A drug can strip half the tau out of a person's spinal fluid and still fail to prove it helps their memory. That is roughly what happened in Biogen's Phase 2 CELIA trial of diranersen, which lowered tau markers substantially but missed the endpoint it was designed to hit — and it is the clearest recent illustration of why a biomarker change is not the same as a clinical benefit.
Tau is one of the two proteins that define Alzheimer's pathology. It builds up inside neurons as tangles, and its spread through the brain tracks with cognitive decline more closely than amyloid plaque does. That correlation is what makes tau-lowering attractive — and what makes the CELIA result worth understanding before anyone reads a headline as a breakthrough.
Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.
Table of Contents
- What CELIA actually tested, and what it found
- Why strong biomarker results did not settle the question
- The lowest dose carried the signal — and why that is a problem
- How specialists are reading it
- What this means if you or a family member has early Alzheimer's
- Frequently Asked Questions
What CELIA actually tested, and what it found
diranersen (BIIB080) is an antisense oligonucleotide: a short strand of engineered genetic material that binds the messenger RNA for the tau gene, MAPT, so the cell makes less tau protein. It cannot be swallowed. It is given intrathecally — injected into the spinal canal — because that is the only practical way to get this class of molecule into the central nervous system. According to the summary from Alzheimer Europe and Biogen, CELIA enrolled 416 people with mild cognitive impairment due to Alzheimer's or mild Alzheimer's dementia, mean age 68, 60% at the MCI stage.
Three regimens ran for 18 months: 60 mg every 24 weeks, 115 mg every 24 weeks, and 115 mg every 12 weeks. The primary endpoint was not "does it work." It was whether a dose–response relationship existed on CDR-SB, the standard clinician-rated scale of cognition and daily function, at Week 76. In plain terms: does more tau lowering produce more benefit? NeurologyLive's topline report states the trial did not show that relationship. The premise the study was built on did not hold.
Why strong biomarker results did not settle the question
The biomarker arm of the trial worked. All three doses produced sustained 50–65% reductions in cerebrospinal fluid total tau and decreases in tau-PET brain imaging signal, which Biogen reported at the Alzheimer's Association International Conference. The drug reached its target and did what it was designed to do biologically. A biomarker like CSF tau is a proxy — a measurable stand-in for a disease process. Proxies earn their standing only when changing them reliably changes outcomes.
Alzheimer's has a long history of proxies that moved without patients improving, which is why regulators and clinicians treat target engagement as a necessary step, not a finish line. The gap matters for how you read any future announcement. "Reduced tau by 65%" describes chemistry. "Slowed decline on a clinical scale" describes a person's life. A press release can contain the first without the second, and CELIA is the case where the two came apart in the same trial.
The lowest dose carried the signal — and why that is a problem
CELIA did produce an efficacy signal, but from an unexpected place. As reported by Biogen and covered in NeurologyLive's follow-up on the full data, the 60 mg every-24-weeks arm — the lowest exposure tested — showed 26% less decline on CDR-SB, 42% on ADAS-Cog13 and 50% on MMSE versus placebo at 18 months. An inverted dose–response like this is not impossible, but it is the pattern most likely to be produced by chance. And the arm carrying it was the smallest in the trial, at roughly 60 participants, per Alzforum's conference coverage. Small groups produce wide, unstable estimates.
This one is the basis for advancing to Phase 3. The imaging evidence has a similar shape. The tau-PET substudy covered 131 of the 416 participants — under a third of the trial — and was not powered to link imaging change to cognitive change. It shows the drug altered pathology. It does not show the pathology change caused the cognitive difference.
How specialists are reading it
Reaction at AAIC was split rather than enthusiastic. Adam Boxer of UCSF told Alzforum the results "seem promising and are in the expected direction, but they are too preliminary to draw any firm conclusions," and other commenters argued the data do not amount to clear validation of tau as a treatment target. That split is the honest state of the evidence, and it is worth holding onto when coverage flattens into either "tau drug works" or "tau drug fails." Both readings are available from the same trial depending on which number you lead with.
The pattern is not unique to Biogen. Eisai reported that etalanetug (E2814) cut CSF MTBR-tau243 by 50% at three months and 75% at nine months, and p-tau217 by 50% at 24 months, in a Phase 1b/2 study in dominantly inherited Alzheimer's — with tau-PET data from only three participants and no cognitive efficacy readout yet. Biomarkers first, clinical answers later, across the field.
What this means if you or a family member has early Alzheimer's
The practical bottom line is short: diranersen is investigational and is not a treatment you can get. It has no FDA approval, it requires spinal injection, and Biogen's stated next step is a Phase 3 trial.
A Phase 2 signal is a reason to run a bigger study, not a reason to change a care plan. What you can do with a report like this: A Phase 3 built on the smallest arm of a trial that missed its primary endpoint is a genuine test of the tau hypothesis, and it has not been run yet.
- Ask which endpoint moved. Biomarker change and clinical change are separate claims; a summary that names only the first has not shown benefit.
- Ask how many people were in the arm that showed the result. Sixty participants supports a hypothesis, not a conclusion.
- Ask whether the result came from the dose the study predicted would work. An unexpected dose pattern usually means more testing is needed.
- If trial participation interests you, raise it with the treating neurologist now, since eligibility for early-stage studies often depends on staging, biomarker confirmation and timing rather than on willingness.
Frequently Asked Questions
Does lowering tau slow Alzheimer's?
Not proven. CELIA lowered CSF total tau by 50–65% across all doses but failed to show that more tau lowering meant more clinical benefit, which was its primary endpoint.
Is diranersen the same kind of drug as lecanemab or donanemab?
No. Those are antibodies targeting amyloid. Diranersen is an antisense oligonucleotide that reduces production of tau protein, given by spinal injection rather than infusion.
Why would a lower dose work better than a higher one?
CELIA was not designed to answer that. The inverted pattern may reflect real biology or may be noise from a small group of about 60 participants; only a larger trial can separate the two.
