Pancreatic cancer is typically a disease of the elderly. It is extremely rare for patients to be diagnosed before the age of 30, and 90% of newly diagnosed patients are aged over 55 years of age, with the majority in their 7^th and 8^th decade of life[9,10]. The age at which the incidence peaks varies between countries. In India, for example, there is a peak in incidence in patients in their sixth decade of life whereas the in the United States this is the seventh decade of life[9].

The worldwide incidence of pancreatic cancer is higher in males than females (Age-standardised rate 5.5 in males compared to 4.0 in females)[1]. This disparity appears to be greater in higher development index countries[7]. Despite the sex difference, a systematic review of 15 studies concluded that reproductive factors were not associated with pancreatic cancer in women[11]. These findings point towards differing exposures in environmental or genetic factors as alternative explanations for the male predominance.

Within the United States, a 50%-90% increased risk of pancreatic cancer in African-Americans compared to Caucasians has been reported, while incidence rates are lowest in Pacific Islanders and Asian-Americans[9]. The higher incidence rates within the African-American population is proposed to be linked to a greater exposure to other risk factors for pancreatic cancer, such as cigarette smoking, alcohol consumption, elevated body mass index and higher incidence of diabetes[12], but there is also evidence for underlying genetic or gene environment interactions to explain at least some of the observed differences in incidence between ethnic groups[13,14].

The risk of developing pancreatic adenocarcinoma has been shown to be associated with different ABO blood groups in several large epidemiological studies. Wolpin et al[15] combined data from the renowned United States Nurse Health Study and Health Professionals Follow-up Study, and found that compared to blood patients with blood group O, patients with blood group A (HR: 1.32, 95%CI: 1.02-1.72), AB (HR: 1.51, 95%CI: 1.02-2.23), or B (HR: 1.72, 95%CI: 1.25-2.38) were at a significantly higher risk of developing pancreatic adenocarcinoma. Results from the Pancreatic Cancer Cohort Consortium which combined data from 12 prospective cohort studies was in agreement with these findings[16]. The proposed mechanisms behind this include alterations in glycosyltransferase specificity and the host inflammatory state across the different ABO blood groups[15]

Multiple studies have been performed examining the role of gut microbiota in pancreatic cancer. A systematic review by Memba et al[17] demonstrated that lower levels of Neisseria elongate and Streptococcus mitis, and higher levels of Porphyromonas gingivalis and Granulicatella adiacens are associated with an increased risk of pancreatic cancer. However, further studies are needed to validate these findings and also to establish if targeted treatment is a therapeutic possibility.

Pancreatic cancer is considered to be familial if two or more first degree relatives have previously been diagnosed with the disease and accounts for 5%-10% of new cases[27]. Patients with familial risk factors have a nine times higher risk of developing pancreatic cancer than those with no family history, and this increases to a thirty-two times greater risk if three or more first degree relatives have been previously diagnosed[28]. A meta-analysis of nine studies has also reported that individuals with a family history of pancreatic cancer were only one first degree relative has been diagnosed with pancreatic cancer, still have an 80% increased risk of developing pancreatic adenocarcinoma (RR: 1.8, 95%CI: 1.48-2.12) compared with individuals with no reported family history[29].

This points towards a strong genetic susceptibility for pancreatic cancer in a subgroup of affected patients. In familial pancreatic cancer, the risk rises exponentially with the number of first degree relatives affected and BRCA2 and PALB are the most commonly implicated mutations in this cohort[2,9]. Specific syndromes are also associated with an increased risk of pancreatic cancer compared to the general population. These are summarised in Table 2[30,31].

Diabetes is a well-established risk factor for pancreatic cancer. Stevens et al[32] performed a meta-analysis which demonstrated that the risk of pancreatic cancer was twice that in patients with type one diabetes compared to those without this condition (RR: 2.00, 95%CI: 1.37-3.01). Another comprehensive meta-analysis of 36 studies also demonstrated a similar magnitude of increased risk of pancreatic cancer in patients with type-2 diabetes (OR: 1.82 95%CI: 1.66-1.89)[33]. However, it must be noted that although diabetes is a risk factor, pancreatic cancer can also manifest itself as new onset of diabetes. This has led to interest in HbA1c as a potential biomarker of early detection in pancreatic cancer[34].

