The research drew on participants from the Clinic substudy of the COcoa Supplement and Multivitamin Outcomes Study, or COSMOS, a large randomized trial run through Brigham and Women’s Hospital and Harvard Medical School that originally examined whether cocoa flavanols and multivitamins could prevent cardiovascular events. From this infrastructure, the team assembled a cohort of 506 older adults with a mean age of 69.7 years, nearly half of whom were women. Each participant underwent blood sampling at baseline and again two years later, with the researchers measuring four of the most scrutinized molecules in contemporary neurodegeneration science: N-terminal tau, known as NT1, a fragment of the tau protein that earlier work has flagged as a specific and early marker of Alzheimer’s pathology; neurofilament light chain, or NfL, which leaks into the bloodstream when neurons suffer structural damage; oligomeric amyloid-beta, or oAβ, the aggregated form of the sticky peptide central to Alzheimer’s disease; and phosphorylated tau 217, or pTau217, which was measured only once because of its more recent arrival in clinical assays.

On the physical side of the equation, the team employed three well-validated benchmarks of motor function. Walking speed, often described in gerontology as the sixth vital sign, captures the integrated output of the nervous system, muscles, joints, and cardiovascular machinery. Grip strength, measured with a dynamometer, serves as a surrogate for whole-body muscle capacity and has repeatedly been linked to mortality and disability risk. The Short Physical Performance Battery, or SPPB, combines gait speed, repeated chair stands, and balance tests into a composite score that predicts future dependency in daily living. The study’s primary analysis asked whether a person’s starting levels of these brain-derived proteins could forecast how much their physical performance would deteriorate over the following two years. A secondary analysis then asked the reverse question: whether changes in the biomarkers themselves tracked with concurrent changes in strength and mobility.

The statistical architecture of the study deserves attention, because it reflects a growing sophistication in how researchers interrogate biomarker data. The investigators adjusted their models for demographics, a comprehensive cardiovascular risk profile, body mass index, and, crucially, the baseline value of each physical performance measure, which strips away the misleading correlations that arise when people who start out weaker remain weaker throughout. They also considered all biomarkers jointly rather than one at a time, reducing the chance that a single protein would absorb variance that actually belonged to its molecular neighbors. This design choice matters, because neurodegeneration markers tend to rise together in aging bloodstreams, and naive single-biomarker models can manufacture associations that dissolve under stricter scrutiny.

When the results arrived, the silence was deafening. Baseline concentrations of NT1, NfL, and oAβ showed no statistically significant relationship with two-year changes in any physical performance measure. The effect estimates were not merely borderline; they hovered around zero with p-values ranging from 0.25 to 0.69 across the fully adjusted models. For NfL, the estimates ran from a beta of −0.046 for SPPB scores to a beta of 0.350 for grip strength, neither remotely close to conventional thresholds of significance. The story repeated itself for NT1 and oAβ across walking speed, grip strength, and the SPPB. In an era when pTau217 assays are being hailed for their diagnostic accuracy in detecting Alzheimer’s pathology, the failure of these circulating proteins to anticipate physical decline is a sobering reminder that prediction of one disease domain does not automatically generalize to another.

Yet the study was not a total null result. In the secondary longitudinal analysis, the researchers found one statistically significant signal: changes in plasma NT1 over the two-year window were associated with concurrent changes in grip strength, with a beta of 1.408 and a p-value of 0.01. This finding is intriguing precisely because it emerged from tracking within-person trajectories rather than comparing people to each other at baseline. It hints that tau dynamics and muscle strength may be coupled at the level of individual aging processes, even if a single snapshot of tau cannot forecast who will weaken. The researchers urge caution in interpretation, however, as this was one significant result among many tests performed, and no other biomarker change showed a meaningful relationship with any physical outcome.

The results complicate a growing literature that has linked amyloid burden and other neurodegenerative markers to gait slowing and motor decline. Previous studies, including analyses of brain amyloid imaging in dementia-free elderly individuals and population-based cohorts tracking plasma amyloid ratios and neurofilament light over longer horizons, have reported associations between Alzheimer-related pathology and slower walking or reduced strength. The new findings suggest that such relationships may depend on the timeframe examined, the specific biomarkers measured, or the composition of the cohort. Two years is a short window in the biology of aging, and the COSMOS participants were, on average, relatively healthy community-dwelling adults rather than patients with established cognitive impairment, which may have compressed the range of both biomarker levels and physical decline available for detection.

There are also mechanistic considerations worth unpacking. Physical performance depends on a sprawling network of systems, including sarcopenia, peripheral nerve function, vascular supply to muscle, vestibular and visual function, and central motor planning. Neurodegenerative proteins circulating in plasma may rise in response to synaptic injury or neuronal death in specific brain regions, but the path from hippocampal tau accumulation to a slower chair stand is neither direct nor necessarily mediated by the molecules measurable in blood. Exercise and physical activity themselves modulate peripheral inflammatory and neurotrophic factors, which means the relationship between brain and body biomarkers is bidirectional and confounded by behavior. The study’s authors acknowledge that overall, these biomarkers provided limited information about short-term physical performance decline, a conclusion that tempers the enthusiasm of clinicians hoping to fold motor-risk stratification into existing dementia screening panels.

What, then, is the takeaway for a field riding a wave of blood-based dementia diagnostics? The work does not diminish the genuine progress these assays represent for detecting Alzheimer’s pathology in symptomatic patients, where trials in primary and secondary care have shown impressive diagnostic performance. Instead, it draws a boundary around their utility, cautioning against the seductive assumption that a molecule diagnostic of one aspect of brain aging will serve as a universal barometer of the aging organism. For older adults, the practical message remains unchanged: walking speed, grip strength, and performance batteries like the SPPB continue to offer accessible, inexpensive, and proven measures of functional trajectory that no blood test currently replaces. For researchers, the study marks out a clearer roadmap, pointing toward longer follow-up periods, cohorts with greater baseline frailty, and integrated models that treat neurodegeneration as one thread in a much larger tapestry of biological aging. The hunt for a single drop of blood that reveals the whole aging body continues, but this study makes clear that the drop has not yet been found.

Subject of Research: Associations between plasma neurodegeneration biomarkers and physical performance decline in older adults

Article Title: Plasma biomarkers of neurodegeneration and physical performance in older adults

Article References: Plasma biomarkers of neurodegeneration and physical performance in older adults. (n.d.). https://doi.org/10.1007/s11357-026-02570-8

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02570-8

Keywords: plasma biomarkers, neurodegeneration, physical performance, aging, Alzheimer's disease, neurofilament light chain, NT1 tau, phosphorylated tau 217, grip strength, walking speed, Short Physical Performance Battery, COSMOS