Pancreatic Carcinoma Surveillance in Patients With Familial Melanoma FREE

IT HAS BEEN estimated that approximately 5% to 12% of melanomas occur in the familial setting.¹ Genetic linkage studies and molecular analyses have established cyclin-dependent kinase inhibitor 2A (CDKN2A, p16, and INK4a) as the major melanoma susceptibility gene in the familial melanoma syndrome.^{2 - 10} Mutations in the CDKN2A gene allow cells to bypass the G₁/S cell cycle checkpoint, leading to uncontrolled cell proliferation and growth.^{11 - 14} Mutations in the CDKN2A gene are found in approximately 20% to 40% of melanoma-prone families^{1 - 2 ,12} and are associated with the familial atypical multiple mole melanoma syndrome (FAMMM).^{15 - 17}

Growing evidence suggests that CDKN2A mutations may also predispose patients to pancreatic carcinoma.^{11 ,17 - 27} A study of Dutch families with melanoma by Vasen et al¹⁷ identified a specific 19–base-pair (bp) deletion in exon 2 of the CDKN2A gene, which they have named the p16-Leiden deletion, in 19 families with FAMMM. The second most frequent cancer in these families, after melanoma, was pancreatic carcinoma. The cumulative risk for development of pancreatic cancer in individuals with the p16-Leiden deletion was 17%, with a mean age at diagnosis of 58 years, whereas no pancreatic carcinoma developed in individuals from families with suspected FAMMM without the p16-Leiden deletions. However, families with other CDKN2A mutations were not studied. A study of Dutch FAMMM families by Hille et al²³ showed nearly equal mortality rates owing to melanoma and pancreatic cancer in these families. One study by Schutte et al²⁵ showed defects in the Rb/p16 pathway in 98% of pancreatic carcinomas. A recent study published by Lynch et al²⁷ proposed a new FAMMM-pancreatic carcinoma syndrome, because the 2 malignancies are so often seen together in patients with CDKN2A mutations. Lal et al¹¹ have suggested that the diagnosis of malignant melanoma and pancreatic adenocarcinoma in the same patient is evidence enough to virtually prove that the patient has an underlying germline CDKN2A mutation. Those authors recommend that relatives of individuals with these double primary malignancies obtain careful screening for melanoma. Vasen et al¹⁷ have also recommended surveillance for pancreatic cancer in families with FAMMM in the setting of a prospective study.

Pancreatic carcinoma is the fourth leading cause of cancer death in the United States.^{28 - 29} It has the lowest 5-year survival rate of any cancer.²⁸ Approximately 50% of patients have metastatic disease at diagnosis.^{28 ,30} It is difficult to screen for pancreatic carcinoma, and screening tests for the general population are not appropriate owing to an unacceptable rate of false-positive findings.³⁰

We herein present the case of a patient from a kindred with familial melanoma and an established functional germline CDKN2A mutation. He survived malignant melanoma with widespread metastases, but died of metastatic pancreatic adenocarcinoma 25 years after his initial melanoma diagnosis. We reviewed the most recent literature surrounding screening methods for early detection of pancreatic tumors, with the goal of evaluating the role of traditional and more innovative surveillance methods in patients with familial malignant melanoma and CDKN2A mutations.

The patient first presented in 1978 at 32 years of age with a clinically atypical melanocytic lesion on the right side of his back. The lesion was excised, and pathological evaluation confirmed superficial spreading malignant melanoma, with a depth of 1.24 mm, or Clark level IV. He underwent a wide local excision 6 months later, and results of pathological evaluation showed no further evidence of melanoma in the surrounding tissue. Fourteen months after his initial diagnosis, at 33 years of age, he presented with painful right axillary lymphadenopathy. Right axillary lymph node dissection was performed, and histological sections revealed metastatic malignant melanoma. One month later, he was referred for local radiation therapy of his right axilla. Staging revealed several pulmonary nodules in the right lung. The patient underwent segmental resection of the right upper, middle, and lower lobes, and pathological analysis confirmed metastatic melanoma. After resection, several pulmonary nodules remained, and new ones appeared within 2 months of surgery. The patient underwent 21 courses of chemotherapy with dacarbazine and dactinomycin (actinomycin D) during the course of 2 years. He had complete regression of metastatic melanoma, with no evidence of residual disease.

