Twice-Yearly Alzheimer’s Injection: Promising Signal, Long Odds

Bottom line: the headline says a twice-yearly “jab” halved cognitive decline in Alzheimer’s. The drug, diranersen, is by Biogen’s account the first tau-directed therapy to lower tau in patients and slow decline at the same time. That is a real advance. But it came from a Phase 2 trial that missed its primary endpoint, the best result came from the smallest dose and the smallest group (60 people), and the 54% figure comes from an open-label extension. Alzheimer’s has one of the worst Phase 2 to Phase 3 records in medicine. Promising, yes. Proven, no. And not available before about 2030 at the very earliest.

What was reported

  • The drug: diranersen (Biogen, discovered by Ionis) is an antisense oligonucleotide that blocks the messenger RNA that tells brain cells to make tau, the protein that forms tangles inside neurons. [1][3][7]
  • The trial: CELIA, a Phase 2 dose-ranging study of 416 people with early Alzheimer’s, presented at the CTAD conference in Boston on 16 to 19 November 2026. Arms: placebo (115), 60 mg every 24 weeks (60), 115 mg every 24 weeks (115) and 115 mg every 12 weeks (116). [2][3][10]
  • The headline numbers: at about two years the 60 mg twice-yearly group showed 54% less decline on the ADAS-Cog (memory and thinking) and 38% less on the CDR-SB (overall severity, a 1.09 point difference). [3][4][5]
  • The biology: cerebrospinal fluid total tau fell 50% to 65% on every dose, and tau PET scans showed reductions across all brain regions measured. No ARIA (the brain swelling and bleeding seen with amyloid antibodies) was observed. [8][10]
  • Convenience: two doses a year, compared with the regular infusions the amyloid antibodies need.

What the headline leaves out

  • The trial missed its primary endpoint. CELIA was designed to show a dose response on the CDR-SB at week 76. It did not (P = 0.21). Biogen is going to Phase 3 anyway, on secondary and exploratory results. [7][9][11][12]
  • The dose response ran backwards. At 18 months the lowest dose slowed the CDR-SB by 26%, the middle dose by 14% and the most frequent high dose by 9%. Biogen cannot yet explain why. [9][10][13]
  • It might be noise. CELIA was sized to test whether effects scale with dose, not to rank individual arms. A gap this large is close to what chance alone can produce. The alternatives are a true ceiling effect or a safety problem at higher doses. Phase 3 has to settle it. [14]
  • The 54% is not the blinded result. The Pharmaceutical Journal describes the two-year figures as coming from an open-label extension, where everyone knew who got the drug. BMO Capital Markets called the placebo comparison “imperfect” because some placebo patients later received diranersen. Alzheimer’s Research UK said the open-label design limits how firmly the results can be read. [3][4]
  • Different scales, different stories. At 18 months the blinded 60 mg result was a 26% slowing on CDR-SB (0.54 points), 42% on ADAS-Cog13 and 50% on MMSE. The cognitive scores reached nominal significance only. The daily-function scale (ADCS-ADL-MCI) showed no separation from placebo, and the functional result at two years is still not statistically significant. [4][10]
  • It is not an arm jab. Diranersen is given intrathecally, into the spinal fluid by lumbar puncture. The commonest side effects were procedural pain, post-lumbar-puncture syndrome and confusional state. Confusional state was 14.8% in the 60 mg group at the latest data cut. Serious adverse events were 23.3% in the highest-dose arm, and completion of the placebo-controlled period was 87.0% on placebo, 86.7% on 60 mg and only 74.1% on the highest dose. [2][4][10]
  • Cross-trial comparisons are shaky. Biogen likes to compare its 1.09 point CDR-SB difference with 0.45 for Leqembi (18 months) and 0.7 for Kisunla (76 weeks). Different populations, different durations, different trial designs. Treat it as illustration, not evidence. [4][14]

Why Phase 2 promise so often dies in Phase 3

This is the part the headlines skip. The base rates are brutal.

  • Alzheimer’s is the worst field. Between 2002 and 2012, 244 compounds went through 413 trials and one was approved, a 99.6% failure rate against about 81% for cancer drugs. [20][21][22] A 2026 review of 497 Phase 2 and 3 trials argues the failure was partly institutional (sunk costs, entrenched hypotheses) and not only biological. That is one author’s argument, so treat it with care. [23]
  • Phase 2 is the biggest filter in all of medicine. Across all diseases (2011 to 2020), only 28.9% of candidates advanced from Phase 2 to Phase 3, and only 7.9% of drugs entering Phase 1 reached approval. [24]
  • Even good Phase 2 results fail to confirm. The FDA studied 22 cases of promising Phase 2 results. Phase 3 did not confirm effectiveness in 14, safety in one, and both in seven, even when Phase 2 was large and used clinical outcomes. [25]
  • Subgroup and low-dose rescues are a known trap. Cummings himself, writing on Alzheimer’s failures, lists inappropriate subgroup analyses among the things to avoid. [20] Companies regularly push on after a primary miss by pointing to a better-looking dose or subgroup. That is exactly the pattern here. [14][16]
  • Aducanumab shows how discordant Phase 3 can be. The twin Phase 3 trials EMERGE and ENGAGE ran the same design. Only EMERGE showed a significant effect from the high dose. [32]

