Experimental Drug Shows First-Ever Reduction in Tau Pathology and Cognitive Benefit in Early Alzheimer's Trial

On May 14, 2026, Biogen announced what it called "compelling" topline results from the Phase 2 CELIA study of diranersen (BIIB080), an antisense oligonucleotide designed to reduce production of the tau protein in the brains of people with early Alzheimer's disease [1]. The company described it as the first randomized Phase 2 study of a tau-directed therapy to demonstrate both biomarker impact and cognitive benefit [1].

But the headline number tells a more complicated story: CELIA did not meet its primary endpoint [1][2].

The trial failed to show a statistically significant dose-response relationship for change from baseline on the Clinical Dementia Rating–Sum of Boxes (CDR-SB) — the standard cognitive scale used in Alzheimer's trials — at 76 weeks [1]. Yet Biogen is pressing ahead to a registrational Phase 3 study, arguing that pre-specified secondary analyses showed meaningful slowing of cognitive decline, particularly at the lowest dose tested [1][3].

The result has split analysts and researchers. Biogen's stock jumped roughly 10% on the news [4], while independent observers described the data as "mixed" [2] and cautioned that the path to approval remains uncertain.

What the CELIA Trial Actually Showed

CELIA enrolled 416 participants aged 50–80 with mild cognitive impairment or mild dementia due to Alzheimer's disease, all with confirmed amyloid positivity [5][6]. The global, randomized, double-blind, placebo-controlled study tested three dosing regimens of diranersen delivered by intrathecal injection (a lumbar puncture into the spinal fluid): 60 mg every 24 weeks, 115 mg every 24 weeks, and 115 mg every 12 weeks [1].

The primary endpoint was dose response — whether higher doses produced greater cognitive benefit on the CDR-SB. They did not. In fact, the lowest dose of 60 mg every 24 weeks showed the most pronounced cognitive signal, an inverted dose-response pattern that caused the study to miss its primary goal [2][3].

Biogen has not yet disclosed the specific magnitude of cognitive benefit versus placebo — no CDR-SB point differences, no p-values, no confidence intervals [1][7]. The company said complete data would be presented at the Alzheimer's Association International Conference (AAIC) in London, July 12–15, 2026 [1].

What was disclosed: diranersen produced "robust reductions" in both cerebrospinal fluid (CSF) tau and tau pathology as measured by positron emission tomography (PET) across all dose groups, with reductions maintained throughout the 76-week dosing period [1]. Earlier Phase 1b data showed CSF total tau and phospho-tau-181 reductions of 30–50% at lower doses and approximately 60% at higher doses, sustained for six months after the last dose [8][9].

How This Compares to Anti-Amyloid Therapies

To understand the significance of the CELIA results, they must be placed alongside the approved anti-amyloid antibodies — lecanemab (Leqembi) and donanemab (Kisunla) — which target a different protein, amyloid-beta, rather than tau.

In Phase 3 trials, lecanemab reduced decline on CDR-SB by 27% compared to placebo over 18 months [10]. Donanemab slowed decline on the iADRS scale by 35% in its primary population, with even better results (33% on CDR-SB equivalent) in patients with low-to-medium tau burden [10][11]. Both drugs demonstrated that clearing amyloid plaques from the brain is associated with modest but statistically significant cognitive benefits.

Critically, anti-amyloid therapies also reduce downstream tau biomarkers — but modestly. Lecanemab reduced CSF phospho-tau-181 by a large effect size (~0.8), while donanemab patients who achieved amyloid clearance saw p-tau-217 concentrations fall by 24% [10][11]. Diranersen, by contrast, appears to produce larger direct tau reductions (30–60% in Phase 1b), but the cognitive benefit from those reductions has not yet been precisely quantified [8][9].

The FDA has generally considered a slowing of CDR-SB decline of 0.3–0.5 points over 18 months as the lower boundary of clinical meaningfulness in Alzheimer's trials, though there is no formal threshold [10]. Until Biogen releases exact numbers, it is impossible to judge whether diranersen's cognitive signal meets that bar.

The Safety Question: No ARIA, But Not Risk-Free

One potential advantage of tau-targeting over amyloid-targeting therapies is the absence of amyloid-related imaging abnormalities (ARIA) — brain swelling and microhemorrhages that occur in a significant fraction of patients on anti-amyloid antibodies. In lecanemab trials, 45% of participants experienced treatment-related adverse events, while donanemab trials reported ARIA in more than one-third of patients [10][12].

Diranersen, as an antisense oligonucleotide (ASO) delivered intrathecally, has a fundamentally different safety profile. The CELIA press release stated that adverse events were "comparable across dose groups," though a higher incidence of serious adverse events was observed at the highest dose (115 mg every 12 weeks) [1]. The overall profile was described as "generally consistent" with the Phase 1b study [1].