Two years after chemotherapy, at 39 years of age, the patient had a new primary melanoma (Clark level I-II) on his left shoulder, which was excised. Another primary melanoma, 0.57 mm in depth (Clark level III), was excised from his left upper back 7 years later, at 46 years of age. Both of these lesions were contiguous with the epidermis, suggesting that they were new primary lesions rather than metastatic in origin. In addition, he was diagnosed as having numerous dysplastic nevi, atypical melanocytic hyperplasia, and benign compound nevi.

At 57 years of age, the patient was enrolled in the Familial Melanoma Research Clinic at the Huntsman Cancer Institute at the University of Utah, Salt Lake City. Sequence analysis of DNA showed that he was heterozygous for a V126D (aspartate substituted for valine at amino acid 126) CDKN2A mutation (Figure 1). He had a family history of melanoma and numerous atypical nevi, as well as pancreatic carcinoma (Figure 2), and was diagnosed as having FAMMM syndrome. This pedigree was evaluated using the familial standardized incidence ratio to establish observed-expected ratios for melanoma and pancreatic cancer (Figure 2).³¹ Two months later, the patient presented to his primary care physician complaining of chronic low back and abdominal pain. Magnetic resonance imaging showed a 2.4-cm enhancing mass in the right kidney. Computed tomography of the abdomen and pelvis showed a 3.3 × 3.2-cm hypoechoic mass in the right kidney and ascites surrounding the liver and spleen. Results of cytology of a paracentesis specimen were consistent with adenocarcinoma, and results of an immunohistochemistry evaluation were negative for melanoma markers (vimentin, S100, and HMB-45) and positive for mucin (Figure 3). This finding was not consistent with renal cell carcinoma or metastatic melanoma. A colonoscopy was performed, but results of all biopsies were negative. Levels of diagnostic serum tumor markers carcinoembryonic antigen, β-human chorionic gonadotropin, and α-fetoprotein were all within reference range, suggesting that the tumor was less likely of colorectal, hepatic, or germ cell origin. However, the cancer antigen 19-9 (CA 19-9) level was markedly elevated to 17 293 U/mL (reference range, 0-37 U/mL). The CA 19-9 result, in conjunction with the negative results of other studies, led to a presumptive diagnosis of pancreatic adenocarcinoma. Laparoscopic exploration revealed widespread metastases. Tissue from the falciform ligament and omentum were positive for cytokeratin 7, carcinoembryonic antigen, and epithelial membrane antigen, and negative for cytokeratin 20, a pattern consistent with pancreatic carcinoma (Figure 3).^{32 - 35} The patient was discharged to hospice care following the laparoscopy and died 11 days later.

Genomic DNA was isolated from blood using a commercially available kit (Puregene DNA Isolation Kit; Gentra Systems, Inc, Minneapolis, Minn). Polymerase chain reaction (PCR) products were generated for sequence analysis, including the promoter region and exons 1, 2, and 3 of CDKN2A, spanning the splice junctions. Primers were synthesized by means of the University of Utah DNA/Peptide core facility, Salt Lake City, and the sequences were as follows (5′ to 3′): distal promoter region (forward, 5′-GTGCAACTCTGCTTCTAGAACA -3′; reverse, 5′-GCCAGCAAAGGCGTGTTT-3′); proximal promoter region (forward, 5′-GCCATACTTTCCCTATGACAC -3′; reverse, 5′-GCCCGCACCTCCTCTA-3′); exon 1 (forward, 5′-CACCTCCTCCGAGCACTC-3′; reverse, 5′-CTCCTCATTCCTCTTCCTTGG-3′); exon 2 (forward, 5′-GACGACAGCTCCGCAGAAGTT-3′; reverse, 5′-GCGTGAGCTGAGGCAAGACC-3′); and exon 3 (forward, 5′-GGTAGGGACGGCAAGAGA-3′; reverse, 5′-TGTAATAGGTGCTCAATAAATGTTGACA-3′). Polymerase chain reactions were performed using 0.5µm forward and reverse primers, 1 × PCR buffer (Idaho Technology Inc, Salt Lake City, Utah); 0.25 mM deoxyribonucleoside triphosphates (Takara Bio Inc, Otsu, Japan), 50- to 100-ng genomic DNA template, and 1.0-U Taq DNA polymerase (Fermentas Inc, Hanover, Md) with Taq antibody (BD Biosciences Clontech, Palo Alto, Calif). Cycling conditions included an initial 5-minute incubation at 94°C, followed by 40 cycles of 94°C for 30 seconds of denaturing, 58°C for 30 seconds of annealing, and 72°C for 60 seconds of extension. A final extension period of 5 minutes at 72°C was performed at completion of the 40 cycles. Before sequencing, each sample was treated with exonuclease I and shrimp alkaline phosphatase (EXOSAPIT; USB Corporation, Cleveland, Ohio). Sequencing was performed using the same primers by the University of Utah Sequencing Core Facility with a fluorescent capillary sequencer (3700 ABI; Applied Biosystems, Foster City, Calif).

