Eli Lilly and AstraZeneca today announced that they will end two ongoing Phase 3 trials of lanabecestat, a small-molecule BACE inhibitor the two companies have been developing jointly since 2014. According to the program’s independent data monitoring committee, a futility analysis showed lanabecestat would most likely miss the trial’s efficacy endpoint. The companies decided to scrap both the AMARANTH trial of 2,202 people with early AD and the DAYBREAK-ALZ trial of 1,899 people with mild AD dementia.

“This is indeed bad news. Together with other recently terminated trials, it emphasizes the view that BACE inhibitors are unlikely to be effective when used to treat symptomatic individuals,” said Stefan Lichtenthaler at the German Center for Neurodegenerative Diseases (DZNE) in Munich (see full comment below).

The setback comes after liver toxicity sank Janssen’s atabecestat (May 2018 news). Merck pulled the plug on verubecestat in prodromal AD when an interim analysis showed no glimmer of efficacy (Feb 2018 news), and Lilly, Roche, Pfizer, and Boehringer Ingelheim all have axed other BACE inhibitors in Phase 1.

Researchers emphasize that it is important to evaluate these BACE inhibitors in symptomatic patients to gather evidence of whether and at what stages of Alzheimer's long disease process they work. At the same time, they say, the approach of reducing Aβ generation stands the greatest chance of success during the presymptomatic 10–15 years of amyloid plaque buildup in the brain. “BACE inhibitors are likely to be more effective when used as a preventive approach,” said Lichtenthaler.

“As with Merck’s verubecestat, it appears that late-stage, i.e. symptomatic, AD is too late for robust benefit from a BACE inhibitor,” Paul Aisen, University of Southern California, La Jolla, wrote to Alzforum. “This does not mean that BACE inhibitors will not find a place in the prevention or very early treatment of AD. Some good news is that lanabecestat—like verubecestat but in contrast to atabecestat—was apparently well-tolerated without major safety concerns. This is particularly important for very early intervention before symptoms,” Aisen wrote (see full comment below). 

Does today’s news mean lanabecestat is dead? Not quite. For the time being, Lilly and AstraZeneca are maintaining the alliance they had forged around this drug. They will close the two trials and analyze their data, according to Lilly spokesperson Nicole Hebert, but have not yet decided whether to launch prevention trials with this compound.

Lilly has a different BACE inhibitor in the clinic. LY3202626 is currently being studied together with Lilly’s anti-plaque N3pG-Aβ monoclonal antibody in the first combination trial of two investigational drugs in the AD field. This Phase 2 study enrolls people with memory loss but not yet dementia, who have both a positive amyloid and tau PET scan. The study will continue, Hebert confirmed, indicating that Lilly is not retreating from BACE inhibitors as a class. Other researchers in the field praise this combination approach, but caution that the patients in this trial also have subtle symptoms, hence have likely had amyloid plaques for years.

Two other BACE inhibitors are still active in single-drug trials. Last week, Eisai and Biogen reported that elenbecestat was safe in a small Phase 2 trial (June 2018 news), though its Phase 3 MISSION AD program of two large trials enrolls people with early symptomatic AD.

Novartis, Amgen, and the Banner Alzheimer’s Institute in Phoenix are going all-out to evaluate CNP520 in a genetically defined population. “My […] colleagues and I remain excited about our ongoing collaboration […] to evaluate […] CNP520 in Alzheimer’s Prevention Initiative (API) Generation Studies 1 and 2, including a proportion of cognitively unimpaired APOE4 homozygotes who have not yet demonstrated significant amyloid plaque burden before the treatment has started,” Eric Reiman of Banner wrote to Alzforum (see full comment below).

“A BACE inhibitor might still be the best pharmacological choice for prevention of patients at risk,” agreed Jochen Herms of the DZNE in Munich. “If long-term treatment with BACE inhibitors is without side effects, then I would take them were I at risk. So these studies, even though they failed in symptomatic AD, are very important for future patients, because they allow the field to get to know the possible side effects of BACE inhibition” (see full comment below).—Gabrielle Strobel


  1. This is indeed very disappointing. How to explain?

    Based on what our lab has seen in mice treated with BACE inhibitors, BACE inhibition very potently inhibits formation of plaques—probably corresponding to about 15 years prior to symptom onset in humans. Here the “physiological” function of BACE in Aβ production is inhibited. This function has mainly been considered in estimating the BACE inhibitor doses needed for the treatment of AD patients.

