Pancreatic Cancer

Meet the patient

A 70-year-old smoker with long-standing type 2 diabetes is referred by his GP. For six weeks his skin has yellowed, his urine has darkened, his stools have gone pale, and itch has kept him awake; he has lost eight kilograms without trying. In the right upper quadrant you palpate a smooth, non-tender, distended gallbladder.^[3]

The two questions that decide his next month are the two that decide every pancreatic cancer: is this resectable? (the only chance of cure) and what is his performance status? (the gate to every chemotherapy regimen). Hold those two questions and the whole page slots into place.^[3]

One cancer, one anatomy, two faces

Why site decides survival, not biology. PDAC is a malignant gland-forming tumour of the exocrine ductal epithelium, wrapped in a dense desmoplastic stroma, that accounts for over 90 percent of pancreatic malignancies. It is the prototype of an almost uniformly lethal solid cancer because most patients are incurable the day they present.^[1][3]

The pancreas sits retroperitoneally behind the stomach and in front of the great vessels. A head tumour is the patient's one piece of luck: by compressing the distal common bile duct it produces painless obstructive jaundice while the disease may still be local. A body or tail tumour has no duct to obstruct, so it grows silently into the retroperitoneum and coeliac plexus until pain and weight loss appear — by which time it has usually encased vessels or seeded the liver. This anatomical asymmetry, not any special aggressiveness, is why the overall outcome is so poor.^[1]

Enlarge

Histology — name the subtype, change the prognosis

Most pancreatic cancer is PDAC, but the rarer subtypes behave very differently, and examiners reward the distinction.^[2]

Site — the clinical classification

Site predicts presentation, resectability and survival — learn the proportions.^[1]

Enlarge

The four horsemen — KRAS, CDKN2A, TP53, SMAD4

Four driver genes, in a fixed order, drive almost every PDAC. Whole-genome sequencing locked the cascade, and the order is examinable.^[5]

1. KRAS — over 90 percent of PDAC, the highest frequency of any solid tumour; the earliest and obligate driver, already present in low-grade precursor lesions. Mutant KRAS locks RAS-RAF-MEK-ERK signalling on, driving proliferation and metabolic reprogramming.
2. CDKN2A (p16) — about 90 percent; the second hit, disabling the p16-CDK4-RB cell-cycle checkpoint and unleashing the G1-S transition. Germline loss causes familial atypical multiple mole melanoma (FAMMM).
3. TP53 — 50 to 75 percent; a later event, disabling DNA-damage-driven apoptosis and cell-cycle arrest.
4. SMAD4 (DPC4) — about 55 percent; lost from chromosome 18q, and its loss is strongly associated with widely metastatic disease and worse prognosis.^[5]

The number rule — the four horsemen: KRAS, then CDKN2A, then TP53, then SMAD4. Say them in that order in a viva and the genetics marks are yours; tack on that SMAD4 loss marks the metastasis-prone tumour.^[5]

Precursor and stroma — why PDAC resists everything

PDAC climbs a PanIN (pancreatic intraepithelial neoplasia) ladder — PanIN-1 through PanIN-3 (carcinoma in situ) — that mirrors the colon's adenoma-carcinoma sequence and tracks the mutation cascade above.^[1][3]

The dominant feature is a dense desmoplastic stroma laid down by activated pancreatic stellate cells, often making up most of the tumour volume. It does three things that define the disease:^[1]

• It raises interstitial fluid pressure, collapsing tumour vessels and limiting drug delivery — the basis of chemoresistance.
• It stays hypoxic and avascular — exactly why PDAC is a hypoenhancing mass against bright normal pancreas on arterial-phase CT.
• It walls the tumour off in an immunosuppressive, T-reg-rich coat — why PDAC is refractory to checkpoint immunotherapy.^[1][2]

Enlarge

The remaining 5 to 10 percent are KRAS-wild-type tumours, more likely to carry actionable alterations — BRCA1/2, PALB2, ATM, mismatch-repair deficiency, NRG1, ALK, ROS1 or NTRK fusions — all worth testing because they open targeted and immunotherapy doors.^[5]

How common, who, and the risks you can change

Pancreatic cancer is the fourteenth most commonly diagnosed cancer worldwide but the seventh leading cause of cancer death — an unusually lethal ratio in which incidence roughly equals mortality. It is projected to become the second leading cause of cancer death in the United States by around 2030, overtaking breast, prostate and colorectal cancer as those improve.^[2][4]

