hero image

Every couple of years, the pharmacology world reads about a phase I trial that has significant problems or is a frank disaster. I discussed the most recent example in this blog last March. It involved a fatty acid amide hydrolase (FAAH) inhibitor, BA-10-2474, from Portuguese company, which was studied in a clinical research organization in France.

During the dose escalation, at a particular dose, several patients had severe adverse reactions leading to one death and hospitalizations of several others with major sequelae. The study was stopped and multiple organizations did analyses of the disaster to try to determine what happened and what could have been done to avoid such problems in the future. Criticisms included questionable study design, questionable dosing and dose elevation schedules, possible inadequate reporting of data and information to the French health agency and the hospital treating the patients. The issue is still being discussed in France and elsewhere.


An earlier example of a phase I disaster involved TGN-1412, a monoclonal antibody that is a strong agonist for the CD 28 receptor expressed by T cells. It was originally developed in a German university. It was felt that it might be useful in preferentially activating “regulatory T cells” to act as an immunomodulator for treatment of autoimmune diseases. A company was created, and they developed  “humanized” antibody for use in clinical trials. It was tested in cynomologus monkeys, which have a CD 28 receptor homologous to the human receptor. Testing in the monkeys showed no serious adverse events. The question of whether TGN-1412 was specific to the CD 28 receptor and whether it might also stimulate other T cells producing unexpected consequences was unclear.

In March 2006, six healthy male volunteers received the TGN-1412 antibody and two healthy male volunteers received placebo in a clinical research unit in the UK. Within an hour of the infusion, several of the TGN-1412 treated subjects complained of nausea, headache and severe lumbar pain. Over the next several hours, hypotension, tachycardia, fever, lymphopenia and monocytopenia occurred in the TGN subjects. Shortly thereafter multi-organ failure, including lungs and kidneys, occurred. They were hospitalized. Neither of the placebo subjects became ill. It was not totally clear how the patients should be treated. Nevertheless all of the 6 men survived, though there were permanent disabilities including necrotic fingers and toes with some amputations. The reaction was determined to be “cytokine storm”  – the release of pro-inflammatory cytokines.  Evidently, TGN-1412 was not as specific as originally thought. The early results were published in a New England Journal of Medicine article.

Lawsuits and monetary settlements occurred and the company went bankrupt. A UK government-established commission investigated and found that all regulatory requirements were met. Others noted, though, that not all of the preclinical data was made available to the health authorities and that clinical trials with similar antibodies did activate T cells leading to cytokine storm.

The starting dose was also analyzed. It was found that the chosen starting dose, though hundreds of times lower than the “safe” dose in the monkeys, was still high enough to cause about 90% of all CD 28 receptors in man to be occupied by TNG-1412 antibodies which can lead to cytokine storm. It also seems that studies of other antibodies by other companies used a starting dose that allows for a receptor occupancy of no more than 10% – much lower than in this study. So using this criterion the starting dose was much too high. Other researchers also discovered that the monkeys, unlike humans, can lose their CD 28 receptors in certain conditions thus preventing the activation of these T cells.  So the monkeys’ T cells could not be activated but the humans’ could be and were activated.  Many other publications followed summarizing the subsequent work.  See, for example, this article.

But everyone agreed that this was a phase 1 disaster.

OK, so where is the good news?

TGN-1412 was not finished in spite of the tragedy that occurred. In 2006, a Russian investor acquired the rights to TGN-1412 and renamed it TAB08.  A Russian company continued the development of the drug. Further lab research on the drug elsewhere resulted in the development of an in-vitro test to determine the effect of TAB08 on T cells. This, plus the work that showed that human CD4+ effector memory T cells retain their CD 28 receptors (unlike the monkeys which lose them), allowed more precise evaluation of TAB08 for use in clinical trials. Information that was not available during the original study.

A new clinical trial was started in Russia in healthy human volunteers. A much much lower starting dose was used which would bind only 1% of the CD 28 receptors (unlike the 90% binding in the first study). Dose elevation was also done with smaller increments and the subjects were very closely monitored. No serious AEs were noted and the study results were published which showed that these low doses selectively activated regulatory T cells but not the inflammation producing T cells – exactly the result desired.

Additional studies are now underway. A phase I PK and tolerability study was done without evident safety issues. A multi-center phase II trial in rheumatoid arthritis patients in whom methotrexate was not effective was begun in Russian in 2014 and was completed in February 2017. A study in psoriasis vulgaris patients has also recently been completed and studies are now recruiting in patients with solid neoplasms and in patients with systemic lupus erythematosis (SLE). These studies are being done in Russia. See the company’s website or Clintrials.gov for further information. So far, no results of the studies are in the public domain but one presumes that we will see them soon enough as two of the studies have recently ended.

A fine editorial summarizing the TGN1412/TAB08 story as of 2015 was published in the British Journal of Clinical Pharmacology.

So what lessons can be drawn from this?

Well, the first is fairly obvious. Testing new drugs, particularly the first in humans, is a risky business. The first dose and doses of the new drug must be done under tight, carefully monitored conditions – ideally as an “in-patient”.

Detailed preparatory work before going into man must be done. This includes animal, in-vitro and clinical pharmacology work. For advanced therapeutic products, biologics and other new products, specialists in the appropriate fields must be consulted and all possible measures (e.g. receptor binding levels) must be studied.  This is best stated in the British Journal article noted above and in   and in another article on biomarkers in research:

“…innovative and potentially risky drugs can be tested safely in humans if researchers have sufficiently detailed insight into the mechanism of action and if informative preclinical studies and biomarkers are available for accurately predicting the effect in humans, thus providing a rational approach for determining a safe starting dose.”

This is true not just for drugs that have had a difficult time in the first study in man but in all studies of new drugs. Credit goes to the researchers in Russia who are working on TAB08 and the results of their studies are eagerly awaited.

The second lesson, is for the drug safety personnel both in the phase I unit and in the Drug Safety unit. Careful attention must be paid to both serious and non-serious AEs in all subjects and patients.

FDA has in the past stated that in phase I studies in normal volunteers in short term studies the likelihood of an SAE happening by chance or due to other causes than the drug is highly unlikely.  Thus, one should strongly consider that any SAE in this setting is possibly due to the drug. This view should be emphasized for all phase I studies and making all SAEs expedited reports (15 day reports) should be the default unless there is clearly another cause for the SAE. The threshold for concern about safety matters in phase I studies and trials should be very low.

Related Articles