AnaptysBio (ANAB) FY Conference June 11, 2025 08:40 AM ET Speaker0 All right. Good morning. And it's my pleasure to introduce Dan Vega, President and CEO of Adaptive Bio as our next speaker. Dan, welcome. It's the time for me to host you at the Goldman Sachs Conference. Conference. So thank you for being here. Before we kick off with the Q and A, I'm going to turn it to you for a quick opening remark. Speaker1 Perfect. Well, thanks for having me here today. It's my pleasure to be here. We have a lot going o ...

Summary of AnaptysBio (ANAB) Conference Call on June 03, 2025 Company Overview - **Company**: AnaptysBio - **Product**: Rozanolimab (ozanilumab) - **Indication**: Rheumatoid Arthritis (RA) Key Points and Arguments Clinical Trial Results 1. **Efficacy of Rozanolimab**: The phase 2b trial demonstrated significant efficacy with statistical significance on primary endpoints (DAS28 CRP and ACR20) at week 12, with all doses showing positive results compared to placebo [4][8] 2. **Durability of Response**: Patients exhibited durable responses off drug for at least two months after six months of treatment, indicating potential for extended dosing intervals [8][63] 3. **Market Potential**: The RA market generates over $10 billion in annual revenue in the US alone, highlighting the commercial opportunity for rozanolimab [8] 4. **Comparison with Competitors**: Rozanolimab showed comparable or superior results to existing therapies like RINVOQ and Orencia in terms of ACR20, ACR50, and ACR70 response rates [9][14][57] Safety Profile 5. **Safety Data**: Rozanolimab exhibited a notably unremarkable safety profile with no treatment-related serious adverse events (SAEs) reported, and a low incidence of injection site reactions [78][80] 6. **Tolerability**: Less than 2% of patients discontinued due to adverse events, indicating high tolerability compared to standard care [80][84] Mechanism of Action 7. **Targeting T Cells**: Rozanolimab demonstrated a rapid and sustained reduction in PD-1 positive T cells, which are implicated in RA pathology, supporting its mechanism of action [39][41] 8. **Gene Expression Changes**: Significant downregulation of genes associated with T cell and B cell activation was observed, indicating a broad impact on immune pathways relevant to RA and ulcerative colitis [42][43] Patient Population Insights 9. **Patient Disposition**: 95% of patients completed the all-active treatment period, demonstrating high acceptance of the treatment [23] 10. **Real-World Implications**: The trial design may have capped the maximum response rates achievable, as many patients who showed improvement were ineligible to continue treatment [21][56] Expert Commentary 11. **Clinical Relevance**: Experts highlighted the importance of the trial design and the implications of excluding patients who showed improvement but did not meet the strict criteria for continuation [66][69] 12. **Long-Term Efficacy**: The consistency of clinical responses and the durability of effects post-treatment were emphasized as significant advantages of rozanolimab [72][73] Additional Important Content - **Market Context**: The RA market has not seen a new mechanism approved since 2012, making rozanolimab's introduction particularly timely [9] - **Comparative Efficacy**: Rozanolimab's performance in the naive patient population was noted to be particularly strong, suggesting a high ceiling for response rates [35] - **Future Directions**: Ongoing studies and follow-ups will provide further insights into the long-term efficacy and safety of rozanolimab [24][62] This summary encapsulates the critical insights from the conference call regarding AnaptysBio's rozanolimab, its clinical trial results, safety profile, and market potential in the context of rheumatoid arthritis treatment.

责编 | 梦依丹 2025 年 2 月发布的 NoLiMA 是一种大语言模型（LLM）长文本理解能力评估方法。不同于传统"大海捞针"（Needle-in-a-Haystack, NIAH）测试依赖 关键 词匹配的做法，它最大的特点是 通过精心设计问题和关键信息，迫使模型进行 深层语义理解和推理，才能从长文本中找到答案。Jina AI 技术团队 受到启发，并进针对向量模型 jina-embeddings-v3 进行了类似实验。 NoLiMa: https://arxiv.org/abs/2502.05167 NoLiMA 的研究结果揭示了一个重要问题：那些号称能处理几十万甚至上百万词元（tokens）的 LLM，在真正需要理解长文本的任务里，性能大打折 扣。比如，在 32K 词元的长度下，有 10 个受测模型，表现还不如处理短文本（小于 1K 词元）时的一半好；就连表现最好的 GPT-4o，性能也从接近完 美的 99.3% 掉到了 69.7%。 【编者按】 2025 年 2 月发布的 NoLiMA 是一种大语言模型（LLM）长文本理解能力评估方法。不同于传统"大海捞针"（Needle-in-a-Haystack ...