Pauline Anderson

July 30, 2015

WASHINGTON, DC — An investigational Alzheimer's disease (AD) drug that a few years ago might have ended up in the disposal heap — or at least in the recycling bin — is now fast tracked for regulatory approval.

The change in direction could be due to its novel neuro–anti-inflammatory mode of action, the strength of its eventual statistical significance, or the fact that there are so few promising pharmaceutical interventions for dementia.

In any case, the drug, TTP488 (Azeliragon, vTv Therapeutics Inc), initially failed a futility analysis but is now fast tracked by the US Food and Drug Administration (FDA), with a phase 3 study already underway.

The lesson is that futility analyses in AD trials may be "misleading," says Marwan Sabbagh, MD, director, Banner Sun Health Research Institute, Sun City, Arizona, and principal investigator for the phase 3 program now investigating TTP488. "They argue for conducting the full analysis plan even when futility criteria are met."

Dr Sabbagh gave an overview of findings with this drug during an oral presentation here at the Alzheimer's Disease International Conference (AAIC) 2015.

Inhibits RAGE

TTP488 works by inhibiting the receptor for advanced glycation end-products (RAGE). "This small orally active molecule blocks RAGE and therefore in consequence reduces the downstream effects," which include vascular dysfunction, metabolic dysregulation, and amyloid deposition, said Dr Sabbagh.

At high levels of expression, RAGE has been correlated with disease severity and progression, and it affects normal microglial and endothelial cells, said Dr Sabbagh. In a RAGE knockout model, mice resist amyloid β plaque formation but are otherwise normal.

Dr Sabbagh provided results of a phase 2b trial of TTP488 that was discontinued a few years ago after a futility analysis. The study included 399 patients with a Mini-Mental State Examination (MMSE) score of 14 to 26 who were receiving background therapy of cholinesterase inhibitors and/or memantine. There was 80% power to detect a treatment effect on the AD Assessment Scale-Cognitive Subscale (ADAS-cog) with a prespecified difference of 3 points.

These patients were randomly assigned to one of three groups: a 60-mg/day load followed by a 20-mg/day maintenance dose; a 50-mg/day load followed by a 5-mg/day maintenance dose, or placebo.

A year after all patients had been randomly assigned, researchers used a decision tree to apply the futility analysis, said Dr Sabbagh.

"Basically, what we said was that if there was conditional probability of less than 10% that there would be a difference between treatment and placebo, there would be a decision to terminate the study. If the conditional probability was greater than 20%, then the study would continue unabated; and if there was somewhere between 10% and 20%, there were secondary triggers that were to be applied."

At that 12-month analysis, the conditional probability of a difference between treatment and placebo was determined to be 9.3%. Researchers decided to terminate the study, and patients discontinued dosing at their next visit.

Dr Sabbagh pointed out that while not a consideration in the futility decision algorithm, there was a 2.3-point difference in median values between the 5-mg dose of the drug and placebo on the ADAS-cog.

And, had the conditional probability reached 10%, the 12-month mean ADAS-cog between-group difference for the 5-mg dose would have been 1.34.

At the end of the 18-month study, the final protocol planned analysis showed that the study did indeed meet its prespecified endpoint, with a 3.1 difference in ADAS-cog of treatment over placebo, with an impressive P value of .008.

Multiple Models

"This was looked at in multiple mathematical statistical models and every single one of them was positive," said Dr Sabbagh. "That is to say that the protocol planned analyses using different methodologies to cope with missing data all showed a statistically significant difference in ADAS-cog favoring 5 mg of treatment over placebo."

An on-treatment analysis, which included all patients with on-treatment data plus 45 days and accounting for a half-life of roughly 18 days, showed a statistically significant value (P = .03) favoring treatment over placebo, with a mean difference of 2.7 points on the ADAS-cog.

A post hoc analysis of patients with mild AD (MMSE score, 21 to 26) showed a between-group ADAS-cog difference of as many as 4 points (P = .018).

This subgroup analysis also found that the difference in the Clinical Dementia Rating sum of the boxes (CDR SB) was statistically significant for the treatment group (a difference of 1 point; P = .02).

How can the "discordance" between these positive results and the prediction of the futility analysis be explained? According to Dr Sabbagh, it was from the use of a statistical model that included only a small sample of participants.

The futility analysis "was based on a single snapshot, a single variable, and a single statistical model that was different from the protocol planned in the final analysis," he said.

"In essence, hindsight is 20/20, and if we were to do it over again, we would not have seen the futility trigger met."

Asked by a delegate whether the parameters for the futility analysis could have been changed, Dr Sabbagh said he "anguishes" over this issue. "I love to speculate about the 'shoulda, woulda, coulda,' but the bottom line is that it was applied. At the end of the day, if we could do it again, I would say there would be no futility analysis, but it is what it is, and that was what was applied at the time."

Now, a phase 3 trial under an FDA-agreed special protocol assessment has gotten under way. The 18-month trial will include 800 patients with an MMSE score of 21 to 26 who will be enrolled in one of two studies, each of which will have two groups: 5 mg/day of the drug or placebo.

The higher dose of the drug has been discontinued because of toxic effects, including confusion, falls, and a greater ADAS-cog decline not seen with the lower dose or placebo.

Weighing in on the topic, Maria Carrillo, PhD, chief science officer, medical and scientific relations, Alzheimer's Association, told Medscape Medical News that TTP488 is "very promising" despite the futility analysis not showing a difference in relation to placebo.

"When the researchers followed the patients out at 18 months, meaning the subjects didn't get any drug for 6 months, they actually found a significant benefit."

She pointed out that the drug company, vTv Therapeutics Inc, is trying to raise money to continue with the research after Pfizer "dropped it." The company has filed a registration statement with the Securities and Exchange Commission.

As the principal investigator for the phase 3 program, Dr Sabbagh is a paid consultant but does not have commercial proprietary ownership in the company. Dr Sabbagh receives research support from Lilly, Avid, Navidea, Functional Neuromodulation, Neuronix, Merck, Takeda, AstraZeneca, Roche, and Genentech. He is an advisor for Biogen Lilly, vTv therapeutics, Avid, and Piramal. He receives royalties from TenSpeed/Random House.

Alzheimer's Association International Conference (AAIC) 2015. Oral presentation 04-09-03. Presented July 22, 2015.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: