The front-runner in the great race to block BACE now has a solid publication to its name. As reported November 2 in Science Translational Medicine, Merck’s verubecestat, a small- molecule inhibitor of β-site amyloid precursor protein cleaving enzyme 1 (BACE1), put the brakes on Aβ production and reduced the amyloidogenic peptide by more than 90 percent in the brain, cerebrospinal fluid, and blood of animal models. More meaningfully, it did so, too, in the cerebrospinal fluid and blood of healthy young to middle-aged people, and in people with Alzheimer’s disease. It did so without causing harmful side effects in people—at least in these short trials. The researchers, led by Matthew Kennedy, Andrew Stamford, and Mark Forman of Merck Research Laboratories in Kenilworth, New Jersey, presented some of this preclinical and Phase 1 trial data at prior meetings (see Apr 2012 conference newsJul 2012 conference news). The paper brings them all together into a peer-reviewed package, and includes a more detailed description of preclinical safety studies. Phase 3 clinical trials of the compound are well underway.

“It is beautiful data,” claimed Robert Vassar of Northwestern University in Chicago. Vassar, who was not involved in this work, was part of the group who cloned BACE1 and discovered its key role in generating Aβ. “The compound works exactly as expected both in animals and humans, and importantly, it works in AD patients,” Vassar told Alzforum.

Michael Wolfe of the University of Kansas in Lawrence commented that the data provide strong support for verubecestat as a safe drug with excellent pharmacokinetic properties. “Ongoing late-stage clinical trials with verubecestat, both in mild-to-moderate AD and in prodromal AD subjects, are therefore well-justified and should provide an important test of Aβ as a therapeutic target for AD,” he wrote.

The published data covers the story of verubecestat, aka MK-8931, from its discovery through Phase 1. The researchers report that the compound latches onto the BACE1 catalytic site with exquisite sensitivity, and unlike other failed BACE inhibitors that caused liver toxicity and/or damaged the retina in the eye, does not inhibit Cathepsin D, a related aspartyl protease. However, like all BACE1 inhibitors in trials, the compound does inhibit BACE1’s sibling, BACE2.

Catalytic Cut-Off.

Verbecestat wedges into the BACE1 substrate cleft, comingling with key catalytic aspartate residues and bringing the enzyme’s APP processing activity to a halt. [Image courtesy of Kennedy et al., Science Translational Medicine 2016.]

In cynomolgus monkeys, single oral doses of 3 mg/kg or 10 mg/kg verubecestat made a major dent in the concentrations of Aβ40 and the BACE1 cleavage product sAPPβ in the plasma, CSF, and cortex—achieving 65 to 81 percent reduction in these products 12 to 24 hours later. What’s more, CSF Aβ40, Aβ42, and sAPPβ all plunged by more than 90 percent when monkeys were given daily doses of 10, 30, or 100 mg/kg for nine months. Similar drops in these peptides also occurred in rats treated with the inhibitor. 

Though not described in the paper, Kennedy told Alzforum that in mice treated with verubecestat, a rise in unprocessed APP as well as in the α-secretase cleavage product, sAPPα, accompanied the reductions in BACE1 activity. Researchers have wondered what fate might befall APP if BACE cleavage faltered. Takaomi Saido of the Riken Brain Science Institute in Wako, Japan, expressed concern about the potential for amyloidogenic peptides arising from α-secretase cleavage, in particular Aβ17-40/42. “If MK-8931 causes overproduction and accumulation of Aβ17-40/42 in vivo, pathological consequences might emerge after some years of treatment,” he wrote.

However, so far no serious adverse events have been reported in ongoing trials. In the paper, Kennedy and colleagues detailed chronic toxicology studies carried out over six months in rats and nine months in monkeys, which had been exposed to verubecestat doses that were 43- and 54-fold higher, respectively, than the equivalent human doses used in current Phase 3 trials. Neither rodents nor monkeys developed overt behavioral symptoms, signs of liver toxicity, changes in blood glucose, defects in nerve myelination, or loss of eye pigmentation—possible side effects previously observed in BACE1 knockout animals or upon treatment with other BACE1 inhibitors.

The only obvious side effect occurred in mice and rabbits: The animals’ dark coats started going gray after a couple weeks on the drug, eventually transforming to silver as treatment continued (see related conference news). However, monkeys treated with the drug for nine months maintained their dark coat color. The lack of fur depigmentation in non-human primates suggests that this side effect may not be a concern for people, Stefan Lichtenthaler of the German Center for Neurodegenerative Diseases in Munich told Alzorum. He added that while this could remove the need for dermatology testing in future clinical trials, such monitoring should continue in ongoing trials, which are of longer duration than the studies reported in this paper. Kennedy told Alzforum that to date in the blinded data from ongoing Merck trials there have been very few reports of hair or skin hypopigmentation. One thousand patients have been on drug for at least a year, he added.

