LONDON New in vitro research in human brain cells has identified off-target effects that may have caused the death of a volunteer, and neurological damage in four others, in the phase I trial of BIA-102474.
Using proteomic methods to determine all the interactions of BIA-102474, the researchers show the fatty acid amide hydrolase (FAAH) inhibitor also engages several other lipases and produces substantial alterations in lipid networks in human cortical neurons.
Writing in the June 9, 2017, issue of Science, Mario van der Stelt and colleagues said that suggests the "promiscuous" lipase inhibitor has the potential to cause metabolic dysregulation in the nervous system.
Bial Portela & Ca SA, the developer of BIA-102474, told BioWorld Today it has been carrying out its own studies, which will be published very soon. The company welcomed the van der Stelt et al research and said, "Generally, our results are aligned with many of [its] findings."
The sudden onset in January 2016 of neurological damage in the highest dose cohort occurred six months into the dose-ascending trial and after the drug had been administered to 84 other volunteers with no ill-effects. The five injured volunteers had received five of six daily 50-mg doses of BIA-102474 when the damage first became evident.
Because a number of other FAAH inhibitors had been tested in other clinical trials with no previous reports of serious adverse effects, the "no observed adverse effect level" criteria were applied in taking BIA-102474 to the first-in-human study.
Two French government investigations into the trial disaster, which was conducted by the CRO Biotrial Research SAS in a unit at Rennes University Hospital, concluded the study protocol was approved in line with regulations and complied with legislation regarding dose escalation.
After examining a number of hypotheses for the cause, an independent committee of 12 pharmacologists, toxicologists and clinicians appointed by the French government concluded the most likely explanation was a gradual accumulation of the drug in the brain, related to specific features of BIA-102474, and which caused unspecified off-target effects. (See BioWorld Today, June 1, 2016.)
The findings prompted the EMA to launch a review of its guideline for first dosing in man, with the new rules due to be published later this month. (See BioWorld Today, Nov. 16, 2016.)
FAAH is a membrane-bound intracellular serine hydrolase, which mediates the breakdown of anandamide, an agonist of the cannabinoid receptors CB1 and CB2. Porto, Portugal-based Bial was initially developing BIA-102474 as a treatment for neuropathic pain.
Van der Stelt has been studying the endocannabinoid system for 20 years. He told BioWorld Today he was intrigued by what happened in the BIA-102474 trial because other FAAH inhibitors have been safe though not effective in human trials.
"When I heard about the toxic effects, I was wondering what the mechanism was. I thought it would be possible to use proteomics to identify off-target effects," van der Stelt said.
Van der Stelt and his colleagues first demonstrated that BIA-102474 reacts with FAAH and other serine hydrolases to form covalent and irreversible addition products, or adducts.
Then, to assess the full extent of BIA-102474's interactions with the serine hydrolase family, the researchers used activity-based protein profiling (ABPP), a technique that makes it possible to assess the functional state of entire classes of enzymes in biological systems.
"In the preclinical studies BIA-102474 had been tested in selected off-targets using recombinant cell lines. We used a more general approach to examine the whole family of serine hydrolases in neurons," van der Stelt said.
For reference, the researchers compared the in vitro activity of BIA-102474 with Pfizer Inc.'s FAAH inhibitor, PF-04457845. In preclinical studies, the Pfizer compound was tested against 68 receptors and 20 hydrolases and was found to be highly selective. In comparison, Bial tested BIA-102474 and one of its metabolites against just three serine hydrolases and five other enzymes.
In a phase IIa study in pain relief in osteoarthritis of the knee, PF-04457845 showed no difference against placebo, despite the fact that FAAH activity decreased by 96 percent. The data were published in the September 2012 issue of Pain.
Van der Stelt said PF-04457845 was chosen as a comparator because the data were in the public domain, highlighting the value of publishing negative clinical trial results.
The ABPP studies confirmed BIA-102474 and PF-04457845 engaged human FAAH at the lowest concentration tested. But while PF-04457845 maintained selectivity at higher concentrations, BIA-102474 engaged numerous off-targets.
Many of those off-targets are involved in cellular lipid metabolism and most show substantial expression in brain tissue. The researchers subsequently showed that BIA-102474 - but not PF-04457845 - causes alterations in lipid metabolism in human cortical neurons.
One of the key issues to be addressed in investigating the BIA-102474 trial was why the preclinical animal studies gave no hint of its toxicity.
While saying "it is difficult to answer" if that new research might offer an explanation, van der Stelt noted one of the off-targets with which BIA-102474 interacts previously has been associated with neurotoxicity in humans but is not toxic in rodents.
As to why there was such a sudden and dramatic manifestation of BIA-102474's toxicity at such a late stage of the trial, van der Stelt suggested a threshold must be reached before there is any effect. "There is a covalent, irreversible binding interaction, but you need quite a high level, which would take several days of dosing," he said.
With EMA due to publish its new guideline, van der Stelt said based on the research he would recommend that human, not recombinant, cells should be used for preclinical testing and that all small molecules that form covalent irreversible adducts should be assessed preclinically using the ABPP proteomic profiling.
"That will allow you to get a better handle on activity and its selectivity profile, and a better feel for dosing," he said.
Van der Stelt and colleagues concluded that far from being specific for FAAH, BIA-102474 is a broad-spectrum inhibitor of serine hydrolases. However, the data "do not allow us to conclude that inhibition of one or more of the identified off-target proteins is responsible for the clinical neurotoxicity caused by this drug," they said.
Bial said it welcomes any study that could help shed some light on the causes of the clinical trial accident with BIA-102474. "We particularly welcome the comparison with another clinically tested FAAH inhibitor as this is essential to understand what might be different about BIA-102474, and has been lacking from some previous speculative publications," the company said.
The research reinforces the off-target hypothesis. However, Bial said, "The clinical implications of the reported off-target interactions and alterations to the lipid profile are not known, so it is not possible to draw any conclusion linking these observations to the neurologic events which occurred in the healthy volunteers who received BIA-102474."