Histological analysis and immunohistochemistry were performed as previously described.^{36 - 37} Briefly, formalin-fixed, paraffin-embedded tissue sections were deparaffinized in xylene and absolute ethanol, then rehydrated by means of successive immersions in 95% ethanol, 70% ethanol, and distilled water. Immunoperoxidase staining was performed on the Ventana ES instrument (Ventana Medical Systems, Inc, Tucson, Ariz) using the avidin-biotin-peroxidase complex according to manufacturer recommendations. We applied primary antibodies to proteins S100 (DAKO Corporation, Carpinteria, Calif) diluted 1:700, cytokeratin AE1/AE3 (Boehringer Mannheim Biochemicals, Indianapolis, Ind) diluted 1:1400, and Melan-A (DAKO Corporation) diluted 1:20 (heat-induced epitope retrieval was performed before primary antibody incubation). The immunostaining was developed using diaminobenzidine as the chromogen, and samples were counterstained with Mayer hematoxylin. We performed hematoxylin-eosin, mucicarmine, and Wright stains according to standard laboratory protocols.

Pedigree analyses were performed through the use of the Utah Population Database. This database contains computerized genealogies from the Family History Library of the Church of Jesus Christ of Latter-Day Saints, which are linked to statewide records from the Utah Cancer Registry, another population-based database that includes clinical and pathological records on all reportable cancers. Through this database, a 6-generation pedigree containing 179 subjects was constructed and evaluated for a history of melanoma and pancreatic adenocarcinoma. The observed-expected ratio of melanoma and pancreatic carcinoma were calculated using the familial standardized index ratio as reported previously.³¹

The involvement of CDKN2A in susceptibility to melanoma is now well established. The estimated gene penetrance for melanoma in those with CDKN2A mutations by 80 years of age is 58% to 92%^{38 - 39} and is geographically variable.³⁸ After melanoma, pancreatic carcinoma is the cancer most commonly associated with the familial melanoma syndrome and CDKN2A mutations.^{17 ,23 - 25 ,27} Penetrance estimations for pancreatic carcinoma in those with CDKN2A mutations have not yet been well established, but a Dutch study estimated a 17% penetrance by 75 years of age in FAMMM families with the p16-Leiden 19-bp deletion.¹⁷ Despite these high penetrance numbers and the fact that 20% to 40% of familial melanoma is associated with CDKN2A mutations,^{1 - 2 ,12} most clinicians take no precautions to prevent this fatal disease.

It is well known that pancreatic cancer is an extremely aggressive tumor, and early detection is exceptionally difficult. Barriers to early detection include absence of symptoms at early stages of disease and the retroperitoneal placement of the pancreas.⁴⁰ The most common presenting symptoms of pancreatic cancer include chronic epigastric pain radiating to the back, jaundice, weight loss, diarrhea, steatorrhea, and new-onset diabetes or hyperglycemia.^{28 ,40 - 42} Once diagnosed, pancreatic cancer is notoriously difficult to treat. It is generally diagnosed after it has metastasized, and patients usually die within 12 months of diagnosis.^{27 ,29 - 30} To date, no known screening method is sensitive and specific enough to screen the asymptomatic general population.³⁰ Experts suggest that surveillance is warranted in high-risk populations, such as those with a family history of pancreatic cancer or melanoma, but optimal screening methods are uncertain.^{17 ,27 ,30 ,41 ,43} Although effective strategies have been developed to screen for early detection of breast, prostate, colon, and cervical cancer, the pancreas creates an obstacle for similar screening methods because of the difficulty in visualizing or obtaining tissue from this organ.