The modifiable and non-modifiable risks:^[1]

• Tobacco smoking — the single biggest modifiable risk, a two-fold increase accounting for 20 to 25 percent of cases; risk falls toward baseline after ten years of abstinence.
• Chronic pancreatitis of any cause — a five- to fifteen-fold increase, highest with hereditary and tropical calcific pancreatitis, accumulating over decades.
• Obesity and physical inactivity — a body mass index over 30 raises risk by about 20 percent; central adiposity and adult-onset weight gain matter most.
• Diabetes mellitus — bidirectional; long-standing type 2 modestly raises risk, but the examinable point runs the other way (see below).
• Heavy alcohol intake — acts largely through chronic pancreatitis.
• Non-O ABO blood group and Helicobacter pylori carriage — small excess risks.
• Advancing age — rare under 40; median age at diagnosis around 70.^[1][4]

New-onset diabetes after 50 — the paraneoplastic trap

The classic trap: new-onset diabetes in a thin older adult with weight loss is paraneoplastic until imaging says otherwise — not "type 2, start metformin". Investigate before you reassure.^[1]

The hereditary 10 percent — name the syndrome, start surveillance

About one in ten pancreatic cancers cluster in families, and each syndrome carries a defined surveillance obligation.^[14]

Familial pancreatic cancer — two or more first-degree relatives with PDAC and no identifiable syndrome — carries a two- to six-fold risk that scales with the number of affected relatives. All high-risk kindreds are referred for surveillance with MRCP and endoscopic ultrasound, because detecting a small T1N0 resectable tumour is the only realistic cure in this population.^[13][14]

Biannual surveillance — every six months, alternating MRCP and EUS — is the schedule now tested for genetic and familial risk, beginning around age 40 to 50 or ten years before the youngest affected relative, and continued indefinitely.^[13]

Head tumour — painless jaundice and Courvoisier's law

The single most common presentation, in over 60 percent of patients, is painless obstructive jaundice. The skin and sclera yellow, the urine darkens, the stools turn clay-pale, and itch dominates the nights. Weight loss, anorexia and fatigue are usually marked and often precede the jaundice.^[1]

Courvoisier's law is the most examined sentence in the topic: in the presence of jaundice and a palpable gallbladder, the cause is unlikely to be gallstones. Stones scar and contract the gallbladder through repeated cholecystitis; a distended, thin-walled gallbladder means a progressive malignant obstruction — pancreatic head, cholangiocarcinoma or ampullary carcinoma — that has built up gradually without inflammation.^[1]

The classic trap: a palpable, non-tender gallbladder in a jaundiced patient is malignancy until proven otherwise, not "just gallstones". Reaching for cholecystitis treatment here costs the patient the staging CT that decides everything.^[1]

Etymology for viva gold: Courvoisier's law bears the name of the Swiss surgeon Ludwig Courvoisier (1843 to 1918); the sign is the palpable gallbladder, the law is the inference that favours malignancy over stones. Examiners test that the two words are not interchangeable.^[1]

Body or tail — pain to the back, and the diabetes you dismiss

A body or tail tumour presents late, with pain and weight loss and no jaundice. Dull epigastric pain radiates through to the back, often worse at night and relieved by leaning forward, from coeliac plexus and retroperitoneal invasion.^[1][3]

Cachexia is marked, new-onset diabetes may be the only clue for weeks, and because no duct is obstructed there is no jaundice to bring the patient in early — so these tumours are locally advanced or metastatic at diagnosis, and almost never resectable.^[1][3]

The paraneoplastic and atypical clues

• Migratory thrombophlebitis (Trousseau syndrome) — recurrent sterile thromboses in unusual sites (subclavian, portal, jugular, superficial), driven by tumour mucins and tissue factor; a paraneoplastic hallmark and a poor prognostic sign.
• Acute pancreatitis from a tumour obstructing the pancreatic duct — any older patient with unexplained pancreatitis is imaged for an underlying mass.
• Upper gastrointestinal bleeding from duodenal invasion, or from sinistral portal hypertension — a body or tail tumour compressing the splenic vein causes isolated gastric fundal varices, and a splenectomy is curative.
• Malignant ascites or a palpable Virchow node (Troisier sign) — peritoneal or nodal metastasis.
• Recent-onset depression — a debated but recognised association.^[1]

