Yesterday, Roche announced a halt to the Phase 3 CREAD 1 and CREAD 2 trials of the anti-amyloid antibody crenezumab. According to a company press release, an interim analysis found the trial was unlikely to reach its primary endpoint of slowing decline on the CDR-SB. No safety issues cropped up. AC Immune developed the antibody and partners with Roche and Genentech. Roche noted that the Alzheimer Prevention Initiative Phase 2 trial for familial AD will continue with crenezumab. The company plans to present full findings at a to-be-determined scientific conference.

“This is certainly disappointing,” Dave Morgan at Michigan State University in Grand Rapids wrote to Alzforum. He noted that crenezumab has unique features, including a low rate of side effects and a purported ability to inhibit the aggregation of Aβ, that initially suggested it might work well (full comment below). However, other researchers noted crenezumab’s limitations. “This is not unexpected, because the Phase 2 studies were not positive,” said Lon Schneider at the University of Southern California, Los Angeles.

Crenezumab has high affinity for oligomeric and fibrillar forms of Aβ, but also recognizes monomers. Notably, Aβ oligomer levels in cerebrospinal fluid fell by half in AD patients who took a low dose of the drug in Phase 2 trials (July 2018 conference news). Despite this, those ABBY and BLAZE trials failed to detect any slowing of cognitive decline on the ADAS-Cog or CDR-SB, nor any drop in amyloid load as seen by PET.

Still, subgroup analyses had hinted at a benefit in participants with mild AD (Jul 2014 conference news; Dec 2014 conference news). This led Roche to go earlier for Phase 3, enrolling people with prodromal to mild AD and evidence of amyloid buildup. Roche also quadrupled the dose of crenezumab to 60 mg/kg, given intravenously once per month (Dec 2016 conference news). CREAD 1 was fully enrolled with 813 participants, while CREAD 2 was still enrolling, with a target of 750 participants. Both studies were planned to run for two years.

Why was crenezumab not working in these trials? Morgan speculated that any inhibitory effect of crenezumab on amyloid aggregation may come too late for people with early AD, whose brains already contain mature plaques. In addition, he noted that anti-amyloid antibodies that mop up plaque, such as aducanumab and BAN2401, do produce significant ARIA in the brain, unlike crenezumab. Perhaps this edema is a sign that antibodies are clearing plaque. “Thus, crenezumab’s failure in early AD is, nonetheless, helping identify antibody features that appear to be key to plaque removal, and, possibly, reduced cognitive decline,” Morgan suggested.

The API trial tests a different paradigm, enrolling 252 cognitively healthy people who carry the PS1 E280A familial AD mutation (May 2012 news). The primary outcome will be slowed decline on a sensitive cognitive composite. Dosing in this trial started low, but was boosted midway through to match the levels tested in the CREAD trials. That study is planned to run for five years.

Roche has two other AD programs: gantenerumab in Phase 3 and the anti-tau antibody RG6100 in Phase 2.—Madolyn Bowman Rogers


  1. This is certainly disappointing. Crenezumab was my initial choice to be an effective antibody. It had two unique features making it distinct from other antibodies against amyloid in clinic testing. First, as an IgG4 variant, it had limited interactions with effector molecules, IgG receptors, or complement activation. This was intended to reduce the risk of ARIA, which it seems to have achieved. Second, it was argued to prevent monomeric amyloid aggregation at substoichiometric ratios (1:100 antibody to Aβ on a molar basis). This implied isomerase activity from the antibody, modifying secondary structure on a catalytic basis to reduce aggregation. Apparently, in individuals with early AD (and presumably mature plaque deposits) this is inadequate to show cognitive benefit.

    Interestingly, the antibodies that recently have shown amyloid reduction and a hint of cognitive benefit, aducanumab and BAN2401, have the property of low affinity for monomer versus aggregates. Presumably this provides greater access to the small amount of IgG that leaks into brain to bind aggregated forms without most of the antibody bound to monomeric decoys. Furthermore, both antibodies produce ARIA, which may imply that activation of effector molecules is a key event in the clearance of plaque amyloid. Thus, crenezumab's failure in early AD is, nonetheless, helping identify antibody features that appear to be key to plaque removal, and, possibly, reduced cognitive decline.

  2. We are disappointed with today’s announcement about crenezumab and the discontinuation of CREAD 1 and CREAD 2 Phase 3 studies. We look forward to reviewing the findings when they are available.

    However, we are pleased to report that the Alzheimer’s Prevention Initiative (API) Autosomal Dominant Alzheimer’s Disease (ADAD) trial in Colombia will continue. That trial is studying whether crenezumab can prevent cognitive decline in unimpaired persons from the world’s largest early onset autosomal-dominant Alzheimer’s disease kindred.

    The API ADAD prevention trial is intended to provide a better test of the amyloid hypothesis than failed trials in the clinical stages of the disease and help build faster ways to find effective Alzheimer’s prevention therapies in everyone at risk for the disease. We remain optimistic that findings and shared data from this trial will have a profound impact on the field.

