Skip to main content

Clinical Trials

Warnings About the French Clinical Trial Disaster

Le Figaro has more details – leaked ones, apparently – on that recent clinical trial disaster in France. Translating, I have it as this (note that the ANSM is the Agence Nationale de sécurité du Médicament, the French equivalent of the FDA):

It is understood that the patients in the hospitalized group have brain damage in an unexpected area. But which one? Asked about this crucial point, the director of the ANSM said, grudgingly, “At the base of the skull.” True, but when pressed, the reply is just “Medical secrecy.” According to our information, this is the hippocampus and pons in the heart of the brain. We understand now the concern expressed at the press conference on 15 January by Professor Gilles Edan, Head of Neuroscience pole of Rennes University Hospital. He explained: “Three people already have symptoms severe enough to fear an irreversible disability but no definitive prognosis can be made.” Today the five patients seem out of danger (tirés d’affaire).

I hadn’t heard that last part – I have some difficulty figuring out how you can have damage in either the hippocampus or pons and be “out of danger”, unless that’s just in the sense of “not in immediate danger of dying”, but we’ll see what details emerge. And then there’s this:

Moreover, in private, the investigative committee (CSST) believes that the protocol should not have given large doses to five people simultaneously. . .Finally, the so-called pre-clinical data, that is to say tests on animals, could well conceal explosive information. Dog studies on the molecule led to the death of several of them, according to the same source. This is information “of considerable importance. This can be a real warning signal,” said Prof. Daniele Piomelli, professor of neurobiology and pharmacology at the University of California at Irvine. But the real details are always hidden.

Deaths in dog tox are indeed very bad news, and warrant great caution in continuing to humans. As Le Figaro notes, though, we don’t know what these animals actually died of. No matter what, though, that can be a show-stopper, and normally you’d only make it on to human clinical trials, if at all, after a very thorough examination of why the animals died and why you don’t expect it to happen in the clinic. Your reasons had better be extremely convincing. The usual rule is “Drugs kill dogs and dogs kill drugs”, so making an exception to that is not to be done lightly. This, one assumes is what the investigative committee is looking into – whether this drug should have gone into human subjects at all, given the preclinical signs. . .

Update: more at The Guardian.

Update 2: not everyone’s quite as worked up about the dog death incidents as I am, pointing out (quite correctly) that we don’t know the dose and the multiple of the toxic dose over the expected effective dose. All true, but in my own experience, a dog’s death in tox has always slowed things down a great deal, as people tried to figure out just what caused it and what level of sensitivity humans would be expected to display from that mechanism. We also, frustratingly, don’t know the actual cause of death – cardiovascular is the usual suspect in a dog tox run, because as a species they’re quite sensitive to some of those mechanisms. In that case, it would be a slowdown, but not a killer for a drug candidate, because toxicologists, although concerned with such an event, feel as if they have a better handle on it. We need more details!

63 comments on “Warnings About the French Clinical Trial Disaster”

  1. Phil says:

    If true, that’s pretty damning information. If they did advance a molecule that killed dogs without the thorough investigation Derek describes, I hope they get the book thrown at them.

  2. Adam says:

    Piomelli is an expert on endocannabinoids, which is certainly why he was consulted (though you wouldn’t know that from the quote in that paragraph).

  3. bmartinmd says:

    The reported information on the neurologic damage is just about useless. But snippets of the Le Figaro article suggest that the victims sustained some kind of direct toxicity to the cerebellum in a very distinctive pattern. That’s my read through the current pea-soup fog anyway.

    1. NJBiologist says:

      “The reported information on the neurologic damage is just about useless.”

      Agreed! How on earth do you get damage to the pons and hippocampus, but not anything else? If you have global ischemia, the hippocampus goes first. If you have a pressure-related compartment syndrome, you can squeeze the pons. But I can’t work out a scenario in which you have both of those, but nothing else.

      1. bmartinmd says:

        Could be a CPM-like phenomenon re possible pontine damage. But the article suggests clinical “coordination” difficulties. My inference, on the basis of that flimsy clue, is that the damage is cerebellar.

