Crenezumab failed to slow cognitive decline in the Alzheimer’s Prevention Initiative’s Colombian study, according to topline findings released June 15. Both primary endpoints were negative, although trends on the primaries and on multiple secondary and exploratory endpoints numerically favored crenezumab. Crenezumab, made by Roche/Genentech, is unique among anti-amyloid antibodies in late-stage trials because it targets Aβ oligomers and does not budge plaque load. Researchers are now analyzing target engagement, dose exposure, subgroup effects, and biomarker data; those data will be presented on August 2 at the Alzheimer’s Association International Conference in San Diego and virtually.

  • Crenezumab did not slow cognitive decline in this autosomal-dominant AD prevention trial.
  • Numeric trends on primary and several secondary and exploratory endpoints favored crenezumab.
  • Biomarker data and dose-exposure and subgroup analyses to be shown at AAIC.

“The [results] are disappointing, but not altogether unpredicted based on earlier failures of crenezumab in sporadic AD, and its inability to clear brain amyloid to any extent,” noted David Knopman at the Mayo Clinic in Rochester, Minnesota (full comment below). His was a common view among researchers, who anticipate that the data to be presented at AAIC will shed more light on the biological effects of the treatment.

The API Colombian study was the first primary prevention trial with a mechanism-based drug in AD. It demonstrated that such trials are feasible and paved the way for many more to come. At a press briefing announcing the result, Eric Reiman at Banner Alzheimer’s Institute in Phoenix noted that API remains committed to working with the families in the Colombian kindred and is exploring options for future studies. Richard Hodes of the National Institute on Aging and Rachelle Doody of Roche said the same. For their part, the families in this kindred expect more trial opportunities to come their way. This is according to Francisco Lopera, who leads the Grupo de Neurociencias at Universidad de Antioquia (GNA). Over the past 40 years, Lopera has built trust and an active research commitment among this population.

The trial kicked off in 2013. It was groundbreaking in many respects, for example by building modern trial infrastructure, and by publicly promising to share data and samples with the research community at large (May 2012 conference news; Aug 2012 news).

The trial enrolled 252 cognitively healthy people from a 6,000-person-strong registry GNA had built with the Colombian kindred, which includes 1,200 carriers of the E280A Paisa mutation in presenilin 1. Trial participants, two-thirds of whom have the mutation, received biweekly subcutaneous injections of crenezumab or placebo for five to eight years. Participants were not told their mutation status, and noncarriers were assigned placebo.

The field’s understanding of how to administer amyloid immunotherapy grew during the course of the trial, with numerous studies showing that initial antibody dosing across the field had been too low. Because of this, API researchers changed the protocol during the trial. At the start, participants received 300 mg crenezumab or placebo, equivalent to about 6 mg/kg, into one arm. This is about as much as can be administered in a single injection. Partway through the trial, the dose was boosted to 720 mg, requiring participants to get a jab in both arms. Then researchers again raised the dose, 10-fold, to 60 mg/kg, which required intravenous infusions.

This further complicated the already-difficult logistics of a trial for people spread out across mountainous rural terrain in addition to the city of Medellín. While subcutaneous injections could be given locally, participants now had to travel to Medellín every two weeks for infusions. Many participants live in remote regions and had to begin their trek by mule. The journey could be dangerous due to civil unrest, said Lopera. Even so, most participants agreed to switch to infusions. They also stuck with the study, posting an astonishing 94 percent completion rate for this five- to eight-year trial. “This is a highly motivated group. They’re amazing,” Reiman said.

Results unblinded this week revealed that the drug missed statistical significance on the endpoint of slowing cognitive decline on the API ADAD cognitive composite. Ditto for the Free and Cued Selective Reminding Test of episodic memory, a dual primary added after the trial had started.

That said, API researchers and outside commenters alike pointed out pressing questions. Because most participants were on the highest dose for only about two years, it is unclear if enough antibody reached their brains for a sufficient period to lower Aβ oligomers. In addition, previously reported data revealed that half the mutation carriers were amyloid-negative at baseline (Aug 2019 conference news). If this group did not decline cognitively during the trial, it would further weaken the trial's already low power to detect a cognitive effect in this small cohort.

Ron Petersen at the Mayo Clinic in Rochester, Minnesota, thinks that because of the drug’s mechanism of action—targeting oligomers rather than plaque—the findings are not translatable to other anti-amyloid antibodies (Jul 2018 conference news; Dec 2019 conference news). Roche had previously halted two sporadic AD trials of crenezumab when the antibody failed to slow decline in symptomatic patients (Jan 2019 news). Familial mutations cause amyloid to accumulate faster than it does in sporadic AD. “One has to wonder if giving a weak-clearance monoclonal antibody still makes sense in an amyloid-overproduction disease,” Marwan Sabbagh at the Barrow Neurological Institute in Phoenix wrote to Alzforum (full comments below).

