Review

Pancreatic cancer incidence and mortality virtually coincide having a five-year survival of less than 5%. Surgical intervention is possible in only about 10% of cases and adjuvant therapies are virtually ineffective. An improved understanding of pancreas cancer genetics is the only means to provide new markers for early diagnosis and to identify potential targets for therapeutic intervention. Molecular analyses of pancreatic cancer have been hindered by the low cancer cellularity of this neoplasm, due to the characteristic host desmoplastic reaction. This has been overcome in large part to the application of various enrichment techniques such as xenografting, cryostat-enrichment, and laser capture microdissection of primary lesions. Consequently, enormous advances have recently been made in our knowledge of the molecular alterations in commonly present pancreatic ductal adenocarcinoma. Not unsurprisingly, as for other cancers, both activation of oncogenes and inactivation of tumor suppressor genes play key roles in pancreatic cancer.

Ductal adenocarcinoma is by far the most common pancreatic neoplasm, comprising around 90% of all pancreatic malignancies. However, other types of tumors arise from both the exocrine or endocrine cellular components. A dissection of the assortment of genetic anomalies has allowed for the important distinction at the molecular level that common ductal carcinoma has a distinct molecular fingerprint compared to other nonductal exocrine or endocrine tumors. Moreover, while endocrine tumors have been traditionally grouped together as a single class of neoplasms, significant molecular heterogeneity exists even among this broad faction of tumors. Accordingly, the present review will give a brief overview of the genetic alterations that have been identified in the various subgroups of pancreatic lesions. The known chromosomal/genetic alterations commonly present in pancreatic tumors are presented in Table 1.

Table 1 Summary of Known Chromosomal/Gene Alterations in Pancreatic Tumors
Full size table

Ductal adenocarcinoma

At present, the well-established molecular events that have been correlated with the pathogenesis of ductal neoplasms include activation of the K-ras oncogene in about 80% of cases and inactivation of the tumor suppressor genes p16INK 4a, p53, and DPC4 in about virtually all, 60%, and 50% of primary cancers, respectively [1, 2]. In addition, anomalies in FHIT gene expression have been shown in about half of cases [3]. The coexistent inactivation of p16, p53, and DPC4 is also common [1, 2]. Telomerase activity has been detected in up to 95% of cases and undetectable in benign tumors [4].

The inactivation of additional genes has been implicated in pancreatic cancer albeit at lower frequencies. These include MKK4, part of a stress-response signaling pathway, in 4% of cases; a TGF-beta receptor in 1% of cases;BRCA2 in nearly 7% and LKB1/STK11 in about 5% of cases [5–9]. Microsatellite instability of the type associated with mutation of DNA repair genes has been observed in a small proportion of ductal cancers and has been associated with an improved prognosis [10, 11]. Microsatellite instability has also been found in a limited number of the rare medullary carcinoma [12].

The knowledge of gene alterations involved in transformation has also allowed for the identification and classification of morphological precursor lesions (PanIN), which has permitted the construction of a model for disease progression [13]. Whereas mutation of K-ras and overexpression of Her2/neu are initial events, alteration of p16 has been associated with progression of malignancy. It is also apparent that mutation of p53 and DPC4 are events that occur relatively late in the transformation process [14, 15]. These are not trivial points as a model of genetic progression forms the basis of our understanding of cancer transformation and progression of malignancy. Telomere shortening has also been recently observed to be a frequent event in PanINs, which is highly suggestive that this genetic event plays an early role in neoplastic progression [16]. The fact that p16 is also methylated in PanINs gives further support to this idea [17, 18]. Such a model has important implications for the development of genetic screening tests, early diagnosis, prognostic genetic markers and chemoprevention.

It is becoming increasingly apparent that the role of DNA methylation has in pancreatic cancer. About 50% of all human genes have 5' CpG islands, which are typically associated with the 5' regulatory regions of genes. Methylation at these 5' CpG islands generally inactivates gene expression, while demethylation has the opposite effect. While part of normal developmental and aging processes, when deregulated, these events may also inactivate tumor suppressor genes or activate otherwise silenced genes. Promoter methylation is implicated in the transcriptional silencing of several tumor suppressor and mismatch repair genes (e.g., p16, Rb, VHL, hMLH1) in many cancers. In pancreatic cancer, de novo methylation of a number of genes has been reported [19]. Among these, ppENK, a gene with known growth suppressive properties, has been associated with transcriptional silencing in over 90% of cases [20]. Larger pancreatic cancers and those from older patients possessed more methylated loci compared to smaller tumors and those from younger patients. The identification of larger numbers of genes controlled by methylation has been undertaken and should provide a larger number of genetic targets, which will be invaluable for diagnosis, prognosis, and therapy [19, 20].