In the past, traditional imaging techniques have been the primary focus of research in this area.^{44 - 45} The American Gastroenterological Association recommends at-risk patients undergo screening with spiral computed tomography, endoscopic ultrasonography (EUS), and measurement of CA 19-9 level as the initial screening tests.²⁸ More recent work has found EUS and endoscopic retrograde cholangiopancreatography (ERCP) to have sensitivity and specificity values in the lower to middle 90s for diagnosis of pancreatic carcinoma.^{42 ,44 ,46} The major limitation of EUS and ERCP is their invasiveness. Patients require conscious sedation for the procedure and have about a 5% risk for complications, including acute pancreatitis and hemorrhage.^{46 - 47} Although these imaging studies appear to be promising for the diagnosis of pancreatic carcinoma, their utility as a screening or surveillance method is uncertain.⁴² Brentnall⁴¹ has suggested an algorithm for surveillance of those at high risk for pancreatic cancer, beginning 10 years before the earliest diagnosis of pancreatic carcinoma in the patient's family or 50 years of age, whichever comes first.

Another promising modality for surveillance is the measurement of tumor markers, which are easily detected in body fluids and tissues. At present, there are no known tumor-specific markers for pancreatic cancer, although CA 19-9 is widely used.^{48 - 49} With a cutoff of 37 U/mL, the CA 19-9 level has sensitivity of 81% to 85% and specificity of 81% to 90% in detecting pancreatic adenocarcinoma.^{45 ,50} Steinberg⁵⁰ has shown that CA 19-9 levels greater than 1000 U/mL approach 100% specificity for diagnosing pancreatic cancer. However, in 1 study, serum CA 19-9 and carcinoembryonic antigen levels were not shown to be sensitive enough markers for early detection of premalignant lesions.⁴² Although measurement of CA 19-9 levels may not be sensitive enough to detect pancreatic dysplasia, changes in CA 19-9 levels over time may be sensitive enough to detect early pancreatic adenocarcinoma, perhaps before the tumor has metastasized or grown large enough to be unresectable. Sensitivity and specificity of 80% to 90%, as estimated for CA 19-9 levels in detection of pancreatic cancer,^{45 ,50} may not be adequate for screening in the asymptomatic population. However, in high-risk patients such as those with CDKN2A mutations, these statistics suggest that an annual measurement of this antigen level may be beneficial as surveillance.

To improve screening methods for early detection of pancreatic carcinoma, a better understanding of the molecular and histological pathology involved in progression of this cancer is needed.^{51 - 52} A model outlining the progression of pancreatic cancer has recently been introduced by Hruban et al.⁵³ This model was patterned after the genetic progression model for colorectal neoplasia.⁵⁴ The model proposes that pancreatic cancer progresses from normal ductal epithelium to ductal hyperplasia to invasive ductal adenocarcinoma. The authors use the term pancreatic intraepithelial neoplasia⁵⁵ to describe this progression from normal epithelium to differing levels of neoplasia. In sporadic pancreatic carcinoma, the progression from level 1 to level 3 disease is often associated with a sequence of genetic alterations, first in K-ras and Her-2/neu, followed by p16 alterations, and finally abnormalities in p53, DPC4, BRCA2, and other tumor suppressor genes.^{53 ,56 - 57} Genetic changes in oncogenes, tumor suppressors, and telomerase are being studied as noninvasive screening tests in individuals at risk for development of pancreatic cancer.^{30 ,58 - 59} Testing of stool, pancreatic juice, and blood for K-ras, p53, p16, and DPC4 mutations or telomerase activity is still in its infancy, but it is hoped that highly sensitive methods such as restriction fragment length polymorphism PCR will soon be available for serological testing.^{58 - 70} Further study in these areas is warranted before these methods can be used as a surveillance method for pancreatic carcinoma.

