噪声还是清除？将阿尔茨海默病中血浆淀粉样蛋白-β变异性重新定义为可测量的清除信号

文献信息

中文摘要

血浆生物标志物，尤其是淀粉样蛋白-β 42/40比值，越来越多地用于阿尔茨海默病的分期以及临床试验的分层和监测，其隐含假设是单次测量反映稳定、稳态的病理水平。最近一项微纵向研究对阿尔茨海默病患者每周采样血浆，持续五周，挑战了这一假设，报告大多数参与者在如此短的窗口内淀粉样蛋白-β 42/40比值变化超过一半。条件反射式的反应是将这种变动视为测量噪声并加以平均消除。在此我认为其中相当一部分不太可能是噪声。在健康成人中，同一比值是最稳定的血浆分析物之一，因此集中于疾病中的不稳定性难以仅归因于检测方法。淀粉样蛋白-β的脑到血清除本身在数天内是动态的，随睡眠而升高，随睡眠碎片化、呼吸暂停、脑灌注不良、自主神经失调以及镇静或抗胆碱能药物而降低。因此，瞬时血浆值部分反映了前一夜的清除状态。这种清除解释与分析前和分析变异以及寡聚体淀粉样蛋白-β动力学保持区分，后者另行考虑。实际问题不是变异本身，而是我们测量了分子却没有测量推动它的生理状态。我提出，一种廉价、可重复、床旁的该状态指数——此处以系统性BenBrik老年评分举例说明——可以与每次血浆抽取一起记录并作为协变量建模，将干扰性变异转化为可解释的信号。该预测是可证伪的，并概述了验证设计。这是一个假设，不是经过验证的结果；此处使用的示例性指数，系统性BenBrik老年评分，不是经过验证的测量工具。

英文摘要

Plasma biomarkers, and the amyloid-β 42/40 ratio in particular, are increasingly used to stage Alzheimer's disease and to stratify and monitor clinical trials, on the tacit assumption that a single measurement reflects a stable, steady-state level of pathology. A recent micro-longitudinal study that sampled plasma weekly for five weeks in people with Alzheimer's disease challenged that assumption, reporting that most participants varied by more than half in the amyloid-β 42/40 ratio over that short window. The reflexive response is to treat such movement as measurement noise to be averaged away. Here I argue that a meaningful part of it is unlikely to be noise. In healthy adults the same ratio is among the most stable plasma analytes, so instability concentrated in disease is difficult to attribute to the assay alone. Brain-to-blood clearance of amyloid-β is itself dynamic over days, rising with sleep and falling with fragmentation, apnoea, poor cerebral perfusion, autonomic dysregulation, and sedating or anticholinergic drugs. An instantaneous plasma value therefore partly reports the clearance state of the preceding night. This clearance account is kept distinct from analytical and pre-analytical variance, and from oligomeric amyloid-β dynamics, which are considered separately. The practical problem is not the variance itself but that we measure the molecule without measuring the physiological state that moves it. I propose that an inexpensive, repeatable, bedside index of that state, illustrated here with the Systemic BenBrik Geriatric Score, could be recorded alongside each plasma draw and modelled as a covariate, converting nuisance variance into interpretable signal. The prediction is falsifiable, and a validation design is outlined. This is a hypothesis, not a validated result; the illustrative index used here, the Systemic BenBrik Geriatric Score, is not a validated measure.