The Corpus Callosum

Usually I cringe when I see yet another newspaper article about suicide.  But I always read them.  This time, I did cringe, but needlessly.  The article turned out to be OK.

F.D.A. Requiring Suicide Studies in Drug Trials
By GARDINER HARRIS
January 24, 2008

After decades of inattention to the possible psychiatric side effects of experimental medicines, the Food and Drug Administration is now requiring drug makers to study closely whether patients become suicidal during clinical trials.

The new rules represent one of the most profound changes of the past 16 years to regulations governing drug development. But since the F.D.A.’s oversight of experimental medicines is done in secret, the agency’s shift has not been announced publicly.

The drug industry, however, is keenly aware of the change. Makers of drugs to treat obesity, urinary incontinence, epilepsy, smoking cessation, depression and many other conditions are being asked for the first time by the drug agency to put a comprehensive suicide assessment into their clinical trials...

All in all, I am not 100% sure that this new attention is a good idea, but it probably is.  The author is correct that the issue had not gotten much attention until recently.  If it weren't for the news media, it would not have gotten as much as it has.  So it is good that the media have brought this issue into the open.  

There is a long history of links or potential links between medication and suicide and suicide-related events.  The oldest I remember -- without taking the time to research it -- is the link with isotretinoin (Accutane®).  There are many others, as mentioned in the article.  They include antidepressants, antiobesity agents, etc.  

One of the lessons is that it is not possible to anticipate what drugs might cause an increase in suicide and suicide-related behaviors.  At first glance, it is not at all obvious that a drug to treat acne could do anything bad in the brain.  That is one problem.  Similarly, it is not obvious that an antiobesity drug could increase such a risk.  Although, with rimonabant, at least there is a theoretical reason to anticipate such a problem, based upon the drug's mechanism of action.  

With antidepressants, the problem is different.  On the one hand, since the drugs clearly affect the brain, it is conceivable that they could have such an effect.  On the other hand, it seems counterintuitive that a drug used to make a particular symptom go away, could actually cause the same symptom to appear.

What this tells us, is that intuition is not a good guide for anticipating possible adverse effects from medication.  So, it might make sense to establish routine screening.  

Are there any potential problems?  Yes, of course.  The most obvious potential problem is that it is not clear that we have a good way to detect this risk.  And once detected, the risk must be quantified.  Then you have to figure out whether the numbers are significant.  This is not as simple as it may seem.  

There are two risks here.  For one, it is possible that the screening method could miss some real risks.  Studies done before a drug is marketed typically are done on a few thousand selected patients.  It is easy to miss a real risk in that context.  This could give a false sense of security.  Second, it is possible that a drug could have no risk, yet a statistical anomaly could falsely indicate the presence of a risk.  Chasing this down could greatly increase the cost of drug development.  It also could, potentially, cause a good drug to fail to make it to the market.  

Ziprasidone (Geodon®) ran into some problems when it was thought that there could be a problem with Q-T interval prolongation.  That turned out to be minimally significant.  Mirtazapine (Remeron®) ran into some trouble because three patients had neutropenia in premarketing studies.  That turned out to be insignificant.  Yet, premarketing studies on fluoxetine (Prozac®) indicated a 1.9% incidence of sexual dysfunction.  That number turned out to be too low, by at least an order of magnitude.  So there are all kinds of problems with premarketing studies.  False positives and false negatives occur fairly often.  

What this means, is that it is important to be selective about the studies that you choose to do, because imperfect studies can cause more trouble than they fix.   So although it sounds like a great idea to screen new drugs for their potential to cause an increased risk for suicide and suicide-related events, I would not jump to that particular conclusion.  Rather, I would want to see to careful consideration or the risks and benefits.  There can be risks inherent in the process of screening for risk.  

Without the benefit of inside information about the process the FDA went through, I do tend to think that they made the right decision.  The problems with false positives and false negatives are well known.  People in the industry and in the government have been struggling with these problems for years.  I assume they know what they are getting into.  Plus, the only way to get a handle on a problem is to study it, even if there can be problems with the studies.  With care, the methodology improves over time.