One man is dead and five other men were hospitalized after participating in a Phase I clinical trial in Rennes, France.
The clinical trial, conducted by the company Biotrial on behalf of the Portuguese pharmaceutical firm Bial, was evaluating a pain relief drug candidate that inhibits fatty acid amide hydrolase (FAAH) enzymes. Blocking these enzymes prevents them from breaking down cannabinoids in the brain, a family of compounds that includes the euphoria-inducing neurotransmitter anandamide and Δ9-tetrahydrocannabinol, the major psychoactive component of marijuana.
Phase I clinical trials are conducted to check a drug candidate’s safety profile in healthy, paid volunteers. In this case, the drug caused hemorrhagic and necrotic brain lesions in five out of six men in a group who received the highest doses of the drug, said Gilles Edan, a neurologist at the University Hospital Center of Rennes, during a press conference on Friday.
All six men were hospitalized by Sunday, Jan. 10, and the clinical trial was stopped the next day. The most severely affected man was pronounced brain dead after hospitalization and then died on Jan. 17.
Prior to these serious adverse reactions, 84 people had taken lower doses of the candidate drug without complications, a hospital press release stated. Over the weekend, ten of those 84 individuals were given MRI scans. The hospital reported that none of the brain anomalies seen in the hospitalized patients were observed in these low-dose trial participants.
A spokesperson for the European Medicines Agency told C&EN that “since 2007, approximately 12,500 Phase I clinical trials have been conducted in the European Union without any major incidents being reported.” The last major Phase I clinical trial catastrophe took place in London in 2006, when six healthy men suffered permanent organ damage, including loss of fingers, during a trial for an immunotherapy drug candidate called TGN1412.
In the case of TGN1412, within minutes of being given the putative arthritis and cancer drug, the volunteers suffered from unanticipated and severe immune reactions. An analysis of the 2006 tragedy concluded that “drugs showing safety and efficacy in preclinical animal models may show very different pharmacological properties when administered to humans. Development of proper preclinical models which can efficiently predict drug behavior in humans is very essential prior to testing a drug in a human subject” (J. Young Pharma 2010, DOI: 10.4103/0975-1483.66810).
French authorities have now launched several investigations to determine whether the Bial clinical trial was conducted properly and whether there were problems with the manufacture or transport of the drug candidate.
In a press release, Bial officials noted that the clinical trial “has been conducted since the beginning in accordance with all the good international practices guidelines, with the completion of tests and pre-clinical trials, particularly in the area of toxicology.”
Multiple pharmaceutical companies, including Merck & Co. and Pfizer, have evaluated FAAH inhibitors as possible treatments for pain, anxiety, and insomnia, among other applications. Until the current tragedy in France, there have been no reports of extreme adverse reactions for FAAH inhibitors. A Phase II clinical trial conducted by Pfizer to evaluate an FAAH inhibitor for post-traumatic stress disorder, for example, was terminated “based on Pfizer portfolio prioritization and not due to safety and/or efficacy concern,” noted an entry on clinicaltrials.gov.
As the news of the clinical trial broke, the online chemical community exploded with speculation about the nature of the putative drug. To date, the company has reported only that it is an FAAH inhibitor developed for pain remedy.
Many observers, however, believe the compound’s name is BIA 10-2474. That’s because the company’s website lists two drug candidates in Phase I clinical trials. One candidate is for pulmonary arterial hypertension—a drug that likely targets the lungs instead of the brain and is therefore unlikely to be an FAAH inhibitor. The second Bial candidate—BIA 10-2474—is listed for “neurological and psychiatric pathologies,” and is thus more likely to be involved in the tragedy. Additional support for BIA 10-2474 being the compound’s name comes from an e-mail forwarded to the French news outlet Breizh-info.com from a man who had tried to enlist in the trial but was denied entry.
“What everybody wants to know now is its structure,” says Christopher Southan, a senior curator for the Edinburgh University-based Guide to Pharmacology database.
Southan has mined the patent literature and thinks he may have figured out the drug candidate’s structure. He says the trick is to look at secondary patent filings. Companies tend to protect a whole series of molecules in their first patent filings without naming their most promising lead compound, Southan explains. In subsequent patents for processing, synthesis, or crystallization techniques, though, it’s possible to surmise which compound is destined for the clinic, he says.
In this case, Southan found a 2014 patent in Bial’s portfolio that he thinks is a “best guess” for the lead compound involved in this clinical trial. Minutes before he posted a blog post reporting his patent mining for BIA 10-2474, Southan says an anonymous Wikipedia editor posted the same structure in the online encyclopedia, giving him confidence that he was on the right track.
Others in the online chemical community have since evaluated the structure proposed by Southan and the anonymous Wikipedia editor for binding partners in the body other than FAAH enzymes. If the drug candidate hits “off target” partners, it might help explain what went wrong with this trial.
For example, computational chemist Sean Ekins, from Collaborative Chemistry, inputted the putative structure for BIA 10-2474 into an in silico screening app called PolyPharma that searches for possible biological targets of a submitted molecule. He also ran the molecule through another ligand-protein modeling program called SEA, developed independently by Brian Shoichet at the University of California, San Francisco.
Both programs found that the proposed structure for BIA 10-2474 “was very promiscuous,” he says. These in silico screening techniques produced statistically significant hits for kinases, histone deacetylases, and a whole array of cytochrome p450 proteins, among other binding partners, he says.
“Using different tools and getting similar results makes me more confident to say that it looks promiscuous. But we don’t even have confirmation that this is the correct structure,” Ekins cautions about drawing too many conclusions from these findings.
“All this goes to show the complexity of biology and that the scientific community needs to do a better job at making predictions about off-target effects,” Ekins says. “It’s a totally tragic way to begin the year.”