Closest precedents: tau and antisense

DrugEarly signalWhat happened next
Tominersen (antisense, Huntington’s)Phase 1/2 lowered the mutant protein in spinal fluid by about 40%, with no serious side effectsPhase 3 halted in 2021. No benefit over placebo, and the more frequent regimen was worse. [31]
Gosuranemab (anti-tau antibody)Entered Phase 2 with a strong tau rationaleFailed on CDR-SB at 18 months; worsened ADAS-Cog decline in Phase 2. Shelved. [27][28]
Semorinemab (anti-tau antibody)Slowed decline on one cognitive battery in one Phase 2The other Phase 2 failed to show meaningful efficacy. Tilavonemab and zagotenemab also failed. [26][29][30]
Tominersen is the sharpest warning: lowering the target protein did not mean the patients got better. Diranersen’s tau lowering is also not dose-related.

The field reads the antibody failures as a reason to expect a different result. Diranersen works inside neurons by cutting tau production, which is a different approach from antibodies mopping up tau outside cells. [14] That is a fair argument. It is still an argument.

What gives me some optimism

  • Every dose lowered tau and every dose slowed decline on the cognitive scales, in a field where earlier tau drugs did neither. [10][11]
  • Cognitive and functional signals pointed the same way in the extension, and 94% of placebo-controlled completers chose to continue in the long-term extension. [4][10]
  • No ARIA, which is the main safety limit on the amyloid antibodies. [4][14]
  • The developers have a record with this drug class. Biogen and Ionis’ earlier antisense drugs Spinraza (spinal muscular atrophy) and Qalsody (ALS) reached the market. [19]
  • The FDA gave diranersen Fast Track designation in 2025, and the UK regulator gave it an innovation passport on 9 November 2026. Neither says anything about whether it works. [3][7]

When could it be released?

  • Not now. The UK passport is not a marketing authorisation. It only helps the development and access pathway. [3]
  • Phase 3 has been promised but not fully detailed. Biogen has said it will advance to registrational development and engage regulators. Reports say it has not disclosed a start date or design. [14][16]
  • Timeline estimates. One investor analysis says Phase 3 starts in 2027 with a readout around 2030 to 2031. Biogen’s CEO has called it high-risk, high-reward with a possible payout around 2030. [17][18] The analysis is a trading blog’s, not Biogen’s.
  • My arithmetic: add regulatory review, and approval and a first launch land in the early 2030s at best, and only if Phase 3 succeeds. If Phase 3 fails, it never arrives.

What this means

  • This is the best tau result Alzheimer’s has had. It deserves a Phase 3, and Biogen is right to run one.
  • “Halves decline” is not a fair summary of a trial that missed its primary endpoint, produced the 54% figure in an open-label extension, and showed a backwards dose response.
  • Watch four things in Phase 3: a blinded primary endpoint met at the chosen dose, a statistically significant functional benefit, the confusional-state rate, and whether lumbar puncture twice a year is workable at scale.
  • Nothing here changes what you should do today. Blood pressure, hearing, exercise and vaccinations still matter. See my post on the shingles vaccine and dementia.

References

Numbers in square brackets above refer to this list. Most trial details come from conference reports and company releases. No peer-reviewed CELIA paper was available when this was written. Header photo: Shawn Day on Unsplash.

The CELIA results and coverage

  1. Donnelly L. Alzheimer’s jab halves rate of decline, study finds. The Telegraph, 18 November 2026 (paywalled); syndicated coverage: Stuff NZ, ‘Thrilling’ results
  2. LBC: Alzheimer’s jab halves decline with one injection every six months (18 November 2026; spinal-fluid injection, larger trial needed)
  3. The Pharmaceutical Journal: Investigational Alzheimer’s drug slows rate of cognitive decline (19 November 2026; open-label extension, MHRA innovation passport)
  4. Taylor NP. Biogen’s tau therapy drives bigger benefits in 2-year Alzheimer’s update. Fierce Biotech, 18 November 2026
  5. Cross R. #CTAD26: Biogen’s tau drug slows Alzheimer’s more after nearly two years. Endpoints News, 17 November 2026
  6. BioSpace: CTAD 2026, Biogen, AriBio lead the way in slowing cognitive decline