  3. It is very difficult to make comments based on the press release only. Having said that, I was never optimistic about this drug. It has a different isoform as all the others (Ig4 vs Ig1); has little ARIA, which, in my opinion, also indicates less efficacy; and had negative Phase 2 studies in mild-moderate AD. Now, the decision to move into prodromal was wise and needed, but the lack of efficacy may thus indicate that the drug is simply not good enough. I did not participate in the crenezumab studies.

  4. Like all inactive Aβ antibodies in clinical studies so far, crenezumab demonstrates a significant cross-reaction with Aβ monomer. As R. Bateman has shown nicely years ago, this molecule is produced at high concentrations and high turnover. So every monomer-reacting Aβ antibody has to saturate the huge excess of monomer in the CNS first, and most of the dose is thereby consumed before it can neutralize misfolded Aβ oligomer species.

    Roche/Genentech correctly tried to compensate for this pharmacokinetic disadvantage by elevating the dose about fourfold in later Phase 3 studies, which is still not enough. Future clinical Aβ antibodies need to demonstrate at least 1,000-fold selectivity for oligomers over monomer. Only with such a profile will we be able to effectively neutralize steadily produced misfolded Aβ oligomer species in the CNS, while not interfering with the default physiological Aβ processing.

  5. The recently announced discontinuation of the crenezumab CREAD1 and CREAD2 studies represents yet another disappointment for AD patients and the field. While the autopsy of yet another negative Phase 3 trial is difficult, such analyses are necessary if progress is to occur. Some optimism for these Phase 3 trials was based on Phase 2 results showing trends for cognitive effects using a dose of 15 mg/kg IV (Cummings et al., 2018), with a subsequent quadrupling of the dose. Safety, perhaps related to the IgG4 backbone of the antibody, appeared to be good.

    Even though both crenezumab and solanezumab bind to the mid-domain of Aβ, unlike solanezumab, crenezumab is said to bind with 10-fold greater affinity to Aβ oligomers (Adolfsson et al., 2012). Elegant recent work has attributed these differences to specific physical-chemical differences between the antibodies (Ultsch et al., 2016). Given these considerations, addressing the following questions may help us to understand the reasons behind the futile results in the CREAD1 and CREAD2 studies.

    • What level of target engagement has been demonstrated for the dose utilized in Phase 3, and what is considered to be the primary target? If Aβ oligomers are considered the primary target, have CSF data shown sufficient target engagement based on Aβ oligomer assays? If direct assays of Aβ oligomers in CSF are not available, what estimates can be made of Aβ oligomer binding? In particular, the relative concentrations of Aβ oligomers, Aβ monomers and crenezumab would need to be considered given a 10-fold difference in binding affinity. An understanding of the overall stoichiometry of crenezumab, Aβ monomers and Aβ oligomers at the dose used in Phase 3 would be useful to understand.
    • Given that a positive visual read of amyloid PET imaging (or CSF analyses) was required for inclusion, how did baseline plaque load based on SUVRs compare to other studies? Could plaque load have been relatively high, even though clinical symptoms were relatively mild, due to the requirement for a visual read?
    • While the populations for the studies included evidence of amyloid positivity, the sample sizes for the studies (approximately 375 patients/arm) were relatively small. Is there any evidence that with a larger sample size a drug effect could have been observed?
    • The CDR-SB in some international studies has had substantial variability. Could that variability have played a role in the futility of the studies, and do other cognitive and functional measures behave similarly?
    • While looking for trends in subgroups is fraught with peril, since these studies did include patients on the disease continuum ranging from mild cognitive impairment to mild dementia, is there any evidence for greater drug effect with either more or less mild clinical symptoms?

    Answers to these and other questions will help the field move forward to find effective disease-modifying therapies for AD.


    . ABBY: A phase 2 randomized trial of crenezumab in mild to moderate Alzheimer disease. Neurology. 2018 May 22;90(21):e1889-e1897. Epub 2018 Apr 25 PubMed.

    . An effector-reduced anti-β-amyloid (Aβ) antibody with unique aβ binding properties promotes neuroprotection and glial engulfment of Aβ. J Neurosci. 2012 Jul 11;32(28):9677-89. PubMed.

    . Structure of Crenezumab Complex with Aβ Shows Loss of β-Hairpin. Sci Rep. 2016 Dec 20;6:39374. PubMed.

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Crenezumab
  2. Aducanumab
  3. BAN2401
  4. Gantenerumab
  5. RG6100

News Citations

  1. On Target: Crenezumab Reduces Aβ Oligomers in CSF
  2. Crenezumab Disappoints in Phase 2, Researchers Remain Hopeful
  3. Immunotherapy I: Baby Steps, but No Breakthroughs
  4. Much ‘Adu’ About a Little: Phase 1 Data Feeds the Buzz at CTAD
  5. NIH Director Announces $100M Prevention Trial of Genentech Antibody

External Citations

  1. CREAD 1
  2. CREAD 2
  3. company press release
  4. Alzheimer Prevention Initiative Phase 2 trial
  5. ABBY 
  6. BLAZE

Further Reading