Loading
Pancreatic cancer has had one operation since 1935 and a handful of chemotherapy drugs since 1997. This roadmap follows the evidence through FOLFIRINOX, chemotherapy before and after surgery, the first inherited-gene drug, the first drugs against the KRAS protein that drives nearly every tumour, personalised vaccines, and the surveillance and blood tests that might catch it earlier, to 2031.
Pancreatic ductal adenocarcinoma is the cancer where progress has been slowest. Whipple's 1935 operation is still the only cure and about one patient in five presents in time for it; Traverso and Longmire's 1978 modification and the centralisation of surgery into high-volume centres made it safer without changing who can have it. The oncogene was found in 1988, when Almoguera and Perucho showed KRAS codon 12 mutations in 21 of 22 tumours, and the blood marker CA 19-9 was already known to be unreadable in the Lewis-negative tenth of patients (Tempero 1987). Gemcitabine (Burris 1997) then set a standard that lasted 14 years, while the European adjuvant trials (ESPAC-1 2004, CONKO-001 2007, ESPAC-3 2010) established that chemotherapy after surgery helped and chemoradiotherapy did not.
The second decade of the century brought combination chemotherapy: FOLFIRINOX (Conroy 2011) and gemcitabine with nab-paclitaxel (Von Hoff 2013) for metastatic disease, liposomal irinotecan second line (NAPOLI-1 2016), gemcitabine with capecitabine (ESPAC-4 2017) and then modified FOLFIRINOX (PRODIGE 24 2018, five-year update 2022: median survival 53.5 months) after surgery, and NALIRIFOX first line (NAPOLI 3 2023). The genomes (Waddell 2015, Bailey 2016) and the classical versus basal-like split (Moffitt 2015) gave the disease a molecular vocabulary; Hu 2018 found a germline fault in 5.5 percent of all patients regardless of family history, and POLO (2019) turned the BRCA subset into the first biomarker-directed approval, without a survival gain. Neoadjuvant treatment became standard for borderline resectable disease after PREOPANC (2020, 2022) but PREOPANC-2 (2025) and NORPACT-1 left the resectable question open, and the first attempt at the stroma failed (HALO-301 2020) after mouse work had warned that removing fibroblasts made tumours worse (Özdemir 2014).
The third decade is the KRAS decade. Ostrem and Shokat's 2013 pocket led to sotorasib in the 1 to 2 percent with G12C (CodeBreaK 100 2023); the RAS(ON) tri-complex chemistry (Holderfield 2024) produced daraxonrasib, which in RASolute 302 (2026) nearly doubled survival in previously treated metastatic disease and became the first RAS inhibitor approved for pancreatic cancer. Personalised mRNA vaccines made T cells that lasted years in responders (Rojas 2023, Sethna 2025), a wearable electric-field device added two months in locally advanced disease (PANOVA-3 2025), and surveillance of germline carriers shifted most detected cancers to stage I (Canto 2018, Dbouk 2022) while the new-onset diabetes score ENDPAC (Sharma 2018) offered a way to enrich the general population for a test. What comes next is on the registry: first-line and adjuvant RAS inhibitor trials, G12D-selective combinations, perioperative chemotherapy trials, the randomised vaccine trial and the 20,000-person PRECEDE surveillance cohort, with dates between 2026 and 2031.
Whipple, Parsons and Mullins described removal of the pancreatic head and duodenum for ampullary cancer in 1935; Traverso and Longmire preserved the pylorus in two patients in 1978 and in their 1980 follow-up of 18 found every patient had pancreatic exocrine insufficiency and needed intensive enzyme replacement. Ninety years on, surgery is the only treatment that cures pancreatic cancer, about one patient in five presents with disease that can be removed, and operative mortality fell through centralisation into high-volume centres rather than through any change in what is removed. The enzyme problem Traverso recorded is still under-treated (Roberts 2019).
Every treatment on this roadmap is either a way to reach this operation, a way to make it work better, or a substitute for patients who cannot have it.
