Pancreatic ductal adenocarcinoma

What remains unsolved: detection (no screening outside high-risk surveillance; MCED tests and new-onset-diabetes enrichment are the leading ideas), the stroma and immune exclusion that have defeated every checkpoint inhibitor trial, resistance to RAS inhibitors (already emerging), and the fact that half of patients are too frail for the most effective regimens. The most promising directions are RAS inhibitors moving into first line and neoadjuvant settings, personalised (autogene cevumeran) and shared-antigen (ELI-002 7P) vaccines in the adjuvant setting despite AMPLIFY-7P's miss, CLDN18.2 and FAP-directed delivery, and blood-based early detection.

How common it is, world and United States. The IARC pancreas fact sheet, served with the GLOBOCAN 2024 estimates, counts 531,318 new pancreatic cancers a year (the 11th commonest cancer, age-standardised rate 4.7 per 100,000) and 490,786 deaths (the 6th commonest cause of cancer death, 4.2 per 100,000). Asia has 45.9 percent of cases, Europe 27.7 percent, Northern America 13.6 percent, Latin America and the Caribbean 7.8 percent, Africa 3.9 percent and Oceania 1.1 percent, and the highest rates are in Western Europe and Northern America, the lowest in South Central Asia. Countries with very high human development had the highest incidence and mortality (age-standardised rates 7.7 and 4.9 in 2018), incidence rose in 14 of 48 countries with registry data for men and 17 for women over the decade to 2017, and higher national rates travel with smoking, alcohol, inactivity, obesity, hypertension and high cholesterol (Huang 2021). In the United States the SEER programme projects 67,530 new cases (3.2 percent of cancers) and 52,740 deaths (8.4 percent of cancer deaths) for 2026; the rate of new cases is 13.9 per 100,000 a year (2019 to 2023) and the death rate 11.3 (2020 to 2024); it is the third leading cause of cancer death; 113,931 people were living with it in 2023. Applying current rates to population growth, pancreatic cancer is projected to become the second leading cause of cancer death in the United States by 2040 (46,000 deaths, behind lung cancer's 63,000), overtaking colorectal cancer (Rahib 2021; the 2014 Rahib projection is in the key papers).

United Kingdom (Cancer Research UK, drawing on the four national registries). Pancreatic cancer is the 10th commonest cancer with 11,479 new cases a year (2019 and 2021 to 2022), 31 a day and 3 percent of all cancers, about 5,600 in females and 5,800 in males; almost half (48 percent) are diagnosed at 75 or over and rates are highest at 90 and over. Incidence rates have risen by about a fifth (21 percent) since the early 1990s and by 7 percent in the last decade, and are projected to rise a further 4 percent by 2038 to 2040 with the count reaching about 16,000 a year; rates are lower in the Asian and mixed ethnic groups and higher in the Black ethnic group than in the White group in England (2013 to 2017). It is the 5th commonest cause of cancer death with about 10,200 deaths a year (2022 to 2024), 28 a day, more than half (54 percent) at 75 or over; mortality rates have stayed flat since the early 1970s, and although rates are projected to fall 4 percent, deaths are projected to reach about 12,600 a year by 2038 to 2040 as the population ages. Mortality is 20 percent higher for men and 25 percent higher for women in the most deprived areas of England than the least deprived (2007 to 2011). The lifetime risk of a diagnosis is 1 in 59 for females and 1 in 55 for males born in 1961. Cancer Research UK's summary page says 37 percent of cases are preventable and its risk page 31 percent; both figures are recorded here. In England in 2022, 26 percent of cases with a known stage (around 1,600) were stage I or II; Northern Ireland reports 30 percent, Scotland 19 percent (2023) and Wales 19 percent. In 2024 there were around 310 curative and 370 palliative radiotherapy episodes for pancreatic cancer in England.

