Blood-Based Circular RNA Biomarkers Outperform pTau217 in Early Alzheimer's Detection, Landmark Nature Medicine Study Shows
Key Insights
A 34-circRNA blood biomarker signature achieved an AUC of 0.945 for detecting biomarker-confirmed Alzheimer's disease (search), significantly outperforming plasma pTau217 (search) (AUC 0.877).
For predicting progression to symptomatic AD in cognitively unimpaired individuals, circRNAs showed a hazard ratio of 2.92 versus 1.81 for pTau217 (search) — a 61% improvement in predictive power.
The integrated circRNA-pTau217 (search) model reached near-perfect diagnostic accuracy (AUC 0.967) and identified 84% of dual-positive individuals who progressed to AD versus 15% of dual-negative individuals.
Blood-based circular RNA (circRNA) biomarkers can detect Alzheimer's disease (search) with unprecedented accuracy and predict progression to symptomatic disease years before clinical onset, according to landmark findings published in Nature Medicine. The study, led by Dr. Carlos Cruchaga at Washington University in St. Louis, represents the first demonstration that circRNAs can outperform existing gold-standard blood biomarkers for both diagnosis and preclinical risk stratification.
The research, validated across nearly 2,400 individuals in three independent cohorts, identified a signature of 34 blood circRNAs that achieved an area under the curve (AUC) of 0.945 for detecting biomarker-confirmed Alzheimer's disease (search) in the Knight-ADRC clinical cohort. This performance significantly exceeded that of plasma pTau217 (search) (AUC 0.877), the current clinical gold standard. When integrated with pTau217, the combined model reached an AUC of 0.967 — approaching near-perfect diagnostic accuracy and rivaling the performance of invasive cerebrospinal fluid tests and expensive PET imaging.
"This study represents the first time we've demonstrated that blood-based circular RNAs can predict progression to symptomatic Alzheimer's disease (search), and they do so with performance that exceeds our current gold-standard biomarkers," said Dr. Carlos Cruchaga, Professor of Psychiatry and Director of the Neurogenomics and Informatics Center at Washington University School of Medicine in St. Louis. "What makes this particularly exciting is that these biomarkers begin changing years before symptoms appear and they capture disease biology that goes beyond amyloid and tau."
Superior Predictive Power for Disease Progression
The circRNA biomarkers demonstrated their greatest advantage in predicting which cognitively unimpaired individuals would progress to symptomatic mild cognitive impairment or Alzheimer's disease (search). The circRNA model showed a hazard ratio (HR) of 2.92 (95% CI 1.63–5.23) compared to pTau217 (search)'s HR of 1.81 (95% CI 1.11–2.94) — a 61% improvement in predictive power.
For identifying individuals who would develop symptoms within five years, the circRNA model achieved an AUC of 0.870 versus pTau217 (search) alone at 0.676 — a dramatic 29% improvement. When the two biomarkers were combined, the integrated model reached an HR of 4.83 (95% CI 2.19–10.68). Notably, 84% of individuals positive for both biomarkers progressed to Alzheimer's disease (search), compared to only 15% of those negative for both.
The combined model also identified two intermediate risk groups: those positive for pTau217 (search) but negative for circRNA (HR 3.25, 41% progressed) and those with the opposite pattern (HR 3.96, 74% progressed), enabling nuanced risk stratification.
Multi-Pathway Biology Captured in a Single Blood Draw
The 34-circRNA signature captures the full complexity of Alzheimer's disease (search) biology through a single blood-based readout. The host genes encoding these circRNAs span multiple critical pathways: GWAS-identified risk genes like PICALM (search), synaptic function regulators like DNAJC6 (search), and amyloid precursor protein (APP) modulators. This multi-pathway mapping encompasses neuroinflammation, synaptic dysfunction, protein aggregation, and neurodegeneration simultaneously.