The pylorus-preserving Whipple became the standard variant, and the 1980 follow-up is an early record of the exocrine insufficiency that still goes untreated in most patients today.
CA 19-9, the serum marker still used to follow the disease, was shown by Tempero and colleagues in 1987 to be unmakeable by patients who lack the Lewis blood group antigens, so a normal value never rules the cancer out; Fahrmann's 2021 pre-diagnostic study later showed it rises about two years before diagnosis and catches half of early cases at 99 percent specificity. In 1988 Almoguera and Perucho found KRAS codon 12 mutations in 21 of 22 exocrine pancreatic carcinomas, present in primary and metastasis alike: the single most uniform driver in any common cancer, and for 33 years an undruggable one.
Burris (1997) made gemcitabine the standard for advanced disease on a clinical benefit endpoint and a modest survival gain over fluorouracil, a standard that held for 14 years. The European adjuvant trials then settled what to do after surgery: ESPAC-1 (2004, 289 patients) found five-year survival of 21 percent with chemotherapy against 8 percent without and 10 percent with chemoradiotherapy against 20 percent without; CONKO-001 (2007, 368 patients) roughly doubled disease-free survival with six months of gemcitabine (13.4 versus 6.9 months), confirmed for overall survival in 2013; ESPAC-3 (2010, 1,088 patients) showed fluorouracil and gemcitabine equivalent. Hidalgo's 2010 review marks where the field stood before combination chemotherapy.
PRODIGE 4/ACCORD 11 (Conroy 2011, 342 fit patients) gave FOLFIRINOX a median survival of 11.1 months against 6.8 with gemcitabine (hazard ratio 0.57) at the price of neutropenia, diarrhoea and neuropathy; MPACT (Von Hoff 2013, 861 patients) gave gemcitabine with nab-paclitaxel 8.5 against 6.7 months and accepted less fit patients, so fitness rather than stage came to decide the regimen. NAPOLI-1 (2016, 417 patients) added a second line, liposomal irinotecan with fluorouracil, 6.1 against 4.2 months after gemcitabine. Rahib's 2014 projection that pancreatic cancer would become the second cause of cancer death in the United States by 2030 (updated in 2021 to 2040, about 46,000 deaths a year) became the funding argument of the decade.
Waddell's 100 whole genomes (2015) sorted tumours by structural variation and noticed that four of five patients with BRCA-type defects responded to platinum; Moffitt (2015) separated tumour from stroma computationally and found the classical and basal-like tumour subtypes and two prognostic stromal subtypes; Bailey (2016) grouped 32 mutated genes in 456 tumours into ten pathways and four expression subtypes, the squamous type with the worst prognosis. Hu's 3,030-patient Mayo series (2018) found a pathogenic variant in one of six genes in 5.5 percent of all patients and 5.2 percent of those without a family history, which moved the guidelines to germline testing for everyone.
ESPAC-4 (2017) added capecitabine to adjuvant gemcitabine; PRODIGE 24 (2018) replaced both with modified FOLFIRINOX in fit patients, and its five-year report (2022, 493 patients) gave a median survival of 53.5 against 35.5 months (hazard ratio 0.68) and five-year survival of 43.2 against 31.4 percent, adopted by ASCO in 2019. For treatment before surgery, PREOPANC (2020, 246 patients) missed its primary endpoint (16.0 versus 14.3 months) but raised clear-margin resection from 40 to 71 percent and showed a five-year benefit in 2022; ESPAC-5 and Alliance A021501 supported neoadjuvant treatment for borderline resectable disease; NORPACT-1 (2024) and PREOPANC-2 (2025, 375 patients: 21.9 versus 21.3 months for FOLFIRINOX against gemcitabine chemoradiotherapy) left resectable disease unresolved. LAP07 (2016) had shown chemoradiotherapy added nothing to survival in locally advanced disease.
