Last year, two issues of Research Practitioner included a discussion about a French trial in which one healthy volunteer died and five others were hospitalized. Researchers and pharmacologists are becoming more critical of the trial as new information become available. Were the researchers negligent? Why was this compound chosen over other similar ones for the study?
BIA 10-2474 was developed by Bial, a Portuguese pharmaceutical company, and was being studied by Biotrial, a contract research organization (CRO) in Rennes, France. Based on information in the trial protocol, BIA 10-2474 was an experimental fatty acid amide hydrolase (FAAH) inhibitor. Documents from a prospective volunteer forwarded to French media described BIA 10-2474 as a "product in development for the treatment of different medical conditions from anxiety to Parkinson's disease, but also for the treatment of chronic pain of sclerosis, cancer, hypertension, or the treatment of obesity.”1 The drug was intended to have an effect on the endocannabinoid system, brain receptors that deal with pain.
The Phase I trial used healthy volunteers, men and women between the ages of 18 and 55.2 The trial recruited 128 volunteers. Ninety received different doses of the drug. The rest received a placebo.3 According to information from the French National Agency for Medicines and Health Products Safety (ANSM), the protocol spelled out three parts of the trial. The first was the administration of single-ascending doses (SAD) of the investigational medicinal product (“verum”) or placebo. ANSM documents show eight ascending doses were tested, each dose being tested in a cohort of eight volunteers (64 subjects total). The volunteers participating in this study received a single treatment dose: verum or placebo. Cohort 1 received a 0.25 mg dose, cohort 2: 1.25 mg, cohort 3: 2.5 mg, cohort 4: 5 mg, cohort 5: 10 mg, cohort 6: 20 mg, cohort 7: 50 mg, and cohort 8: 100 mg. No adverse events were noted in this part of the trial. The second part involved a cohort of 12 volunteers receiving one 40 mg verum dose on two occasions, once in the morning on an empty stomach and once after a “fat-rich breakfast.” Again, no adverse events were noted.
The third part of the trial involved the administration of multiple-ascending doses (MAD). Five ascending doses (2.5 mg, 5 mg, 10 mg, 20 mg, and 50 mg) were tested, each in a cohort of eight volunteers. Two volunteers in each cohort were given a placebo. Each volunteer was to receive one treatment once a day for 10 days. On Wednesday, Jan. 6, 2016, the fifth cohort began treatment with a 50 mg dose of the verum (19 days after the end of cohort 4). Sunday, January 10 was the fifth day of administration of a 50 mg dose to the eight male volunteers. A volunteer was hospitalized in the evening with “stroke-like” neurological symptoms.
By the next morning, this volunteer, who had now lapsed into a coma, received an MRI. The results of the scan showed deep hemorrhagic and necrotic lesions. A report issued in April 2016 by ANSM’s Temporary Specialist Scientific Committee (TSSC) says that Biotrial did not initially consider the relationship between the acute symptoms presented by the subject and the molecule tested “to be possible.” Therefore, without waiting for the results of the MRI and other tests on the hospitalized volunteer, the other five volunteers received their sixth dose that morning, January 11, at 8 a.m. (total dose: 300 mg).4
The sponsor and Biotrial then agreed to discontinue the trial on January 11, and the hospitalized volunteer was said to be brain-dead. By Friday, January 15, five other volunteers treated with the verum in that cohort had been hospitalized with similar neurologic symptoms. Biotrial advised ANSM of the adverse effects on January 14. None of the remaining volunteers have shown neurological symptoms similar to those hospitalized, it said at the time. “The clinical symptoms in the five subjects were remarkable in that they were purely neurological, suggesting that only the central nervous system was affected (but with no seizures), without any other symptoms to suggest other organ damage and without the least sign of infection,” the TSSC report says.4
The report noted several findings in regard to the official investigation. Unlike what has been “wrongly cited” by several information sources, there was no time overlap between the various cohorts, especially in the MAD ones, the report says. Instead, there was an interval between the end of one cohort and the start of the next.
The committee members say they “can only be astonished” that, especially as it is a molecule acting via the central nervous system, volunteer selection, inclusion, and follow-up did not include a neuropsychological assessment with clinical interview and cognitive tests. “Such assessments have been carried out using suitable scales among others for some other FAAH inhibitors.” This finding led to one of the recommendations made by the TSCC in the report (see Table 1).
The dose escalation method (arithmetic, geometric, and Fibonacci sequences, etc.) in Phase I trials is based neither on a consensus nor on clearly established international recommendations, the report says. In the case of the trial in Rennes, the dose escalation appears to be “too abrupt” at the end of escalation, “whereas common sense would be in favor of the reverse.” The type of sequence used in this trial does not appear to have been used in trials on other FAAH inhibitors, the committee says.
