July 14, 2026 | Researchers from Washington University School of Medicine, St. Louis have showed that elevated levels of a set of circular RNAs (circRNAs) in the blood nearly tripled patients’ risk of developing Alzheimer’s Disease (AD) symptoms, suggesting these molecules are more sensitive to symptom onset than traditional AD biomarkers. Their work was published this month in Nature Medicine (DOI: 10.1038/s41591-026-04485-5).
Biomarker-confirmed diagnosis of AD has traditionally used biomarkers that measure amyloid beta (Aβ), Tau, and phosphorylated Tau (pTau) in cerebrospinal fluid (CSF) or via amyloid-positron emission tomography (PET), the authors write. But CSF collection through lumbar puncture is painful, and PET scans are expensive.
Blood biomarkers, therefore, are crucial, particularly ones that can capture overall disease status and symptomology independent of amyloid plaque. “Plasma pTau217 is widely regarded as the leading plasma-based biomarker of AD in preclinical stages and has indeed been shown to be a marker of amyloid pathology that changes 15–20 years before clinical onset,” the authors acknowledge. “However, pTau217 can be unreliable as a biomarker in patients treated with anti-amyloid therapies, as it may normalize with brain Aβ plaque removal without corresponding changes in cognitive improvement,” the authors write.
“Patients being treated with novel Aβ-removal therapies, can become pTau negative but still have Alzheimer’s disease. These circular RNAs may grant us a more complete perspective of someone’s overall disease biology,” said Cruchaga, the study’s corresponding author, in a press release.
“Biomarkers independent of Aβ and Tau pathology are essential to monitor overall neurodegeneration in this therapeutic area,” the authors conclude.
Unlike amyloid plaques, which accumulate slowly in the brain, circRNAs are more dynamic, reflecting the brain’s more recent activity. Research published in 2022 reported a small number of circRNAs in blood that were associated with AD in a 40-person dataset (DOI: 10.1186/s40364-022-00405-0). Cruchaga and his colleagues, then, looked for a targeted list of circRNAs that could differentiate individuals with symptomatic AD from cognitively unimpaired individuals.
The team generated RNA-sequencing (RNA-seq) data and longitudinal clinical data from 816 cognitively unimpaired individuals and 405 participants with AD to identify circRNAs associated with AD clinical status, Aβ and Tau stages, and progression to symptomatic AD. Any circRNAs to survive multiple tests were considered significant.
The team identified a set of 34 circRNAs that were associated with AD, three of which were nominally significant in all neurodegenerative diseases. Interestingly, the authors highlight that 2 of the 34 circRNA biomarkers, “are derived from the gene PICALM, which is a genome-wide association study-significant AD risk gene associated with Aβ, Tau, APOE4 function, synaptic dysfunction and microglia-derived neuroinflammation.”
They built predictive models for this set of circRNAs that successfully identified individuals with AD pathology, performing similarly to models trained on the protein pTau217 data.
Based on longitudinal clinical data for participants, the team found that 78 participants progressed from cognitively unimpaired to symptomatic AD after blood collection. The researchers were able to perform survival analyses to determine if the 34 blood circRNAs could predict progression to symptomatic AD. When they analyzed progression to symptomatic AD within 5 years, the circRNA model showing a significantly higher AUC than pTau217 alone.
The researchers replicated their findings in independent samples from the Knight-Alzheimer Disease Research Center (ADRC) and the Anti-Amyloid Treatment in Asymptomatic Alzheimer Disease (A4) cohorts, they explain. In these datasets, too, the circRNA model was better than plasma pTau217 when predicting progression to symptomatic AD. “These results suggest the potential of blood circRNAs as a noninvasive, high-precision tool for early AD diagnosis and progression monitoring in clinical practice,” the authors write.
The test may not be sufficient for patient stratification for clinical trials or treatment, the authors concede, but they found that combining the pTau217 and circRNA models further improved the predictive power of which patients would proceed to symptomatic AD in both the discovery dataset and the replication datasets.
Together with commercial partners, the researchers are currently working to develop translatable clinical assays for blood-based circRNAs.
“It’s nice to have good science and models, but we’re ultimately doing this to help people,” Cruchaga said.