The Alzheimer’s field was dealt another body blow yesterday with the announcement of a premature end to testing of Novartis/Amgen’s BACE inhibitor CNP520, aka umibecestat, in two Phase 2/3 trials. Part of the Alzheimer’s Prevention Initiative’s Generation program, the trials evaluated the ability of this BACE inhibitor to prevent Alzheimer’s dementia in cognitively unimpaired people who are at high risk for the disease. As a reason for the decision, announced on July 11, the sponsors cited worsening of cognitive function in the treatment groups as measured during a preplanned interim analysis.

Apparently, participants taking umibecestat declined on the RBANS cognitive composite, had more brain atrophy, and lost more weight than did people on placebo. The results appear broadly similar to those of Merck’s verubecestat, Janssen’s atabecestat, and Lilly’s lanabecestat (April 2019 news; Nov 2018 conference news; May 2019 conference news). 

“We were extremely disappointed to learn about the early worsening of some cognitive measures in our research participants, and the need to discontinue testing of the BACE1 inhibitor CNP520 (umibecestat) in our API Generation Program,” wrote Banner Alzheimer’s Institute directors Pierre Tariot, Eric Reiman, and Jessica Langbaum in a statement to Alzforum. “Our research participants are pioneers in Alzheimer’s prevention research, and their participation in these studies has already had a positive impact on how to conduct prevention trials. Trial data and biological samples from these studies will provide vitally important new information to help us better understand how to treat or even prevent AD.”

Other site leaders were equally dismayed at having to cut back another big Alzheimer’s program, noting that enthusiasm and compliance for the CNP520 trials had been high. Both Phase 2/3 trials included at least one treatment group that received CNP520 (Lopez et al., 2017). The Generation Study 1 had been on track to enroll 1,340 homozygous ApoE4 carriers aged 60 to 75. It included one group taking 50 mg CNP520 daily, one group treated with the active anti-Aβ vaccine CAD106, and a placebo group. The trial was slated to run for 60 to 96 months, with change on the Alzheimer's Prevention Initiative Composite Cognitive or time to diagnosis of mild cognitive impairment or AD as dual primary endpoints. The CAD106 portion of the API program will continue.

The Generation Study 2 aimed to enroll 2,000 people who either carried two copies of ApoE4, or who carried one copy and had evidence of amyloid deposition. This trial tested two daily doses—15mg and 50mg—of CNP520 compared with placebo.

The decision to end these two CNP520 studies leaves the field with only one large BACE inhibitor program, Eisai/Biogen’s two Phase 3 Mission trials of elenbecestat.

Eisai announced in May 2019 that the NIA-funded Alzheimer’s Clinical Trials Consortium had chosen elenbecestat for the upcoming A3 prevention trial. Whether Eisai and ACTC will stick with this decision in light of umibecestat’s failure is now in question. The announcement of a cognitive liability with this reportedly more BACE1-selective compound also complicates the choice of drug for the DIAN primary prevention trial. That said, Randy Bateman of Washington University noted that while BACE inhibitors do seem to have a class problem, each drug is different. Other indications also went through crisis periods where most investigational drugs in a given class failed—until one succeeded.

Umibecestat’s downfall raises pressure on the field to develop a greater variety of new treatments. Attention is shifting toward tau-based drugs, ASOs, and other genetic approaches.—Jessica Shugart and Gabrielle Strobel


  1. We were extremely disappointed to learn about the early worsening of some cognitive measures in our research participants and the need to discontinue testing of the BACE1 inhibitor CNP520 (umibecestat) in our Alzheimer’s Prevention Initiative (API) Generation Program. Our research participants are pioneers in Alzheimer’s prevention research, and their participation in these studies has already had a positive impact on how to conduct prevention trials. Trial data and biological samples from these studies will provide vitally important new information to help us better understand how to treat or even prevent Alzheimer’s disease (AD).

    We will be working with our trial partners from Novartis, Amgen, the National Institute on Aging (NIA), philanthropic partners, and other colleagues on plans to analyze, interpret, and report findings from these trials as soon as possible and clarify whether or not the observed worsening of some cognitive measures is reversible. We will also work on a plan to conduct appropriate follow-up assessments, and eventually share valuable resources of data and biological samples according to the Collaboration for Alzheimer’s Prevention (CAP) principles.

    At this time, it is not known whether all BACE inhibitors will have the same early cognitive effects observed in this study. It is also not known whether CNP520 or other BACE inhibitors could be safe and effective at lower doses.

    We realize that our valued research participants, colleagues, and other stakeholders have many questions, some of which may take more time to address fully. Some answers will require us to access, analyze, and report findings in a more comprehensive way. We and our industry colleagues hope to provide those answers as quickly as we can.

    We will continue to study CAD106, an active immunotherapy, in high-risk cognitively healthy older adults who have two copies of the APOE4 gene, the major genetic risk factor for Alzheimer’s disease, as part of Generation Study 1.