The report also questions the choice of maximum dose to be tested on the volunteers. “In effect, we see that FAAH inhibition (alleged mechanism of the pharmacological activity of BIA 10-2474) is achieved in humans at 1.25 mg and is almost complete at 5 mg. In these conditions, the choice of 100 mg is tantamount to testing a dose 20 to 50 times higher than that presumed to be effective, which seems absolutely excessive.... This major safety issue could not be anticipated when the trial was approved or when it started (only data in animals were known). However, it would have been logical and expected that the dose escalation plan be reviewed in the light of the pharmacokinetic data collected from the volunteers, as it was collected for other FAAH inhibitors. This was not the case in this study.”
The investigator brochure did not contain information, especially data on toxicology, suggesting a specific risk during first-in-human use, the report says. “We should recall, however, that the brochure contains many mistakes, inaccuracies, figure inversions, or incorrect translations of source documents,” often making them difficult to understand. “This is highly surprising given the regulatory importance of this document.”
Overall, the report pointed to the “toxicity of the molecule” and the specific pharmacokinetic features of BIA 10-2474 as its most likely hypothesis to date of the adverse events in the volunteers. Problems include the likelihood of the compound’s binding to other brain cell structures, the use of multiple doses in the study higher than those leading to complete and lasting FAAH inhibition, and the compound’s probable gradual accumulation in the brain.4
Bial issued a statement on May 23, 2016, reiterating that it had made proper decisions on the compound’s escalating doses and noting that the safety and tolerability profile of BIA 10-2474 was favorable up to 20 mg. “There were no alerts, or signals in any of the safety parameters collected from any of the previous cohorts that could have anticipated the tragic accident. The integrated analysis of single (up to 100 mg) and multiple doses of drug exposure did not reveal any unexpected behavior of the molecule. Therefore, given the data collected in the previous phases of the trial, there was no reason to modify the escalation of doses forecasted and approved by the authorities in the trial protocol.”5
A report in Nature refers to a December 2016 presentation by Helena Gama, head of the pharmacovigilance & drug safety office at Bial. Gama spoke at the British Pharmacological Society (BPS) meeting in London. During her talk, she gave an update on the volunteers who were injured. She said they continue to struggle with the same neurological symptoms they have had since the trial: memory loss, headaches, motor disorders, and tremors.6
That these volunteers are still experiencing symptoms points to a long-term effect, says BPS President David Webb. “That’s the first time we know they are not well—that they are still damaged in some way.”6
One of the survivors, Stéphane Schubhan, told the newspaper Le Maine Libre that he “sleeps badly, has nightmares, sees double at all times, walks with difficulty, and succumbs to dizziness and nausea if he stands more than 10 minutes at a time.”7 He says he wishes he had known about the dogs and a monkey that had died after being given the compound.8 “They didn’t tell us the truth about the dogs,” Schubhan said. “If I’d known the dogs were dead, I wouldn’t have risked my life for €1,900. I wouldn’t have signed up. I’m not crazy.”
François Peaucelle, director of Biotrial, told French news channel BFMTV that the death of the dogs was not significant. “The conclusions of this study were sufficiently clear and clean to rule out any particular ambiguity about proceeding with human tests.”9
The French daily newspaper Le Figaro said that according to a confidential internal investigation report drafted on January 18, 2016, by ANSM’s Director of Evaluation, Cécile Delval, a nonclinical evaluator alerted his clinical colleague to a “central neuro-nervous system effect,” in particular to the lesions observed in dogs, mice, rats, and monkeys after being given the drug. Still the clinical evaluation of the ANSM was that “patient safety is assured in this study,” according to an English translation of the article.10 The Le Figaro article says that Delval notes that the clinical evaluation report does not include the non-clinical evaluator's statements on the neurological effects observed in the animals.
In her presentation at the BPS meeting, Gama also listed 10 instances of neurological side effects, including dizziness and blurred vision, in other volunteers who were given lower doses of the drug, although she said that these were “mild and transient.”5 The Le Figaro article, however, says the internal report is more detailed.10 Two volunteers from the MAD cohort who had taken 10 mg of the compound reported blurred vision and diplopia (double vision) about five hours after taking it. The four episodes they reported each lasted between one hour and two-and-a-half hours. However, according to the article, this was not considered a “relevant event” by the investigator and the monitoring committee and, in fact, was not found in volunteers in the 20 mg cohort. Headache also was reported in volunteers in the 10 mg and 20 mg cohorts. One headache lasted almost 29 hours, another for 47 hours. When contacted by the newspaper, a lawyer for Biotrial responded that "signs that happen in everyday life and that are neither generalized nor repeated are not disturbing.”10
Once the accident happened, clinical researchers and pharmacologists looked for information as to why the compound caused such a reaction. In a Forbes article, scientists said that any time a small molecule is designed to irreversibly bind to an enzyme, “an opportunity arises for the immune system to recognize it as foreign,” resulting in the immune system mounting an inflammatory response against it.12 The drug could also have “off-target” effects, meaning it could have had an effect on an unintended biochemical process. Daniele Piomelli, a pharmacologist and professor of anatomy and neurobiology at the University of California in Irvine, told NPR that the compound also likely acted on parts of the brain that didn't have anything to do with endocannabinoid system.13 “The January tragedy was a surprise and a shock to everybody,” he says. “All the other FAAH inhibitors that had been tested in the past had been shown to be totally safe.” He told NPR that even mild side effects were rare. “What happened in Rennes didn't match with anything we knew.”