Phase 1b safety data were more granular: no serious adverse events were reported, though mild adverse effects included nausea, vomiting, diarrhea, fatigue, confusion, and musculoskeletal pain [8]. Three cases of mild tinnitus occurred in treated patients versus none in the placebo group [8]. ASOs as a drug class carry known risks including post-lumbar-puncture headache, injection site reactions, and in rare cases, neuroinflammation [13].

Biogen has not disclosed specific serious adverse event rates or numbers from CELIA. An ongoing long-term extension study continues to monitor durability and safety [1].

A 416-Patient Trial: Is It Enough?

The CELIA trial's 416 participants is a typical size for a Phase 2 dose-ranging study in Alzheimer's disease but small by Phase 3 standards [5]. For context, lecanemab's confirmatory CLARITY AD trial enrolled 1,795 patients, and donanemab's TRAILBLAZER-ALZ 2 enrolled 1,736 [10][11].

The fact that CELIA missed its pre-specified primary endpoint while showing benefit in secondary analyses raises standard concerns about multiple comparisons and the risk of false positives. When a trial tests multiple doses and endpoints, the probability of finding at least one positive signal by chance increases.

Independent biostatisticians have not yet weighed in publicly on the CELIA data, largely because the full dataset remains undisclosed. One analyst noted that "without a dose response from BIIB080 treatment and no indication of the magnitude of cognitive benefit," more work is needed in Phase 3 to confirm the signal [4]. The inverted dose-response — where the lowest dose worked best — is unusual and will require a mechanistic explanation, though it is not unprecedented in ASO pharmacology.

The dropout rate has not been disclosed [7]. In Alzheimer's trials, dropout rates of 15–25% are common over 18-month treatment periods, and differential dropout between treatment and placebo arms can bias results.

The Tau Hypothesis: Cause, Consequence, or Collaborator?

Diranersen's results arrive at a pivotal moment in the long-running scientific debate over whether tau or amyloid is the primary driver of Alzheimer's disease.

The amyloid cascade hypothesis — that amyloid-beta accumulation triggers tau pathology, which in turn causes neurodegeneration — has dominated the field for three decades [14]. The approvals of lecanemab and donanemab provided partial validation: clearing amyloid produces cognitive benefit, at least modestly [10].

But the correlation between tau pathology and cognitive decline has always been stronger than the correlation with amyloid. Neuropathological studies show that the number of neurofibrillary tau tangles, not amyloid plaques, tracks closely with cognitive impairment [15]. Some individuals harbor extensive amyloid deposits yet remain cognitively normal [15].

Source: OpenAlex

Data as of Jan 1, 2026CSV

Research into tau-targeting approaches in Alzheimer's has accelerated in recent years, with publications on tau ASOs and related topics rising from 54 papers in 2011 to 410 in 2025 [16]. Meanwhile, amyloid-focused research, while still dominant, has plateaued — with nearly 9,800 papers published in 2023, the field may have reached saturation [16].

Source: OpenAlex

Data as of Jan 1, 2026CSV

A 2025 review in Cell Death & Disease argued that amyloid-beta and tau "exert synergistic effects" on Alzheimer's pathogenesis and that framing the disease as driven by one protein alone is reductive [14]. Stanford neuroscientists have advocated for giving tau pathology equal weight in therapeutic strategy, noting that tau tangles spread through the brain in a pattern that mirrors the progression of clinical symptoms far more closely than amyloid plaques do [15].

If CELIA's cognitive signal holds up in Phase 3, it would provide the strongest evidence yet that directly reducing tau — rather than waiting for amyloid clearance to indirectly lower tau — can slow Alzheimer's progression. If it fails to replicate, the tau-targeting approach could face the same crisis of confidence that the amyloid hypothesis endured after repeated failures in the 2010s.

Biogen's Financial Gamble

Biogen's decision to advance diranersen to Phase 3 must be understood against the company's troubled Alzheimer's history.

Aducanumab (Aduhelm), Biogen's first Alzheimer's drug, received a controversial accelerated FDA approval in 2021 despite ambiguous efficacy data and a 41% rate of brain swelling in clinical trials [17]. The drug was a commercial failure — generating only $13 million in revenue in 2022 against an initial annual price tag of $56,000 per patient — and was withdrawn from the market in January 2024, costing Biogen approximately $60 million in discontinuation charges alone [17][18].

Biogen pivoted to lecanemab (Leqembi), developed with Eisai, which received traditional FDA approval in July 2023. Leqembi's commercial ramp has been slow but accelerating: global in-market sales reached approximately $134 million in Q4 2025, up 54% year-over-year, with full-year 2025 revenue from Leqembi and other new products approaching $1 billion [19].

