The biotech company BioArctic Neuroscience announced positive topline results at the 18-month mark of a Phase 2b trial of BAN2401. This anti-Aβ protofibril immunotherapy was developed originally by Lars Lannfelt at Sweden’s Uppsala University following his discovery of the “Arctic” APP mutation E693G in a family in Swedish Lapland. The current trial enrolled people with early Alzheimer’s disease, and the highest antibody dose reportedly slowed cognitive decline and reversed accumulation of amyloid as seen on positron emission tomography (PET) scans. It is the largest proof-of-concept study of a disease-modifying drug in AD.

  • Anti-Aβ protofibril antibody BAN2401 shows positive results in Phase 2b.
  • It slowed cognitive decline in MCI and mild AD during 18 months.
  • At the highest dose, the antibody appears to clear Aβ plaques.

“I’m encouraged about these preliminary findings,” said Paul Aisen, University of Southern California, La Jolla. He cautioned that while a treatment benefit is plausible, researchers need to see overall study results, and subject them to multiple analyses to avoid a false positive result, before they can be sure the effect is real. “I’m hoping that these results will hold up under scrutiny and will reinforce the attention on the amyloid hypothesis in the face of so many negative studies,” Aisen said.

BAN2401—which binds selectively to large, soluble Aβ protofibrils—was developed by Stockholm-based BioArctic, then licensed to Eisai, and subsequently Biogen acquired a stake through an AD partnership with Eisai. The Phase 2b trial tested five dose regimens in people who had mild cognitive impairment due to AD or mild AD and who had evidence of brain amyloid pathology by cerebrospinal fluid analysis or PET. Patients received a 60-minute infusion of 5 or 10 mg/kg once a month, or 2.5, 5, or 10 mg/kg twice a month.

After years of discussion about the challenges of clinical trials in AD research, researchers led by Andy Satlin, who led AD research at Eisai but has since left the company, took a stab at innovating on both design and outcome measures for early AD progression. For the first time in a large AD trial, researchers employed a Bayesian statistical model, where they tried to enrich enrollment of patients into the more effective dose regimens based on frequent interim analyses (Satlin et al., 2016). The hope was that this approach would yield a result sooner, with fewer than 800 patients needing to be enrolled.

The primary outcome was change at 12 months on the ADCOMS, which combines all six items of the CDR sum of boxes, four items of the ADAS-Cog, and two items of the MMSE. This composite was developed by Veronika Logovinsky and colleagues at Eisai as a more sensitive measure to track progression at early disease stages (Wang et al., 2016). An 80 percent chance of a 25 percent slowing of cognitive decline would have triggered a Phase 3 trial.

This did not pan out. By 12 months, the trial did not meet its primary endpoint or futility criteria, indicating the antibody neither had a large effect size nor was clearly ineffective (Dec 2017 news). Hence the trial continued to 18 months, eventually enrolling 856 participants. 18-month amyloid PET was a secondary outcome.

For this time point, the researchers applied a prespecified, conventional statistical analysis. They now report a statistically significant reduction in cognitive decline at six, 12, and 18 months for people receiving 10 mg/kg of BAN2401 twice monthly. In addition, at 18 months patients on this dose reportedly had a significant reduction in standard uptake value ratio for the amyloid PET ligand, and some converted from amyloid-positive to -negative.

“It’s encouraging that there was a significant effect in the correct direction, both for the clinical endpoint and the biomarker,” said Lawrence Honig, Columbia University, New York. While this is potentially promising, it is worth remembering that this was a Phase 2 study, he added. Nevertheless, Honig was optimistic. “To me, it follows a trend that we are getting closer to potentially having an agent that shows some efficacy,” he said.

Overall, the treatment appeared safe. Patients most commonly reported mild to moderate infusion-related reactions, while MRI scans showed ARIA-E (edema) in fewer than 15 percent.

The researchers will present trial data at AAIC this month. Kramer said Eisai/Biogen plan to start a Phase 3 trial of BAN2401 during this fiscal year. He noted that this is the first large study to find both a reduction in cognitive decline and reduction in amyloid plaque in a substantial number of patients. “That correlation is very important,” Kramer said. “There is a biomarker change along with a clinical one.” The press release did not disclose how many of the 856 participants received the 10 mg/kg dose.

Since these are topline, descriptive results devoid of methods or numbers, one cannot interpret these findings, cautioned Lon Schneider, University of Southern California, Los Angeles. “There’s no reason to doubt that the cognitive outcomes and PET data are statistically significant in the models they used, but scientists will have to wait for formal presentations—hard data on the effect sizes, sample sizes, and p-values—to make inferences,” he told Alzforum. “This is especially true because it’s a complex Bayesian study, not just a simple parallel group, dose-ranging design.”

For company press releases, see BioArctic and Biogen.—Gwyneth Dickey Zakaib


  1. As a note of caution: A press announcement is not a scientific report. There are no protocol, methods, and data presented from which inferences can be made. It’s carefully written for investors betting on prices, to fulfill compliance obligations, and doesn’t say too much or overcommit, and uses affirmative quotes to sound hefty. It is not clear how this announcement on secondary outcomes sorts out with the December release on the primary outcomes.

    What is clear is that the BAN2401 antibody does what it is designed to do, knock down Aβ, and so could very well be effective. This is an achievement, itself.

    We’ve been through the drill many times, trying to divine the meaning of outcomes when none was actually presented. Biogen, Eisai, and BioArctic, however, have the perfect opportunity for a full scientific presentation at the Chicago AAIC in July.