The mimics — separate the malignant from the curable

Two mimics must not be missed: autoimmune pancreatitis and ampullary carcinoma. Autoimmune pancreatitis is steroid-responsive and curable — never operate without tissue and an IgG4 if the picture is atypical. Ampullary carcinoma obstructs early, behaves far better, and is often confused with PDAC on imaging.^[1][3]

CT first, then tissue — the staging ladder

Investigation is staged and goal-directed: confirm the mass and stage it with CT, obtain tissue with EUS-FNA, complete staging with laparoscopy if resecting, and use CA 19-9 for monitoring rather than screening.^[1][3]

• Multiphase CT pancreas protocol (triphasic) — the single most important test. Performed as non-contrast, late arterial (pancreatic) phase and portal venous phase, thin slices. PDAC appears as a focal hypoenhancing mass against brightly enhancing normal pancreas. Look for the double-duct sign (dilated CBD and pancreatic duct together), vascular involvement, liver and peritoneal deposits, and regional nodes.
• MRCP — the best non-invasive ductal map; preferred when CT is indeterminate, in iodine allergy, and to characterise cystic lesions.
• EUS with FNA or FNB — the definitive tissue-sampling test, especially for small tumours missed on CT; sensitivity for tissue diagnosis over 90 percent.
• ERCP — now mainly therapeutic (biliary stent placement, brush cytology), not diagnostic.
• 18F-FDG PET-CT — for occult metastases before planned resection and for restaging after neoadjuvant therapy.
• Staging laparoscopy — immediately before laparotomy, to detect occult peritoneal or superficial liver deposits; upstages about 10 to 15 percent of patients planned for resection.^[1]

The classic trap: operating on apparently resectable disease without a staging laparoscopy misses the occult peritoneal and liver deposits that upstage one in seven patients — and turn a curative laparotomy into a futile one.^[1]

Bloods and CA 19-9

CA 19-9 (carbohydrate antigen 19-9) is the traditionally used pancreatic cancer tumour marker — used as a diagnostic adjunct and to assess chemotherapy response, determine prognosis and predict recurrence. Three limitations decide how you use it:^[12]

• It has significant limitations in sensitivity on its own.
• It gives false-positive elevation in obstructive jaundice — cholestasis from any cause, benign or malignant, inflates the value.
• It gives false-negative results in certain blood groups — so a normal CA 19-9 never excludes cancer.^[12]

The classic trap: a normal CA 19-9 in a patient who clearly has PDAC is a false negative — certain blood groups simply do not produce the marker — not reassurance. The marker supports diagnosis, prognosis and monitoring; it does not rule disease out.^[12]

The resectability fork — the 180-degree rule

Resectability is set by vascular contact, measured in degrees of circumference, and it decides everything downstream.^[1]

AJCC staging (8th edition)

T4 (involvement of the coeliac axis, superior mesenteric artery or common hepatic artery) defines locally advanced, unresectable disease; M1 defines metastatic disease. Both are surgically incurable.^[1]

The bedside round — look for stage and fitness

Examination rarely diagnoses PDAC; its job is to stage the disease and assess fitness for the Whipple.^[1]

• General — cachexia, temporal and muscle wasting (sarcopenia), conjunctival pallor from anaemia of chronic disease.
• Skin — jaundice, excoriations from pruritus, migratory thrombophlebitis (Trousseau), and clues to a hereditary syndrome (Peutz-Jeghers mucocutaneous pigmentation, FAMMM atypical naevi).
• Abdomen — palpable gallbladder (Courvoisier sign), hepatomegaly from metastases or biliary congestion, an epigastric mass, ascites.
• Nodes — left supraclavicular (Virchow, Troisier sign) and periumbilical (Sister Mary Joseph) nodes indicate advanced disease.
• Fitness — exercise tolerance, frailty (Clinical Frailty Scale) and nutrition, all of which must be optimised before a major operation.^[1]

The only cure is the Whipple — resectable disease

Surgical resection is the only potentially curative treatment, and only about 15 to 20 percent are resectable at presentation. Everything else is palliation.^[1][3]

Enlarge

The Whipple — what comes out, and the three joins

Pancreaticoduodenectomy (Whipple) is for tumours of the pancreatic head, uncinate process, duodenum, distal bile duct and ampulla — one of the largest abdominal operations, and the most examined procedure in GI oncology.^[1]

Adjuvant chemotherapy after R0 or R1 resection improves survival and is now standard, the regimen chosen by fitness.^[8][9]