POLO (Golan 2019, 154 patients) gave maintenance olaparib to germline BRCA carriers whose metastatic disease had not progressed on 16 weeks of platinum: progression-free survival hazard ratio 0.53, the first biomarker-directed approval in the disease, and in the 2022 final analysis no overall survival gain (hazard ratio 0.83). ASCO's 2020 update made germline and tumour testing for BRCA, mismatch repair deficiency and TRK fusions routine, each pointing at a drug for a few percent of patients (olaparib, pembrolizumab, larotrectinib, entrectinib), and zenocutuzumab later added NRG1 fusions. The same year HALO-301 (494 patients) showed that dissolving the tumour's hyaluronan raised response rate (47 versus 36 percent) without changing survival (11.2 versus 11.5 months), as Özdemir's 2014 mouse work had warned that removing fibroblasts made tumours worse.
Chari's Minnesota cohort (2005) found pancreatic cancer in 0.85 percent of 2,122 people diagnosed with diabetes after 50 within three years, eight times the expected rate; Sharma's ENDPAC score (2018) used weight change, glucose change and age at onset to concentrate that risk into a group with 3.6 percent prevalence, and Fahrmann (2021) showed CA 19-9 rising from two years before diagnosis. For people with inherited risk, the CAPS programme reported in 2018 that 9 of 10 cancers found under surveillance were resectable, published consensus rules in 2020 (start at 50 or 55, endoscopic ultrasound and MRI annually, research settings only) and in 2022 (CAPS5, 1,461 people) found 7 of 9 cancers at stage I, with a median survival of 9.8 years for screen-detected against 1.5 years for cancers found outside surveillance. The 20,000-person PRECEDE cohort is the scale-up.
Ostrem and Shokat (2013) found the switch-II pocket on KRAS G12C; CodeBreaK 100 (Strickler 2023) gave sotorasib a 21 percent response and 6.9-month survival in the 1 to 2 percent of pancreatic cancers with that mutation, and KRYSTAL-1 did the same for adagrasib. The common alleles needed different chemistry: Holderfield (2024) described the RAS(ON) tri-complex inhibitors that clamp active mutant and wild-type RAS, and daraxonrasib, the clinical compound, nearly doubled survival in RASolute 302 (2026, 500 patients after one line of chemotherapy: 13.2 versus 6.7 months, hazard ratio 0.40), becoming the first RAS inhibitor approved for pancreatic cancer in August 2026. The G12D-selective zoldonrasib and the combination with daraxonrasib (RASolute 309) follow; resistance through secondary RAS mutations and receptor bypass is already described.
Rojas (2023) showed an individualised mRNA neoantigen vaccine, autogene cevumeran, raised T cells in half of 16 resected patients, and Sethna (2025) that at 3.2 years those responders had mostly not relapsed (median recurrence-free survival not reached versus 13.4 months) with vaccine-induced clones estimated to live 7.7 years on average; the randomised IMCODE003 is enrolling 260 patients. The off-the-shelf KRAS vaccine ELI-002 7P missed in AMPLIFY-7P (2026). NAPOLI 3 (2023) made NALIRIFOX a first-line option over gemcitabine with nab-paclitaxel, and PANOVA-3 (2025, 571 patients) gave tumour treating fields with that chemotherapy a survival of 16.2 against 14.2 months in locally advanced disease (hazard ratio 0.82), the basis of the 2026 Optune Pax approval, without improving progression-free survival.
The RAS inhibitor moves earlier: RASolute 303 (daraxonrasib alone or with gemcitabine and nab-paclitaxel first line, 900 estimated participants, primary completion June 2028), RASolute 304 (adjuvant daraxonrasib after resection, 500, May 2029) and RASolute 309 (zoldonrasib with daraxonrasib against chemotherapy first line in G12D disease, 400, March 2029), with Incyte's G12D inhibitor in DAWN-303 (588, September 2028). The perioperative question is being answered by PREOPANC-3 (perioperative against adjuvant modified FOLFIRINOX, 378 estimated, January 2027) and Alliance A021806 (358, December 2028). IMCODE003 tests the vaccine (260, January 2031) and PRECEDE follows 20,000 high-risk people to December 2030. RASolute 302 itself lists study completion for December 2027 and AMPLIFY-7P for November 2026.