Cigarette smoking is considered the most important modifiable risk factor in pancreatic cancer with multiple individual and combined studies demonstrating a strongly positive association. The Panc4 study combined data from 12 case-control studies of which there were 6507 cancer cases and 12890 controls. The results demonstrated a dose responsive significantly increased risk of pancreatic cancer in ever smokers[18]. A meta-analysis of 82 published studies found that there is a 74% increased risk of pancreatic cancer in current (OR: 1.74, 95%CI: 1.61-1.87) and a 20% increased risk in former smokers (OR: 1.20, 95%CI: 1.11-1.29) compared to never smokers[19]. This study also found that following smoking cessation the risk remains for at least 10 years[19] while others have shown it may take up to 20 years following smoking cessation for the risk to return to baseline[9]. The Pancreatic Cancer Cohort Consortium has reported similar findings, and also found the risk increased with both duration of smoking (> 50 years OR: 2.13, 95%CI: 1.25-3.62) and number of cigarettes smoked (> 30 cigarettes/d, OR: 1.75, 95%CI: 1.27-2.42)[20].

A novel area for future research remains unanswered in relation to e-cigarettes and pancreas health. E-cigarettes deliver heated nicotine, but fewer chemicals than tobacco smoking, and have generally been promoted as safer (but not necessarily safe) alternatives to traditional cigarettes[35]. New studies are required to determine the risk/benefit balance of e-cigarettes as an exposure with unknown carcinogenic potential, or as a helpful smoking cessation tool contributing to pancreatic cancer prevention[35].

Multiple studies have investigated the impact of alcohol consumption on the development of pancreatic cancer but thus far results have been mixed[9,21,22]. A pooled analysis of 14 cohort studies with 2187 cases of pancreatic cancer found an increased risk when patients consumed > 30 g of alcohol per day (RR: 1.22, 95%CI: 1.03-1.45)[23]. The most recent meta-analysis found that low and moderate alcohol consumption was not associated with pancreatic cancer risk, however, in those with a high alcohol consumption there was a 15% increased risk of pancreatic cancer (RR: 1.15, 95%CI: 1.06-1.25; P = 0.001)[24]. This increased risk was strongest in heavy male drinkers and heavy drinkers of spirits[24].

Excessive alcohol consumption is also the main cause of chronic pancreatitis, which is a known risk factor for pancreatic cancer and therefore alcohol in this setting is a risk factor for pancreatic cancer[36].

Chronic pancreatitis is a progressive inflammatory condition of the pancreas leading to fibrosis and loss of acinar and islet cells. Significant variety exists in the reported incidence of this disease, ranging from 2-14/100000 of the United States population[37]. Approximately 5% of these patients will develop pancreatic cancer a during their lifetime[38]. Pooled results from seven studies investigating chronic pancreatitis and found significantly 13-fold higher risk of pancreatic cancer (RR: 13.3, 95%CI: 6.1-28.9) in these patients, compared with the general population or controls[38]. The relatively low incidence and greater risk infers that of chronic pancreatitis patients could be a potential target group for pancreas cancer screening, if an effective test can be found and long latency period accounted for.

The worldwide prevalence of obesity is increasing with an estimated 1.97 billion adults and 338 million children and adolescents categorised worldwide as overweight or obese in 2016[25]. The World Cancer Research Fund in the pancreatic cancer report from 2012 identified 23 studies which assessed for an association between a raised body mass index (BMI) and pancreatic cancer. Nineteen of these individual studies reported an increased risk of pancreatic cancer in obese patients and in the meta-analysis performed of these studies there was a 10% increased risk of pancreatic cancer for every 5 BMI units (RR: 1.10, 95%CI: 1.07-1.14) with no difference in outcomes between males and females[25].

