AstraZeneca has entered a strategic collaboration with BostonGene to integrate the latter’s multimodal AI foundation model for tumor and immune biology into its early-stage oncology pipeline. The agreement will see BostonGene’s pre-trained foundation models, which combine cell-free RNA (cfRNA) analysis, tumor microenvironment profiling and advanced omnimodal data analytics, applied to AstraZeneca’s investigational compounds. By leveraging predictive insights on patient-level safety and efficacy outcomes, AstraZeneca expects to shorten development timelines by up to 20%, reduce trial attrition and refine biomarker-driven patient selection across multiple tumour indications. Jorge Reis-Filho, MD, PhD, Chief of AI for Science Innovation in AstraZeneca’s Enterprise AI Unit, noted the collaboration aims to ‘predict which patients will benefit from treatment’ and design safer, more effective therapies from the outset.

In separate news, AstraZeneca reported positive Phase III TULIP-SC trial results for Saphnelo (anifrolumab) subcutaneous injection in systemic lupus erythematosus. The trial met its primary endpoint, demonstrating a 56.2% reduction in disease activity as measured by the SLE Responder Index compared with 37.1% for placebo (p<0.001). Safety data showed comparable rates of serious adverse events (6.5% vs 7.2%) and no new safety signals over 52 weeks. Saphnelo is already approved in over 70 countries and has been administered to more than 40,000 patients globally. AstraZeneca plans to submit these data to additional regulatory authorities in H2 2024, targeting label expansion and increased market access for the self-injectable formulation.