Researchers are awaiting AAIC presentations on dose exposure and subgroups, as well as biomarker data, before drawing conclusions. Participants underwent scanning with florbetapir amyloid PET, FDG PET, and volumetric MRI at baseline, two years, five years, and at the final visit. Partway through the trial, researchers added tau PET with Genentech/Roche’s in-house tracer GTP-1. Cerebrospinal fluid was drawn on the same schedule, and evaluated for Aβ, p-tau181, and total tau.

One set of data unlikely to be ready by AAIC is the plasma analysis. All participants donated blood annually, meaning API researchers have thousands of samples to process for multiple markers. That work is ongoing, Reiman said.

However, researchers will present results of multiple clinical and cognitive endpoints, which include CDR, CDR-SB, and MMSE scores, executive-function measures, neuropsychological testing, the time to a diagnosis of MCI, and functional assessments. For details on the trial design and a complete list of outcomes, see Tariot et al., 2018.

“Any directional changes in clinical measures or biomarkers would be of great interest,” Eric Siemers at Siemers Integration LLC wrote to Alzforum (full comment below).

Reiman expects the dataset to be highly informative. It enables scientists to compare amyloid-positive and amyloid-negative carriers and noncarriers over as long as eight years' time, and to compare treated and untreated subgroups. “We will be able to clarify which cognitive and biomarker endpoints give us the most power, and in which subgroups. That will be invaluable for informing the size and design of future prevention trials,” Reiman told Alzforum.

The GNA and API plan to continue working together. API has recruited Robert Alexander, who had previously worked at Takeda, Pfizer, AstraZeneca, and other pharma companies, to bring his early phase trial design expertise to bear on defining next steps.

Until a decision is made, the former trial participants are continuing to receive crenezumab. Despite the high dosing in the trial, no new safety issues cropped up, suggesting the treatment is well-tolerated. In addition, Lopera noted the potential to begin trials in other Colombian kindreds carrying different AD mutations.

“This is not the end of the story, it is the beginning,” Pierre Tariot of Banner said in a press briefing. Lopera agreed, saying, “We are happy because we were able to finish this prevention trial in a good way, even if we have not met the clinical outcome.”—Madolyn Bowman Rogers


  1. The crenezumab trial in the Colombian dominantly inherited AD cohort is disappointing, but not altogether unpredicted based on earlier failures of crenezumab in sporadic AD, and based on the inability of crenezumab to clear brain amyloid to any extent (Salloway et al., 2018). Whether a higher dose would have given a different result, or whether a different anti-amyloid monoclonal antibody that actually reduced brain amyloid would have fared better, is unknown.

    I applaud the investigators, especially our Colombian colleagues and the patients, for completing this trial under the difficult circumstances of the past few years.


    . Amyloid positron emission tomography and cerebrospinal fluid results from a crenezumab anti-amyloid-beta antibody double-blind, placebo-controlled, randomized phase II study in mild-to-moderate Alzheimer's disease (BLAZE). Alzheimers Res Ther. 2018 Sep 19;10(1):96. PubMed.

  2. Unfortunately, it was not unexpected. Crenezumab has struggled given the negative results from CREAD and ABBY in symptomatic AD trials. BAI and Roche were highly innovative in designing a prevention trial in a cohort which was assuredly destined to develop AD. However, hindsight is always 20/20, and one has to wonder if giving a weak-clearance monoclonal antibody still makes sense in an amyloid-overproduction disease.

  3. The results from this study of crenezumab in patients with preclinical autosomal-dominant AD are certainly a disappointment for the field. GNA, Banner, Roche, and the NIA are to be congratulated for performing such a challenging trial. Nevertheless, there are some caveats in the study that likely limit broad conclusions taken from it.

    First, the dose used in the study is substantially lower than those tested in the Phase 3 trials of crenezumab for sporadic AD. The Colombia study started with 300 mg Q2W and then increased to 720 mg Q2W after Phase 2 data. In the Phase 2 crenezumab studies ABBY and BLAZE, doses of 300 mg subcu every two weeks or 15 mg/kg IV every four weeks were studied. In the Phase 3 CREAD studies of sporadic AD, the dose was increased to 60 mg/kg (~4200 mg) monthly. Careful review of additional data including biomarker data will be necessary to determine if measurable target engagement was achieved with the low dose used in the Colombia study.   

    Additionally, with only ~125 people per arm, the study was likely to be substantially underpowered unless the assumption is made of a very large drug effect. Powering analyses can be complicated in that, for very large and perhaps unrealistic assumed effect sizes from a drug, even with a small study statistical power can be shown. The assumption about the effect size of the drug becomes very important. Powering for most Phase 3 studies in sporadic AD results in at least 500 people per arm.