Given the strength of the evidence linking obesity to pancreatic cancer, it is likely that the rising incidence of obesity is a major factor for the increasing incidence of pancreatic cancer in the developed world. There have been large public health campaigns around some of the other major lifestyle with a subsequent decrease in alcohol consumption and cigarette smoking. Similar campaigns need to focus on educating the public on the health risks associated with obesity.

Table 3 provides a concise review of the impacts of diet and nutrition on the risk of pancreatic cancer according to the World Cancer Research Fund global report. There is limited suggestive evidence that red and processed meat consumption are association with pancreas cancer development. This is biologically plausible given that excessive consumption of red and processed meat has been shown to potentially cause DNA damage and the formation of carcinogens such as N-nitroso compounds[25]. Other dietary factors with limited suggestive evidence in pancreatic cancer aetiology include foods and beverages containing fructose, or foods containing saturated fatty acids; while no conclusions could be made with regards to other dietary exposures. This reflects the difficulties in nutritional epidemiology and appropriate study designs for investigating pancreatic cancer risk.

The relationship between several infections and pancreatic cancer has also been investigated, with increased risks observed in patients with Helicobacter pylori (H-pylori)[26] or hepatitis C infections[39]. Further studies are necessary to strengthen these findings[26]. The potential association for H-pylori raises interesting speculation about H-pylori eradication (intended to reduce gastric cancer risk) having potentially negative consequences for increasing pancreas cancer incidence, as has been noted for oesophageal adenocarcinoma trends[40].

The worldwide 5-year survival rate for pancreatic cancer patients is approximately 6%, but this ranges from 2% to 9% in published literature[2,41,42]. Factors that impact on survival include age, sex, quality of healthcare available, presence of co-morbidities and lifestyle habits and some of these account for the difference in survival rates between countries. However, the main factor influencing disease outcome is the tumour stage at the time of diagnosis[43]. Unfortunately pancreatic cancer often presents late and only 20% of patients with pancreatic cancer have surgically resectable disease at time of presentation[2,43]. In patients who are able to undergo successful surgical resection, 5-year survival is quoted as 27% whereas if the patient has locally advanced or metastatic disease the median survival is six to eleven months and two and six months respectively[44]. Despite advances in surgical and medical treatment of pancreatic cancer there has been a minimal improvement in the 5-year survival rates. For example, population-based Northern Ireland cancer registry data revealed minimal improvements in five-year survival from 2.5% to 5.2% in cases diagnosed between 1993-1999 compared with 2005-2009[45]. The rising incidence and ongoing poor survival figures highlight the need to identify methods of screening patients at high risk, develop methods of early detection and improve both surgical and medical management of these patients.

Pancreatic intraepithelial neoplasia is a non-invasive microscopic lesion that occurs in the small (usually less than 0.5 cm) pancreatic ducts. It has been proposed that PanIN may have a role in the development of localised pancreatitis and that the resultant epithelial injury and repair cycles may further propagate the neoplastic process[55]. These lesions were first categorised in 2001 and initially graded from 1-3, reflecting progressive neoplastic morphological changes[56]. More recently there has been a move to simplify the classification using a two-tiered system, with the suggestion that the historical grades of 1a/1b and 2 be classified as low grade PanIN (Figure 2B), and the original PanIN 3 revised to high grade (Figure 2C)[57].

A recent microsimulation model, using the original PanIN classification, has sought to shed further light on the natural history of these lesions. Based on this model, the authors estimate an overall chance of 1.5% for men and 1.3% for women progressing from PanIN 1 to detectable pancreatic adenocarcinoma over their lifetime[58]. It was also estimated that it will take 11.3 years for men and 12.3 years for women to transform from PanIN 3 to pancreatic adenocarcinoma[58]. This represents a possible window for screening prior to the development of invasive malignancy as will be discussed later.

IPMNs are also well recognised as precursor lesions for pancreatic cancer[59]. They represent a broad group of pathology, being mainly classified as arising from the main pancreatic duct or one of the side branches. This distinction is important as the risk of malignancy is significantly different. For example, several studies found malignant cells, including carcinoma in situ, present in a mean of 70% of resected main duct IPMNs compared to a mean of 25% of side branch lesions that were removed[60].