The questions about the drug led to incredulity when Bial’s Gama revealed at the BPS conference in December that the company did not include pharmacodynamics data in its dose-escalation decision. When asked why, Nature reports she said, “The evaluation before starting a new drug escalation is based on its safety evaluation and pharmacokinetic data. We did not have any profile that precluded the path to the further dosage.”6
Although Bial legally did not have to use this data, Webb called the decision negligent, according to Nature. “Without the pharmacodynamics data, they were flying blind. That’s when accidents happen.”6 It was important to collect pharmacodynamics data, he says, because BIA 10-2474 is “relatively unselective, meaning that it could have effects in addition to its intended target in the body.” The pharmacodynamics data could have revealed when the copies of the enzyme in the brain were close to being saturated with the drug, the scientists say.
A Bial company spokesperson does not agree. “All official reports about the trial were inconclusive and all have considered this event as unpredictable, and that there was no indication whatsoever, including in the pre-trials, that could have predicted this outcome,” the spokesperson told Nature.
Why BIA 10-2474?
The question has been asked of why even study this compound? It hadn’t shown much promise for pain relief, says one pharmacologist. “I still don't understand why they would bother,” Steve Alexander, associate professor of molecular pharmacology at the University of Nottingham Medical School in England, told Nature. “Most drug companies ... would move on and go and look for something with a higher potency and a better selectivity.”
“As it is usual, at this early stage of evaluation in healthy volunteers no specific therapeutic target had been selected,” says an email from Bial communication manager Susana Vasconcelos to Nature. “However, as described in the clinical trial protocol, it aimed to evaluate the potential effects of the compound on the endocannabinoid system, namely in pain sensitivity.”
Two bioethicists from McGill University in Montreal, Canada, use the Bial trial as a reason why ethics boards should consider potential efficacy as well as safety before allowing clinical trials to proceed. Since the trial events, independent experts have told the media that they believe there was little evidence to support a trial, and that at least five other drugs designed to act in a similar way had been tested in people without success, say Jonathan Kimmelman and Carole Federico. The TSSC report identifies only two studies in Bial’s investigator brochure that are presented as evidence for efficacy, and they are “both problematic,” they say.14 “In one, Bial had data for a different marketed drug showing it was more effective than Bial’s drug at relieving pain in animals, but did not include that information in a summary figure. Both preclinical studies showed only ‘moderate’ positive effects. Moreover, Bial’s drug had been tested at a range of doses in mice that made it impossible to estimate the most likely effective dose in humans.” (For their recommendations about how to ensure clinical trials only proceed for drugs that show clinical promise, see p. 54.)
“I've been doing this research about clinical trials for over a decade now, and when I first read that regulators don't evaluate a drug for efficacy before a company can proceed, I just didn't believe it at first,” Kimmelman tells CNN. “But there is no regulation in place that adjudicates the clinical promise of a new drug when the data is submitted. It is up to the sponsor to determine that.”15
Citing trade secrets, Bial has been reluctant to release information about BIA 10-2474. Although Gama said at the conference that Bial did not have any issue regarding releasing data to the scientific community, she would not make any commitment for the company to do so about BIA 10-2474. Scientists are particularly interested in the investigator’s brochure and the “full human data.”6
Researchers even published an “open letter” in The Lancet on Dec. 9, 2016, asking that the data be made available.11 Justice is now unintentionally interfering with science, they say. “Regulatory changes should be based on a thoughtful analysis of the raw data of the BIA 10-2474 trial…. A pharmacokinetic and pharmacodynamics analysis is needed to examine the drug and metabolite concentrations that might have led to the adverse events. A careful analysis of all study periods is also needed to determine what type of non-serious adverse events occurred in the earlier studies, and how this might be used in the future to recommend a strategy for dose increments across the different study phases of an FIH study.
“We ask the French justice authorities to let a small group of European experts in clinical pharmacology—us or others—have full access to the raw data of the BIA 10-2474 trial to try to determine if the exact cause of the adverse events can be identified,” they say. “This analysis would ensure that any new regulatory recommendations are based on robust evidence.” As of the time of publication, the data had not been released.
By Sue Coons, MA