    In symptomatic AD, however, most of the Aß that might be relevant for progressive cognitive decline is produced within axonal dystrophies at the border of neuritic plaques. Here, BACE accumulates dramatically and may not be sufficiently inhibited with the BACE inhibitor doses that have been used in all these clinical trials. At least in AD mice treated with BACE inhibitors, new plaques still appear close to existing neuritic plaques and alter synapses. 


    . BACE1 inhibition more effectively suppresses initiation than progression of β-amyloid pathology. Acta Neuropathol. 2018 May;135(5):695-710. Epub 2018 Jan 11 PubMed.

  2. This is indeed bad news. Together with other recently terminated trials, it emphasizes the view that BACE inhibitors are unlikely to be effective when used to treat symptomatic individuals. They are likely to be more effective when used as a preventive approach. In my view even the Eisai/Biogen inhibitor elenbecestat, which appeared to slow cognitive decline in a Phase 2 study, will be more effective when used in a prevention paradigm.

    At the same time, this trial termination must not be used to argue that BACE1 is too difficult a target for drug development. So far, all BACE trials really only tested for side effects of the individual drugs, not for efficacy in a prevention trial. Thus, it is important to save BACE1 as a drug target and to retest the safe BACE inhibitors in prevention trials. 

  3. This is another disappointment for the field. The Phase 3 lanabecestat trials were both discontinued based on futility analyses. As with Merck’s verubecestat, it appears that late-stage, i.e. symptomatic, AD is too late for robust benefit from a BACE inhibitor.

    This does not mean that BACE inhibitors will not find a place in the prevention or very early treatment of AD. Some good news is that lanabecestat, like verubecestat, but in contrast to atabecestat, was apparently well-tolerated without major safety concerns. This is particularly important for very early intervention before symptoms. I continue to think that BACE inhibitors will be effective in the primary prevention of AD, i.e., treating the metabolic derangement before amyloid accumulation in brain, and may well be effective as very early treatment, alone or in combination with an amyloid-reducing antibody.

    While we must continue to explore all plausible therapeutic strategies, the amyloid hypothesis remains the best-supported theory and is still likely to yield effective treatments.

  4. I was disappointed to learn this news, but the findings are not entirely surprising. Merck’s BACE inhibitor verubecestat demonstrated no significant effects in clinically affected patients, and we have long suspected that this particular drug class might need to be initiated before the onset of symptoms, by which time amyloid plaques are already extensive.

    There is both the opportunity and a critical need to put safe and well-tolerated BACE inhibitors—and the amyloid hypothesis itself—to the test. For instance, BACE inhibitors need to be evaluated in cognitively unimpaired persons at genetic risk for AD, including those who do not yet have extensive amyloid plaque burden. In addition, a BACE inhibitor could be evaluated in combination with plaque-busting immune therapies in later preclinical and early clinical stages of the disease, as Lilly has begun to do in its proof-of-concept Trailblazer-ALZ trial.

    My Banner Alzheimer’s Institute colleagues and I remain excited about our ongoing collaboration with Novartis and Amgen to evaluate the BACE inhibitor CNP520 in Alzheimer’s Prevention Initiative (API) Generation Studies 1 and 2, including a proportion of cognitively unimpaired APOE4 homozygotes who have not yet demonstrated significant amyloid plaque burden before the treatment has started. 

    In addition to providing more definitive tests of the amyloid hypothesis, the field needs to clarify disease mechanisms, establish additional therapeutic targets, and develop a more diversified portfolio of promising treatments that could be used alone or in combination to treat and prevent AD. These and other efforts are critically important in the scientific fight against this devastating disease, and we need to see them through. I am extremely grateful to our colleagues in academia and industry for persevering in the face of recent trials and tribulations.