These extended toxicology studies were critical to reassure an anxious field—still reeling from past failures in both γ-secretase and BACE1 inhibitors—that the compound will be safe, commented Riqiang Yan of the Cleveland Clinic in Ohio. Yan has reported defects, including seizures and a slowing of nerve remyelination following injury, in BACE1 knockout mice. He added that while it remains to be seen whether some side effects will pop up in people after extended periods on the drug, he is optimistic about verubecestat and other BACE1 inhibitors in the pipeline.

The study also laid out the results of three human studies—single- and multiple-dose studies in healthy people, and a multiple-dose study in people with AD. In the first study, the researchers gave single doses of placebo or 20, 100, or 550 mg verubecestat to 24 healthy volunteers, aged 24-45, and observed maximal drops in CSF Aβ40 of 48, 77, and 93 percent, respectively, within 12 to 24 hours. Similar reductions occurred for Aβ42 and sAPPβ. For the second study, the researchers measured BACE1 products in 31 volunteers treated with daily oral doses of placebo or 10, 40, 150, or 250 mg of verubecestat for 14 days. Compared to baseline, maximal reductions of 66, 87, 94, and 95 percent in CSF Aβ40 were achieved for these respective doses. The story was similar in 30 AD patients who were treated for seven days with daily doses of 12, 40, or 60 mg: They had reductions in CSF Aβ40 of 70, 83, and 90 percent, respectively.

Pharmacokinetic data indicated the drug level increased dose-dependently in plasma and CSF following ingestion, and declined with a half-life of 14 to 24 hours. People with AD had slightly higher concentrations of the drug in their CSF and plasma than did people without the disease, a difference the researchers attributed to the disease process or to the older age of the AD patients.

Except for one volunteer who broke out in hives a few days after taking a single 550 mg dose of verubecestat, participants tolerated the drug well. Any adverse events were mild to moderate in nature, and occurred with similar frequency between drug and placebo groups. No serious adverse events or deaths occurred, and no one dropped out of any of the studies due to health complaints.

The researchers pooled their data from the three human studies to generate predictions about how patients would respond to the drug in the much larger Phase 3 trials. From these models, they settled on doses of 12 mg and 40 mg, which are predicted to drop Aβ40 by an average of 67 percent and 83 percent, respectively, in most AD patients.

Two Phase 3 trials of verubecestat are ongoing (see Dec 2012 newsDec 2013 newsOct 2016 conference news). The EPOCH trial has already enrolled 2,221 people with mild to moderate AD, while the APECS trial is currently recruiting an estimated 1,500 people with prodromal AD as diagnosed by positive amyloid PET scans. Treatment in the trials will run for 18 months and two years, respectively. The EPOCH trial results are expected in the summer of 2017, while those from APECS are due in 2019.

These trials will provide an opportunity to test the amyloid hypothesis, Kennedy told Alzforum. While it is unclear what effect BACE1 inhibition will have on Aβ plaques that already riddle the brains of people with AD, Kennedy hypothesized that all species of Aβ would drop to some extent. “Dramatically shutting off production of Aβ could allow the brain’s natural clearance mechanisms to take hold, and we may then see reduction of plaques,” he said.

Will halting Aβ production in people who already carry a substantial Aβ burden curb the symptoms of disease? The field will have to wait and see, but Vassar is hopeful. He pointed out that dystrophic neurites surrounding plaques seem to heal after plaques are removed (see Oct 2016 conference newsJan 2005 news). Still, he and most other researchers acknowledge the alternative possibility—that the drugs may only work in the presymptomatic stages of AD. This makes the safety profile of verubecestat all the more important, Yan said. “We need people to take this drug as they would a vitamin, so they can take it early enough to stop the factors that initiate the disease,” Yan said. Some scientists in the field conceptualize a BACE inhibitor as a sort of “maintenance” drug that would keep Aβ levels in check after a one-time course of immunotherapy has removed the bulk of the pre-existing amyloid deposits.—Jessica Shugart


  1. It is great for the whole field that the Merck group publishes their preclinical and Phase I data. The results look very promising for the ongoing Phase 3 trials. A very interesting finding is that a potential side effect—the fur depigmentation—only occurred in rodents, but not in the non-human primates, suggesting that this side effect (resulting from BACE2 inhibition) may not be a concern for humans. As such, the regular dermatology testing may no longer be required for future clinical trials. However, this needs to be carefully monitored in the ongoing clinical trials, which will also provide other key outcomes, such as insights into how safe BACE inhibitors are upon long-term treatment and how efficiently they slow cognitive decline in patients. The latter point will inform us about whether BACE inhibition can still be effective at the onset of symptoms or whether they are likely to be even more effective as a preventive treatment. The recent data on aducanumab—although acting through a different mechanism—gives hope that amyloid reduction at the onset of symptoms may still be effective.

