The pancreatic cancer proteome

---

Pancreatic cancer is the 11th most common cancer globally and is associated with poor prognosis. Estimated 5- year survival rate is less 5% and pancreatic cancer accounts for 4% of all deaths that occur worldwide each year. The risk of pancreatic cancer increases with age and the tumor is slightly more common in women than in men. Most patients suffering from pancreatic cancer are above 50 years of age and pain, jaundice and weight loss are the most common symptoms. The cause of pancreatic cancer is unknown. However, pancreatic cancer is more common in people with diabetes and chronic pancreatitis (persistent inflammation in the pancreas) as well as in tobacco smokers.

Ductal adenocarcinoma is the most common form of pancreatic cancer, accounting for 85% of cases. Pancreatic endocrine tumors constitute the remaining 5% of reported cases. Due to diffuse symptoms pancreatic cancer is often detected at such a late stage of the disease that curative surgery is not possible. In addition to local spread in the pancreas and to surrounding tissues, pancreatic cancer often spreads to regional lymph nodes and to the liver. Over 80% of all patients have metastasis at time of diagnosis.

Here, we explore the pancreatic cancer proteome using TCGA transcriptomics data and antibody based protein data. 1527 genes are suggested as prognostic based on transcriptomics data from 176 patients; 670 genes associated with unfavourable prognosis and 857 genes associated with favourable prognosis.

TCGA data analysis

In this metadata study we used data from TCGA where transcriptomics data was available from 176 patients in total, with 80 female patients and 96 male patients. 84 patients were alive and 92 patients were deceased at the time of data collection. The stage distribution was stage i) 21 patients, stage ii) 145 patients, stage iii) 3 patients, stage iv) 4 patients and 3 patients with missing stage information.

Unfavourable prognostic genes in pancreatic cancer

For unfavourable genes, higher relative expression levels at diagnosis gives significantly lower overall survival for the patients. There are 670 genes associated with unfavourable prognosis in pancreatic cancer. In Table 1, the top 20 most significant genes related to unfavourable prognosis are listed.

MUC1 is a gene associated with unfavourable prognosis in pancreatic cancer. The best separation is achieved by an expression cutoff at 65.9 fpkm which divides the patients into two groups with 17% 5-year survival for patients with high expression versus 60% for patients with low expression, p-value: 4.06e-4. Immunohistochemical staining using an antibody targeting MUC1 (CAB000036) shows differential expression pattern in pancreatic cancer samples.

KPNA4 is another gene associated with unfavourable prognosis in pancreatic cancer. The best separation is achieved by an expression cutoff at 8.6 fpkm which divides the patients into two groups with 6% 5-year survival for patients with high expression versus 64% for patients with low expression, p-value: 1.03e-6. Immunohistochemical staining using an antibody targeting KPNA4 (CAB034336) shows differential expression pattern in pancreatic cancer samples.

Favourable prognostic genes in pancreatic cancer

For favourable genes, higher relative expression levels at diagnosis gives significantly higher overall survival for the patients. There are 857 genes associated with favourable prognosis in pancreatic cancer. In Table 2, the top 20 most significant genes related to favourable prognosis are listed.

PELP1 is a gene associated with favourable prognosis in pancreatic cancer. The best separation is achieved by an expression cutoff at 9.3 fpkm which divides the patients into two groups with 74% 5-year survival for patients with high expression versus 14% for patients with low expression, p-value: 4.07e-5. Immunohistochemical staining using an antibody targeting PELP1 (HPA060760) shows differential expression pattern in pancreatic cancer samples.

The pancreatic cancer transcriptome

The transcriptome analysis shows that 73% (n=14256) of all human genes (n=19571) are expressed in pancreatic cancer. All genes were classified according to the pancreatic cancer-specific expression into one of five different categories, based on the ratio between mRNA levels in pancreatic cancer compared to the mRNA levels in the other 16 analyzed cancer tissues. 194 genes show some level of elevated expression in pancreatic cancer compared to other cancers (Figure 1). The elevated category is further subdivided into three categories as shown in Table 3.

Figure 1. The distribution of all genes across the five categories based on transcript abundance in pancreatic cancer as well as in all other cancer tissues.