Mucinous cyst neoplasms also represent premalignant lesions of the pancreas. They account for 25% of pancreatic cysts undergoing resection and are significantly more common in women[53]. A retrospective study of 163 patients undergoing pancreatic resection for MCN, found malignancy in 17.5% of the lesions removed[61].

Given that 1% of abdominal computed tomography (CT) scans will identify a cystic lesion of the pancreas, it is imperative that clear guidelines exist to ensure the appropriate management of these potentially premalignant abnormalities[53]. European[62] and international[63] consensus papers have recently been published and are an important point of reference for clinicians dealing with these lesions. However, there is a lack of high-quality population-based studies investigating all premalignant lesions of the pancreas and future work is needed to progress our understanding of aetiology, trends in incidence and factors affecting progression to malignancy. This is particularly urgent given the known rise in pancreatic cancer incidence, and that a diagnosis of PanIN and/or pancreatic cysts represents a potential opportunity for intervention and patient management to minimise this risk of progression. On the other hand, this must be balanced with better understanding of which patients could be considered low-risk, which could provide reassurance both to the patient and minimise unnecessary burden on healthcare systems.

PanIN is the most common precursor of pancreatic adenocarcinoma and this is supported by molecular studies that show that these lesions have genetic abnormalities that are common to adjacent pancreatic adenocarcinoma and the histological progression of PanIN parallels the accumulation of molecular abnormalities[46]. Lower grade PanIN lesions have mutations in the KRAS oncogene and exhibit telomere shortening, suggesting these are early changes on the pathway to invasive malignancy[64]. Mutations in p16, CDNK27, p53 and SMAD4 appear later and are present in higher grade PanIN and pancreatic adenocarcinoma. The rate of KRAS mutation also increases in relation to the grade of PanIN[53,65]. Abnormalities in notch signalling and sonic hedgehog pathways have also been implicated in pancreatic adenocarcinoma development and 80% of these mutations appear to be sporadic[9,43].

Recent genomic analysis identified 32 recurrently mutated genes in pancreatic adenocarcinoma and these were able to be stratified into four sub-groups namely squamous, pancreatic progenitor, immunogenic and aberrantly differentiated endocrine exocrine, each of which has a unique genomic signature which corresponded to histopathological findings and prognosis[66]. The squamous sub-type was associated with the adenosquamous histological variant of pancreatic adenocarcinoma and found to carry an independently poor prognosis. The pancreatic progenitor group highly expressed transcription factors involved in determining pancreatic cell lineage. Significant immune infiltration was found in the immunogenic tumours and the aberrantly differentiated endocrine exocrine tumours were associated with acinar cell carcinomas[66].These findings shed further light on the complex and heterogenous nature of pancreatic cancer and may aid the development of more targeted, personalised therapy based on individual tumour biology.

Given that the majority of pancreatic tumours express androgen receptors (AR), the role of these in the pathogenesis of this disease has been an area of study for many years[67-69]. Some in vitro and mouse models have shown reduced cell line proliferation and tumour shrinkage with androgen receptor blockade[69]. However, there is discordance between studies regarding AR expression in pancreatic adenocarcinoma and its possible role in pathogenesis[70]. A retrospective cohort study of 60 patients who underwent pancreatic resection found that AR expression was not related to the grade of tumour or prognosis[70]. The relative lack of robust evidence prevents any current recommendation that therapies targeting AR be used in pancreatic adenocarcinoma and this is another area in need of further study[68,71].

A significant proportion of patients with pancreatic cancer present with jaundice. This can have implications with regards to coagulopathy and increased peri-operative infective complications[90]. Traditionally, patients with obstructive jaundice would have this relieved prior to resection taking place.

A Cochrane review comparing the outcomes of five studies investigating pre-operative biliary drainage (4 via percutaneous transhepatic cholangiography and 1 using endoscopic retrograde cholangiopancreatography (ERCP) found no evidence for or against drainage, although the evidence was acknowledged to be poor[91]. However, a recent multi-centre randomised trial of ERCP and drainage vs immediate surgery found a higher rate of peri-operative complications in the drainage group[92]. This suggests that a select group of patients may do better with expedited surgery rather than biliary decompression, followed by resection.

