Pancreatic cancer preprint uses machine learning to identify an adaptive target after KRAS inhibition
Published in Artificial Intelligence, Drugs development & discovery, Journals.
Published October 5, 2026. Preprint posted: October 2, 2026.
A bioRxiv preprint explores how pancreatic ductal adenocarcinoma adapts to KRAS inhibition. Researchers mapped changes in proteins, phosphorylation and cell-surface proteins over time, then used machine learning with additional public datasets to prioritize targets for combination treatment.
The analysis identified increased placental alkaline phosphatase, ALPP, after KRAS inhibition, linked to compensatory JAK2/STAT3 signaling. The team developed an antibody-drug conjugate targeting ALPP/ALPPL2 and carrying the payload MMAE.
In pancreatic cancer models, the conjugate showed activity on its own. Combining it with KRAS inhibitors increased internalization and antitumour activity, suggesting a strategy to investigate against adaptive resistance.
This is a preprint reporting experimental-model findings. It does not demonstrate clinical benefit, prevention of resistance in patients, or an approved combination therapy. Peer review and further translational and clinical testing are needed.
Source: DeLiberty and colleagues, bioRxiv; DOI 10.64898/2026.10.01.756025.