Friday, May 06, 2005

More on the Twellman-Dieppa Case

Jeffrey King has some strong, negative feelings about my essay on the Twellman-Dieppa case.

Calling current scientific methodology "unconscionable" and "unacceptable and cruel" is a cold slap in the face to the many passionate, God-fearing and scrupulously ethical people engaged in medical research. How many men and women who devote their lives to finding treatments for life-threatening, even terminal illnesses, are driven by the purest of motivations: the loss of a loved one in their own lives?

If you had daily contact with such people (and the victims of the diseases they study) as I do, you might be surprised.

Reading sinister, hidden agendas into the actions of researchers doesn't add to the dialogue. It is instead, important to consider the unintended consequences of the decisions we make. I also don't appreciate the characterization of current methodology as "convenient" unless that's how one regards a demand for factual clinical data.

King doesn't believe in randomized clinical trials (RCT's) in the setting of fatal illness. And what are the consequences of not performing such trials in this setting? Never knowing for sure if your intervention is truly helpful or deleterious. Without a randomized control group as outlined in my essay, conclusions regarding safety and efficacy are not possible. To eliminate the use of control groups truly invites indiscriminant propagation of unproven, potentially harmful interventions into the medical armamentarium. Such interventions will never be properly studied and no framework for the follow-up of fruitful pathways or for the rejection of deleterious ones are possible.

Patients are not served by this. King's approach essentially ends the acquisition of key knowledge in the dynamics of drugs and their actions in real-life settings.

In the past, I've written about the drug Iressa (gefitinib). This promising drug has been used to treat advanced lung cancer. It had been shown to shrink tumor size dramatically and for this reason received fast-track provisional approval by the FDA with the understanding that a safety and efficacy RCT would be ongoing.

Unfortunately, when results of this study became available, it turned out that while tumors did in fact shrink, mortality was not improved. There are probably several reasons for this one of which is that Iressa may cause a previously unknown and fatal interstitial lung disease. Another reason is that the drug might only work on a subset of patients whose tumors have particular markers. Without the RCT, none of this would be known.

If it hadn't been for the RCT, some patients with terminal lung cancer would be taking Iressa and would be directed away from the drug Tarceva (erlotinib) which has been proven by RCT's to improve mortality!

In addition, by going back to the basic science, researchers will seek a deeper understanding of the dynamics Iressa's side-effects and improve the drug by eliminating them. This opportunity would never have arisen were it not for the RCT. Also, by studying the markers of the tumors that Iressa does seem to work on, future patients may be more precisely selected to maximize the benefit of the drug. This is in fact happening.

There's one other thing I can't resist mentioning. As an example of things we "just don't do", you mentioned testing drugs on pregnant women to see if they cause fetal abnormalities. To a certain extent, you're correct. It is very difficult to enroll such patients into clinical trials because of the emotionality involved. What's unfortunate is that almost no new drugs are categorically approved specifically for use in pregnancy. It becomes almost impossible to use them on these patients for this reason and many pregnant women are being denied the use of medications that may very well have favorable risk-benefit profiles.

Some women's advocacy groups are actually lobbying the government to promote more RCT's on pregnant women!

Anyone who has followed my blog will realize that I'm not, in general, an apologist for big pharma. But I do think that in this case there is more to this than meets the eye.

And Jeffrey, you're absolutely right about one thing. You don't need a fancy label after your name to understand these issues. Just an open mind.


Blogger Svolich said...

RTCs for pharma studies may be - for reasons I don't understand - good medicine. But they're lousy engineering.

In engineering terms, there's no reason that the two cohorts have to be studied at the same time. There's no reason that the study should be limited to two cohorts. One could easily study 2 drugs - or 10, or 50 - and compare them against a placebo group. The results for the single pairing of drug "A" to the placibo group will be the same whether or not there are other drugs being studied at the same time.

Similarly, if the placibo group runs this year and the drug runs next, they're going to get the same results. I know the rest of the world will be different next year - the weather may be different, there will be a different strain of flu going around. Realistically, those things aren't going to matter.

So study a placibo group. Establish the universal baseline. Let them die. Then whenever a new drug comes along, study it and compare it to the baseline.

May 07, 2005 5:18 PM  
Blogger The Medicine Man said...


The process you're suggesting is what epidemiologists refer to as the use of a historical control. You call it a "universal baseline".

This is sometimes done but unfortunately does not lead to quantitatively sound results. The potential errors that can result are legion.

Read the section in my original essay describing the pitfalls of historical controls and why randomization is absolutely necessary to evaluate drug safety and efficacy.

It is an unfortunate reality that boundary conditions cannot be controlled in clinical trials with the same degree of precision as in an engineering experiment. For this reason, comparing current data with historical controls is akin to comparing apples to oranges.

Also reread the section on Iressa in this post to see that these issues are not merely theoretical and in fact impact on patient care.


May 07, 2005 11:27 PM  

Post a Comment

<< Home