Who gets it. Age is the main risk factor: the US median age at diagnosis is 71, 32.7 percent of cases are diagnosed at 65 to 74 and only 3 percent under 45 (SEER 21, 2019 to 2023); the US rate is slightly higher in men (15.7 against 12.4 per 100,000) and highest in non-Hispanic Black men and women (18.0 and 15.4) against 16.3 and 12.3 in non-Hispanic White, 13.6 and 12.1 in Hispanic and 11.5 and 9.8 in Asian and Pacific Islander people. Smoking: current smokers have 2.2 times the risk of never smokers, former smokers 1.2 times, rising to 3.4 times at 35 or more cigarettes a day, and the risk returns to that of never smokers about 20 years after quitting (Bosetti 2012, 6,507 cases in the PanC4 consortium); 22 percent of UK cases are caused by smoking (Cancer Research UK). Body size: obesity (body mass index 30 or more) carries 1.47 times the risk of a normal weight, being overweight in early adulthood and obese later 1.54 times, and a high waist-to-hip ratio 1.35 times, in 846,340 people across 14 cohorts (Genkinger 2011); Cancer Research UK puts 12 percent of UK cases down to overweight and obesity, with a 10 percent rise in risk per 5 units of body mass index. Diabetes: type 2 diabetes carries about 1.8 to 1.9 times the risk (odds ratio 1.82 in 36 studies, Huxley 2005; relative risk 1.94 in 35 cohorts, Ben 2011), and the association is strongest when the diabetes is recent, which is where cause and consequence blur: 47 percent of 512 newly diagnosed patients had diabetes against 7 percent of controls, much of it new-onset and resolving after resection, so the tumour itself is diabetogenic (Pannala 2008). Chronic pancreatitis: about 16 times the risk within two years of the pancreatitis diagnosis, falling to about 8 times at five years and 3.5 times at nine years or more, part of the early excess being cancers first misread as pancreatitis (Kirkegard 2017); hereditary pancreatitis with a PRSS1 mutation raises risk more than 50-fold (Cancer Research UK). Alcohol at 3 or more units a day raises risk by 15 to 19 percent; gallstones by 25 percent, mostly soon after diagnosis; metabolic syndrome by 58 percent in women; red meat by 21 percent in men (Cancer Research UK risk page). The same factors, with alcohol most strongly, apply to cancers diagnosed before 60 (McWilliams 2016).

Family history and inherited genes. A first-degree relative with pancreatic cancer raises risk by 62 to 76 percent (Cancer Research UK); in the Johns Hopkins registry, members of familial kindreds (at least one pair of affected first-degree relatives) had 9.0 times the expected incidence, rising to 6.4 times with two and 32 times with three affected first-degree relatives, while spouses and unrelated relatives had no excess (Klein 2004). Germline testing of 3,030 unselected patients found a pathogenic variant in one of six genes in 5.5 percent (7.9 percent with a family history of pancreatic cancer, 5.2 percent without): ATM 2.3 percent of cases (odds ratio 5.71), BRCA2 1.9 percent (6.20), BRCA1 0.6 percent (2.58), CDKN2A 0.3 percent (12.33), TP53 0.2 percent (6.70) and MLH1 0.13 percent (6.66) (Hu 2018); 3.9 percent of 854 apparently sporadic resected cancers carried a deleterious germline mutation, and only 3 of those 33 patients had reported a family history (Shindo 2017), which is why guidelines now offer germline testing to every patient. Syndromes: Peutz-Jeghers more than 100 times the risk, familial atypical multiple mole melanoma (CDKN2A) 13 to 38 times, Lynch syndrome up to about 9 times, BRCA2 3.5 times and BRCA1 up to 2.3 times (Cancer Research UK). Pancreatic cysts: intraductal papillary mucinous and mucinous cystic neoplasms are precursors, and cysts of 2 mm or more were found on MRCP in 49.1 percent of 1,077 adults in a German population study, though only about 6 percent of cysts were over 1 cm and no cancer arose in five years of follow-up (Kromrey 2018); the cyst pages of this record carry the surveillance rules.

