰ Additionally, a large case-control study conducted by the Department of Veterans Affairs in the United States among hospital discharges from 1970 until the mid-1990s observed a pancreatic cancer odds ratio of 2.31 among chronic pancreatitis patients.³¹ The mislabeling of pancreatic cancer or lack of recognition of intraductal papillary mucinous neoplasms (IPMNs) as chronic pancreatitis may also be responsible for the disparity of the degree of observed association.

Factors Associated With Modestly Increased Risk

In large-scale studies, obesity and anthropometric variables have displayed a positive association with pancreatic cancer incidence, as have diets high in meat and/or fat content and low in fruits and vegetables.^32,33 Although a direct causative role for alcohol consumption has not been observed, alcohol may indirectly contribute to risk by its association with pancreatitis, as well as with glucose intolerance and diabetes.³⁴ Occupational exposure to chlorinated hydrocarbon solvents used in the dry-cleaning industry, as well as polycyclic aromatic hydrocarbons in some dyes and pigments, may increase pancreatic cancer risk.^35,36

Genetic Risks

Individuals with a family history of pancreatic cancer have an increased risk of developing pancreatic cancer,³⁷ and it has been estimated that as many as 17% of all pancreatic cancers may have a hereditary component.³⁸ Although a minority of all pancreatic cancer cases may be hereditary in nature, genetic predisposition carries a much greater degree of risk than environmental factors (Table 2).³⁹

Table 2.

Genetic Risk Factors/Syndromes for Pancreatic Cancer

Condition	RR or OR	Genetic Mutation	Location
Hereditary pancreatitis40–42	54+	PRSS1	7q35
Cystic fibrosis44–46	2.6–61	CFTR	7q31.2
Peutz-Jeghers syndrome50	~132	STK11/ LKB1	19p13.3
Familial atypical multiple mole melanoma51	13–22	p16 INK4a/ MTS1	9p21
Hereditary breast or ovarian cancer52,53	∼2–9	BRCA1 and BRCA2	17q21-24, 13q12-13
Familial adenomatous polyposis55	∼5	APC	5q21
Hereditary nonpolyposis colorectal cancer54	Unknown	MSH2 and MLH1	2p22-21

RO

odds ratio

RR

relative risk.

Diseases That Increase Pancreatic Cancer Risk: Hereditary Pancreatitis and Cystic Fibrosis

Hereditary pancreatitis, caused by mutations in the cat-ionic trypsinogen gene, PRSS1, has a lifetime pancreatic cancer risk of approximately 40% in individuals in whom the disease follows a chronic course.⁴⁰ The two most common PRSS1 mutations, R122H and N29I, are responsible for approximately two thirds of all cases of hereditary pancreatitis.⁴¹ These mutations result in premature activation or dysfunction of the failsafe autolysis mechanism of intrapancreatic trypsinogen and result in an extensive duration of enzyme-induced pancreatic inflammation.²⁷ Hereditary pancreatitis displays approximately 80% of phenotypic penetrance⁴⁰; thus, some mutation carriers never experience an episode of pancreatitis, whereas others have a fulminant course. One large-scale study demonstrated that patients with hereditary pancreatitis had a 54-fold increased risk of developing pancreatic cancer over the general population that increased to a 154-fold risk if they were cigarette smokers.⁴²

Cystic fibrosis is an autosomal recessive disease caused by mutations in the cystic fibrosis transmembrane receptor (CFTR) gene.⁴³ The CFTR gene product is a chloride ion channel protein responsible for normal production of sweat, digestive secretions, and mucus. Some studies have associated cystic fibrosis with an increased risk of developing pancreatic cancer. An early British cohort study of 412 cystic fibrosis patients observed 2 pancreatic cancers versus the expected 0.008 cancers, with an odds ratio of 61.⁴⁴ Neglia and colleagues⁴⁵ evaluated 28,000 individuals with cystic fibrosis and found 2 cases of pancreatic cancer that developed in the third decade of life—a rare event— with an odds ratio of 31.5. In contrast, a recent study of 28,858 cystic fibrosis patients whose data were reported to the Cystic Fibrosis Foundation patient registry from 1990 to 1999 noted significantly increased risks of biliary, colon, and small-bowel cancer, but a nonsignificant 2.6-fold increased risk of pancreatic cancer.⁴⁶ The variation in results of these studies likely reflects the relative rarity of cystic fibrosis. Heterozygous CFTR mutation carriers may have an increased pancreatic cancer risk, as well as subclinical pancreatitis.^47,48