We propose an algorithm to be used for surveillance of patients with familial melanoma patients and known CDKN2A mutations, based on the surveillance program recommended by Brentnall⁴¹ in 2000 (Figure 4). Although no screening method has been effective in reducing the mortality rate for sporadic pancreatic cancer, patients with a known CDKN2A mutation are at high risk and surveillance is appropriate in this subpopulation. Confirmed CDKN2A mutations are present in only 20% to 40% of families with familial melanoma,^{1 - 2 ,12} and there is no known or proven correlation between families with non-CDKN2A familial melanoma and increased predisposition to pancreatic cancer. Therefore, we cannot recommend pancreatic cancer surveillance on any member of a kindred with FAMMM without a known CDKN2A mutation. At present, CDKN2A results are obtained primarily through research protocols, making referral to one of these studies important for these family members.

Surveillance should begin 10 years before the age at onset of pancreatic carcinoma in the youngest member of the patient's family who has been diagnosed as having pancreatic carcinoma. If no family history of pancreatic carcinoma is known, surveillance should begin at 50 years of age. Initial screening with EUS is suggested in asymptomatic patients. The annual EUS examination should be performed and interpreted by an individual who is an expert in EUS and diseases of the pancreas. If the EUS finding is abnormal, ERCP will be performed. If the patient is symptomatic or has a history of pancreatitis, alcohol abuse, or cholelithiasis, which can influence the EUS findings, ERCP should be performed initially in addition to the EUS. If suspicious changes are detected on ERCP, total pancreatectomy should be considered. Spiral computed tomography should be considered before surgery to detect large masses or metastases. Frozen sections should be obtained in the operating room and evaluated by a pathologist who is a pancreatic specialist to make a histological diagnosis of pancreatic dysplasia or carcinoma before the pancreatectomy. Such a surveillance program is best performed at a center specializing in the endoscopy, pathology, and surgery of the pancreas.

At present, the CA 19-9 level is not used as a screening method for high-risk patients or those with sporadic disease because of concern regarding insensitivity and lack of specificity in sporadic pancreatic cancer. However, in certain high-risk subsets of patients, annual testing in conjunction with EUS may increase the capacity for early detection. We suggest an annual measurement of the CA 19-9 level. If a patient is found to have a CA 19-9 level near or greater than 37 U/mL or experiences a dramatic increase in the level during a 1-year period, ERCP is suggested. Until K-ras, p53, or DPC4 mutations or telomerase levels (or a combination of these tests) have been validated as screening tools, these are not recommended outside research protocols. In the future, we envision a screening test for patients with familial melanoma and CDKN2A mutations using an annual quantitative serum measurement of CA 19-9, carcinoembryonic antigen, K-ras, or telomerase level to detect early pancreatic neoplasia, much as the prostate-specific antigen level is used to detect prostate cancer. Future research on the cost-benefit ratio and efficacy of these screening recommendations in families with the CDKN2A mutation is planned. If benefit is demonstrated, clinical CDKN2A testing in families with familial melanoma may be warranted.

Corresponding author and reprints: Sancy A. Leachman, MD, PhD, Hunstman Cancer Institute at the University of Utah, 2000 Circle of Hope, Room 5242, Salt Lake City, UT 84112 (e-mail: sancy.leachman@hci.utah.edu).

Accepted for publication February 18, 2003.

This study was supported by the Huntsman Cancer Foundation and the Huntsman General Clinical Research Center (Salt Lake City); grant MO1 RR00064 from the Salt Lake City General Clinical Research Center Public Health Service; grant SR03CA096429-2 from the National Cancer Institute (Bethesda, Md); and the Utah Cancer Registry, which is funded by contract NCI-CN-67000 from the National Cancer Institute with additional support from the Utah Department of Health and the University of Utah (Salt Lake City).

We thank the Doris Duke Charitable Foundation (New York, NY), Hunstman Cancer Foundation, The Tom C. Matthews Jr Familial Melanoma Research Clinic (Salt Lake City), the Pedigree Population Resource (Salt Lake City), and the following individuals: Jahn Barlow, Thomas Day, Heidi Hanson, Marybeth Hart, MS, Richard Kerber, PhD, Patricia Porter-Gill, and Lisa Wadge, MS, CGC, for their support of and contributions to this manuscript.