In 1906, a 50-year-old woman in Germany died of a mysterious illness. Before her death, she presented with a combination of symptoms that stumped doctors—progressive memory loss, paranoia, confusion, and aggression. A closer look into her brain post-mortem revealed abnormal clumps and tangled bundles of fibers. This was the first documented case of Alzheimer’s disease, described in detail by Alois Alzheimer, a clinical psychiatrist and neuroanatomist.1 His characterization of the disease pathology is still used for diagnosis of this neurodegenerative disorder. Scientists now know that the clumps are plaques formed by the protein fragment amyloid-beta (Aβ) and the tangles are abnormal accumulations of the protein tau within neurons.
Although researchers have since made substantial strides towards uncovering the mechanisms of the disease, certain aspects remain unclear. People with Alzheimer’s disease can show cognitive impairment—difficulty with thinking, making decisions, learning, and remembering—between the ages of 40 and 100 years. The conversion of mild cognitive decline into dementia can take anywhere between two and 20 years after the onset of first symptoms.2 Clinicians rely on the appearance of Aβ plaques and tau tangles in diagnostic imaging to predict the chances and timing of cognitive impairment. However, these parameters explain only 20–40 percent of the variation in the population, making them unreliable.3 In fact, some people with Aβ plaques maintain normal cognitive abilities. Now, based on experimental and machine learning analysis of data collected over decades, a cohort of researchers have reported a potential signature to estimate the tendency for cognitive decline in individuals with early Alzheimer’s disease.4 This predictor, which is independent of Aβ and tau tangles, is a significantly stronger indicator of cognitive decline than the current gold-standard biomarkers.
To identify proteins that are strongly associated with Alzheimer’s diseases and cognitive decline, Tony Wyss-Coray, a neurologist at Stanford Medicine, and his colleagues performed a large-scale proteomic analysis (measuring more than 7,000 proteins per sample) on the cerebrospinal fluid of more than 3,000 people with Alzheimer’s disease across Sweden, Finland, and the US. The team also had access to detailed information on the cognitive status of the individuals over multiple years, their age, sex, Aβ and tau biomarkers, and Alzheimer’s disease risk genes. They observed changes in the expression of hundreds of proteins with respect to cognitive impairment, the most significant of which were proteins present at neuronal connections, or synapses. Using machine learning, the researchers arrived at a new signature of cognitive impairment—a ratio of two synaptic proteins, YWHAG:NPTX2.
Wyss-Coray and his team observed that an increase in YWHAG:NPTX2 was a robust predictor of cognitive decline across all cohorts, suggesting its potential use in clinical settings. Individuals with high YWHAG:NPTX2 had a 15-times higher risk of suffering cognitive decline than those with a low ratio. When the researchers compared its performance to currently used predictors—Aβ, tau, or neurodegeneration markers—the YWHAG to NPTX2 ratio was substantially better at estimating the risk of cognitive dysfunction. Moreover, individuals with high levels of tau tangles had high YWHAG:NPTX2, but the reverse was not true, indicating the ratio of these proteins changes before tau tangles appear in Alzheimer’s disease.
The team wanted to know if the ratio between YWHAG and NPTX2 could predict the progression of Alzheimer’s disease in people with a diagnosis or those with a genetic risk of the disease. They determined that YWHAG:NPTX2 starts increasing 30 years before plaques and tangles form. While everyone showed an age related increase in YWHAG:NPTX2, individuals with Alzheimer’s disease-associated genes had a steeper rise in the ratio. Symptoms in these people generally manifested 20 years after YWHAG:NPTX2 increased.
Wyss-Coray and his colleagues speculate that YWHAG:NPTX2 is a measure of synapse dysfunction, which could be restored by therapeutically targeting these proteins.
- Hippius H, Neundörfer G. The discovery of Alzheimer’s disease. Dialogues Clin Neurosci. 2003;5(1):101-108.
- Dujardin S, et al. Tau molecular diversity contributes to clinical heterogeneity in Alzheimer’s disease. Nat Med. 2020;26(8):1256-1263.
- Tosun D, et al. Contribution of Alzheimer’s biomarkers and risk factors to cognitive impairment and decline across the Alzheimer’s disease continuum. Alzheimer’s & Dementia. 2022;18(7):1370-1382.
- Oh HSH, et al. A cerebrospinal fluid synaptic protein biomarker for prediction of cognitive resilience versus decline in Alzheimer’s disease. Nat Med. 2025.