Primary sources and detailed data

  1. Biogen. Topline results from Phase 2 CELIA study of diranersen (BIIB080). Press release, 14 May 2026 (primary endpoint not met; FDA Fast Track 2025)
  2. Biogen. Biogen presents Phase 2 CELIA data at AAIC. Press release, 14 July 2026 (CSF tau reductions; plan for Phase 3)
  3. NeurologyLive. Topline Phase 2 CELIA results show diranersen misses primary end point, 14 May 2026
  4. Meglio M. Anti-tau agent diranersen slows cognitive decline across multiple endpoints. NeurologyLive, 14 July 2026 (arm sizes, scale-by-scale results, safety, completion rates)
  5. Alzheimer Europe. Biogen announces topline results from Phase II CELIA study, 14 May 2026
  6. Alzheimer’s Association. Statement on topline results from the CELIA Phase 2 clinical trial, 14 May 2026
  7. BioCentury: Biogen finds Alzheimer’s path for tau ASO despite mixed data, 15 May 2026 (low dose outperformed higher doses; unexplained)
  8. GeneOnline: Biogen’s Alzheimer’s tau drug splits experts (noise versus biology; ARIA; Leqembi comparison)
  9. Kartal Allen A. Biogen forges on with Alzheimer’s therapy development despite Phase II miss. Clinical Trials Arena, 15 May 2026
  10. BioSpace: Biogen takes tau Alzheimer’s therapy to Phase 3 despite mid-stage fail, May 2026
  11. BioSpace: Biogen validated the Alzheimer’s tau theory. CEO Viehbacher still isn’t betting the store (potential payout around 2030)
  12. Merlin Trader. Biogen and Ionis: diranersen lowers tau, but CELIA leaves clinical risk unresolved (investor blog; Phase 3 start 2027 and readout 2030 to 2031 are its estimates, not Biogen’s)
  13. Fierce Biotech: AAIC, Biogen bolsters hopes for Alzheimer’s pivotal push, July 2026 (Spinraza and Qalsody background)

Base rates: Alzheimer’s and drug development in general

  1. Cummings J. Lessons learned from Alzheimer disease: clinical trials with negative outcomes. Clinical and Translational Science 2018;11(2):147-152. doi:10.1111/cts.12491
  2. Scientific American: Why Alzheimer’s drugs keep failing (244 compounds, 413 trials, 99.6% failure versus 81% for cancer)
  3. Cleveland Clinic. New analysis highlights urgent need for Alzheimer’s disease drug development (original analysis in Alzheimer’s Research & Therapy, 2014)
  4. Guth MAS. The highest clinical trial failure rate for any disease in modern medicine. Published 27 August 2026 (review of 497 Phase 2 and 3 trials; the institutional-failure argument is the author’s)
  5. BIO, Informa Pharma Intelligence and QLS Advisors. Clinical development success rates and contributing factors 2011 to 2020 (28.9% Phase 2 to Phase 3; 7.9% overall likelihood of approval)
  6. US FDA. 22 case studies where Phase 2 and Phase 3 trials had divergent results

Precedents: tau drugs, antisense, aducanumab

  1. Imbimbo BP et al. Initial failures of anti-tau antibodies in Alzheimer’s disease are reminiscent of the amyloid-beta story. 2022. PMID 35799522
  2. Alzforum. Tau antibody trial results at CTAD (gosuranemab worsened ADAS-Cog decline; semorinemab mixed)
  3. Alzforum. Biogen shelves gosuranemab after negative Alzheimer’s trial
  4. Alzforum. TAURIEL Phase 2 data published (semorinemab, tilavonemab, gosuranemab, zagotenemab all fell short)
  5. Florian H et al. Tilavonemab in early Alzheimer’s disease: results from a Phase 2, randomized, double-blind study. Brain 2023;146(6):2275-2284
  6. Tominersen: Nature news report (2021); Scientific American; Fierce Biotech; Cureffi (Phase 1 lowered the protein about 40%)
  7. Aducanumab EMERGE and ENGAGE discordance analysis. Journal of Prevention of Alzheimer’s Disease 2023. springer.com
  8. Cummings J et al. Alzheimer’s disease drug development pipeline: 2025. Alzheimer’s & Dementia: Translational Research & Clinical Interventions 2025;11(2):e70098 (182 trials, 138 drugs)

Follow the serialisation. Free.

Energy on Mondays. The Debunk on Wednesdays. AI, food and health long reads on weekends. Delivered to your inbox.

Scroll to Top