How it presents, and why late. The pancreas sits deep behind the stomach and the early tumour is silent; roughly 75 percent of cancers start in the head (Cancer Research UK types page), where they block the bile duct and cause painless jaundice, dark urine, pale stools and itching, while body and tail tumours present later with pain in the upper abdomen or back that is worse after eating or lying down and eased by leaning forward, weight loss, loss of appetite, fatty stools (steatorrhoea), nausea, indigestion, new diabetes, fever and blood clots (Cancer Research UK symptoms page; NHS). In the SYMPTOM pancreatic study of 391 people referred with suspected pancreatic cancer, 119 (30 percent) had it and 34 percent of those were already metastatic; no first symptom separated cancer from no cancer, but jaundice as a later symptom did (49 percent against 12 percent), and the authors asked clinicians to stay alert to evolving gut and systemic symptoms, particularly in people with diabetes or mental health conditions (Walter 2016). NICE NG12 (1.2.4) sends anyone aged 40 or over with jaundice on the suspected cancer pathway and (1.2.5) asks for an urgent direct-access CT, or ultrasound if CT is unavailable, for anyone 60 or over with weight loss and any of diarrhoea, back pain, abdominal pain, nausea, vomiting, constipation or new-onset diabetes; the NHS page asks for an urgent GP appointment or NHS 111 for yellowing of the eyes or skin, vomiting for more than two days or diarrhoea for more than seven. Even so, in England in 2019 only 22 percent of cases were diagnosed through an urgent suspected cancer referral and 45 percent through an emergency presentation (Cancer Research UK), the route that carries substantially lower one-year survival across cancers (Elliss-Brookes 2012); a 2026 US Veterans Affairs measure found 60.9 percent of 4,415 patients presenting as emergencies, with 1.38 times the odds of advanced stage and 1.64 times the mortality, and missed diagnostic opportunities in nearly one in five (Br J Cancer 2026). In the United States 51 percent of cases are distant at diagnosis, 28 percent regional and 15 percent localised (SEER). Weight loss is part of the presentation, not only a complication: 85 percent of patients meet the definition of cancer cachexia (Hendifar 2019) and 40.5 percent of 227 patients scheduled for surgery were already cachectic, which halved their chance of a resection (48.9 against 77.8 percent) because metastases were found more often (Bachmann 2008).

How it is diagnosed. NICE NG85 (1.1) sets the UK sequence: for obstructive jaundice with suspected pancreatic cancer, a pancreatic protocol CT before the bile duct is drained (1.1.1); if the diagnosis is still unclear, FDG PET-CT and or endoscopic ultrasound with EUS-guided tissue sampling (1.1.2), and a biliary brushing for cytology if ERCP is used to relieve the obstruction without a tissue diagnosis (1.1.3); for pancreatic abnormalities without jaundice, the same CT then PET-CT or EUS (1.1.4 to 1.1.6); for cysts, CT or MRI with MRCP, the other test if more is needed, EUS with fine-needle aspiration and CEA on the fluid when malignancy is in question (1.1.7 to 1.1.13). The pancreas protocol CT is a thin-slice, dual-phase scan read against a standard template that names each artery and vein the tumour touches and by how many degrees (Al-Hawary 2014); Cancer Research UK adds that a biopsy is often not taken when the tumour looks removable, the diagnosis then resting on imaging until the specimen is examined, and that EUS, ERCP, ultrasound- or CT-guided biopsy or laparoscopy are the routes when tissue is needed. Blood tests include liver function and the tumour markers CA 19-9 and CEA. CA 19-9 has a sensitivity of 79 to 81 percent and specificity of 82 to 90 percent in symptomatic patients but a positive predictive value of 0.5 to 0.9 percent as a screening test, is falsely negative in the 5 to 10 percent of people who are Lewis antigen-negative, and levels above 100 U/mL suggest unresectable or metastatic disease (Ballehaninna 2012); in 1,482 Chinese patients 8.4 percent were Lewis-negative and, contrary to the textbook, 27.4 percent of them still had a raised CA 19-9, while Lewis-negative status itself predicted shorter survival (hazard ratio 1.30, Luo 2018). Germline testing and tumour sequencing (KRAS status, fusions when KRAS is wild-type, mismatch repair) are offered at diagnosis (NCCN; the subtype pages carry the detail). The specialist pancreatic multidisciplinary team makes the shared decision (NG85 1.2.1) and partners with local units to deliver it.