• Modified FOLFIRINOX (mFOLFIRINOX) — the preferred adjuvant regimen since PRODIGE 24 (Conroy 2018, NEJM). Regimen: oxaliplatin 85 mg per square metre, leucovorin 400 mg per square metre, irinotecan 180 mg per square metre reduced to 150 mg per square metre after a protocol-specified safety analysis, and fluorouracil 2400 mg per square metre, every two weeks for 24 weeks. Result: median overall survival 54.4 months versus 35.0 months for gemcitabine.^[8]
• Gemcitabine plus capecitabine — the ESPAC-4 (Neoptolemos 2017, Lancet) alternative. Regimen: six four-week cycles of gemcitabine 1000 mg per square metre once weekly for three of every four weeks, plus capecitabine 1660 mg per square metre orally daily for 21 days followed by 7 days' rest. Result: median overall survival 28.0 months versus 25.5 months for gemcitabine alone.^[9]
• Gemcitabine monotherapy (1000 mg per square metre weekly) — the older standard, now reserved for less fit patients.^[6][7]

Distal pancreatectomy, total pancreatectomy

Distal pancreatectomy with splenectomy is for body and tail tumours. The spleen is removed en bloc because the splenic artery and vein run along the upper border of the body and tail and are sacrificed; draining lymph nodes are taken. A spleen-preserving distal pancreatectomy is reserved for benign or low-grade lesions, not PDAC — and losing the spleen means pneumococcal, meningococcal, Haemophilus and influenza vaccination before and after surgery.^[1]

Total pancreatectomy is rarely used (multifocal disease, positive margin after Whipple, or hereditary IPMN) because it produces brittle diabetes and exocrine insufficiency.^[1]

Borderline and locally advanced — neoadjuvant first, never cure by force

A T4 tumour encasing the SMA, coeliac axis or common hepatic artery is not surgically curable. Forcing a resection through encased vessels harms the patient; the answer is systemic therapy first.^[1][3]

Borderline resectable tumours (venous reconstruction needed, or arterial contact under or equal to 180 degrees) are managed with neoadjuvant chemotherapy for two to four months — usually FOLFIRINOX (preferred if fit) or gemcitabine plus nab-paclitaxel — then restaging CT and resection if there is no progression. The rationale is threefold: treat occult micrometastases early, downsize to an R0 margin, and spare non-responders a futile operation.^[1]

Locally advanced disease gets FOLFIRINOX (fit patients) or gemcitabine plus nab-paclitaxel (less fit), with consolidative chemoradiotherapy in selected responders, and biliary obstruction managed by endoscopic or percutaneous stenting. The Murphy total-neoadjuvant regimen — FOLFIRINOX plus losartan (an angiotensin-receptor blocker that softens the desmoplastic stroma to improve drug delivery) followed by chemoradiotherapy — converted many tumours previously thought unresectable.^[11]

Metastatic disease — FOLFIRINOX or nab-paclitaxel, by fitness

Metastatic PDAC is treated for palliation, and the first-line regimen is chosen by performance status. Fit patients (ECOG 0 to 1) get one of two regimens; the discriminator is fitness and toxicity, not efficacy alone.^[6][7]

The one-line discriminator: fit patient equals FOLFIRINOX; less fit equals gemcitabine plus nab-paclitaxel. The single regimen that is wrong is the one chosen without measuring performance status first.^[6][7]

Patients well enough for combination therapy get one of the two regimens above; those not fit for combination chemotherapy receive single-agent gemcitabine 1000 mg per square metre weekly, the comparator used in both landmark trials.^[6][7]

Biomarker-driven therapy — maintenance olaparib for germline BRCA

• Maintenance olaparib for germline BRCA1 or BRCA2-mutated metastatic PDAC — POLO (Golan 2019, NEJM). Olaparib tablets 300 mg orally twice daily as maintenance in patients whose disease had not progressed during first-line platinum-based chemotherapy significantly extended progression-free survival (7.4 versus 3.8 months with placebo).^[10]

Acute presentations — drain, decompress, relieve pain

Palliation is treatment — start it early

For the 80 percent who are unresectable, early specialist palliative care improves quality of life and survival, and runs in parallel with disease-directed therapy — not after it.^[1]