A major source of morbidity following Whipple’s procedure is leak from the pancreatic anastomosis and formation of a pancreatic fistula[93]. It is possible to reconstruct the alimentary tract following Whipple’s by anastomosing the pancreatic remnant to the stomach or jejunum. A recent Cochrane review found no difference in outcome when these two techniques were compared to each other[94]. Variations in anastomotic technique have also been described but a recent meta-analysis failed to demonstrate reduced rates of pancreatic fistula with the “duct-to-mucosa” anastomosis vs the “invagination” technique[94].

In line with other areas, interest has grown in minimally invasive techniques for pancreatic surgery. Laparoscopic distal pancreatectomy was the first minimally invasive pancreatic resection to be described. One meta-analysis found comparable morbidity and mortality between laparoscopic and open distal pancreatectomy, with reduced blood loss and length of stay in the minimally invasive group. There was no difference in the rate of positive resection margins[95]. A further meta-analysis stated that laparoscopic distal pancreatectomy was at least non-inferior to the open procedure, but lack of level one evidence meant that it could not be deemed superior[96].

Attempts have also been made to use robotic techniques to improve Whipple’s procedure. When compared to open pancreatectomy, a meta-analysis of retrospective cohort studies found a lower complication rate and less margin involvement in the robotic group[97]. However, the lack of randomisation in these studies leaves them open to selection bias. Robotic surgery also requires a significant capital investment and no cost-effectiveness evaluations were included in any of the papers[98].

The relationship between any pancreatic tumour and the surrounding vasculature is an important determinant of resectability[99]. Whilst it is often technically feasible, the benefit of resection of mesenteric and portal vessels invaded by tumour remains a controversial topic[99].

Meta-analysis of studies involving patients undergoing Whipple’s procedure with or without major arterial resection found higher rates of peri-operative mortality and poor outcomes at year one and three in the group undergoing arterial reconstruction[100]. For this reason, invasion of the superior mesenteric artery or coeliac trunk remains a largely accepted contraindication to resection. Outcomes from venous resection, however, may be more promising. A meta-analysis of 22 retrospective cohort studies found no difference in perioperative morbidity, one or three-year survival in those undergoing resection of the portal or superior mesenteric vein when compared to those in whom no vascular intervention was undertaken. Unsurprisingly, there was an increased operative time and blood loss recorded in the venous resection group[101]. As stated previously, the lack of randomisation leaves these studies at risk of selection bias. However, combined pancreatectomy and venous resection may have a role in a select group of patients.

The use of adjuvant chemotherapy was supported by the landmark randomised CONKO-001 study which compared adjuvant gemcitabine after complete surgical resection against surgery alone. This study demonstrated a significantly improved median disease free survival (13.4 mo vs 6.7 mo) and overall survival with a five year survival of 20.7% vs 10.4% and ten year survival of 12.2% vs 7.7%[102]. However, despite these promising results the median overall survival only improved from 20 to 23 mo (P = 0.01)[102]

Further studies have sought to identify the best chemotherapy regime. The ESPAC-3 trial demonstrated that gemcitabine was the chemotherapy agent of choice when compared to 5-fluorouracil[103]. Although survival outcomes were comparable in both groups, the latter was less well tolerated[103]. Due to the success of dual therapy of capecitabine and gemcitabine in both advanced and metastatic disease, Neoptolemos et al[103] performed the ESPAC-4 trial in patients with resected disease and found that the median overall survival was 28 mo (95%CI: 23.5-31.5) in dual therapy compared to 25.5 mo (22.7-27.9) in gemcitabine alone (HR: 0.82, 95%CI: 0.68-0.98; P = 0.032).

