Amatya Bhasker, Khan Fary
Department of Rehabilitation and Australian Rehabilitation Research Centre, Royal Melbourne Hospital, Parkville, Victoria, Australia.
Department of Medicine (Royal Melbourne Hospital), the University of Melbourne, Parkville, Victoria, Australia.
NeuroRehabilitation. 2025 Feb;56(1):78-80. doi: 10.1177/10538135241303581. Epub 2025 Feb 25.
BackgroundPrognostic models have the potential to support people with Multiple Sclerosis (pwMS) and clinicians in treatment decision-making, enable stratified and precise interpretation of interventional trials, and offer insights into disease mechanisms. Despite many researchers being involved in developing these models to predict clinical outcomes in multiple sclerosis (MS), no widely accepted prognostic model is currently used in clinical practice.ObjectiveCommentary on the review by Reeve et al. (2023) to identify and summarise multivariable prognostic models, and their validation studies for quantifying the risk of clinical disease progression, worsening, and activity in pwMS.MethodsThis review included studies evaluating statistically developed multivariable prognostic models aiming to predict clinical disease progression, worsening, and activity, as measured by disability, relapse, conversion to definite MS, conversion to progressive MS, or a composite of these in adult individuals with MS.ResultsThe review included 57 studies, comprising 75 model developments, 15 external validations, and six author-reported validations. Only two models were validated multiple times externally, and none by independent researchers. The outcomes evaluated included disease progression (41%), relapses (8%), conversion to definite MS (18%), and conversion to progressive MS (28%). All models required specialist skills, 59% needed specialized equipment, and 52% lacked sufficient details for application or independent validation. Reporting quality was poor, and most models had a high risk of bias. The findings suggest increases in the number of participants on treatment, diverse diagnostic criteria, the use of biomarkers, and machine learning over time.ConclusionsDespite the development of many prognostic prediction models in pwMS, current evidence is insufficient to recommend any of these models for clinical use due to the high risk of bias, poor reporting, and lack of independent validation. The review's findings necessitate a cautious approach to integrating existing MS prognostic models into rehabilitation practice.
背景
预后模型有潜力在治疗决策方面为多发性硬化症患者（pwMS）和临床医生提供支持，实现对干预性试验进行分层且精确的解读，并深入了解疾病机制。尽管许多研究人员参与开发这些模型以预测多发性硬化症（MS）的临床结局，但目前临床实践中尚未广泛采用被认可的预后模型。
目的
对Reeve等人（2023年）的综述进行评论，以识别和总结多变量预后模型及其用于量化pwMS临床疾病进展、恶化和活动风险的验证研究。
方法
本综述纳入了评估通过统计学方法开发的多变量预后模型的研究，这些模型旨在预测临床疾病进展、恶化和活动，其衡量指标包括残疾、复发、转化为确诊MS、转化为进展性MS或这些指标的综合情况，研究对象为成年MS患者。
结果
该综述纳入了57项研究，包括75个模型开发、15项外部验证和6项作者报告的验证。只有两个模型在外部进行了多次验证，且没有一个经过独立研究人员的验证。评估的结局包括疾病进展（41%）、复发（8%）、转化为确诊MS（18%）和转化为进展性MS（28%）。所有模型都需要专业技能，59%需要专用设备，52%缺乏足够的应用或独立验证细节。报告质量较差，大多数模型存在高偏倚风险。研究结果表明，随着时间的推移，接受治疗的参与者数量增加，并采用了多样化的诊断标准、生物标志物和机器学习方法。
结论
尽管在pwMS中开发了许多预后预测模型，但由于存在高偏倚风险、报告质量差和缺乏独立验证，目前的证据不足以推荐将这些模型中的任何一个用于临床。该综述的结果要求在将现有MS预后模型整合到康复实践中时采取谨慎的方法。
