Postoperative Staging of Pancreatic Cancer

 

Pancreatic cancer is a devastating disease with a 5-year survival rate of 3%-5%. The mortality of pancreatic cancer is almost identical with its in-cidence. The vast majority are pancreatic ductal adenocarcinomas. It is typically a tumour of the elderly. The main risk factor is smoking. Clini¬cal and histopathological studies have identi¬fied pancreatic cancer precursor lesions. These include pancreatic intraepithelial neoplasia (PanIN), intraductal papillary mucinous neo¬plasia (IPMN) and mucinous cystic neoplasm (MCN). To improve patient prognosis, surgical interventions have become more aggressive, in¬cluding pancreaticoduodenectomy and more or less radical lymphadenectomy. Following sur¬gery, it is the surgical pathologist who provides valuable information regarding the exact tumour localization, histological tumour type, grading, completeness of resection, nodal status and the presence of precursor lesions. Although many tissue-based prognostic biomarkers have been characterized and can be studied by immunohis-tochemistry or molecular biological techniques, their impact on patient management and treat¬ment is still limited. More recently proteomic profiling has raised hopes for early cancer detec¬tion, thereby improving the prognosis of pancre¬atic cancer patients.

5.1      Pancreatic Cancer
Epithelial tumours of the pancreas arise from or share similarities with the duct epithelium, aci-nar cells or endocrine cells (Table 5.1). The vast majority of pancreatic cancers are pancreatic ductal adenocarcinomas (PDAC), which account for 85%-90% of pancreatic tumours (Table 5.1). The incidence ranges from 3.1 to 20.8/100,000 per year for men and from 2.0 to 11.0/100,000 per year for women. Within the last 40 years the incidence of PDAC has tripled. It is typically a tumour of the elderly. Of PDAC patients, 80% are in their seventh to ninth decade of life. PDAC occurs rarely in patients younger than 40 years. The median survival of PDAC is less than 6 months and the 5-year survival rate is 3%-5%. The mortality of PDAC is almost identical with its incidence (Hamilton and Aalton 2000).
The main risk factor for PDAC is smoking. It is estimated that 25% of pancreatic cancers are related to smoking. Smokers have a twofold in-creased risk of pancreatic cancer compared with non-smokers. Less well defined risk factors are dietary factors, chronic pancreatitis, and diabe¬tes mellitus. While numerous inherited germline mutations are associated with pancreatic cancer, only 10% or less of pancreatic cancers are caused by an inherited disorder. The commonest in¬herited genetic disorder is caused by mutations in the BRCA2 (breast cancer type 2 susceptibil¬ity protein) gene which, in addition to causing breast and ovarian tumours, can increase the frequency of pancreatic cancer. Epigenetic alter¬ations also contribute to pancreatic cancer biol¬ogy and pathogenesis (Hezel et al. 2006; Karhu et al. 2006; Maitra et al. 2006; Sato and Goggins
2006).
PDAC mainly affects the head (60%-70%) and less commonly the body and tail of the pancreas. PDAC are firm and poorly defined tumours. The cut surface is yellow to grey-white. The mean size is 2.5-3.5 cm in diameter (range 1.5-5.0 cm). Tumours of the head are usually smaller than tu-mours occurring in the body and tail (Hamilton
and Aalton 2000).
5.2    Precursor Lesions
Clinical and histopathological studies have identified pancreatic cancer precursor le¬sions, i.e. pancreatic intraepithelial neoplasia (PanIN), intraductal papillary mucinous neo-plasia (IPMN) and mucinous cystic neoplasm (MCN). PDAC probably develops from PanINs and IPMNs. The four categories of PanIN, i.e. PanIN-1A, PanIN-1B, PanIN-2 and PanIN-3, and the three categories of IPMN, i.e. IPMN ad¬enoma, IPMN borderline and intraductal papil¬lary mucinous carcinoma, constitute a spectrum of cytological and histological alterations of the pancreatic duct epithelium which is character¬ized by increased atypias and an accumulation of genetic alterations that finally lead to invasive cancer (Hruban et al. 2001, 2004). MCN is a dis-tinct lesion which typically occurs in women and is almost always located in the tail or body of the pancreas and may progress to mucinous cystad-enocarcinoma (Klimstra 2005).