      2. Wile E. Coyote, Genius says:

        It happens all the time. MPTP toxicity is incredibly specific as to anatomic location in the brain. Could this one be that specific? Possibly. Also look up the Olney lesion. That is an excitotoxic lesion which is also very specific. Could similar changes happen in the pons and/or hippocampus exclusively? Possibly. Hard to eliminate anything based on the sketchy info.

        1. bmartinmd says:

          Yes, I thought about the MPTP toxicity as an analogy as well. Is there something about the drug or metabolite that’s specifically toxic to Purkinje or some other vulnerable cells? WTF knows? This definitely should be reported in the medical literature at some point. There is one human death, so there should be pathologic data.

  4. anchor says:

    Death during the dog toxicity should have stopped the show!! Not that it matters but did they did a postmortem on dogs? Was it cardiac related? Dog is a good model for CV studies hence my curiosity. The human clinical trial after the canine death would have been a non starter for the FDA in the US and I wonder how the French allowed it?

  5. drsnowboard says:

    Err… deaths in dog safety studies are inevitable as without an effect level, there is no No Effect level. Very sensationalist reporting and a temporary forgetting of the Dose makes the Poison?

    1. JJ says:

      Most effects are not deaths.

    2. TX Raven says:

      I guess today we would rephrase it as “the unbound exposure at the target makes the poison”?

  6. AF says:

    On its website ansm has posted a statement saying investigative committee has studied all information ansm was given prior to authorization ( incl. dog studies showing death of some subjects), and has concluded that ansm was correct in authorizing human trials.
    Statement also says committee is asking Bial for more info on dog studies.
    A french website also alludes to death of a monkey
    Other websites also state that doses used in dead animals were 50 to 100 times clinically relevant doses.

    1. Cellbio says:

      50 fold multiple over death is not enough, at least without a lot of reasons to make you feel better, like strong evidence that the mechanisms for toxicity leading to death are species specific, other species have much higher multiples and bioavailabilty or clearance are not going to vary much in humans. And was that a dose multiple or an exposure multiple, if you know?

  7. Andrew says:

    The other thing that has not been explained- for a drug that was in phase 1b studies, why were there extensive tox studies done already in 4 species (up to 3 months in mice, dogs and monkeys plus up to 6 months in rats according to the protocol document)?

    It is hard to draw firm conclusions on the dog deaths without knowing more of the details, including most importantly what dose/exposure they occurred at compared to the human dose.

  8. Julien says:

    According to someone from the ANSM, such a cerebral damage had not been found on the dogs. (Le Point). The ANSM therefore granted an authorization to carry on the trial on humans.

  9. dudley dont says:

    Deaths in dog tox studies are not unusual and they do not necessarily stop a drug from going into man. As had been said many time, the dose makes the poison. Just need to have a no effect level that your efficacious exposure is likely to fall within. It is helpful if there is some reversible tox that can be monitored that will show up well before a lethal exposure.

  10. Hap says:

    If you’re going to ignore what comes out of dog studies, why bother doing them? If they didn’t ignore the results, then they must have had some reason to know that the effects were specific to dogs, or that they didn’t understand what the toxic dose was before giving it to the dogs (but the second would seem to make one kind of hesitant – if you don’t have any idea what the poison dose is then dosing’s going to be kind of hard). Seeing that the dogs that died didn’t have brain damage doesn’t seem sufficient to tell you that the drug didn’t kill the dogs, and that a similar effect won’t happen in people.

    There’s a lot we don’t know, but the hesitancy doesn’t make it look very good.

  11. Me says:

    I think the ‘pea soup’ comment has it.

    Dogs died in tox – big deal. It’s the dose they died at that’s the problem. The months of animal tox studies is the only flag for me that someone was worried about running this trial.

  12. Peter Kenny says:

    As Lady Bracknell might have observed, to lose one dog is unfortunate…

    1. Matthew K says:

      Gosh I hope this is a Trouble With Lichen reference.

      1. Matthew K says:

        Rats – Wilde. I should have Googled first … Trouble With Lichen seems like a perfect book for the readers of this blog … !