Cancer Syndromes

A number of cancer syndromes have been shown to carry an elevated pancreatic cancer risk. Peutz-Jeghers syndrome (PJS) is an autosomal dominant disorder with a multicancer phenotype that is caused by a germline mutation in the STK11/LKB1 tumor suppressor gene. Individuals with PJS develop intestinal and extraintestinal hamartomatous polyps and mucocutaneous-pigmented lesions on various bodily regions. In a meta-analysis, PJS showed an estimated relative risk of pancreatic cancer of 132 and a lifetime risk of developing pancreatic cancer of 36%.⁴⁹

Familial atypical multiple mole melanoma, caused by mutations in p16/CDKN2A, has demonstrated an increased risk of benign and malignant skin lesions, as well as an increased risk of a variety of cancers, including pancreatic cancer.⁵⁰ Hereditary breast and ovarian cancer is associated with mutations in the BRCA1 and BRCA2 genes. We recently described the role that these mutations play in familial aggregation of pancreatic cancer, as well as in seemingly sporadic cases.⁵¹ BRCA2 mutations may be responsible for a large portion of the pancreatic cancer burden of familial pancreatic cancer (FPC). For instance, Murphy and associates⁵² studied candidate genes in FPC and identified germline BRCA2 mutations in 17.2% (5/29) of families with 3 or more relatives with pancreatic cancer. Individuals from kindreds with hereditary non-polyposis colorectal cancer⁵³ and familial adenomatous polyposis⁵⁴ have also demonstrated modestly increased pancreatic cancer risk.

Familial Pancreatic Cancer

FPC is a term that has been applied to families with 2 or more first-degree relatives who have pancreatic cancer that is not associated with a known cancer syndrome. The genetic mutations for FPC in large part have not been identified, although there is evidence to indicate that mutations common to cancer syndromes, especially BRCA2, may play a role.⁵⁵ The National Familial Pancreas Tumor Registry at the Johns Hopkins University prospectively evaluated 3,957 FPC kindreds and noted an observed-to-expected pancreatic cancer risk of 4.6 for kindreds with 1 affected relative, 6.4 for those with 2 affected relatives, and 32.0 for those with 3 relatives affected with pancreatic cancer.²⁰

Pathogenesis of Pancreatic Adenocarcinoma

Like all forms of cancer, pancreatic cancer is a genetic disease associated with the accumulation of germline and sporadic mutations in proto-oncogenes, tumor suppressor genes, and maintenance genes, such as the mismatch repair genes. Akin to colon cancer, pancreatic cancer has been shown to undergo a stepwise progression from dysplasia to invasive adenocarcinoma. Nearly one decade ago, studies demonstrated that pancreatic intraepithelial neoplasia (PanIN) lesions were a precursor to invasive carcinoma.⁵⁶ The PanIN pathway is believed to be responsible for the majority of pancreatic tumors. Two mucinous forms of pancreatic neoplasia that may progress to invasive carcinoma, intraductal papillary mucinous neoplasia (IPMN) and mucinous cystic neoplasms (MCN), are being diagnosed with greater frequency due to the standard use of diagnostic imaging (Table 3).^57,58 Although mucinous neoplasms were previously considered relatively uncommon, increasing clinical awareness, their comparatively superior rates of long-term survival, and the recent recognition that they often manifest as a part of FPC⁵⁹ support describing them in detail.

Table 3.