5.3    Surgical Pathology Report
for Pancreatic Cancer
Different surgical procedures are in use for the surgical treatment of pancreatic cancer. Among these, pancreaticoduodenectomy with or with¬out pylorus preservation has become the stan¬dard surgical procedure for resectable pancreatic cancers of the head (Alderson et al. 2005). To im-prove patient prognosis, surgical interventions have become more aggressive, including more or less radical lymphadenectomy (Yeo et al. 2002). Following surgery, it is the surgical pathologist who provides valuable information regarding the exact tumour localization, histological tumour type, grading, completeness of resection, nodal status and the presence of precursor lesions. Ide-ally, the surgical pathology report will enclose all these clinically useful and relevant data, includ¬ing tumour node metastasis (TNM) staging ac-cording to Unio Internationale Contra Cancrum (UICC) (Table 5.2). Several proposals and rec-ommendations have been published describing requirements for the examination and report¬ing of pancreaticoduodenectomy specimens,
forming the basis for an adequate post-operative staging and assessment of patient prognosis (Al-bores-Saaverda et al. 1998; Compton and Hen-son 1997; Luttges et al. 1999).
5.4     Macroscopic Examination
Duodenopancreatectomy specimens should be examined in a fresh unfixed state. The bile duct and the main pancreatic duct should be probed and the whole specimen should be cut horizon-tally along the probes (Luttges et al. 1999). Fresh unfixed samples from tumour and non-lesional tissue for subsequent molecular biological in-vestigations or research purposes should only be obtained by a trained surgical pathologist. The pancreatic carcinoma's site of origin must be identified exactly. Pancreatic cancer is a neo¬plasm localized in the head of the pancreas. The ampullary carcinoma has its centre in the region of the ampulla. It should be specified whether the ampullary carcinoma predominantly in¬volves the ampulla, the intraduodenal portion of the common bile duct or the pancreatic duct. The peri-ampullary carcinoma is usually an ad¬vanced tumour that eludes any definition of its precise site of origin. Finally, the terminal bile duct carcinoma may also extend into the head of the pancreas and stems from the lower third of the bile duct. Ampullary carcinomas usually have a significantly better prognosis than pan¬creatic cancer. However, most tumours present in the advanced stages, often preventing specifi¬cation of the exact anatomical origin.
Cystic tumours should be assessed with re-gard to the relationship with the main pancreatic duct, presence or absence of a pseudocapsule, whether the cystic lesion is uni- or multilocular, the character of the cystic content (serous, muci-nous, bloody) and the internal surface (smooth, papillary projections), and the presence or ab¬sence of mural nodules.
Macroscopic examination should also specify the local tumour extension, i.e. tumour size at least in two dimensions, the distances from the closest margin and towards the ampulla, the dor-sal resection margin, and the resection margins of the pancreatic and common bile ducts. The retroperitoneal resection margin is prognosti-cally important and should be thoroughly sam¬pled and labelled. In case of suspected tumour infiltration of the portal vein or superior mesen-teric artery, these should be detached separately from the specimen, serially sectioned and sub¬

mitted in their entirety to histology. Both vascu¬lar ends and the dorsal perivascular tissue are of high prognostic significance (Luttges et al. 1998, 1999).
Any lymph node attached to the resection specimen should be recorded with regard to its localization and submitted to histology.