  13. eub says:

    On the previous post, somebody commented with an article that described the dosing schedule. If I understood it, the effect must have been cumulative (or interacting with an additional factor) because they were redosing 50 mg: no reported effects at 4 x 50 mg cumulative, one catastrophic response at 5 x 50, majority by 6 x 50. Does this add any clues to the pea soup here?

    “It really does not sound from this like an aggressive ramp-up of dosage to the disaster. Even a cumulative or delayed effect requires a rather narrow window — no reported effects at 4 x 50 mg cumulative, one catastrophic response at 5 x 50, majority by 6 x 50.

    I wonder about possibility of an additional factor in common across the affected group, something that had not been present earlier in the 50 mg series but now took effect. An environmental factor or something.

    Article says dosed up to 100 mg as early as October, but by implication those subjects were not in the group harmed later.

    The first hospitalization was Jan 10 following a 50 mg dose, which I believe they’re saying was the fifth 50 mg dosing, to an eight-patient group.

    Jan 11 the other seven volunteers received the sixth 50 mg dose (and I’m sure that decision will be revisited in hindsight). Later that day the first patient was in a coma. Dosing was then stopped.”

  14. eub says:

    Don’t know if folks saw this, “ANSM Summary of conclusions, 16 February 2016”: just a one-page brief, but some interesting sentences.

    “The Committee concluded that, on the basis of a detailed analysis of the toxicological data available, the results of the studies carried out on animals met the necessary prerequisites and, as a result, allowed testing on humans to begin. Furthermore, the pharmaceutical quality of the product also met the expected requirements for a clinical trial.”

    “Anomalies found under radiology imaging (MRI) show micro cerebral tissue damage of varying severity with very unusual topography but, like with the clinical symptomology, consistent among volunteers. No warning signs were identified among the other trial volunteers, be it those who received daily administrated doses under 50 mg or those who received doses up to 100 mg in a single dose. The appearance of serious negative adverse events after 5 or 6 daily doses of 50 mg therefore demonstrates a threshold effect, 5between 20 mg and 50 mg. Among the various hypothesis put forward, two held preference: [cumulative depletion of all the FAAH, accumulation of a metabolite

    (That PDF is served with a strange MIME type or something? The Google cache of it acts more normally in the browser –)

  15. milkshake says:

    If I understand the mechanism of action, this is a covalent binder, an enzyme inactivator -right? I wonder if this drug candidate interferes with metabolism of complex phospholipids found in neuronal membranes… For a CNS drug they should have used rhesus monkeys (I remember MPTP does not have much neurotoxicity and dopaminergic neuron selectivity in mice because the selective transport and metabolic activation to MPP+ does not happen in mice.)

    1. Erebus says:

      There are a few neurological diseases that damage and atrophy the brainstem and cerebellum preferentially. These are, possibly among others, Pelizaeus–Merzbacher disease and Krabbe disease. I suppose it’s not out of the question that this drug, in addition to inhibiting FAAH and increasing endocannabinoid concentrations in the brain, might also induce galactosylsphingosine accumulation in the brain… thus mimicking Krabbe disease.

  16. Anon says:


    the dogs were sacrificied due to pulmonary complications.

  17. I find it difficult to justify the secrecy of the deliberations surrounding this case, given the enormity of the events. As drug-developing scientists, we have a *need* to know, promptly and in detail, how this drug slipped through the safety net.

    I was deeply troubled by the protocol’s sole reference to the NOAEL levels, without mention of what the observed pre-clinical AE’s or findings had actually been: it robs the physicians of any specific symptom to monitor for. That should be a no-no: we must fairly and squarely inform physicians of where to look, rather than just to waive them off with “we saw nothing at doses below the NOAEL”…it’s a dangerous tautology. It should not have occurred, and should not have been allowed to slip through.

    I also note, with dismay, that the Investigator’s Brochure has still not been released. This, too, is difficult to justify to the drug-developing scientists of the world. The pre-clinical safety and ADME facts in that IB can go a long way to explain what happened, and how to avoid it happening again. And if the relevant facts were *not* in the IB, then Bial and the ANSM have an even more serious problem on their hands, but at least the drug-developing scientists in the world know that the world is still spinning in the same direction it was before these patients were hurt.

