文献速递银屑病与炎症性皮肤病

银屑病生物制剂个体化治疗:精准给药仪表盘的开发、验证与用户测试

Personalizing Biologic Therapy in Psoriasis: Development, Validation, and User Testing of a Precision-Dosing Dashboard.
PubMed 文献 · J Invest Dermatol · 2025-02-19

研究开发并验证了一种基于贝叶斯估计的精准给药仪表盘,用于指导银屑病生物制剂(以risankizumab为例)的个体化减量。该仪表盘整合了来自I-III期试验的群体药代动力学模型,并经英国真实世界数据外部验证,预测性能良好。英国医护人员评价其用户友好且可接受,平均2分钟可生成给药间隔建议。

要点

  • 部分银屑病患者在血药浓度较高时可用较低剂量维持疾病控制,提示精准减量可行。
  • 基于贝叶斯估计的仪表盘预测个体药代动力学参数,性能优于基础模型。
  • 医护人员认为仪表盘易用可接受,平均2分钟生成给药间隔建议。

核心结论

  • 基于贝叶斯估计的精准给药仪表盘在银屑病生物制剂个体化减量中具有潜力,预测性能良好且用户接受度高。

研究设计与样本

  • 研究开发、验证并用户测试了一种精准给药仪表盘,采用模型引导的精准给药方法,利用贝叶斯估计预测个体药代动力学参数。
  • 以risankizumab为示例生物制剂,药代动力学模型来源于I-III期银屑病试验数据(1899名患者,13123个观测值),并使用英国真实世界数据进行外部验证。

主要发现

  • 贝叶斯模型的后验预测平均绝对误差为0.89 mg/l,平均百分比误差为19.55%,均方根误差为1.24 mg/l,R2为0.86。
  • 贝叶斯模型的预测能力优于基础药代动力学模型(先验预测)。
  • 英国医护人员评价仪表盘用户友好且可接受,平均生成给药间隔所需时间为2分钟。

关键数值与推荐

  • 贝叶斯模型后验预测:平均绝对误差=0.89 mg/l,平均百分比误差=19.55%,均方根误差=1.24 mg/l,R2=0.86。
  • 仪表盘允许输入个体患者数据(血清药物浓度和模型协变量)以生成个体化给药建议。

临床怎么用

  • 该仪表盘有潜力整合其他生物制剂并扩展至其他疾病场景(如无应答和其他炎症性疾病),以优化精准给药对健康和成本结局的现实影响。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

An increasing number of individuals receiving psoriasis biologics achieve clear/nearly clear skin (disease control). Clinical trial data indicate that some maintain disease control with lower doses, especially those with higher serum drug concentrations. This indicates the potential of model-informed precision dosing, an advanced therapeutic drug-monitoring technique, to guide dose minimization. We developed, validated, and user-tested a precision dosing dashboard. We applied a model-informed precision-dosing approach that leveraged Bayesian estimation to predict individual pharmacokinetic parameters for personalized dosing recommendations. A pharmacokinetic model of the exemplar biologic risankizumab derived from phases I-III psoriasis trial data (13,123 observations from 1899 patients) was externally validated using real-world data from the United Kingdom. The Bayesian model (posterior prediction: mean absolute error = 0.89 mg/l, mean percentage error = 19.55%, root mean square error = 1.24 mg/l, R2 = 0.86) had superior predictive power to the basic pharmacokinetic model (prior prediction). The model was incorporated into an interactive dashboard that enabled input of individual patient data (serum drug concentrations and model covariates). Healthcare professionals in the United Kingdom rated the dashboard as user friendly and acceptable. The mean time required to generate a dosing interval was 2 minutes. Our dashboard has the potential to incorporate other biologics and extend across disease contexts (nonresponse and other inflammatory diseases) for optimal real-world impact of precision dosing on health and cost outcomes.

来源:J Invest Dermatol,查看原文。版权归原作者所有。

引用

  1. Thomas CM, Baudry D, Arkir Z, et al. Personalizing Biologic Therapy in Psoriasis: Development, Validation, and User Testing of a Precision-Dosing Dashboard. J Invest Dermatol. 2025. PMID: 39983977. doi:10.1016/j.jid.2025.01.031

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。