Four things no trial on this page has fixed. Four in five patients present with disease that cannot be removed, and neither the new-onset diabetes score nor carrier surveillance has yet been shown in a prospective trial to change that at population scale. Half of patients are not fit for FOLFIRINOX-class chemotherapy, and the pivotal trials enrolled the fit half; whether RAS inhibitors change that is being measured. Cachexia and exocrine insufficiency stop treatment being delivered, enzyme replacement reaches about one patient in five in UK primary care data (Roberts 2019) and cachexia has its first mechanism-based drug (ponsegromab 2024) but no phase 3. And in England and Wales the National Pancreatic Cancer Audit reports each year how many patients receive any active treatment and how quickly, which the UK and NHS page holds; each has an idea on this page.
Enzyme replacement is cheap, guideline-recommended (NICE NG85) and under-prescribed; the National Pancreatic Cancer Audit now reports the prescribing rate as a performance indicator, which the UK and NHS page tracks.
The first drug to reverse cancer cachexia mechanistically rather than by appetite stimulation, in a disease where weight loss stops chemotherapy being delivered; the phase 3 programme and the question of survival remain.
One term page, one bottleneck page and eleven idea pages on OnCo cite this paper by its DOI; this record gives the citation a page of its own so a reader can follow it without leaving OnCo. Read the abstract above alongside the citing pages listed under Related; the record was created automatically from the Europe PMC entry and its figures have not been checked by hand.
Every era's records, trial outcomes and papers, and every watch item, as JSON.
Readouts, decisions and registry completion dates ahead. Each date is quoted from its source, not inferred; a missing date means no source states one.
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
Loading this step…
One of the most cited trial reports Europe PMC returns for Daraxonrasib in Pancreatic ductal adenocarcinoma, so it is a natural first reading for anyone weighing the idea it is linked from. The record was linked automatically by title and abstract; read the abstract above and the paper itself before relying on any figure.
The strongest human evidence that a cancer vaccine can make durable T cells in a tumour with few mutations; the randomised phase 2 IMCODE003 (260 patients, primary completion listed for January 2031) is the test of whether that translates into fewer relapses.
The evidence behind the 2026 approval of Optune Pax for locally advanced disease, the first new approval in that setting in decades, and an unusual case of a survival gain without a progression-free survival gain.
Together with NORPACT-1 this left the neoadjuvant question for resectable disease open: FOLFIRINOX before surgery is not proven superior to alternatives, and the comparison against upfront surgery with adjuvant modified FOLFIRINOX is what PREOPANC-3 and Alliance A021806 are running.
The mechanism that let one drug address G12D, G12V and G12R together, which is why the RASolute 302 trial could enrol unselected pancreatic cancer and nearly double survival.
The first drug to reverse cancer cachexia mechanistically rather than by appetite stimulation, in a disease where weight loss stops chemotherapy being delivered; the phase 3 programme and the question of survival remain.
For fit patients with newly diagnosed metastatic pancreatic cancer, a FOLFIRINOX-type regimen is now proven to be better than gemcitabine plus nab-paclitaxel, settling a long-standing debate. The absolute gain is about two months of median survival, and the regimen is more toxic for the gut. Whether liposomal irinotecan adds anything over conventional irinotecan (standard FOLFIRINOX) has never been tested head-to-head.
Pancreatic cancer was thought to be immunologically inert because of its low mutation burden; this study showed that with the right vaccine platform its few neoantigens can still be targeted. It is the strongest human evidence so far that personalised cancer vaccines can generate durable, tumour-specific immunity, and it justifies the randomised trials now running in pancreatic cancer and melanoma. Benefit is not yet proven, because responders may simply have had more immunogenic tumours.
This is the European standard the UK and NHS page for pancreatic cancer is compared against; NICE guideline NG85 (2018) covers the same ground with an older evidence base and a narrower set of funded drugs.