Precursors of Pancreatic Cancer and Degree of Histologic Differentiation

	Ductal Neoplasia	Mucinous Neoplasia
Degree of cellular atypia	Pancreatic intraepithelial neoplasia (PanIN)	Intraductal papillary mucinous neoplasms and mucinous cystic neoplasms
Benign	PanIN I: intraepithelial ductal hyperplasia PanIN IA: flat PanIN IB: papillary	Benign mucinous cystadenoma
Low-grade	PanIN II: low-grade dysplasia	Borderline cystadenoma
High-grade	PanIN III: high-grade dysplasia/ carcinoma in situ	Noninvasive proliferative cystic mucinous neoplasms

Pancreatic Intraepithelial Neoplasia

The PanIN model describes how pancreatic cellular atypia advances from hyperplasia to low-grade dysplasia and eventually to carcinoma in situ.⁶⁰ Pancreatic cancer has also been characterized by a sequential acquisition of mutations, first in the K-ras proto-oncogene and then in stepwise fashion in the p16, p53 and Smad4 tumor suppressor genes.⁶¹ This mutation accumulation is reflected in the evolving dysplastic changes that occur as PanIN progresses from early to later stages. For example, K-ras and p16 mutations have been found in PanIN IA and PanIN IB lesions, but p53 and Smad4 mutations are not evident until the appearance of PanIN II and PanIN III pancreatic specimens.⁶² Mutations in BRCA2, LKB1/STK11, MKK4, tumor growth factor-beta receptors, and mismatch repair genes such as MLH1, occur at low frequencies in pancreatic adenocarcinoma (Table 4).^61,63

Table 4.

Mutation Rates in Sporadic Pancreatic Ductal Adenocarcinoma

Mutation	Type	Location	Percentage affected
K-ras	Oncogene	12p	95
p16 INK4a	Tumor suppressor	9p	90–95
p53	Tumor suppressor	17p	55–75
DPC4/Smad4	Tumor suppressor	18q	55
AKT	Oncogene	19q	10–20
MYB	Oncogene	6q	10
BRCA2	Tumor suppressor	13q	7–10
STK11/LKB1	Tumor suppressor	19p	∼5
MKK4	Tumor suppressor	17p	4
TGF-beta receptor 1	Tumor suppressor	9q22	<5
TGF-beta receptor 2	Tumor suppressor	3p22	<5

TGF

tumor growth factor.

Adapted from Sohn et al⁶¹ and Kern.⁶³

Mucinous Premalignant Neoplasms

IPMNs are mucin-producing, epithelial neoplasms arising from the main pancreatic duct or branch ducts that produce varying amounts of duct dilation.⁵⁸ IPMNs occur with approximately equal frequency among men and women and are typically diagnosed during the seventh decade of life.⁵⁸ The majority of small IPMNs are asymptomatic lesions discovered incidentally on cross-sectional imaging performed for other reasons. Tree large surgical series described the common presenting symptoms of resected IPMNs as abdominal pain, weight loss, jaundice, acute pancreatitis, diabetes, and steatorrhea.^64–66 Independent predictors for the presence of carcinoma in these studies were weight loss, jaundice, and new-onset diabetes. Invasive IPMNs recur frequently even after a complete “curative” resection and portend poor survival; in contrast, some studies demonstrate that noninvasive IPMNs recur approximately 10% of the time after resection, and survival is excellent, regardless of the degree of epithelial dysplasia in the tumor.⁶⁷

MCNs are distinct, self-contained, mucin-producing neoplasms of epithelial origin consisting of ovariantype stroma.⁶⁸ Approximately 90% of MCNs occur in women, with 50 years of age being the average age of diagnosis.⁶⁹ In the majority of series, MCNs in younger patients tend to be noninvasive (adenomas), whereas MCNs are more likely to be invasive in older patients.⁷⁰ The presenting symptoms of MCNs are typically vague and include anorexia, mild abdominal pain, and a feeling of fullness.⁶⁸ MCNs typically do not communicate with the pancreatic duct, and intraductal growth is not observed. In bulk, MCNs appear as large, multilocular cystic masses, most often in the tail of the pancreas.^71,72 At the molecular level, oncogenic point mutations in K-ras occur early; inactivation of the tumor suppressor genes p53 and Smad4 has been observed in invasive MCNs.⁷² An invasive ductal adenocarcinoma coexists in approximately one third of all surgically resected MCNs. In most series, when an invasive carcinoma is not found and the noninvasive MCN is completely resected, the patient can be considered cured.^69,72–74 However, resected patients with an invasive adenocarcinoma have an estimated 5-year survival rate of 60%.^57,71,75