    Taken together, a negative result in an underpowered study with a very low dose unfortunately may not be very informative. Any directional changes in clinical measures or biomarkers would be of great interest, so we’ll need to see the AAIC presentation to understand any possible drug effect in this study.

  4. This is disappointing news for patients and their families. As with all disappointments, we must not subsequently succumb to reactive and glum thinking, but rather make a reasoned inquest into the nature of the failure and take a measured stock of the implications. Objective introspection is now crucial to building a stronger knowledge base and bringing about changes in scientific practices so that future therapeutic approaches are better informed and have a greater chance of succeeding.

    In this regard, a potential reflex response to the failure of crenezumab to bring about therapeutic benefits in patients might be to infer that Aβ is not relevant to AD symptoms or that mouse models of AD (which are often reported to respond positively to anti-Aβ antibodies) are not relevant to the clinical condition. Yet, we contend that this view lacks important nuance, and that mouse-model-derived data has, in fact, provided valuable clues and insights all along that rationalize the failure of this and other anti-Aβ therapies. For example, we have previously demonstrated that anti-Aβ antibodies not dissimilar from crenezumab worsened neuronal dysfunction in AD model mice despite reducing brain Aβ burden (Busche et al., 2015). 

    In addition, some of the best-powered studies have failed to show an unequivocal improvement in behavior in AD mouse models following Aβ immunotherapy (see, e.g., Chen et al., 2007). This latter report, in particular, fuels ongoing concerns of the possibility of publication bias in the field toward positive therapeutic outcomes, and further emphasizes the need for negative therapeutic findings to be treated as equally impactful and for preclinical behavioral studies to be robustly statistically powered.

    In humans, the situation is even more complex, since AD patients manifest both Aβ and tau (and other) pathologies. Here, it is becoming increasingly recognized that tau impairs neuronal function which dominates that of Aβ, and that both proteinopathies show synergy throughout the disease (Busche et al., 2019Busche and Hyman, 2020). Thus, targeting Aβ alone may not, in and of itself, be sufficient to bring about therapeutic benefits.

    We conclude that our understanding of the effect of anti-Aβ antibodies on brain (dys)function in humans with AD is rudimentary. The fact that patients on such treatments continue to undergo cognitive decline ostensibly suggests the absence of a major therapeutic benefit, though the possibility remains that any potentially valuable positive outcomes are concealed by off-target or secondary detrimental effects. To disambiguate these, and to appraise treatment outcomes more accurately, we must better understand the normal physiological function and role of the proteins being targeted (e.g., APP and Aβ) and, perhaps most importantly, develop sensitive and dynamic markers of human brain function in AD that provide greater specificity and granularity than simple measures of proteinopathic burden and cognition.


    . 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.

    . Active beta-amyloid immunization restores spatial learning in PDAPP mice displaying very low levels of beta-amyloid. J Neurosci. 2007 Mar 7;27(10):2654-62. PubMed.

    . Tau impairs neural circuits, dominating amyloid-β effects, in Alzheimer models in vivo. Nat Neurosci. 2019 Jan;22(1):57-64. Epub 2018 Dec 17 PubMed.

    . Synergy between amyloid-β and tau in Alzheimer's disease. Nat Neurosci. 2020 Oct;23(10):1183-1193. Epub 2020 Aug 10 PubMed.

  5. By way of disclosure: I was one of the two neurologists on this study who adjudicated the cases from Columbia when they progressed from cognitively unimpaired to MCI. So I was involved with the study.

    Having said that, it is disappointing that crenezumab did not show a benefit on the primary outcomes of the study. Crenezumab is designed to attack the oligomeric stage of AD but probably does not influence the plaque load. In that sense, it is probably not comparable to aducanumab, lecanemab, donanemab or gantenerumab. We await the results of these other trials.

Make a Comment

To make a comment you must login or register.


Therapeutics Citations

  1. Crenezumab

News Citations

  1. NIH Director Announces $100M Prevention Trial of Genentech Antibody
  2. Collaborative Umbrella CAPs Three Prevention Trial Initiatives
  3. Crenezumab Update: Baseline Data from Colombian Prevention Trial
  4. On Target: Crenezumab Reduces Aβ Oligomers in CSF
  5. Amyloid Clearance: Check. Cognitive Benefit: Um … Maybe.
  6. Roche Pulls Plug on Two Phase 3 Trials of Crenezumab

Mutations Citations

  1. PSEN1 E280A (Paisa)

Paper Citations

  1. . The Alzheimer's Prevention Initiative Autosomal-Dominant Alzheimer's Disease Trial: A study of crenezumab versus placebo in preclinical PSEN1 E280A mutation carriers to evaluate efficacy and safety in the treatment of autosomal-dominant Alzheimer's dise. Alzheimers Dement (N Y). 2018;4:150-160. Epub 2018 Mar 8 PubMed.

External Citations

  1. topline findings

Further Reading