  2. This study demonstrates that verubecestat is a potent, selective, brain-penetrating BACE1 inhibitor with excellent pharmacokinetic properties. The compound also appears to be safe by a number of measures, avoiding some of the toxic effects seen with genetic knockout or other pharmacological inhibitors. Ongoing late-stage clinical trials with verubecestat, both in mild to moderate AD and in prodromal AD subjects, are therefore well-justified and should provide an important test of Aβ as a therapeutic target for AD. The primary concern, however, is that even the prodromal phase may be too late. Amyloid deposition can begin up to 25 years before the onset of symptoms, and it is not yet clear if and when the disease process becomes Aβ-independent and primarily tau-driven. Thus, there are no guarantees that verubecestat will show clear efficacy in these trials.

  3. First of all, I would like to express my respect for the tremendous amount of work performed by Kennedy et al. Protease inhibitors have proven to be effective and safe medications. ACE inhibitors for hypertension are a good example, although they are being replaced by angiotensin receptor blockers.

    MK-8931 potently inhibits both BACE1 and BACE2 without major side effects. The authors also show that the inhibitor reduces sAPPβ in vivo (in animals and humans). Why do BACE inhibitors show much fewer side effects than γ-secretase inhibitors? This is probably because there exists essentially no alternative enzyme for γ-secretase, while at least one, α-secretase, can substitute for the β-secretases (BACE1 and BACE2).

    The above explains why the selective BACE inhibitors do not exhibit major side effects up to a few months. My only concern is about the α-secretase product(s), Aβ17-40/42, which is extremely amyloidogenic in vitro. I am aware, because I have synthetized these peptides by my own hand. Normally, Aβ17-40/42 is not abundant in AD brain (Saido et al., 1996; Iwatsubo et al., 1996). Although the reason for this remains elusive, it is possible that Aβ17-40/42 is easier to catabolize than Aβ1-40/42 in vivo.

    Quantification of sAPPα in the paper would have given an estimate of Aβ17-40/42 production. If MK-8931 causes overproduction and accumulation of Aβ17-40/42 in vivo, pathological consequences might emerge after some years of treatment in people. I suggest that the authors examine the effect of MK-8931 on our single NL-F App knock-in mice (Saito et al.,  2014), which show extensive pathology at six months without overexpressing APP. We probably are the only ones who possess Aβ17-40/42-specific antibodies, extremely difficult to generate, and are willing to help in this context.


    . Amino- and carboxyl-terminal heterogeneity of beta-amyloid peptides deposited in human brain. Neurosci Lett. 1996 Sep 13;215(3):173-6. PubMed.

    . Full-length amyloid-beta (1-42(43)) and amino-terminally modified and truncated amyloid-beta 42(43) deposit in diffuse plaques. Am J Pathol. 1996 Dec;149(6):1823-30. PubMed.

    . Single App knock-in mouse models of Alzheimer's disease. Nat Neurosci. 2014 May;17(5):661-3. Epub 2014 Apr 13 PubMed.

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Verubecestat

News Citations

  1. Keystone: Therapies Around ApoE—Has Their Time Come?
  2. Wave of New BACE Inhibitors Heading to Phase 2
  3. Will Next-Gen BACE Inhibitors Dodge Side Effects?
  4. Merck Launches Largest Trial of BACE Inhibitor in AD
  5. Merck BACE Inhibitor Clears a Safety Hurdle, Gets New Trial
  6. At 2nd Kloster Seeon Meeting, Renewed Optimism for Targeting BACE1
  7. Does BACE Drive Neurites into Dystrophy, Shorting Circuits?
  8. Window to the Brain Shows Dystrophic Neurites Shrinking

External Citations

  1. EPOCH trial
  2. APECS trial

Further Reading


  1. . Discovery of an Orally Available, Brain Penetrant BACE1 Inhibitor that Affords Robust CNS Aβ Reduction. ACS Med Chem Lett. 2012 Nov 8;3(11):897-902. PubMed.
  2. . Stepping closer to treating Alzheimer's disease patients with BACE1 inhibitor drugs. Transl Neurodegener. 2016;5:13. Epub 2016 Jul 14 PubMed.
  3. . Inhibiting BACE1 to reverse synaptic dysfunctions in Alzheimer's disease. Neurosci Biobehav Rev. 2016 Jun;65:326-40. Epub 2016 Apr 1 PubMed.

Primary Papers

  1. . The BACE1 inhibitor verubecestat (MK-8931) reduces CNS β-amyloid in animal models and in Alzheimer's disease patients. Sci Transl Med. 2016 Nov 2;8(363):363ra150. PubMed.