Other chemotherapy regime have been studied, for example, in the PRODIGE24/CCTG randomised clinical trial which compared the outcomes of gemcitabine or mFOLFIRONOX (a combination of oxaliplatin, irinotecan, and leucovorin) in patients with an R1 or R0 resection of pancreatic adenocarcinoma[104]. The results at a median follow up time of 33.6 mo have shown that administration of mFOLFIRONOX was associated with a significantly improved disease-free survival (21.6 mo vs 12.8 mo), and overall survival (54.4 mo vs 35 mo) compared to gemcitabine[105]. Administration of mFOLFIRONOX was associated with a significantly increased risk of complications although the only death that occurred was within the gemcitabine treatment group[105]. The current standard of care is guided by post-operative fitness and mFOLFIRONOX is used for very fit patients with tumours of the head, body and tail of the pancreas whereas in less fit patients dual therapy with gemcitabine and capecitabine is given[105]. Single agents (usually 5-Fu) are used for periampullary tumours as there is insufficient evidence for the same treatment as the previously mentioned tumours[106].

Although there has been shown to be a survival benefit with adjuvant treatment, between 71% and 76% per cent of patients still relapse within two years up. Furthermore, due to complications associated with surgery up to 40% of patients are not suitable for progression to adjuvant therapy[105]. Such figures coupled with the success seen with neo-adjuvant treatment in several other cancers including rectal, oesophageal, and gastric cancer have led to the exploration of the impact of neo-adjuvant treatment in pancreatic cancer[107].

The theoretical advantage of neo-adjuvant therapy includes eliminating micro-metastases and shrinkage of the primary tumour and both these factors are associated with a decreased incidence of tumour recurrence[108]. However, patients receiving neo-adjuvant treatment may develop complications which can delay or prevent the progression to surgery and tumours may be unresponsive to the chemoradiotherapy leading to disease progression and previously resectable disease becoming unresectable. Furthermore, the administration of chemo radiotherapy induces fibrosis within the pancreas which can increase the complication rate associated with pancreatectomy[109].

Studies looking at the impact of neo-adjuvant treatment have been performed in patients with resectable or borderline resectable disease. The definition of resectable disease is in those patients who have no involvement of the superior mesenteric artery, coeliac axis, portal vein or superior mesenteric vein whereas the definition of borderline resectable disease is based on the degree of involvement of these major venous and arterial structures[110].

Multiple meta-analyses have been performed studying the impact of neoadjuvant treatment on survival in pancreatic adenocarcinoma. The most recent was by Versteijne et al[110] which included 38 studies with a combination of 3 randomised controlled trials, 9 phase one or phase two trials, 12 prospective cohort studies and 14 retrospective cohort studies. In intention-to-treat analysis there was a median overall survival of 18.8 mo in the neo-adjuvant group compared to 14.8 mo in the surgery first group. For those who actually underwent surgery the median survival time was 15 mo in the surgery-first group, compared to 26.1 mo in the neoadjuvant treated group. The overall resection rate was lower in the neoadjuvant group compared to those who had surgery first (66% vs 81.3%; P < 0.001) however the R0 resection rate was higher in patients who had neo-adjuvant treatment compared to those who had surgery first (86.8% vs 66.9%; P < 0.001)[111].

The ongoing Preopanc-1 trial is a Dutch study which recruited 246 patients with resectable or borderline resectable disease[112]. Patients were randomised to either immediate surgery or to pre-operative chemoradiotherapy followed by surgery. There was an increased rate of resection in the immediate surgery group (72%) compared to the group which received preoperative chemoradiotherapy (60%) although this did not reach a level of statistical significance. There was an improved survival in intention to treat analysis with 17.1 mo in the neoadjuvant group compared to 13.7 mo in the immediate surgery group although this did not reach statistical significance (P = 0.07) either. In patients who underwent an R0 or R1 resection there was a significantly improved overall survival in the neo-adjuvant group (42.2 mo vs 16.8 mo; P < 0.001) and there was also a significantly increased time until distant metastases (P = 0.01) and loco regional recurrence (P = 0.002)[112]. It should be noted that the evidence base for neo-adjuvant treatment in pancreatic cancer is based on phase two trials and meta-analysis while the results of phase three trials are awaited.