    Purely as a historical note, this is the 10-year anniversary of Elan’s and Wyeth’s June 17, 2008, press release, “Top line results from phase 2 clinical trial of bapineuzumab for Alzheimer’s disease,” which was the first Aβ antibody.

    Their bullet points were:

    • Safety and Efficacy Findings Support Design of Phase 3 Program
    • Primary Efficacy Endpoints in Overall Study Population Not Statistically Significant
    • Statistically Significant and Clinically Meaningful Benefits Seen in ApoE4 Non-Carriers
    • Overall Results Support Prior Decision To Initiate Phase 3
    • Detailed Data Presentation At ICAD July 29, 2008

    Kelly Martin, Elan’s president and CEO, said at the time:

    “The preliminary analyses of the Phase 2 study are a continued validation of the amyloid approach to Alzheimer's disease and an important milestone in our companies’ ongoing commitment to bring new treatment options to patients. These results clinically support our decision to move into Phase 3 last year."

    Sid Gilman’s presentation at the July 2008 Chicago AAIC was to an overflowing, jostling crowd. A more detailed, somewhat different report was published subsequently (Salloway et al., 2009). 

    Déjà vu?


    . A phase 2 multiple ascending dose trial of bapineuzumab in mild to moderate Alzheimer disease. Neurology. 2009 Dec 15;73(24):2061-70. PubMed.

  2. The results of the 18-month study with BAN2401 in 856 subjects with prodromal AD or mild AD has raised a lot of interest and prudent hope. We know very little about the trial and will learn more in a couple of weeks at AAIC. In the meantime, will can make some observations:

    1) The protocol and the statistical methodology of the study were fully published in 2016 (Satlin et al., 2016).

    2) The rationale and performance of the primary outcome of efficacy of the study (ADCOMS) was published in 2016 (Wang et al., 2016). The performance of the scale was tested also in biomarker-positive prodromal AD subjects and in subjects with mild AD.

    3) According to, secondary outcomes of the study include hippocampal volume and brain amyloid levels. Apparently, no other cognitive or clinical or functional scales were planned. On the other hand, ADCOMS includes all six items of the CDR-SB (memory, orientation, judgment and problem-solving, community affairs, home and hobbies, personal care), four items of the ADAS-Cog (delayed word recall, orientation, word recognition, word-finding difficulty), and two items of the MMSE (orientation to time, constructional praxis).

    4) Since the study was carried out according to an adaptive design that allocated subjects to the most favorable treatment group while ongoing, we should expect that the size of the 10 mg biweekly group and the placebo group is the most numerous, maybe around 200. A hypothetical example could be the following: 203, 100, 100, 125, 125, 203, could be the respective sample sizes in the placebo, 2.5 mg biweekly, 5 mg monthly, 5 mg biweekly, 10 mg monthly, 10 mg biweekly groups.

    5) Hypothesizing group size around 200 subjects in the 10 mg biweekly group and the placebo group, the difference between the two groups could be in the range of 0.005 to 0.050 points for the ADCOMS. This difference should equate to a delta range of −0.18 to 1.59 for the ADAS-Cog, −0.94 to −0.07 for the MMSE, and −0.03 to 0.36 for the CDR-SB (Wang et al., 2016).

    6) The crucial missing information is the number of dropouts in the different treatment groups and their potential bias on the outcome of the study. At 18 months, dropouts could be considerable. Importantly, we need to know the distribution of the staging at baseline of the dropouts, i.e., whether they had prodromal AD or mild AD, in the different treatment groups. In other words, we need to understand if there is a dose-dependent loss of subjects based on their staging at baseline. If indeed BAN2401’s adverse events (i.e., ARIA and other negative behavioral or systemic effects), dose-dependently affect the abandoning of the study by participants most severely impaired at baseline, then we might have a dose-dependent enrichment with “benign” patients in the treatment groups, especially in the 10 mg biweekly group.


    . Design of a Bayesian adaptive phase 2 proof-of-concept trial for BAN2401, a putative disease-modifying monoclonal antibody for the treatment of Alzheimer's disease. Alzheimers Dement (N Y). 2016 Jan;2(1):1-12. Epub 2016 Feb 4 PubMed.

    . ADCOMS: a composite clinical outcome for prodromal Alzheimer's disease trials. J Neurol Neurosurg Psychiatry. 2016 Sep;87(9):993-9. Epub 2016 Mar 23 PubMed.

  3. I'd like to thank Lon Schneider for his thoughtful and informative comments. All I would add is, "Past performance is no guarantee of future failure."

Make a Comment

To make a comment you must login or register.


Mutations Citations

  1. APP E693G (Arctic)

News Citations

  1. No Man’s Land: Neither Early Success nor Failure for BAN2401

Paper Citations

  1. . Design of a Bayesian adaptive phase 2 proof-of-concept trial for BAN2401, a putative disease-modifying monoclonal antibody for the treatment of Alzheimer's disease. Alzheimers Dement (N Y). 2016 Jan;2(1):1-12. Epub 2016 Feb 4 PubMed.
  2. . ADCOMS: a composite clinical outcome for prodromal Alzheimer's disease trials. J Neurol Neurosurg Psychiatry. 2016 Sep;87(9):993-9. Epub 2016 Mar 23 PubMed.

Other Citations

  1. BAN2401

External Citations

  1. Phase 2b
  2. BioArctic 

Further Reading