5.5     Microscopic Examination
Adequate post-operative staging of pancreatic tumours requires a thorough histological exami-nation and classification according to WHO cri-teria (Table 5.1). The histological type has a sig-nificant impact on patient prognosis. Although the majority of pancreatic tumours are PDACs, other less common varieties have to be ruled out, including endocrine tumours, acinar cell carcinomas and metastases. Histological tumour typing of cystic tumours has to separate MCN and IPMN, which have a much better prognosis than cystically degenerated PDACs. While a mi-nority of patients with PDAC survive for at least 5 years, unexpected long-term survival should raise suspicion about whether a correct diagnosis was reached by histological examination (Car-pelan-Holmstrom et al. 2005). Thus, proper rec¬ognition of variants of PDAC and other malig-nant tumours of the pancreas requires specialist pathological expertise (Alderson et al. 2005).
Grading of PDAC is essential and an inde-pendent prognostic factor. The WHO criteria entail the combined assessment of glandular dif-ferentiation, mucin production, nuclear atypia, and mitotic activity, in which each of these four
categories is evaluated separately from 1 to 3, in the highest grade areas of the tumour and a score is obtained by summation of the results and its division by 4 (Tables 5.3 and 5.4). Grade 1 tu-mours have a score of less than 1.7; grade 2, 1.7 to 2.3; and grade 3, greater than or equal to 2.3
(Hamilton and Aalton 2000; Luttges et al. 2000).
This system has proved to be prognostically rel-evant. Patient prognosis significantly correlates with WHO tumour grading (Luttges et al. 2000). However, it is a cumbersome and complicated grading system, which also relies on counting mitotic figures in high-grade areas and requires observer experience.
Recently a new grading system was proposed that is similar to the Gleason's scoring system used for prostate cancer (Adsay et al. 2005). The Gleason's system divides the histoarchitecture of prostate adenocarcinomas into five patterns. Since the morphology of PDAC and prostate carcinoma share certain similarities, this grad¬ing system can be transferred to pancreatic can¬cer. However, since Gleason patterns 1 and 2 are practically non-existent in PDAC, this leaves only three different patterns attributable to PDAC. PDAC pattern 1 is characterized by well-formed tubular units with complete, easily discernable borders. Pattern 2 shows incomplete, ill-defined borders, fusion of glands or irregular multi-lu-mina formation (cribriform architecture). Pat¬tern 3 has non-glandular patterns including cord-like areas, individual cell infiltration, nested or solid (sheet-like) growth patterns. Finally, a score is obtained by the summation of the pre-dominant and the secondary patterns, which is then translated into an overall grade 1 (score < 3), grade 2 (score = 4) and grade 3 (score > 5). Interestingly, this modified grading system has a moderately good reproducibility (kappa value of 0.43) and can be easily applied even by surgi¬cal pathologists with limited exposure to PDACs (Adsay et al. 2005). Furthermore, it seems to more accurately predict tumour biology and pa-tient prognosis than the WHO grading system (Adsay et al. 2005).
However, both systems have clearly demon-strated that grading of PDAC matters and pre-dicts patient prognosis. Future confirmatory studies are required to prove which of these dif-ferent grading systems is more practicable and reliable or whether they can be used in a comple-mentary way.

5.6     Prognostic Factors
Apart from grading, many studies, comprising more than 3,000 patients collectively, provide strong evidence that tumour size, lymphatic in-vasion, presence of lymph node and distant me-tastases, UICC tumour stage, resection margins, and infiltration of large vessels and veins corre¬late significantly with patient prognosis (Benassai
et al. 2000; Brennan et al. 2004; Gebhardt et al. 2000; Kuhlmann et al. 2004; Lim et al. 2003;
Luttges et al. 2000; Millikan et al. 1999; Moon et al. 2006; Sohn et al. 2000; Takai et al. 2003; Tseng
et al. 2004; Wenger et al. 2000). Table 5.5 gives
a selection of studies which investigated inde-pendent prognostic factors of pancreatic cancer. Thus, post-operative staging necessitates a surgi-cal pathology report that provides all this infor-mation.
However, following surgical resection, the vast majority of pancreatic cancers return, even those with R0 status and supposedly tumour-free lymph nodes. This observation has raised concern that the routine surgical examination is not sensitive enough and fails to detect minimal residual disease. Sensitivity can be increased by immunohistochemistry and molecular assays (Niedergethmann et al. 2002; Ridwelski et al. 2001). Niedergethmann et al. (2002) showed in a prospective study that immunohistochemical detection of tumour cells in paraaortic lymph nodes and PCR-based assays with respect to mutated K-ras in codon 12 are superior to con-ventional histological examination. Tumour cells were found by conventional histology in 3 out of 69 patients with PDAC, by immunohistochemis-try in 5 and, using molecular assays, K-ras mu¬tations identical to those of the primary tumour were found in 12 paraaortic lymph nodes. All of the latter patients had recurrence after sur¬gery and a significant poorer survival than those without detection of mutated K-ras in paraaortic lymph nodes. This study supports the contention that recurrence may be related to incomplete re¬section of e.g. lymph node metastases.

5.7    Ancillary Techniques
Considerable research has focussed on identify¬ing molecular events in pancreatic carcinogen-esis, and their correlation with clinicopathologi-cal variables of pancreatic tumours and survival that can be used as an adjunct to predict patient prognosis (for a review, see (Garcea et al. 2005)). Using immunohistochemistry, the expression of oncogenes (K-ras, cyclin D1), tumour sup¬pressor genes (p53, p16, p21, SMAD4/DPC4, p27), proteins involved in apoptosis (Bcl-2, Bax, Survivin), growth factors (TGF(3, EGF, FGFs) and growth factor receptors (EGFR-1 to -4, bFGFR), proteases (MMPs, uPA, cathepsin B, hepa-ranse) cell-cell adhesion molecules (E-cadherin, a-catenin, (3-catenin, -y-catenin), and angiogenic biomarkers (VEGF, VEGFR, PDGF, TSP-1) has been investigated (Garcea et al. 2005). The ex-pression of several of these markers was shown to correlate significantly with patient prognosis and survival. However, their impact on patient management and treatment is still limited.

