AstraZeneca has entered a strategic collaboration with BostonGene, leveraging the latter’s multimodal AI foundation model for tumor and immune biology to generate patient-level safety and efficacy predictions in early-stage oncology trials. Under the agreement, AstraZeneca will integrate BostonGene’s pre-trained foundation models—which combine genomic, transcriptomic, cell-free RNA and tumor microenvironment profiling—into its R&D pipelines. The AI-driven insights are expected to shorten development timelines by up to 20%, reduce trial risk through more precise biomarker-informed cohort selection, and support regulatory submissions for label expansions. BostonGene’s platform is already deployed across five global biopharma programs, and this collaboration marks its first application within AstraZeneca’s oncology portfolio, spanning both solid tumors and hematologic malignancies.

In the late-stage TULIP-SC trial, AstraZeneca’s Saphnelo (anifrolumab) self-injectable formulation demonstrated a 56.2% reduction in disease activity at Week 52 compared to 37.1% for placebo, achieving statistical significance on the primary Composite Lupus Responder Index endpoint. The subcutaneous delivery showed comparable pharmacokinetics to the intravenous infusion, with injection-site reactions in 8% of patients and no new safety signals reported. Saphnelo is currently approved in more than 70 countries and has treated over 40,000 patients to date. These data support an anticipated U.S. supplemental filing later this year, with peak annual sales for the subcutaneous formulation projected to exceed $1.2 billion by 2030, reinforcing AstraZeneca’s position in systemic lupus erythematosus treatment.