    API was established at the Banner Alzheimer’s Institute in Phoenix to accelerate the evaluation of promising prevention therapies and the timely discovery of treatments that work. It includes current and future prevention trials, enrollment registries, and other programs to support participation in prevention research. The API Autosomal Dominant AD Colombia Trial, API Generation Program, and other existing and planned API programs have helped launch a new era in Alzheimer’s prevention research. They have also helped to establish public-private partnerships and data sample-sharing programs to advance prevention research in ways that will have the greatest public benefit.

    Despite disappointing findings from this and other recent AD trials, we are encouraged by the scientific progress that continues to be made in AD research. We need to stay the course and do all we can find answers to Alzheimer’s as soon as possible.

    We are determined to do everything we can to learn from these studies, advance our trials and other programs, share data and samples following CAP principles, and to do so in ways that will benefit our research participants and everyone else involved in the fight against AD.

  2. Why do some BACE inhibitors worsen cognition? Because BACE1 regulates proliferation and neuronal differentiation of newborn cells in the adult hippocampus (Carney et a., 2018).

    If an intervention can be found which can be used together with a BACE inhibitor to promote neurogenesis and differetiation of newborn nerve cells, the result may be different.


    . Partial, Rather than Full, BACE1 Inhibition May Be a Better Therapeutic Strategy for Alzheimer's Disease Due to Effects of Complete Loss of BACE1 Activity on Adult Hippocampal Neurogenesis. eNeuro. 2018 Sep-Oct;5(5) Epub 2018 Oct 17 PubMed.

  3. The discontinuation of umibecestat in the API Generation studies is a terrible disappointment for the participants, investigators, and the entire field. It is fortunate that the data will be shared so that we can learn as much as possible from the trial experience. This is an essential responsibility: We must be thorough in our analysis to guide us forward.

    With at least four BACE inhibitor programs now discontinued, and with cognitive worsening (along with some behavioral symptoms and weight loss) observed in multiple trials, some are suggesting that we abandon this strategy just as we did γ-secretase inhibition. I think this would be a costly mistake.

    AD is a complex disorder, but it has a well-specified beginning: cleavage of APP to generate Aβ42. BACE inhibitors are orally bioavailable, brain-penetrant drugs that dramatically inhibit this cleavage. This is the strongest strategy for the primary prevention of AD (once we learn to identify those at high risk in middle age). It is also an attractive strategy for early intervention against AD pathophysiology, before the disease is complicated by extensive irreversible neurodegeneration and co-pathologies.

    But with four BACE inhibitors down, and only one presently standing, isn’t the message clear: This approach is too toxic to be viable? No. We have not adequately explored this strategy. The trials have generally used high doses to achieve a high degree of inhibition of BACE. The atabecestat data suggest that the cognitive toxicity is dose-related and reversible. In the verubecestat data, an adverse cognitive signal appears early along with a bump in atrophy on MRI, but neither effect increases with time. While some scientists have demonstrated neurotoxicity in vitro with BACE inhibitors that may explain the toxicity, much remains to be learned about the mechanism.

    Safety concerns will always be paramount. But we now know well how to monitor individuals in trials for cognitive toxicity; we know which measures are sensitive to this effect. We must continue our work to explore regimens with lower-potency BACE inhibition, obviously with the informed consent of participants and independent safety oversight. It remains quite plausible that BACE inhibition, at the right dose and disease stage, alone or in combination with other therapeutics, will have an important beneficial effect on disease progression.

  4. This is sad news—both for the patients and for drug discovery using BACE1 as a drug target. A major question that arises from the stopped CNP-520 trials is: What is the mechanistic cause of the cognitive decline observed for a few different BACE1-targeted inhibitors? Given that CNP520 appears more selective for BACE1 than BACE2, it was assumed that BACE2 inhibition is a potential cause of the cognitive worsening. Now that CNP520 shows the same cognitive worsening, this argument does not appear to be valid anymore, or the inhibitors need to be even more specific for BACE1 (over BACE2) than CNP520.

    It would be very helpful to have an easily measurable biomarker for in vivo BACE2 target engagement, which, however, is still lacking. Another issue: An increasing number of scientists and managers start questioning whether BACE1 is a reasonable AD drug target. While this concern is understandable, it is too early to give up on BACE1—in particular because there is no other drug currently available and because it may take years before other drug targets yield effective drugs. What can we do with regard to BACE1 …? Lower the dose of the drug and use it for prevention rather than for treatment of AD.

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Umibecestat
  2. Elenbecestat

News Citations

  1. Results from Verubecestat APECS Trial Published
  2. Bump in the Road or Disaster? BACE Inhibitors Worsen Cognition
  3. BACE Inhibitors: Postmortem on One, Live Updates on Two

Paper Citations

  1. . The Alzheimer's Prevention Initiative Generation Program: Evaluating CNP520 Efficacy in the Prevention of Alzheimer's Disease. J Prev Alzheimers Dis. 2017;4(4):242-246. PubMed.

External Citations

  1. Generation Study 1
  2. Generation Study 2
  3. announced in May 2019

Further Reading