How it is staged. Two systems run side by side. The TNM 8th edition (2017) sizes the tumour: T1 is 2 cm or less (T1a to 0.5 cm, T1b over 0.5 to 1 cm, T1c 1 to 2 cm), T2 over 2 to 4 cm, T3 over 4 cm, T4 growing into the coeliac axis, superior mesenteric artery or common hepatic artery; N1 is one to three regional nodes, N2 four or more; M1 is spread to the liver, lungs, peritoneum or distant nodes (Cancer Research UK TNM page). Stage 1 is T1 or T2 without nodes (1A and 1B by size), stage 2 is T3 without nodes (2A) or T1 to T3 with one to three nodes (2B), stage 3 is four or more nodes or a T4 tumour, and stage 4 is any metastasis (Cancer Research UK stages page). The 8th edition's size-based T and count-based N categories were validated in 2,318 margin-negative resections from three US centres (Allen 2017) and in 8,960 SEER resections, where they discriminated survival about as well as the 7th edition (c-index 0.60 against 0.59) while stratifying resected patients more finely by nodal burden (Kamarajah 2017). The second system is resectability, which is what decides treatment: the NCCN classes of resectable, borderline resectable and locally advanced (unresectable) are defined by the tumour's contact with the superior mesenteric artery, coeliac axis and common hepatic artery (arterial contact of less than 180 degrees without stenosis is borderline; more, or encasement, is locally advanced) and with the superior mesenteric and portal veins (contact with narrowing or occlusion that can be reconstructed is borderline; unreconstructable occlusion is locally advanced), and the 2017 international consensus added a biological dimension (suspected but unproven metastases or nodes on PET-CT or biopsy, or a CA 19-9 above 500 U/mL) and a conditional one (performance status 2 or worse) (Isaji 2018; Tempero 2021). Staging tests in the UK are a pancreatic protocol CT of chest, abdomen and pelvis for every new diagnosis (NG85 1.3.1), FDG PET-CT for everyone with localised disease on CT who will have treatment (1.3.2), and, when the decision still turns on it, MRI for suspected liver metastases, EUS for tumour and node staging, or laparoscopy with laparoscopic ultrasound for small-volume peritoneal or liver disease before an attempted resection (1.3.3).

What the pathologist sees. About 80 percent of pancreatic cancers are ductal adenocarcinomas (Cancer Research UK types page), the entity this record describes: gland-forming tumours in a dense desmoplastic stroma, with the KRAS, CDKN2A, TP53 and SMAD4 alterations set out in the biomarkers list. The 2019 WHO classification of digestive tumours (5th edition, Nagtegaal 2020) lists its variants, each with a page linked from this record: adenosquamous carcinoma, colloid (mucinous non-cystic) carcinoma, undifferentiated carcinoma with or without osteoclast-like giant cells, and the carcinomas arising in intraductal papillary mucinous and mucinous cystic neoplasms; acinar cell carcinoma, solid pseudopapillary neoplasm and pancreatoblastoma are separate non-ductal tumours, and the neuroendocrine tumours of the islets are a different disease with their own record. Precursors: pancreatic intraepithelial neoplasia (PanIN), the microscopic duct lesion named and graded in 2001 (Hruban 2001) and reduced to two tiers, low grade and high grade, by the 2015 Baltimore consensus, which also set 0.5 to 1 cm as the boundary between a large PanIN and a small (incipient) IPMN (Basturk 2015); high-grade PanIN was present in 4 percent of 173 autopsy pancreata without cancer, always multifocal and more often in the body and tail, in people with diabetes and in the old (Matsuda 2017). Margins are reported by the 1 mm rule: the Royal College of Pathologists calls tumour within 1 mm of a margin R1, a definition that, applied with inked margins and axial slicing, raised the reported R1 rate to 76 to 85 percent of head resections (Verbeke 2006; Esposito 2008; Campbell 2009), and in the ESPAC-3 trial 43.9 percent of 1,151 patients were R1, with direct margin involvement carrying the worst survival and more local recurrence (Ghaneh 2019).