The quantitative and comprehensive analysis of cellular gene expression profiles has been performed using serial analysis of gene expression (SAGE) and DNA microarrays. These expression profiling studies and SAGE studies have already allowed for the identification of several candidate genes which may be useful as potential diagnostic/prognostic markers as well as therapeutic targets [21–31]. Nonetheless, these studies are far from complete and are just beginning to unravel the enormous complexities that underlie pancreatic cancer.

One hallmark of pancreatic ductal cancers is the characteristic host desmoplastic reaction that has hindered genetic studies on this neoplasm. Several reports have examined in detail this interaction. Recently, the stromal reaction has been investigated by DNA array analysis using normal pancreas, bulk pancreatic tumor tissues and pancreatic tumor aspirates that contain more than 95% tumor cells. In fact, fine needle aspiration of cancer provides a fast and efficient way of obtaining samples highly enriched in tumor cells with sufficient yields of RNA. Altered expression of genes not previously associated with pancreatic adenocarcinoma was found, including rac1, GLG1, NEDD5, RPL-13a, RPS9 and members of the Wnt5A gene family [25].

Little is known about the genetic factors responsible for familial pancreatic cancer, although familial aggregation and genetic susceptibility may play a role in as many as 10% of pancreatic ductal adenocarcinomas. Germline mutations have been reported in BRCA2 and p16, although these account for only a fraction of cases [32–34].

Other tumors arising from the exocrine pancreas

Less is known about the molecular abnormalities in the more rare epithelial neoplasms of the exocrine pancreas. This heterogeneous group of neoplasms includes tumor entities with distinct clinicopathologic and prognostic features that comprise the acinar cell carcinomas, the serous cystic tumors, mucinous cystic tumors, intraductal papillary-mucinous tumors, the solid-pseudopapillary tumors, and ampullary cancers. In general, the available data suggests that these neoplasms have molecular pathogenetic pathways that are different from those occurring in common pancreatic adenocarcinoma.

Acinar cancers

In these neoplasms, mutations in K-ras are exceedingly rare and p53 mutations have not been found [2, 35]. Likewise, alterations in p16 or DPC4 are absent [2]. A recently performed genome wide allelotyping of these tumors has shown a high degree of allelic loss [36]. Chromosomes 1p, 4q, and 17p show LOH in >70% of cases and chromosomes 11q, 13q, 15q, and 16q show allelic loss in 60–70% of cases. The resulting allelotype of acinar carcinoma is markedly different from that of either ductal or endocrine tumors of the pancreas and the involvement of chromosome 4q and 16q seems characteristic of this tumor type. Interestingly, alterations in the APC/β-catenin pathway have been found in 4 of 17 cases of acinar carcinoma studied [37].

Serous cystic tumors

These lesions may present as sporadic forms or in association with von Hippel Lindau syndrome. A molecular characterization of these tumors consisting in genome-wide allelic loss analysis, assessment of microsatellite instability, and mutational analysis of the VHL, K-ras and p53 genes has been recently reported [38]. While no case showed microsatellite instability, a relatively low fractional allelic loss of 0.08 was found. The allelotype demonstrated that losses on chromosome 10q were the most frequent event, observed in about 50% of cases, followed by allelic losses on chromosome 3p, found in almost 40% of cases. The VHL gene was found to possess somatic inactivating mutations in two of nine (22%) cases analyzed, while no mutations were found in either K-ras or p53. Thus, involvement of chromosome 10q is characteristic of serous cystic tumors and the VHL gene is involved in a subset of sporadic cases.

Mucinous cystic tumors

K-ras mutations have been reported in a variable proportions among mucinous cystic tumors. The expression of p53 and Dpc4 has been reported as frequently altered and is likely to be related to progression in this malignancy [39, 40].

Intraductal papillary-mucinous tumors

To date, K-ras mutations have been found in intraductal papillary-mucinous tumors (IPMT) at varying frequency, while no alterations of p16, p53 or DPC4 have been found [2, 41]. There is also a low frequency of LOH found on chromosomal arms 9p, 17p, and 18q further substantiating the supposition that these are entities separate from ductal cancers. IPMT have also been recently studied by cDNA microarray analysis, which led to the observation that several genes are differentially expressed both in IPMTs and pancreatic carcinomas. This suggests that they may be involved at an early stage of pancreatic carcinogenesis [42].

Solid-pseudopapillary tumors

These neoplasms are generally low-grade malignancies primarily affecting girls and young women and characteristically show progesterone receptor immunostaining [39]. Neither alterations in ras-family genes, p53 gene/protein, p16, or DPC4 have been found. Similarly, allelic losses on chromosomal arms 9p, 17p, or 18q have not been detected in these tumors [2]. Immunohistochemistry showed positive staining for Dpc4, as expected from microsatellite analysis [2]. Recently, it has been shown that the vast majority of cases harbor mutations in β-catenin [43, 44].

