Pancreatic cancer: the failed and stopped programmes and why

Stroma and microenvironment. HALO-301 (492 patients, hyaluronan-high metastatic disease): pegvorhyaluronidase alfa with nab-paclitaxel and gemcitabine gave median overall survival 11.2 against 11.5 months (hazard ratio 1.00), more responses (47 against 36 percent) and no progression-free benefit, and Halozyme ended the programme. LAPIS (284 patients, locally advanced): pamrevlumab with chemotherapy missed overall survival. RESOLVE (424 randomised): ibrutinib with nab-paclitaxel and gemcitabine gave 9.7 against 10.8 months and shorter progression-free survival (5.3 against 6.0 months), with patients on ibrutinib receiving less chemotherapy. Immunity. SEQUOIA (567 patients, second line): pegilodecakin with FOLFOX gave 5.8 against 6.3 months (hazard ratio 1.045). ECLIPSE (303 patients, previously treated): GVAX pancreas with cyclophosphamide and CRS-207 gave 3.7 months against 4.6 with chemotherapy. IMPRESS: adjuvant algenpantucel-L added nothing. AMPLIFY-7P (2026): the KRAS peptide vaccine ELI-002 7P missed its disease-free survival endpoint. Checkpoint inhibitors have failed outside mismatch repair deficient disease. Metabolism and stemness. CanStem111P (1,134 patients): napabucasin gave 11.4 against 11.7 months and was stopped for futility. AVENGER 500: devimistat with modified FOLFIRINOX added nothing. TRYbeCA-1: eryaspase second line missed overall survival. Targeted therapy. Erlotinib with gemcitabine is licensed but added nothing in the adjuvant CONKO-005 (disease-free survival 11.4 months in both arms), in RTOG 0848 step 1 (median survival 29.9 against 28.1 months) or in LAP07 (13.6 against 11.9 months); MRTX1133, the first KRAS G12D inhibitor in the clinic, was stopped after 63 patients (NCT05737706 terminated, March 2025). Radiotherapy. LAP07 (449 patients): capecitabine chemoradiotherapy after four months of chemotherapy gave 15.2 against 16.5 months with fewer local progressions (32 against 46 percent); CONKO-007 (336 randomised): chemoradiotherapy after induction did not raise the overall R0 resection rate (25 against 18 percent) or survival (hazard ratio 0.94) but raised R0 rates among those operated (69 against 50 percent); ESPAC-1 (2004) found adjuvant chemoradiotherapy harmful (five-year survival 10 against 20 percent); Alliance A021501 closed its radiotherapy arm at interim analysis. Second line. CONKO-003 showed oxaliplatin with fluorouracil and folinic acid (OFF) lengthened survival (5.9 against 3.3 months) but PANCREOX (108 patients) found modified FOLFOX6 inferior to fluorouracil and leucovorin (6.1 against 9.9 months) with far more grade 3 to 4 toxicity (63 against 11 percent). Neoadjuvant chemotherapy for clearly resectable disease. NORPACT-1 (140 patients): 18-month survival 60 percent with neoadjuvant FOLFIRINOX against 73 percent with upfront surgery; PREOPANC-2 (375 patients): neoadjuvant FOLFIRINOX no better than gemcitabine chemoradiotherapy (21.9 against 21.3 months); APACT (866 patients): adjuvant nab-paclitaxel with gemcitabine missed its independently assessed disease-free survival endpoint (19.4 against 18.8 months) despite a later overall survival difference (41.8 against 37.7 months). The common reading is that unselected additions to chemotherapy fail in this disease, and that the successes since 2011 have been better chemotherapy (FOLFIRINOX, NALIRIFOX), biomarker selection (olaparib, zenocutuzumab, daraxonrasib for RAS) and a device (tumour treating fields).