Additional information

Most patients with pancreatic cancer displayed diffuse symptoms which poses a major obstacle for early detection. Moreover, imaging techniques and biopsy often give inconclusive results due to malign lesion being indistinguishable from chronic pancreatitis or benign cysts. There are a great need for biomarkers to facilitate early detection and help establishment of diagnosis. The only available treatment for pancreatic cancer are surgical resection. However, less than 25% of all patients have tumors that are eligible for resection at the time of diagnosis.

Ductal adenocarcinomas are poorly differentiated tumors that are characterized by atypical cells forming irregular, often complex tubular or glandular structures, embedded in a dense tumor stroma. Two-thirds of tumors arise in the head of the pancreas. This type of pancreatic cancer grow rapidly and have often spread beyond the pancreas at time of diagnosis. It is often accompanied by chronic pancreatitis showing infiltrates of inflammatory cells, fibrosis and atrophy of normal exocrine pancreatic structures. Tumor cells vary in shape and size of tumor cell nuclei, often with distinct nucleoli. Ductal carcinomas are classified as well, moderately or poorly differentiated dependent on morphology. The degree of differentiation may vary within a tumor so that occasional anaplastic foci of tumor can be present within an otherwise well-differentiated ductal carcinoma. The differentiation grade has not proven helpful for predicting prognosis and likewise has staging only limited value as the vast majority of patients have an advanced stage of the disease when. Immunohistochemical analysis shows that most cases of ductal adenocarcinomas show expression of CK7 and are also positive for CK8, CK17, CK18, CK19, CEA, CA19-9, Dupan-2, MUC1, MUC4 and MUC5AC.

Pancreatic endocrine tumors (PETs) are categorized clinically as either "functioning" or "non-function" depending on whether there are any symptoms indicating elevated hormone production. Around 40 % of PETs are classified as "non-functioning" and these tumors may still produce hormones detectable by immunohistochemistry or blood samples despite lack of symptoms. These lesions are often noticed by clinicians once they have reached a size that affect other organs in the vicinity or when they metastasize. These tumors are classified according to WHO classification into three categories based on size, mitotic count, Ki-67 proliferation index, angioinvasion and metastasis. Cells in PETs are typically round with granular pattern in cytoplasm and chromatin pattern in nuclei. Synaptophysin and chromogranin are use for identifying neuroendocrine differentiation in cancers.

In addition to the typical ductal carcinoma there are uncommon variants of pancreatic cancer, including adenosquamous carcinoma, mucinous carcinoma, papillary carcinoma and acinar cell carcinoma. Moreover, mucinous cystadenocarcinoma and tumors corresponding to the endocrine compartment of the pancreas also exist with various symptoms, microscopical features and often less severe prognosis.

Relevant links and publications

Uhlen M et al, 2017. A pathology atlas of the human cancer transcriptome. Science.
PubMed: 28818916 DOI: 10.1126/science.aan2507

Cancer Genome Atlas Research Network et al, 2013. The Cancer Genome Atlas Pan-Cancer analysis project. Nat Genet.
PubMed: 24071849 DOI: 10.1038/ng.2764

Uhlén M et al, 2015. Tissue-based map of the human proteome. Science
PubMed: 25613900 DOI: 10.1126/science.1260419

Danielsson A et al, 2014. The human pancreas proteome defined by transcriptomics and antibody-based profiling. PLoS One.
PubMed: 25546435 DOI: 10.1371/journal.pone.0115421

Freelove R et al, 2006. Pancreatic cancer: diagnosis and management. Am Fam Physician.
PubMed: 16477897 

Ilic M et al, 2016. Epidemiology of pancreatic cancer. World J Gastroenterol.
PubMed: 27956793 DOI: 10.3748/wjg.v22.i44.9694

Asa SL. 2011. Pancreatic endocrine tumors. Mod Pathol.
PubMed: 21455203 DOI: 10.1038/modpathol.2010.127

Wong HH et al, 2012. Immunohistochemical features of the gastrointestinal tract tumors. J Gastrointest Oncol.
PubMed: 22943017 DOI: 10.3978/j.issn.2078-6891.2012.019

Herreros-Villanueva M et al, 2016. Non-invasive biomarkers in pancreatic cancer diagnosis: what we need versus what we have. Ann Transl Med.
PubMed: 27162784 DOI: 10.21037/atm.2016.03.44

Histology dictionary - Pancreatic cancer