The best survival ever recorded in a pancreatic cancer trial, and the benchmark every perioperative trial (PREOPANC-3, Alliance A021806) and every adjuvant RAS inhibitor or vaccine trial (RASolute 304, IMCODE003) now has to beat or add to.
The stage shift the pancreatic cancer page quotes (about three in four surveillance-detected cancers at stage I) and the strongest argument for offering surveillance to every germline carrier found by universal testing, which the NHS does not yet do outside research.
NCCN is the source of the category grades on the pancreatic cancer page and the guideline that first placed germline testing for every patient and universal tumour profiling in routine care.
Fixes the window in which a blood test could plausibly work and shows why CA 19-9 alone is not enough: half of early cases are missed even at diagnosis, and Lewis-negative patients are missed entirely.
Pancreatic cancer's share of cancer deaths rises because it is standing still while others improve; the 2014 projection to 2030 already made it the second cause of cancer death in the UK's peer countries, and this is the funding argument charities on both sides of the Atlantic use.
The trial that made neoadjuvant treatment standard for borderline resectable disease and opened the still unresolved question for resectable disease, which PREOPANC-2, NORPACT-1, PREOPANC-3 and Alliance A021806 inherited.
The definitive negative result for first-generation stromal targeting: a biomarker-selected population, a drug that did what it was designed to do to the matrix, and no survival benefit; the stroma ideas on this page start from here.
The rulebook behind the CAPS cohorts and the UK EUROPAC programme; it is also the reason surveillance for carriers is not yet a routine NHS service, because the consortium itself asked for it to stay within research until benefit was shown.
The document that made biomarker testing part of routine pancreatic cancer care in the United States; the 2018 version it built on fixed the second-line chemotherapy sequence still used in most guidelines.
The formal adoption of PRODIGE 24 into practice; every later debate about giving chemotherapy before rather than after surgery starts from this recommendation.
Germline testing for every pancreatic cancer patient, platinum first line for BRCA carriers, and olaparib maintenance for those who respond are all downstream of POLO.
Enzyme replacement is cheap, guideline-recommended (NICE NG85) and under-prescribed; the National Pancreatic Cancer Audit now reports the prescribing rate as a performance indicator, which the UK and NHS page tracks.
Modified FOLFIRINOX is the adjuvant standard for patients who recover well from a pancreatic resection and can tolerate combination chemotherapy; gemcitabine-based regimens remain for those who cannot.
Family history misses most carriers, so the NCCN and ASCO guidelines moved to testing every patient; each carrier found is a family that can be offered surveillance and a patient who may be eligible for platinum and PARP inhibition.
The first long-term evidence that surveillance in high-risk people finds operable cancers, and the source of the imaging features that trigger surgery in the CAPS recommendations.
A scoring tool that needs no new test and can run in primary care records; it turns the 1 percent of Chari's cohort into a 3.6 percent group in whom imaging or a blood test becomes defensible.
With Moffitt's classical and basal-like split (the squamous and basal-like groups overlap) this fixed the molecular vocabulary of the disease and gave Precision-Panc and the UK's Glasgow group their trial framework.
The first approved second-line regimen and the basis of NALIRIFOX, which NAPOLI 3 later moved to first line; it fixed the sequence (gemcitabine-based first, then liposomal irinotecan with fluorouracil) written into the 2018 ASCO guideline.
The genomic case for platinum in BRCA-type pancreatic cancer, four years before POLO built a maintenance strategy on top of it.
The classical versus basal-like split, later tied to GATA6 expression and to chemotherapy response, is the subtype scheme most likely to reach the clinic; the stromal subtypes are why the desmoplastic stroma is treated as a partner in the disease rather than inert scar.
Together with the negative HALO-301 trial of hyaluronidase this ended the first, blunt version of stromal targeting; second-generation ideas aim to reprogramme rather than remove the stroma.