  5. The Eli Lilly/AstraZeneca decision to discontinue the global Phase 3 trials of lanabacestat was a disappointment, but, in view of similar decisions by Merck, perhaps not a complete surprise. Both of the lanabecestat programs enrolled persons with early Alzheimer’s disease and/or mild Alzheimer’s disease dementia. However, our field has yet to answer the very different and critically important questions of whether BACE inhibitor treatment can delay, or even prevent, the emergence of clinical manifestations of Alzheimer’s disease in cognitively healthy people who are at increased risk of the disease by virtue of age and genetics. These are the very people being enrolled in the two international placebo-controlled trials of the BACE inhibitor CNP520 that is currently being tested in the Alzheimer’s Prevention Initiative Generation Program, a collaboration among Novartis, Amgen, and Banner Alzheimer’s Institute. Additionally, our field is still looking to answer safety-related questions, and at this time we have no reason to believe the previously reported safety concerns surrounding Janssen’s atabacestat are a general property of all BACE inhibitors, nor that they will occur with CNP520. Answering these questions of efficacy and safety is of tremendous public health significance. For this reason, our commitment to this Alzheimer’s prevention research program remains steadfast even in the face of the recent announcements.

  6. The reported futility of the lanabecestat AMARANTH and DAYBREAK-ALZ studies is of course a huge disappointment for AD patients and their families. While one possible reason is that even mild symptoms may be too late in the disease continuum for a BACE inhibitor to be efficacious, firm conclusions cannot be made until the actual data are presented. The AMARANTH study included people with mild cognitive impairment or mild dementia due to Alzheimer’s disease or Alzheimer’s pathologic change, so these patients were earlier in the AD continuum than those in the Merck EPOCH study.   

    The Phase 2 results announced recently for elenbecestat were promising and the verubecestat EPOCH results do provide useful information regarding the mechanism.  Both elenbecestat and verubecestat cause robust inhibition of the BACE enzyme with substantial reductions in CSF Aβ. Further, in the verubecestat EPOCH study, small but statistically significant reductions in SUVRs were reported using the PET ligand flutemetamol. In the 78 weeks of the trial, essentially no change was seen for patients treated with placebo, while SUVR values declined for patients taking active treatment. 

    In the press release from Eisai statistically significant “differences” between active treatment with elenbecestat and placebo were described; however, the recent Alzforum article noted Eisai researchers said that “... amyloid reductions were larger than those in Merck’s verubecestat study, which reported declines of 0.02 to 0.04 in standard uptake value ratios.” If plaque was reduced using both verubecestat and elenbecestat, that would indicate robust target engagement for both molecules such that not only was CSF Aβ decreased but also that equilibria shifted enough to cause a downstream effect on plaque.  

    In additional to the lack of clinical efficacy, a worrisome result from the EPOCH study was that hippocampal atrophy was greater for both the 12 mg (-5.6 percent) and 40 mg (-5.7 percent) groups than for the placebo group (-5.0 percent). Why would a robust effect on Aβ result in greater hippocampal atrophy? While not known to have as many substrates as γ-secretase, BACE does cleave transmembrane proteins other than APP, e.g. neuregulin 1. In a Phase 3 study, the γ-secretase inhibitor semagacestat was found to cause unexpected cognitive worsening, and very similar results were seen in a Phase 2 study using the BMS γ-secretase inhibitor avagacestat. The similarities between results for semagacestat and avagacestat clearly suggested that simple inhibition of γ-secretase would not provide positive results. The reductions in hippocampal volume in the EPOCH study may or may not be seen in the AMARANTH study, but until the data are presented, this possibility should be considered. 

    In the worst-case scenario, a conclusion is made that BACE inhibition is not a viable mechanism for AD treatment due to untoward effects. In that event, this conclusion should not be taken as weighing on the Aβ/amyloid hypothesis more broadly. Even if both γ- and β-secretase inhibitors generally are difficult to develop for AD, monoclonal antibodies remain a viable option. Antibodies are being studied that bind primarily to soluble monomers, primarily to deposited plaques, and more recently to Aβ oligomers.   A variety of these antibodies with different binding epitopes and characteristics should be studied before making firm conclusions about the Aβ/amyloid hypothesis broadly. 

    But for the time being, we will need to wait to see actual data from the lanabecestat studies before making conclusions about BACE generally.  The team at Lilly is no doubt working to present those findings as soon as possible.