5.8    Proteomics
More recently, the investigation of the pancreatic cancer proteome has gained considerable atten-tion. After the first two drafts of the complete hu-man genome were published in 2001 (Lander et al. 2001; Venter et al. 2001) it was evident that there is not even half of the number of chromo-somal genes that had been expected originally: both groups identified 30,000-35,000 genes in-stead of the expected 100,000 (Lander et al. 2001; Venter et al. 2001). The "true" number of genes is surpassed by an estimated number of proteins of several millions (Anderson et al. 2004; Ander¬son and Anderson 2002; Pieper et al. 2003). The genome is basically the same in every cell type and is relatively static. Mutations, chromosomal instability and epigenetic modifications do not contribute significantly to physiological cell and tissue homeostasis. The latter is accomplished by the proteome. Apart from a low gene to gene-product ratio, several studies have indicated that mRNA expression levels do not necessarily cor¬relate with protein expression or disease progres¬sion, whereas profiles of proteins and their vari¬ous isoforms are able to more accurately identify disease states, such as cancer (Wulfkuhle et al. 2003). Cancer may be genetically based, but on the functional level, it is a proteomic disease: tumour progression, invasion and metastasis de-pend on the functional activity of proteins, such as growth factors and proteases. Additionally, the vast majority of drug targets, including those for cancer, are proteins (Wulfkuhle et al. 2003). Furthermore, transcriptomics cannot predict the activation of key signalling molecules in impor-tant protein networks. These developments and considerations have brought the proteome back into focus (for a review see Rocken et al. 2004).
Four different sources have been searched for novel protein-based biomarkers for pancre¬atic cancer, i.e. serum (Bhattacharyya et al. 2004; Koopmann et al. 2004; Xia et al. 2005; Yu et al. 2005a,b), pancreatic juice (Gronborg et al. 2004; Rosty et al. 2002), pancreatic tissue (Chen et al. 2005; Shekouh et al. 2003) and culture super-natants of pancreatic cell lines (Gronborg et al. 2005). Few groups searched for biomarkers using samples from pancreatic cancer and correspond¬ing non-neoplastic tissues (Chen et al. 2005; Shekouh et al. 2003). However, sampling of pan¬creatic tissue depends on adequate sampling. Tis¬sue homogenates of undissected non-neoplastic pancreatic tissue enclose ductal and acinar cells, various neuroendocrine cells and mesenchymal cells, among others. Normal ductal epithelial cells, from which the cancer is believed to arise, represent as little as 5% of the normal pancreas. The differences between the proteomes of undis-sected pancreatic tissue and ductal epithelium, enriched by laser capture microdissection, was elegantly demonstrated by Shekouh et al. (2003). Thus, using undissected non-neoplastic pancre¬atic tissue can generate highly misleading results
(Rocken and Ebert 2006).
In recent years many elaborate studies have shown that proteomics is suitable to search for novel biomarkers for pancreatic cancer. The number of differentially expressed or secreted proteins in pancreatic cancer is overwhelming. However, a meticulous analysis of their suitabil-ity in a large series of pancreatic cancer patients and an equal number of adequate controls is missing so far. To date, none of the biomarkers has reached a clinical stage.
5.9     Conclusion
Post-operative staging of pancreatic cancer ne-cessitates a thorough surgical pathological exam-ination of the pancreaticoduodenectomy speci-men, since it harbours information that has been shown to correlate with patient prognosis and survival, i.e. tumour size, histological tumour type, tumour grade, lymphatic invasion, presence of lymph node and distant metastases, UICC tu-mour stage, resection margins, and infiltration of large vessels and veins. Ancillary techniques such as immunohistochemistry and molecular analysis can be used to detect micrometastases and predict patient prognosis more accurately. However, post-operative staging can only influ-ence post-operative management of cancer pa-tients. It does not have any impact on the major culprit of pancreatic cancer, i.e. its recognition in advanced stages. In the future, proteomic analy-sis of patient serum or pancreatic juice may have the potential to improve early diagnosis of pan-creatic cancer patients (Ebert et al. 2006).