  18. Anonymous Researcher snaw says:

    I also find the lack of transparency here very disturbing, especially since this wasn’t the only FAAH inhibitor in the pipeline. If I worked for another company with an FAAH compound, I’d surely want to know as much as possible about this disaster as soon as possible because I would really need to know ASAP whether this is likely to be a mechanism-based tox inherent to FAAH inhibition or something specific to this compound.

    1. TX Raven says:

      how long do you think it would take to conclude whether this effect was *likely* due to mechanism-based toxicity or not? Are you willing to stop developing your compound until then?

  19. Mfernflower says:

    Turns out this may have confirmed my suspicions – that the aryl-aryl(n-oxide) on the right hand side of the molecule may behave like orellanine – A mycotoxin found in certain mushrooms

    1. Istvan says:

      I would not worry about the N-oxide moiety. The reversible formation in vivo of N-oxides from tertiary amine drugs is not uncommon. Orellanine is a bipyridine with chelating hydroxy groups hence its toxicity. For that matter, BIAL has launched the COMT-inhibitor opicapone being also a pyridine N-oxide:

      As for the dogs: I read it somewhere that the death of the dogs was NOT due to the effect of BIA 10,2474: the animals were killed / sacrificed after showing the toxic symptoms (and most likely to carry out postmortem study).

      1. Mandrake says:

        With the n-oxide ruled out – Could the drug be uncerimonusly carbamolaying (sp?) Serines in enzyme active sites? Almost like aldicarb but somehow worse!

        1. Istvan says:

          There has been a lot of speculations about the compound being an ‘irreversible inhibitor’ of FAAH. There is no evidence, however. In fact, such heterocyclic ureas are not typical reactive serine hydrolase inhibitors. Such an effect should have been noted – and was certainly looked for – during precilinical studies. (Researchers at BIAL are not stupid!) So, I would rule out irreversible FAAH inhibition.

          1. TX Raven says:

            well… you could indeed have interrogated the irreversible binding thru simple in vitro studies…. not sure whether that was done… I hope they did….

  20. Mandrake says:

    This is proving to be a tough nut to crack..
    What is you therory of toxicity for this molecule?

  21. cdsouthan says:

    Its clear that Bial dont have the resources, let alone the inclination, to do the detailed mechanistic and tox studies that might have at least a chance of illuminating what might have gon wrong via generating new data. For example, it looks like they never even worked with purified enzyme, let alone did the type of detailed Kinact studies done for JNJ-42165279 and PF-04457845 ( Logically, only others could do this and quite a few might be willing (especially if they have FAAH projects running). The daft situation is that Bial must be sitting on at least a kilo of BIA 10-2747 that could be distributed (even brokered by a vendor) since it aint going nowhere at that scale internally. Sure, this dishing out of a lead (maybe even with a few of the patent analogues) has no precedent, but then neither does the trial tradgedy. Without breaking this and their frankly mystifying declared secrecy precedents the entire pharma enterprise will endure even more reputational damage

  22. Me says:

    Guys don’t play the ‘Bial doesn’t have much resources’ card. In my old company, we partnered a Bial molecule all the way to market – a pretty well-received epilepsy treatment. I’m sure the royalites they get for that would pay for the pre-clin tox on this particular molecular nasty.

  23. Mandrake says:

    I feel like the only thing Bial can do now is come clean and release their tox studies and what ever amount of the molecule they have in storage (Chris said 1kg?) to vendors like Cayman.

  24. Istvan says:

    So far all speculations have focused on the molecule but this is just one side of the coin: conspicuously, it was the oldest of the volunteers who died. Could age or, in a broader sense, genetic predisposition behind the particular vulnerability of this victim behind the tragedy? Personalized toxicology, or the like 😉 Metabolic issues must be considered and this Phase I study was, in part, conducted to get such data. It has been over a month since the tragedy so researchers must have the metabolic profile of the substance in humans, including the ones experiencing serious adverse effects.