  7. Just a note of appreciation and support for this BACE inhibitor prevention trial. We would all like an AD pharmacologic approach that would stop disease progression even after significant amyloid accumulation. This would be similar to attacking LDL with a statin and PCSK9 antibody to stop and even reverse atherosclerosis. However, these current results indicate that if a BACE inhibitor will impact AD, treatment must be started earlier.

    Writing as a bench scientist, I expect that there is temptation to move to another target, and so kudos for pushing BACE through to the prevention format. If 20 years from now people are getting their AD risk assessment when they turn 55 so that they can begin appropriate medicines, we’ll look back and wonder why we didn’t move to BACE inhibitor prevention trials sooner. 

  8. The continuing failure of Aβ-based therapies in AD and MCI, such as the recent results of the Phase 3 lanabecestat trials, are leading to the abandonment of these approaches and are posing a serious challenge to the Aβ/amyloid hypothesis. However, in preclinical experiments, removal of brain amyloid plaques can result in an accentuation of neuronal hyperactivity in areas such as the cortex and hippocampus (Busche et al., 2015), which might cause a worsening of cognition including memory.

    Thus, before abandoning these potentially effective disease-modifying treatments, it is imperative that the effects of these drugs on neuronal hyperactivity (Pomara et al., 1986) be studied in AD and MCI, to determine if this treatable or preventable complication emerges in conjunction with these treatments and if so, whether it contributes to a worsening or to a lack of significant improvements in cognition which have been observed in clinical trials in these patient populations.


    . Decreased amyloid-β and increased neuronal hyperactivity by immunotherapy in Alzheimer's models. Nat Neurosci. 2015 Dec;18(12):1725-7. Epub 2015 Nov 9 PubMed.

    . Prospective strategies for cholinergic interventions in Alzheimer's disease. Prog Neuropsychopharmacol Biol Psychiatry. 1986;10(3-5):553-69. PubMed.

  9. Together with the trial data on verubecestat, these latest data with lanabecestat underscore the dangers of assuming a linear simple hypothesis based on genetics and a host of preclinical animal models. First, amyloid has a complex non-linear biology that has often been overlooked. Secondly, AD patients in clinical trials present with a large variability in β-amyloid load. To assess the differential biology of short and long forms of β-amyloid on glutamate and nicotinic neurotransmission, we used a mechanism-based and ADAS-Cog calibrated Quantitative Systems Pharmacology computer model (Geerts et al., 2018). We showed that the introduction of a beneficial effect of shorter Aβ isoforms on glutamate neurotransmission, originally reported in preclinical studies, is absolutely necessary to explain three clinical data sets in human AD patients. This leads to the conclusion that the clinical effect of amyloid modulators that do not discriminate between short and long forms of the Aβ peptide, is very dependent upon baseline amyloid load, with subjects at low baseline experiencing a cognitive worsening and subjects at high baseline improving but with a maximal effect very similar to the effect of current approved AD medications. The model also suggests that interventions that spare shorter Aβ forms and reduce the toxic Aβ42 could have a modest clinical benefit. I acknowledge that this comment falls completely outside of the current thinking in the AD community, so it would be great if we could test these predictions in a post hoc analysis of the failed trials. This is extremely important, because if these assumptions hold true, indiscriminate amyloid reduction in cognitively normal subjects with low or zero amyloid will likely worsen outcome, therefore questioning the rationale to start treatment of healthy elderly subjects with low amyloid levels “before the pathology starts.”


    . Impact of amyloid-beta changes on cognitive outcomes in Alzheimer's disease: analysis of clinical trials using a quantitative systems pharmacology model. Alzheimers Res Ther. 2018 Feb 2;10(1):14. PubMed.

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Lanabecestat
  2. Atabecestat
  3. Verubecestat
  4. LY3202626
  5. Donanemab
  6. Elenbecestat
  7. Umibecestat

News Citations

  1. Liver Tox Ends Janssen BACE Program
  2. Merck Axes Verubecestat for Prodromal AD, Researchers Say ‘Go Earlier’
  3. Hanging in There: Another BACE Inhibitor Still Safe in Phase 2

Further Reading

No Available Further Reading