Researchers have identified a set of 34 circular RNAs in blood that can detect Alzheimer’s disease with striking accuracy and signal when cognitive symptoms are likely to appear. The blood-based signature outperformed the leading plasma biomarker, p-tau217, in distinguishing people with biomarker-confirmed Alzheimer’s from those without it. For the roughly six million Americans living with the disease and millions more in its silent preclinical phase, the ability to narrow the window between a positive biomarker result and actual memory loss could reshape treatment timing and clinical trial enrollment.
Why a circRNA blood test changes the Alzheimer’s timeline
Most Alzheimer’s biomarkers turn positive years or even decades before a person notices memory problems. A study published in The New England Journal of Medicine documented biomarker changes spanning 15 to 20 years before clinical diagnosis. That long preclinical window creates a practical problem: telling someone their brain shows signs of Alzheimer’s pathology is very different from telling them when they will start to struggle. Current blood tests can confirm that amyloid and tau proteins are accumulating, but they offer limited precision about the pace of decline.
The new circRNA panel addresses that gap. In a study published in Nature Medicine, researchers analyzed blood samples from participants in the Knight Alzheimer Disease Research Center cohort and found that the 34-marker signature achieved an AUC of approximately 0.945 for identifying biomarker-confirmed Alzheimer’s status. By comparison, plasma p-tau217, the current front-runner among blood-based markers, reached an AUC of approximately 0.877 in the same analysis. An AUC of 1.0 would represent perfect discrimination, so the circRNA panel’s edge is clinically meaningful and suggests that noncoding RNA patterns in blood capture disease-relevant biology that protein markers may miss.
The National Institutes of Health noted that elevated circRNA levels nearly tripled the risk of developing symptomatic Alzheimer’s disease among individuals who were cognitively unimpaired at baseline but already had biomarker evidence of pathology. That risk estimate moves the marker beyond simple detection and toward prognosis, giving clinicians a tool that could help answer the question patients and families ask most often: how much time do we have?
A separate line of research has built plasma p-tau217 “clock” models that estimate the age at which a person’s p-tau217 level crosses a positivity threshold, then relates that estimate to time until symptom onset. Those models predict onset age with a median absolute error of roughly 3.0 to 3.5 years within the cohorts studied. The circRNA findings raise a natural question: could combining the two approaches, one excelling at detection and the other at temporal prediction, produce a hybrid model that narrows that error below 2.5 years in an independent validation set? No published study has tested that combination head-to-head in the same individuals, but the complementary strengths of each marker make the hypothesis worth tracking.
Brain tissue atlas and Knight ADRC data behind the circRNA findings
The biological case for using circRNAs as Alzheimer’s biomarkers did not emerge overnight. An earlier atlas of cortical circular RNA expression in Alzheimer’s disease brains established that these molecules are systematically associated with Alzheimer’s pathology in human brain tissue. That foundational work showed circRNAs are not random bystanders but are linked to disease-related changes in the cortex, providing the rationale for looking at whether the same signals appear in blood and whether they track with disease stage.
The Knight Alzheimer Disease Research Center at Washington University in St. Louis supplied the clinical and multi-omic data that made both the circRNA study and the p-tau217 clock work possible. A resource paper describing the Knight ADRC cohort details the genetic, proteomic, and clinical variables available to researchers, including longitudinal follow-up that allows tracking of participants from cognitively normal status through symptom onset. That depth of data is what enables investigators to move from cross-sectional snapshots to predictive models that estimate not just who is at risk, but when symptoms are likely to emerge.
The circRNA panel’s performance metrics were confirmed in a PubMed entry listing the same AUC values for the circRNA signature and for p-tau217. That record anchors the study’s authorship, publication date, and quantitative headline figures in the formal literature, reducing the chance of misattribution and helping clinicians interpret the results in the context of other biomarker work.
Gaps in replication and what to watch next
Strong as the circRNA results are within the Knight ADRC cohort, several questions remain open. No independent replication in a separate population has been reported yet, leaving uncertainty about how well the 34-marker panel will generalize across different ancestries, comorbidities, and clinical settings. The Knight cohort is deeply characterized but relatively homogeneous compared with the broader population, so external validation in community-based samples and in health systems outside academic memory clinics will be critical.
Another unresolved issue is assay standardization. The Nature Medicine study used a specific sequencing and analysis pipeline to quantify circRNAs, but routine clinical deployment would likely require more streamlined, reproducible platforms. Factors such as sample handling, storage time, and coexisting inflammatory conditions could influence circRNA levels, and those pre-analytic variables have not yet been fully mapped. Without harmonized protocols and reference standards, different laboratories could obtain non-comparable results from ostensibly similar tests.
There are also questions about how circRNA testing would fit alongside existing biomarker tools. Plasma p-tau217 and related assays are already moving toward clinical use for triaging patients and deciding who needs confirmatory imaging. If circRNA panels primarily improve risk stratification among people who are already biomarker positive, they may be best deployed as a second-line test to refine prognosis rather than as a first-line screen. Health systems will need cost-effectiveness analyses comparing strategies such as “p-tau217 alone,” “circRNA alone,” and “combined model” to determine where each approach adds enough value to justify its expense.
Ethical considerations loom as well. A blood test that can tell an asymptomatic person that their risk of developing Alzheimer’s symptoms has tripled, and that those symptoms may arrive within a specific multi-year window, carries psychological and practical consequences. Counseling frameworks developed for genetic risk disclosure, such as APOE status, offer a starting point but may not fully capture the impact of time-specific prognostic information. Policymakers and clinicians will need to consider how to protect individuals from discrimination in employment and insurance if highly predictive preclinical tests become widely available.
Despite these caveats, the circRNA work marks a notable step in the evolution of Alzheimer’s biomarkers-from static indicators of pathology toward dynamic tools that map the trajectory from silent disease to overt dementia. If future studies replicate the high AUC values in more diverse cohorts, demonstrate stable performance across laboratories, and show that prognostic information meaningfully changes clinical decisions or trial design, circRNA panels could become a standard component of precision neurology. For now, they offer a glimpse of a future in which a simple blood draw does not just confirm that Alzheimer’s is present in the brain, but helps define when its most visible symptoms are likely to appear.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