Biogen's total revenue in Q3 2025 was $2.5 billion, up 3% year-over-year [19]. The company has been aggressively cutting R&D costs through "Fit for Growth" initiatives while prioritizing its most promising pipeline assets [19].

A Phase 3 trial for diranersen would likely cost hundreds of millions of dollars. Analysts project peak sales of approximately $2 billion by 2030 if diranersen is approved [4] — a significant opportunity in a global Alzheimer's therapeutics market projected to grow from $6.49 billion in 2025 to $33.62 billion by 2034 [20].

Source: Precedence Research

Data as of Jan 1, 2025CSV

Biogen's stock surged roughly 10% on the CELIA announcement, but Jefferies reiterated its existing rating, suggesting the data are encouraging but not yet conclusive enough to change the investment thesis [4][21].

Source: Alzheimer's Association

Data as of Jan 1, 2026CSV

Tau-targeted agents now represent approximately 20% of the Alzheimer's drug development pipeline, up from 6% in 2018, reflecting the field's diversification beyond amyloid [22]. Diranersen is not the only tau-targeting compound in development — TauRx Pharmaceuticals has reported positive results for hydromethylthionine mesylate, an oral tau aggregation inhibitor, and is pursuing regulatory filings [22].

The Subgroup Question

One of the most critical unanswered questions is which patients drove the cognitive benefit signal. Did the effect hold across APOE4 carriers (who carry the strongest genetic risk factor for Alzheimer's) and non-carriers? Across different levels of baseline tau burden? Across age brackets?

Biogen's topline press release did not break out results by APOE4 status, tau-PET stratification, or age [1]. The Phase 2 baseline characteristics paper confirmed that participants were aged 50–80 with confirmed amyloid positivity, CDR global scores of 0.5–1, and MMSE scores of 21–30 [5], but subgroup efficacy data remain under wraps.

This matters because of a pattern seen with anti-amyloid drugs. Donanemab's TRAILBLAZER-ALZ 2 trial found that patients with low-to-medium tau burden benefited significantly, while those with high tau pathology saw minimal cognitive benefit despite robust amyloid clearance [11]. If diranersen's signal is similarly concentrated in a narrow patient subpopulation, its real-world applicability could be limited.

The requirement for intrathecal delivery (lumbar punctures every 12–24 weeks) adds another practical constraint. Unlike intravenous infusions or subcutaneous injections, intrathecal administration requires specialized clinical settings and carries procedure-related risks, potentially limiting the eligible patient population.

What a Skeptic Would Say

A rigorous skeptic would argue that the CELIA results, as disclosed, prove little.

First, the trial missed its primary endpoint. The dose-response test was pre-specified to guard against exactly the kind of cherry-picking that post-hoc subgroup analyses can produce. Failing it means the positive cognitive signals in secondary analyses must be treated with caution.

Second, an inverted dose-response — where less drug works better — needs a biological explanation. One possibility is that higher doses cause off-target effects or trigger compensatory mechanisms that blunt efficacy. Another is that the lowest-dose result is a statistical artifact. Without the full data, neither explanation can be ruled out.

Third, 416 patients over 76 weeks in a progressive neurodegenerative disease leaves room for placebo effects, regression to the mean (patients enrolled during a cognitive dip may naturally improve), and patient selection bias (the stringent inclusion criteria may have selected for a population that declines more slowly than the general early Alzheimer's population).

Fourth, robust biomarker changes do not guarantee clinical benefit. The Alzheimer's field has been burned before by drugs that hit their biological target — clearing amyloid, in the case of several failed antibodies in the 2010s — without translating to meaningful cognitive improvement.

For Phase 3 to be convincing, independent researchers would likely want to see: a pre-specified primary endpoint that diranersen can meet, a larger sample (likely 1,000+ patients), stratified subgroup analyses by APOE4 status and tau burden, and a direct demonstration that the cognitive benefit exceeds the FDA's informal threshold for clinical meaningfulness.

What Comes Next

Biogen has stated it will present full CELIA data at AAIC in July 2026 [1]. The FDA granted Fast Track designation to diranersen in April 2025, which allows for rolling submission of a future application but does not lower the evidentiary bar for approval [23].

The Phase 3 trial design will be critical. If Biogen selects the 60 mg every-24-weeks dose — the one that showed the strongest cognitive signal — and powers the study adequately with a clear primary CDR-SB endpoint, a positive result would be a landmark in Alzheimer's research: the first direct validation that reducing tau protein can slow cognitive decline.

If it fails, the consequences extend beyond Biogen. The broader tau-targeting field, which has grown to 20% of the Alzheimer's pipeline, would face hard questions about whether tau reduction can translate from impressive biomarker data to the thing that actually matters: keeping people's minds intact longer.

The data presented so far are a reason for cautious attention, not celebration. The full picture arrives this summer.