Most of the 34 circRNAs showed high expression in the brain, with 30 of 34 having brain expression ranked in the 80th percentile or above across 33 tissues in the CircAtlas 3.0 database. These circRNAs are brain-enriched and brain-derived, crossing the blood-brain barrier to reflect central nervous system disease biology directly in a simple blood draw.
Pseudotrajectory analyses revealed that circRNA levels begin changing 2–4 years before clinical onset, with a linear and consistent increase starting in the presymptomatic phase and continuing until symptomatic Alzheimer's disease (search) emerges.
Rigorous Validation Across Three Independent Cohorts
The 34-circRNA biomarker signature was discovered and validated across three independent cohorts: the Knight-ADRC discovery cohort (n=1,221), a Knight-ADRC replication cohort (n=551), and the Anti-Amyloid Treatment in Asymptomatic Alzheimer's Disease (search) (A4) study cohort (n=676). In the Knight-ADRC replication cohort, the circRNA model showed an AUC of 0.863 for biomarker-confirmed status, increasing to 0.955 when integrated with pTau217 (search). In the A4 cohort, the circRNA model replicated progression prediction with an HR of 2.93, significantly higher than pTau217 alone (HR 1.87).
The model demonstrated robust performance across ancestries, with AUCs of 0.906 in European, 0.934 in African, and 0.922 in mixed ancestry populations. Importantly, the 34-circRNA model showed high specificity for Alzheimer's disease (search), with low AUC values when applied to non-AD dementias: Parkinson's disease (search) (AUC 0.413), dementia with Lewy bodies (search) (AUC 0.545), and frontotemporal dementia (search) (AUC 0.404).
CircPATH™ Platform Launch
Concurrent with the Nature Medicine publication, Circular Genomics (search) announced the launch of CircPATH™, the company's proprietary circular RNA discovery and validation engine that serves as the technological backbone for all future diagnostic product offerings.
"Circular Genomics (search) is advancing the kind of bold innovation the Diagnostics Accelerator was created to support — technologies that can fundamentally transform how we detect, monitor, and understand Alzheimer's," said Laura Nisenbaum, PhD, Executive Director of Drug Development at the Alzheimer's Drug Discovery Foundation (ADDF). "These findings suggest circRNA biomarkers may provide a more comprehensive window into disease processes through a simple blood test."
Dr. Nikolaos Mellios, co-founder and Chief Scientific Officer of Circular Genomics (search), noted: "What sets circular RNAs apart is their ability to map multiple disease pathways simultaneously while remaining highly specific to Alzheimer's disease (search). This isn't just incremental improvement — it's a fundamentally different approach to biomarker development."
The CircPATH™ platform enables enrichment of clinical trial populations with individuals most likely to progress, reducing sample sizes and trial duration. It provides objective endpoints for monitoring treatment response across multiple biological pathways and enables stratification strategies to identify which patients are most likely to benefit from specific therapeutic mechanisms.
Clinical Implications
The study authors note that plasma pTau217 (search) predominantly tracks amyloid pathology and can be unreliable in patients treated with anti-amyloid therapies, as it may normalize with brain Aβ plaque removal without corresponding cognitive improvement. Biomarkers independent of Aβ and tau pathology are essential to monitor overall neurodegeneration, and the 34 blood circRNAs appear to capture Alzheimer's disease (search) biology beyond these pathologies.
Limitations acknowledged by the researchers include small sample sizes in the time-to-onset subsets, limited CDR data primarily at very early disease stages, and the need for further investigation of three circRNAs (circRBM23, circEPB41, and circNUP54) that were also associated with other neurodegenerative diseases. Future studies are needed to examine the impact of Alzheimer's comorbidities on the blood circRNA models.
Circular Genomics (search) is advancing a pipeline of circRNA-based diagnostic products targeting Alzheimer's disease (search), Parkinson's disease (search), and other neurodegenerative conditions, and is actively pursuing partnerships with biopharma companies, clinical research organizations, and academic medical centers.