The natural history of cystic lesions has not been well defined, and some clinicians feel that surgical resection of all cystic tumors is warranted, considering the possibility of their malignant transformation.⁵⁷ Differentiating among mucinous subtypes and determining their malignant potential is an active area of investigation.⁷⁶ Currently, a definitive diagnosis for the majority of pancreatic cystic lesions is possible only by surgical resection and histologic examination.

How to Screen

The ultimate goal of screening for pancreatic cancer is to detect a premalignant, surgically resectable, neoplastic growth that is potentially curable. Ideally, a screening tool for early stage cancer should be relatively inexpensive and noninvasive. It should be sensitive and specific enough to have the ability to perceive a precancerous lesion, as well as predict with reasonable certainty that one does not exist. For pancreatic cancer, this tool should also have high positive predictive and negative predictive values, as the screened individual may subsequently choose whether or not to undergo surgery that could be curative but is also associated with significant risks and morbidity, including lifelong diabetes.

Because of the relative rarity of pancreatic cancer, population-based screening is not a viable option. However, screening of high-risk cohorts may have a role in prevention. Conducted from a third-party payer perspective and modeled on life expectancy and lifetime costs of medical care, one study determined that screening was cost-effective if the prevalence of dysplasia was greater than 16%.⁷⁷ Unfortunately, noninvasive means of pancreatic evaluation are not sensitive enough to serve as a worthwhile screening method. The one recognized serum marker of pancreatic cancer, carbohydrate antigen 19-9 (Ca 19-9), is insensitive for early lesions and is best used as a marker of response to neoadjuvant therapy or postsur-gical tumor progression.^47,78,79 Currently, multimodality screening combining the strengths of various evaluative techniques appears to be the most effective way to detect precancerous pancreatic lesions.

Screening Modalities

The majority of studies evaluating sensitivity and specificity of screening methods have been performed on patients with suspicion of pancreatic cancer, and their effectiveness for screening asymptomatic patients is undetermined. The following sections describe several modalities that have been used to screen high-risk cohorts.

Endoscopic Ultrasound

Endoscopic ultrasound (EUS) has gained in popularity as a screening tool, due to its ability to discern multifocal and very small pancreatic lesions and its low risk of complications. EUS can detect pancreatic masses as small as 5 mm in size, and it is also an effective tool for demonstrating vascular invasion.⁸⁰ An experienced and well-trained endosonographer can perceive areas of pancreatic abnormality that reflect PanIN as well as premalignant mucinous tumors, including parenchymal heterogeneity, and the presence of echogenic foci or hypoechoic nodules. EUS can also demonstrate pancreatitis-like changes that may occur simultaneously with dysplasia. Investigators at the University of Washington used EUS over CT, magnetic resonance imaging, or positron emission tomography in the evaluation of a high-risk pancreatic cancer family, as the other methods did not detect abnormalities in this family.⁸¹ EUS-fine needle aspiration (EUS-FNA) can be used to detect dysplasia, as well as measure levels of carcinoembryonic antigen, which predicts the presence of MCNs with high sensitivity.⁸²

Computed Tomography

Multidetector CT is one of the most effective means of diagnosing and staging pancreatic cancer, due to its ability to show very high-resolution images of pancreatic and abdominal organ anatomy. Clinically, CT is most useful in visualizing pancreatic malignancies and ductual and parenchymal changes, as well as evaluating for the presence of liver or others peritoneal cavity metastases and arterial or venous invasion. The utility of CT in screening asymptomatic patients for the presence of pancreatic cancer, however, is limited by its low sensitivity for small lesions. In a small study of 35 patients, pancreatic adenocarcinoma was visualized by dynamic CT as a low-density mass in 97% of cases, although only 3 of these tumors were smaller than 2 cm in diameter.⁸³ Although CT is not ideal for detecting PanIN changes, it is an effective imaging tool for visualizing mucinous precursor lesions.⁶⁸

