AstraZeneca taps Pathos for breast cancer degrader
AstraZeneca has put Pathos AI in charge of the early clinical development of AZD4241, its preclinical ERα degrader for ER-positive, HER2-negative breast cancer, under a co-exclusive licensing agreement.
Why it matters: The deal allows AstraZeneca to retain shared rights to a next-generation estrogen receptor therapy while placing early trial work with an AI-focused developer. It also gives AstraZeneca another approach to ER degradation alongside Etcamah (camizestrant), its oral selective estrogen receptor degrader recently approved in the EU for certain patients with advanced ER-positive breast cancer and emerging ESR1 mutations.
How it works: AZD4241 is an oral PROTAC designed to recruit the cell’s protein-disposal machinery to remove both wild-type and mutated forms of ERα.
- AstraZeneca researchers reported at AACR 2026 that the drug produced dose-dependent antitumor activity and, in some cases, tumor regressions in preclinical models, including one resistant to the CDK4/6 inhibitor palbociclib.
Backstory: AstraZeneca and Pathos already have close ties. In April 2025, the companies partnered with Tempus AI to build a multimodal oncology foundation model using de-identified patient data. The agreements included $200 million in data licensing and model-development fees to Tempus.
- Pathos now plans to use its Foundry platform – whose backbone is the oncology model being built with AstraZeneca and Tempus – to guide patient selection, dosing and trial design for AZD4241.
Yes, but: The partners did not disclose financial terms, explain how their co-exclusive rights are divided or provide a date for starting clinical trials.
Also notable: In a parallel deal, Pathos licensed rights outside Greater China to Alphamab Oncology’s JSKN016, a TROP2/HER3 bispecific antibody-drug conjugate already in Phase 3 testing in China for triple-negative breast cancer. Pathos will pay $125 million (€108 million) upfront, with total potential payments reaching $2.093 billion (€1.814 billion), plus royalties.
The bottom line: For AstraZeneca, the agreement turns last year’s data-and-model collaboration into an asset-level test. Pathos will now try to use that AI infrastructure to move an AstraZeneca molecule into humans.