With FOLFIRINOX this trial defined the two chemotherapy standards for metastatic pancreatic cancer that still apply, and the gemcitabine and nab-paclitaxel backbone is the comparator in most current first-line pancreatic trials.
Sotorasib and adagrasib (approved in lung cancer 2021 to 2022, active in the 1 to 2 percent of pancreatic cancers with G12C) descend from this chemistry; the non-covalent G12D inhibitors and the pan-RAS(ON) drugs that matter for pancreatic cancer came from the confidence it created.
FOLFIRINOX and, soon after, gemcitabine plus nab-paclitaxel ended the era of single-agent gemcitabine for metastatic pancreatic cancer. The regimen's modified form later improved survival after surgery in PRODIGE 24, and its toxicity is why fitness, not just stage, decides which treatment a patient is offered.
A fixed point for readers who want to see how much, and how little, has changed since 2010.
The trial that made adjuvant gemcitabine standard in Europe and the control arm PRODIGE 24 and ESPAC-4 later beat.
The observation behind every new-onset diabetes strategy: a 1 percent prevalence is a hundred times the general population's and high enough for a blood test or scan to have a useful positive predictive value.
The origin of the European position that adjuvant treatment means chemotherapy alone; CONKO-001, ESPAC-3, ESPAC-4 and PRODIGE 24 all built on it, and the role of radiotherapy remains contested (LAP07, PREOPANC).
This trial introduced a patient-centred composite endpoint and a drug that remained the backbone of pancreatic cancer treatment for a generation. Its small survival gain also shows how low the bar was, which is the context for the FOLFIRINOX and MPACT trials that followed.
The reason pancreatic cancer is the proving ground for RAS drugs: nearly every tumour depends on the same mutant protein, so a drug that works against it works for nearly every patient.
The reason a normal CA 19-9 never rules pancreatic cancer out, why Lewis-negative patients need a different marker (CA 125, CEA or CA 19-9-independent panels), and a constraint on every blood-based detection idea on this page.
The pylorus-preserving Whipple became the standard variant, and the 1980 follow-up is an early record of the exocrine insufficiency that still goes untreated in most patients today.
Every treatment on this roadmap is either a way to reach this operation, a way to make it work better, or a substitute for patients who cannot have it.
Shares RAS(ON) inhibitors to convert unresectable pancreatic cancer to resectable, PREOPANC-1, Resectable, borderline resectable and unresectable, PRODIGE 24 / CCTG PA6 and the tags pancreatic, gi.
Shares Blood-based pancreatic cancer detection in new-onset diabetes, CA 19-9, Galleri, High-risk pancreatic surveillance (CAPS / PRECEDE) and the tags pancreatic, gi.
Shares PRECEDE, High-risk pancreatic surveillance (CAPS / PRECEDE), Germline BRCA mutation (gBRCA), Germline vs somatic mutations and the tags pancreatic, gi.
Shares Desmoplasia (tumour stroma), Gemcitabine + nab-paclitaxel, FOLFIRINOX / mFOLFIRINOX, Resectable pancreatic ductal adenocarcinoma and the tags pancreatic, gi.
Shares Resectable, borderline resectable and unresectable, Whipple procedure (pancreaticoduodenectomy), Borderline resectable pancreatic ductal adenocarcinoma, Robotic & minimally invasive surgery and the tags pancreatic, gi.
Shares Resectable, borderline resectable and unresectable, CA 19-9, Borderline resectable pancreatic ductal adenocarcinoma, Resectable pancreatic ductal adenocarcinoma and the tags pancreatic, gi.
Shares Resection margins (R0 / R1 / R2), Whipple procedure (pancreaticoduodenectomy), Borderline resectable pancreatic ductal adenocarcinoma, Resectable pancreatic ductal adenocarcinoma and the tags pancreatic, gi.
Shares Resection margins (R0 / R1 / R2), Whipple procedure (pancreaticoduodenectomy), Robotic & minimally invasive surgery, Resectable pancreatic ductal adenocarcinoma and the tags pancreatic, gi.