The fatal BIA 10-2474 trial and lessons learned (?)

The Phase I trial with the BIA 10-2474 drug of the Portugese company Bial done in French Rennes was stopped in January 2016 after occurring serious complications in 6 volunteers from the same trial arm. One man died.

ANSM, the French regulatory authority, released on 19^th April 2016 a final report which summarizes findings of the BIA 10-2474 trial by the special scientific committee (CSST) established in January 21^st, 2016. Aim of this committee was to analyze mode of action and possible toxicity of the BIA 10-2474 (direct or indirect), to formulate hypothesis to explain toxicity and to adopt global recommendations for first-in-human (FIH) trials.

The BIA 10-2474 belongs to a class of fatty acid amid hydrolase (FAAH) inhibitors which act by degrading an enzyme anandamide, a mediator of endocannabinoid system. There have been more than 6 drug candidates in development, no one yet marketed due to low efficacy. These drugs have been developed for such indications like pain, vomiting, anxiety, mood disorders, Parkinson´s disease, Huntington´s disease and some cardiovascular indications by companies like Sanofi-Aventis, Astellas, BMS, Pfizer and Janssen. Janssen voluntarily suspended its Phase II trials with JNJ-42165279 until they will have more information. Bial has tested its drug for neuropathic pain. Bial has 6 candidates in its pipeline, with one product already marketed for one indication in EU and USA and one product registered, both in central nervous system indications.

The committee at ANSM had a closer look on preclinical findings on the BIA 10-2474 and the Investigational Brochure (IB) accompanied the FIH filing. The BIA 10-2474 showed in in vitro studies quite low activity, in other words IC₅₀ lies in a micromolar range. For those involved in small molecule drug development it is obvious that this compound was quite weak since usually companies move to further development compounds with a nanomolar range found in vitro. Actually, when the molecule was compared to an existing reference molecule in animal pain killing tests, BIA 10-2474 was less effective. Interestingly, the reference molecule values were removed from one figure in the IB.

Further, the drug was not very specific and it also expressed a very narrow concentration range between an absence of inhibition and total inhibition. The latter fact means that even a small concentration change can lead to a significant effect.

When compared to other developmental molecules in the class, based on preclinically obtained characteristics on specificity and efficacy, the BIA 10-2474 was not very competitive to stronger candidates (e.g from Pfizer or Janssen).

What was quite unusual was toxicology testing of BIA 10-2474 in 4 different animal models- rat, mouse, dog and monkey and in addition, in rabbits for fertility and reproduction studies. Toxicology studies are prerequisites for clinical testing, they are done in a GLP (Good Laboratory Praxis) environment and are quite expensive. There was no noticeable toxicity found in particular organs except of gametes (observed also with other FAAH inhibitors). On the basis of NOAEL (no observed adverse effect level), 100 mg dose was chosen as the highest dose in humans. However, the study in dogs had to be redesigned since serious pulmonary events and signs of motor coordination problems occurred. In dogs, the NOAEL was set to 20 mg.

The CSST committee in its report concluded based on information provided that the BIA 10-2474 could not be a priori regarded as a risk product according to valid criteria and therefore was allowed to continue to a first-in-human trial.

The BIA 10-2474 trial started in 2015 (no information in usual clinical trials registers) and had several arms, consisting of single doses (ranging from 0,25 to 100 mg) and repeated doses (from 2.5 mg to 50 mg). Analysis during the trial showed 50% of FAAH inhibition in humans with a 0,25 mg dose and the dose 5 mg showed nearly 100% FAAH inhibition. This was however far less (10x) than the extrapolated dose from animal studies. The volunteers were involved in the injection schedule sequentially, based on results from previous treatments to increase safety of participants.

In January 10^th, 2016, one man was hospitalized after the 5^th injection of the 50 mg dose (cumulative dose 250 mg). Other 5 were at that time without apparent problems. They got their 6^th injection next morning (cumulative dose 300 mg) without waiting for results of the hospitalized man. Then, 2 to 4 days after the last injections these 5 people were also hospitalized due to brain bleeding and other neurological symptoms and the first man died at the 3^rd day after first symptoms occurred. No other volunteers except of this cohort #5 (repeated 50 mg dose) showed any marks of toxicity, especially not of neurological origin.

The committee has concluded that the toxic effect was clearly linked to the BIA 10-2474 molecule. It is likely caused by the molecule itself outside of the endocannabinoid system (not connected to the drug class effect). Toxicity of the BIA 10-2474 can be caused by its binding to other brain cell structures:

1. Due to low drug specificity for the target enzyme
2. Due to use of repeated dose well above the dose causing in human complete inhibition (5mg, found during the trial)
3. Due to gradual accumulation in brain (effect on brain functions occurred after the 5^th injection of 50 mg dose. A single 100 mg dose did not cause any effect).

The committee, despite concluding that there was a rationale behind allowance of the candidate for clinical testing, mentioned that there was a fair number of errors, inaccuracies, reversing numbers, erroneous translations of document sources in the IB provided by the company.

According to the report in Nature, many researchers believe that BIA 10-2474 is acting ‘off-target’. After publication of the Clinical Study Protocol, there were efforts to assess these “off-targets”, e.g. one company involved its proprietary software and predicted a mechanism by which the drug candidate could cause the death and serious adverse effects in the trial.

The most important are lessons we could learn and recommendations of the committee:

1. Focus on preclinical efficacy
2. Include neuropsychological evaluation and cognitive tests for drug candidates affecting central nervous system
3. For the first-in-human studies do not expose all subjects of a cohort/arm to experimental treatment at the same time (as happened in the BIA 10-2474 trial)
4. Dose escalating strategies for first-in-humans studies should involve considerations based on clinical and pharmacological common sense
5. A debate is needed on European and international level about access to data of first in human administration and to toxicological data in order to protect persons involved in future trials.

Especially the last point is very important- should be a trade secrecy more than a human life? I am sure we agree it should not.

Update: “A French health ministry (on 23.5.2016) ruling faulted both Portugal-based Bial-Portela and the French laboratory Biotrial for a failed drug trial that left one patient dead and five others hospitalized earlier this year. Marisol Touraine, France’s health minister, said that although neither company violated health legislation in France, both share responsibility over the dosage of the medication given to patients as well as criticism over the time it took to seek medical assistance after the first patient became ill, Reuters reported Monday”. More  information here.

Update 2: 27.5.2016 “The European Medicines Agency (EMA) has started a review of the guidelines that describe first-in-human clinical trials and the data needed to enable their appropriate design and allow initiation. The review will identify which areas may need to be revised in the light of the tragic incident which took place during a Phase I first-in-human clinicial trial in Rennes, France, in January 2016.” This action is directly connected to the recommendation No. 5 of the CSST as mentioned above.