Ampullary cancers

Compared with other exocrine pancreatic tumors, only ampullary cancers have a molecular fingerprint that may involve some of the same genes most frequently altered in common ductal cancers. In fact, alterations have been found in K-ras, p53, p16, and DPC4, with p53 inactivation being the most frequent event (60%) [45]. K-ras mutations are seen in about one-half of cases [46, 47]. Inactivation of DPC4 was found in about 50% of cases as shown by negative staining for the protein by immunohistochemistry [2]. There is however, no correlation between the lack of expression of Dpc4 and survival [48]. However, allelic losses on chromosomal arm 17p (63%) have been previously found to be an independent prognostic factor among ampullary cancers at the same stage [49]. Taken together, this data reinforces the hypothesis that ductal tumors and ampullary cancers share common molecular pathways related to tumorigenesis and, possibly, progression of malignancy. Mutations in the APC gene have also been found in a proportion of these cancers [50]. Unlike ductal carcinoma however, about 10% of ampullary tumors show microsatellite instability, a feature that significantly correlates with increased survival [51].

Pancreatic endocrine tumors

Much progress has been made in the understanding of pancreatic endocrine tumors (PETs). Nonfunctioning (NF) PETs do not lead to clinical symptoms due to hormonal hypersecretion by the neoplasm, while functioning PETs are in fact a heterogeneous group of malignancies that give rise to clinical symptoms due to hormonal hypersecretion by the neoplasm. To date, nine distinct functional PETs have been designated, although detailed molecular studies of these neoplasms are lacking, in contrast to NF-PETs.

In reality, the elucidation of the molecular events involved in PET carcinogenesis has in part been hindered by the fact that these neoplasms have been considered a single disease entity. The emergence of novel molecular characterization strategies has made it apparent however that these lesions exhibit diverse molecular fingerprints (see [52] for review). Studies involving the genes most frequently altered in exocrine pancreatic tumors (i.e., p53, K-ras, p16 and DPC4) have confirmed that PETs arise from distinctly different molecular pathways and are unrelated to ductal cancers [2]. Mutations in K-ras and p53 are extremely rare and p16 and DPC4 alterations are virtually absent [2, 53]. The rare involvement of Dpc4 in either primary or metastatic PETs has also been confirmed by immunohistochemistry [54], but reinforces the fact that these tumors have pathogenetic pathways distinct from ductal adenocarcinoma.

To date, mutation of MEN-1 is the most common genetic alteration found in PETs, but with markedly different frequencies among insulinomas (7%), other functioning PETs (44–67%) NF-PETs (27%), giving the first genetic clue that PETs might be divided into the three above-mentioned subgroups (see [55]). The fact that mutations in MEN-1 are found in NF-PETs is not surprising when considering that NF-PETs are fairly common in MEN1 patients. Mutations in VHL are extremely rare in sporadic PETs [55, 56].

A recent high resolution allelotype for NF-PET has suggested the existence of two subgroups: those showing frequent, large allelic deletions and those showing a small number of random losses, designated high or low FAL, respectively [57]. Chromosomes 6q and 11, 20q, and 21 show frequent LOH. The allelotype of NF-PET is moreover markedly different from that of ductal, acinar, or serous tumors of the pancreas as well as from that of functional PETs [57–61]. Moreover, the two genetic phenotypes also show correlation with ploidy status: high-FAL tumors are aneuploid, while low-FAL neoplasms are diploid. When utilized in conjunction with the Ki-67 cellular proliferation index, ploidy status provides powerful, independent statistically significant information that predicts long-term survival, even among metastatic cases [57]. It is likely that the eventual separation of PETs into distinct clinical and pathological groups will facilitate an even more precise delineation of PET prognosis, histopathology, and carcinogenesis.

More recently, the study of sex chromosome abnormalities in PETs by both microsatellite and FISH analysis identified different frequencies of loss and gain of sex chromosomes in female and male cases [62]. The loss of a sex chromosome significantly correlated with the presence of local invasion, metastasis, and higher proliferation status. Moreover, sex chromosome loss is significantly associated with poor survival and increases the risk of death by approximately two-fold [62].

Conclusion

The demonstration and realization that ductal adenocarcinoma has a distinct molecular fingerprint compared to other nonductal or endocrine tumors has important implications for the development of desperately needed genetic screening tests, as well as for markers for early diagnosis and prognosis. Unfortunately, the results of genetic studies have not yet been translated into significant clinical applications that show either diagnostic or therapeutic benefit for patients with pancreatic malignancies. Much promise is held for the use of new technologies such as expression profiling and proteome analysis.