复发缓解型多发性硬化患者认知轨迹与残疾进展的关联
本研究纳入460例复发缓解型多发性硬化患者,使用计算机化反应时任务串行评估精神运动速度、视觉注意和工作记忆,采用潜在类别混合模型识别纵向反应时轨迹。每个任务最优模型均含3个潜在类别,视觉注意和工作记忆任务识别出的反应时减慢组患者发生6个月确认残疾进展的风险分别高3.7倍和2.6倍,仅5次测试即可分类预测认知轨迹(准确率64%-89%)。
要点
- 计算机化反应时任务可识别复发缓解型多发性硬化患者中反应时恶化的认知轨迹组。
- 视觉注意和工作记忆任务识别的反应时减慢组,6个月确认残疾进展风险分别高3.7倍和2.6倍。
- 仅5次测试即可分类预测认知轨迹,准确率64%-89%,提示临床监测可行性。
核心结论
- 基于计算机化电池的纵向认知数据潜在类别建模可识别反应时恶化且残疾进展风险增加的患者。
- 基线反应时较慢、年龄和残疾程度增加该轨迹的归属概率。
- 临床实践中用计算机化测试监测认知可能有助于发现认知变化轨迹和残疾进展风险患者。
研究设计与样本
- 研究纳入460例复发缓解型多发性硬化患者,串行完成计算机化反应时任务(MSReactor电池)评估精神运动速度、视觉注意和工作记忆。
- 参与者至少完成3次测试,跨度至少180天。
- 采用潜在类别混合模型建模纵向反应时,以多项逻辑回归检验基线关联,用生存分析评估类别间反应时恶化和6个月确认残疾进展概率差异。
主要发现
- MSReactor电池每个任务的最优模型均包含3个潜在类别。
- 所有MSReactor任务均能识别出反应时减慢高概率组。
- 视觉注意任务识别的反应时减慢组发生6个月确认残疾进展的可能性高3.7倍,工作记忆任务识别的组高2.6倍。
- 仅5次测试即可将参与者分类至预测认知轨迹,准确率64%至89%。
关键数值与推荐
- MSReactor电池每个任务最优模型为3个潜在类别。
- 视觉注意任务反应时减慢组6个月确认残疾进展风险高3.7倍。
- 工作记忆任务反应时减慢组6个月确认残疾进展风险高2.6倍。
- 5次测试分类预测认知轨迹的准确率为64%至89%。
临床怎么用
- 临床可考虑使用计算机化反应时测试(如MSReactor电池)对复发缓解型多发性硬化患者进行认知监测。
- 关注基线反应时较慢、年龄较大和残疾程度较高的患者,其更可能归入反应时恶化轨迹。
- 认知轨迹分类或有助于识别残疾进展风险较高的患者,辅助临床决策。
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
BACKGROUND AND OBJECTIVES: Longitudinal cognitive trajectories in multiple sclerosis are heterogeneous and difficult to measure. We aimed to identify discrete longitudinal reaction time trajectories in relapsing-remitting multiple sclerosis using a computerized cognitive battery and to assess the association between trajectories of reaction time and disability progression. METHODS: All participants serially completed computerized reaction time tasks measuring psychomotor speed, visual attention, and working memory. Participants completed at least 3 testing sessions over a minimum of 180 days. Longitudinal reaction times were modeled with latent class mixed models to identify groups of individuals sharing similar latent characteristics. Optimal models were validated for consistency and baseline associations with class membership tested using multinomial logistic regression. Interclass differences in the probability of reaction time worsening and the probability of 6-month confirmed disability progression were assessed with survival analysis. RESULTS: A total of 460 people with relapsing-remitting multiple sclerosis were included in the analysis. For each task of the MSReactor battery, the optimal model comprised 3 latent classes. All MSReactor tasks could identify a group with high probability of reaction time slowing. The visual attention and working memory tasks could identify a group of participants who were 3.7 and 2.6 times more likely to experience a 6-month confirmed disability progression, respectively. Participants could be classified into predicted cognitive trajectories after just 5 tests with 64% to 89% accuracy. DISCUSSION: Latent class modeling of longitudinal cognitive data collected by a computerized battery identified patients with worsening reaction times and increased risk of disability progression. Slower baseline reaction time, age, and disability increased assignment into this trajectory. Monitoring of cognition in clinical practice with computerized tests may enable detection of cognitive change trajectories and people with relapsing-remitting multiple sclerosis at risk of disability progression.
引用
- Merlo D, Stankovich J, Bai C, et al. Association Between Cognitive Trajectories and Disability Progression in Patients With Relapsing-Remitting Multiple Sclerosis. Neurology. 2021. PMID: 34556562. doi:10.1212/WNL.0000000000012850
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。