Endoscopic Retrograde Cholangiopancreatography

PanIN typically arises in small- and medium-sized pancreatic ducts, and endoscopic retrograde cholangiopan-creatography (ERCP) findings of irregular or ectatic ducts and evidence of sacculations have been noted with PanINII and PanIN III dysplastic changes.⁸⁴ ERCP is best atevaluating periampullary malignancies and disordersinvolving the biliary tree. Brush cytology and pancreaticjuice sampling during ERCP allow for the histologic andmolecular analysis of pancreatic lesions. ERCP is currentlybecoming more widely used as a therapeutic option. Thereare a number of risks involved with ERCP, the most common of which is post-ERCP pancreatitis, which has beenestimated to occur in approximately 4–7% of patients.^85,86Additional risks include bleeding, perforation, sepsis,and cholangitis.

Results of Recent Screening Studies

Few studies have prospectively evaluated screening of individuals at high risk of pancreatic cancer. A comprehensive algorithm for managing potentially high-risk pancreatic cancer patients was recently published by Dr. Teresa Brentnall at the University of Washington in Seattle.³⁸ High-risk individuals were screened first with EUS. If the EUS was abnormal, it was typically followed by ERCP for patients without clinical suspicion or evidence of pancreatitis. If the ERCP was abnormal, individuals were offered the choice of either continued surveillance or a tissue diagnosis performed by a histo-logic evaluation of the pancreatic tail (pancreatic tailec-tomy). High-risk individuals who demonstrated PanIN III lesions on histologic examination of the pancreatictail specimen were offered a total pancreatectomy versus surveillance.

The results from studying 75 high-risk individuals who participated in this screening program have been recently reported.⁸¹ Fifteen patients underwent pancre-atectomy—12 for total pancreatectomy and 3 for distal pancreatectomy. Pathologic evaluation of all pancreatec-tomy specimens demonstrated widespread dysplasia that mainly involved small- and medium-sized ducts. All 15 patients survived the procedure; 5 patients had PanIN II lesions and 10 patients had PanIN III lesions. None of the patients displayed evidence of invasive adenocarcinoma, and none had a normal pancreas. The 12 patients who underwent total pancreatectomy became diabetic, and 1 patient required pancreatic transplant, due to the inability to detect hypoglycemia.

The largest high-risk pancreatic-cancer screening study to date in the United States was conducted at the Johns Hopkins University.^59,87 In a two-part series, a total of 116 high-risk pancreatic cancer patients were screened with EUS and CT for evidence of pancreatic neoplasia. The study population included 7 patients from kindreds affected with PJS and 109 patients from FPC kindreds. The screening protocol of FPC individuals was defined either as: a part of families with at least 2 affected first-degree relatives and who were more than 40 years of age or 10 years younger than the age at which their youngest relative with pancreatic cancer was diagnosed; or a part of families from a kindred with at least 3 affected family members, 1 of whom was a first-degree relative. The first section of their study involved 38 patients, and the second included 78 patients, in addition to a control population. High-risk patients were prospectively screened using a combination of CT and EUS. An abnormal EUS led to the use of EUS-FNA, multidetector CT, and ERCP.

Overall, 29 patients displayed neoplastic-type lesions and 15 patients underwent surgery, although how these patients were selected is not clear. Final pathologic examination demonstrated 6 high-grade or invasive lesions (PanIN III, IPMN with carcinoma in situ, and 1 frank adenocarcinoma), 11 low-grade lesions (PanIN II or IPMN), and 6 nonneoplastic lesions. Additionally,6 extrapancreatic lesions were detected via screening,including 1 malignant ovarian tumor.