• Biliary decompression — initial medical treatment with antibiotics and early endoscopic biliary drainage for cholangitis; percutaneous transhepatic drainage when endoscopic drainage fails or is unavailable.
• Pain — WHO analgesic ladder up to strong opioids; coeliac plexus block or neurolysis, by the percutaneous approach or under EUS guidance, for difficult pancreatic-cancer pain.^[15][16]

When the Whipple goes wrong — the complications clock

Complications separate by timing, and the timing is the most examinable single fact about each — the clock matters more than the murmur.^[1]

The sentinel bleed is the one you must not dismiss. A small haematemesis days after a Whipple heralds a pseudoaneurysm eroding into the gastroduodenal artery stump; the response is urgent mesenteric angiography and embolisation. Treating it as a minor ooze and observing is how a salvageable complication becomes a fatal one.^[1]

The classic pitfalls

• Mistaking autoimmune pancreatitis (IgG4) for PDAC and resecting — always check IgG4 and obtain tissue in an atypical case; AIP is steroid-responsive and curable.
• Missing ampullary carcinoma by assuming any periampullary mass is PDAC — it has a far better prognosis.
• Operating on a T4 locally advanced tumour that encases the SMA — it is unresectable; check the CT for vessel contact.
• Forgetting Lewis-negative CA 19-9 — a normal value does not exclude PDAC.
• Ignoring a sentinel post-Whipple bleed — it heralds a pseudoaneurysm and death if untreated.
• Failing to germline-test metastatic PDAC — misses the small but important BRCA subgroup eligible for olaparib.^[1]

Prognosis — set by stage at presentation

Five-year survival is about 10 to 12 percent overall, barely improved over decades — though adjuvant mFOLFIRINOX is finally changing this for the resected subgroup.^[1][2]

• After R0 resection plus adjuvant mFOLFIRINOX — median overall survival now exceeds 54 months and five-year survival reaches 30 to 40 percent.
• After R0 plus gemcitabine or no adjuvant — five-year survival 20 to 25 percent.
• Borderline, downstaged and resected — five-year survival 15 to 30 percent.
• Locally advanced (T4) — median overall survival 15 to 20 months with FOLFIRINOX.
• Metastatic — median overall survival 8 to 11 months (FOLFIRINOX), 6 to 7 months (gemcitabine plus nab-paclitaxel), 6 months (gemcitabine alone), under three months with best supportive care alone.
• Untreated — median survival 3 to 6 months.^[1]

Poor prognostic factors: advanced stage at diagnosis, cachexia and low performance status (ECOG over 1), liver or peritoneal metastases, raised baseline CA 19-9, SMAD4 loss or basal-like molecular subtype, positive resection margin (R1 or R2), node-positive disease, vascular invasion, and new-onset diabetes.^[1]

Disposition is to a specialised hepatopancreatobiliary (HPB) multidisciplinary team for resectable and borderline disease, to medical oncology and palliative care for locally advanced and metastatic disease, and to MRCP and EUS surveillance for hereditary high-risk kindreds. After resection, follow-up is clinical with CA 19-9 and CT every three to six months for two to three years.^[1]

The trials that changed practice

Each trial below set a regimen that is now standard at its stage of the disease — name the trial, name the regimen, name the result.^[1]

Preventable harm — the public-health list

Most pancreatic cancer is not preventable, but the modifiable drivers and the high-risk kindreds are where prevention and early detection actually live.^[4]

• Smoking — the single biggest modifiable risk and 20 to 25 percent of cases; cessation lowers risk toward baseline over ten years. This is the one intervention with population-level impact.^[4]
• New-onset diabetes after 50 with weight loss — investigate, do not reassure; it is the earliest clue in a minority who are still resectable.^[1]
• Chronic pancreatitis — manage the cause (alcohol, tobacco, hereditary) and surveil the high-risk forms.^[1]
• Family history and hereditary syndromes — refer Peutz-Jeghers, hereditary pancreatitis, BRCA2, FAMMM, Lynch and familial PDAC kindreds for MRCP and EUS surveillance, because finding a small T1N0 tumour is the only realistic cure in this group.^[13][14]

Regional differences

The diagnostic framework — CT pancreas protocol, EUS-FNA, CA 19-9, resectability criteria, and stage-driven chemo — is globally consistent; resource and access differences drive the deltas.^[1]

The mantra, and the mnemonic

The mantra: Jaundice, weight loss, and a dilated duct — scan, stage, and only the fit get the Whipple.^[3][6]

Ward-round test — three stems