  25. pharmacologist says:

    I wonder about the “irreversible” FAAH inhibition that resurfaces all the time. The protocol of the trial is available and the links posted here. There they say that the compound is reversible. Would be interesting to see the actual data, but obviouly that is in the files submitted to the French health authorities.

    The other thing is that they give the NOAELs in the protocol. The MTD (which they do not give) would be likely higher and possibly related to the dogs dying.

  26. Janex says:

    The secrecy around this is disturbing. There are a lot of brilliant toxicology minds around the world that could be working on this if only they weren’t blindfolded. They should publicly release the entire IND. I’d like to know more details about the dog deaths. My experience is that there is no safe margin to sudden death – i.e. where death is the first clinical sign. Generally when dog deaths are acceptable it’s because at a lower dose there were clear clinical signs, blood work etc indicating that a MTD was approaching. If dogs are fine one day and dead the next with no advance warning, that is a very different situation. I’d love to see the details of the dog tox studies and see if that was the case. On another issue my thought is that this may actually be an off target effect. Multiple other companies have compounds against this target. None of them have seen anything like this.

  27. Seb says:

    For those who can read French, there is another article in Le Figaro today, where one of the volunteers (who was part of the high dose group) tells the symptoms he had (and still has as of today) during the clinical trial… quite impressive and frankly scary.

    (Sorry I don’t have time to translate now because of work and stuff… 🙂

    1. bmartinmd says:

      It’s still difficult to know what the neurologic injury is, even with a decent translation. But this latest case report suggests that the damage could be vascular, at least in part.

  28. Istvan says:

    Assuming that inhibition of FAAH leads to elevated (brain) level of the endocannabinoid anandamide, there could be analogies in the neurological / cardiovascuaer adverse effects observed for the other traditional exogenous cannabinoid receptor agonist THC. A PubMed search for ‘cannabis stroke’ retrieved, among others, a recent article:
    Tetrahydrocannabinol induces brain mitochondrial respiratory chain dysfunction and increases oxidative stress: a potential mechanism involved in cannabis-related stroke.
    One can speculate about or even doubt its relevance to the tragic clinical trial but it seems that sustained activation of CB receptors by exo- or endogenous receptor agonist is not without adverse physiological consequences.

    1. AF says:

      On the Ouest-France website, the volunteer (42 yrs old) indicates that
      he was given Endoxan at the Rennes CHU as an exploratory measure, when he was in fairly desperate condition. Two days later his health
      improved a bit.
      Endoxan (cyclophosphamid) -listed as used for some auto-immune indications.
      To Cellbio above, from the various information available on websites, the dogs were given on the order of 1500 mg/day for a 15kg dog for 3 months.

      1. AF says:

        Any chance the FAAH inhibitors could catastrophically combine with acetaminophen?
        There is dispersed information that the volunteers had already developped headaches on the saturday (9/1), that paracetamol was certainly distributed after the 11th to one of the volunteers, and that headaches was considered a common side-effect of clinical trials (enclosed environment?)

        1. Lars says:

          @AF Could be. Acetaminophen metabolizes to AM-404, which is a cannabinoid.

          1. Istvn says:

            Simultenous submission on the same 😉 !

        2. Istvan says:

          Interesting observation! It is hoped that acetaminophen did not exacerbate the adverse effects. It is known that the medicine, after deacetylation, is converted in the body to an amide conjugate with multiple activities.
          Conversion of acetaminophen to the bioactive N-acylphenolamine AM404 via fatty acid amide hydrolase-dependent arachidonic acid conjugation in the nervous system

  29. Cecilia Hillard says:

    But if FAAH was inhibited, that conversion would also presumably be blocked.

    1. Istvan says:

      Oh, yes, indeed (and obviously 😉 you are right, Cecilia!* I think, apart from analgesia in inflammatory pain, acetaminophen does have other activity not requiring FAAH (COX inhibition, etc?)? In the case of BIA poisoning the use of the drug must have served purposes other then analgesia. Anyway, from a scientific point of view it would be interesting to look for AM404 in autopsy.