Overall, these screening studies demonstrated that multimodality pancreatic evaluation could predict the presence of dysplastic or premalignant disease in high-risk groups. It is worth mentioning that although abnormalities noted on screening prompted surgery, approximately one half of the malignant or potentially malignant lesions discovered in the Johns Hopkins study were detected at surgery and not by the initial screening tests. In addition, approximately one third or more of the patients in both parts of the series who underwent pancreatic resection were found to have low-grade lesions (PanIN II or IPMN), whose natural history is unknown. Finally, we cannot determine with certainty either the sensitivity or specificity of the particular examinations used in the screening protocol, as only a portion of the cohort underwent resection.

Screening Recommendations

Participants of the Fourth International Symposium of Inherited Diseases of the Pancreas recently developed counseling and surveillance guidelines for individuals at high risk for developing pancreatic cancer.⁸⁸ Table 5 summarizes their screening recommendations for pancreatic cancer. They strongly advised that pancreatic-cancer screening studies be performed as part of a peer-reviewed protocol whenever possible and always with the protection of the human subjects. They also recommended that studies be conducted at one or more centers with demonstrated expertise in pancreatic diseases, as well as a comprehensive genetics counseling program.

Table 5.

Recommendations for Pancreatic Cancer Screening

A member of a Peutz-Jeghers family (a verified germline mutation carrier, if possible) Tree or more first-, second-, or third-degree relatives with pancreatic cancer in a single pedigree, 1 of whom is a first-degree relative of the candidate A known BRCA1, BRCA2, or p16 mutation carrier with at least 1 first- or second-degree relative with pancreatic cancer Families with 2 relatives in the same lineage with pancreatic cancer, 1 of whom is a first-degree relative of the candidate An individual affected by hereditary pancreatitis

Adapted from Brand et al.⁸⁸

The Future of Screening

Molecular-based strategies that incorporate DNA micro-array technology have provided insight into the genetic and epigenetic changes involved in tumor growth and development and may offer novel ways of screening. Clinical researchers at the University of Pittsburgh recently demonstrated that multianalyte profiling of selected serum cytokine markers could discriminate pancreatic cancer from chronic pancreatitis with a sensitivity of 98% and specificity of 96.4%.⁸⁹ Some of the numerous biomarkers recently evaluated in association with pancreatic cancer are described in Table 6.^90–100 These molecular studies are currently in their infancy and need confirming evidence of their reproducibility as effective indicators of early pancreatic neoplasia. In the future, the combinatorial usage of certain serum or genetic markers and imaging studies may prove to be an effective screening protocol for pancreatic cancer.

Table 6.

The Future of Screening: Potential Biomarkers

Test or marker Macrophage inhibitory cytokine190 Carcinoembryonic antigen cell adhesion molecule 191 DNA methylase transferase 192 Tissue inhibitor of metalloproteinase 193 Regenerating islet-derived family factor 494 Osteopontin95 NAD(P)H quinone reductase 196 Mucin 497 S100A698 Beta 4 integrin99 Insulin-like growth factors and binding proteins100

Summary

Complete pancreatic resection is the only method of obtaining long-term survival for pancreatic cancer patients. Currently, any patient who achieves long-term survival is diagnosed with early disease and has a small tumor and no lymph node involvement at the time of presentation. At this time, our only cost-effective screening protocol can be applied to high-risk patients who have a recognized predisposition for developing pancreatic cancer. PanINs, IPMNs, and MCNs have all been shown to be precursor lesions to invasive disease, and research has demonstrated that high-risk individuals often have multifocal premalignant pancreatic lesions that can be detected at screening.

Multimodality screening that includes EUS appears to be effective at identifying pre-invasive lesions, such as IPMNs, and detecting chronic pancreatitis-like changes throughout the pancreatic parenchyma. Future efforts should focus on research into the molecular evolution and pathogenesis of pancreatic cancer. Furthermore, collaborative, prospective trials of screening modalities will be crucial for reinforcing the feasibility and benefit of surveillance regimens for pancreatic cancer in high-risk groups.