Screening and surveillance. There is no population screening: the UK National Screening Committee does not recommend it and Cancer Research UK explains why (the cancer is relatively uncommon, so many people would have unnecessary tests, and the benefits do not outweigh the costs), and the US Preventive Services Task Force reaffirmed a grade D recommendation against screening asymptomatic adults in 2019 (Owens 2019). Surveillance of high-risk people is different. NICE NG85 offers it (1.1.15) to people with hereditary pancreatitis and a PRSS1 mutation, to carriers of BRCA1, BRCA2, PALB2 or CDKN2A with one or more first-degree relatives with pancreatic cancer, and to Peutz-Jeghers syndrome, and asks clinicians to consider it (1.1.16) for two or more first-degree relatives across two generations and for Lynch syndrome with an affected first-degree relative, by MRI with MRCP or EUS (1.1.17), or by pancreatic protocol CT in hereditary pancreatitis where EUS is not used (1.1.18, 1.1.19). The International CAPS consortium (2020) sets the goal as high-grade dysplasia and T1N0M0 cancer, starts familial surveillance at 50 or 55 or ten years before the youngest affected relative, uses EUS and MRI/MRCP yearly, and added ATM carriers with one affected relative (Goggins 2020). The programmes: in CAPS, 7 percent of 354 high-risk people progressed over 16 years, 1.6 percent a year, and 93 percent had worrisome features first (Canto 2018); in CAPS5, 1,461 people were enrolled, 7 of 9 cancers found under surveillance were stage I (77.8 percent), and across CAPS1 to 5 (1,731 people) 57.9 percent of the 19 surveillance-detected cancers were stage I against 85.7 percent stage IV for the seven found outside surveillance, with five-year survival of 73.3 percent and a median of 9.8 years for screen-detected disease (Dbouk 2022); in the Netherlands the cumulative incidence was 9.3 percent in mutation carriers and 0 percent in mutation-negative familial kindreds over an average of 63 months, with EUS outperforming MRI and four of ten cancers presenting between visits (Overbeek 2022); in 347 CDKN2A carriers 8.9 percent developed cancer (20.7 percent by age 70), 83.3 percent resectable at imaging and 33.3 percent stage I, with five-year survival of 32.4 percent (Klatte 2022). EUROPAC, the European Registry of Hereditary Pancreatitis and Familial Pancreatic Cancer based in Liverpool, screened 321 people from 258 kindreds and found one cancer, two low-grade neuroendocrine tumours and 41 cystic lesions, the branch-duct IPMNs being unrelated to inherited risk (Sheel 2019). PRECEDE (Pancreatic Cancer Early Detection consortium) published its recommendations for an organised approach in 2021 (Gonda 2021) and its 20,000-person registry study has a trial record on this site. New-onset diabetes after 50 is the other route: about 1 percent are diagnosed with pancreatic cancer within three years (Chari 2005), the ENDPAC score (weight change, glucose change, age) picks out a group with 3.6 percent prevalence (Sharma 2018), and in over a fifth of cancers the diabetes began while the cancer was still asymptomatic, a median of 6.5 months before diagnosis (Aggarwal 2012); the glossary term and the ideas linked from this record carry the detail.