      *Supra-spinal FAAH is required for the analgesic action of paracetamol in an inflammatory context
      Acetaminophen; from liver to brain: new insights into drug pharmacological action and toxicity

    2. Istvan says:

      Damn right, Cecilia! Conceivably, acetaminophen was used not as an analgesic, acting indirectly via the endocannabinoid system, but for its other therapeutic effects (COX inhibition, etc.).
      Acetaminophen; from liver to brain: new insights into drug pharmacological action and toxicity.
      It would be very interesting to see whether postmortem studies detected any AM404 (at lease from a scientific point of view).

      1. Istvan says:

        Apologies for the duplicate comments. The first one failed to appear for hours.

  30. pipelineGSK says:

    There are some observations: Antibodies against L1 receptor worked in healthy mice and have no effect in lung cancer.
    French patient got anti-pain drug and died from stroke. Wishful thinking of mechanism of actions (MOA) which did not meet the reality.

  31. Cymantrene says:

    This link says that two dogs perished during tox due to lung lesions, but they can not link it directly to this drug, or rule out a link:

    1. AF says:

      A detailed interim report has been produced by ANSM committee
      Committee concludes to the likelihood of some off-target effect, in particular because the molecule is described as fairly non-specific and used at concentrations that would have largely inhibited all FAAH present in the brain (and above).
      Injuries to volunteers are confirmed to have been limited to the brain, pons and hippocampus. Committee says that injuries correspond to no known pathology or type of brain injury ever observed.
      It is mentioned in the report that one of the injured volunteers had had some sort of “brain related incident” before the trial (how serious or when is not quantified), another source says it is the man who died.
      Some more information on preclinical studies therein.
      Another report is due around March 25th.

  32. Istvan says:

    I have tried to read the above minutes of the Temporary Specialist Scientific Committee (TSSC) meeting but gave up after discovering it is replete with speculations and errors regarding the chemistry, reactivity and pharmacology of the substance. I have to admit though, that these errors are due to insufficient information in the dossiers provided by the company. Yet, some open issues were/are supposed top be clarified by this Phase I trial such as human metabolic pathway and the formation of (re)active metabolites.
    For example, on pages 3-4 the statement that BIA 10-2474 is a micromolar inhibitor of FAAH fails to consider the that this pyridine N-oxide could reductively be converted into the pyridine species (Compound No. 335 in the WO2010074588 patent) that has to be evaluated as well.
    As for the irreversible nature of the BIA compound: from a chemical point of view it is hard to imagine how an N-cyclohexyl-N-methylurea could be converted into a reactive isocyanate: it has to be demethylated first to be able to undergo such a conversion (in fact, the corresponding N-desmethyl susbtance is described in the patent as Cpd. No. 561, while its reduced pyridine form is Cpd. No. 458; there apperars to be no bioadata, however). I wish to pint out that this N,N-disubstituted imidazolecarboxamide distinguishes the BIA compound from the PF and JNJ inhibitors, both containing a partially monosubstituted urea which, in turn, can indeed be an acylating (irreversible) species. What a difference a methyl could make!
    So again, a lot of mystery/misery and speculations. We need more data.

  33. bmartinmd says:

    So damage was to the hippocampus and pons (and not the cerebellum), per the TSSC report of March 7th (

    “brain imaging (MRI) showed damage of highly variable severity but also of the same form in terms of its characteristics, and essentially affects the hippocampus and the pons.”

    “the entire picture, both clinical and radiological, was therefore completely unusual, with no relatedness to a known disease or toxicity.”

  34. AF says:

    A french newspaper (le Figaro in its Figaro-Santé edition) has published today
    some information on a CT-scan and MRI-scan done at Rennes CHU on the
    patient who died (on the evening he was admitted and the following morning).
    Some information also on the treatment he received after admission.

    The newspaper has also obtained MRI scans from a patient having taken the
    Bial compound in november. The scans show a stroke. Figaro Santé has had the scans reviewed by specialists who say the stroke was concomitant to the patient receiving the drug (ie 2 months prior to the MRI scan in January). The patient had noted vision problems at the time.

  35. AF says:
    No conflict of interest with above newspaper on my part!

Comments are closed.