Letter to the Editor as a mechanism of post-publication scientific discussion

The rather stimulating discussion that arose following Isis the Scientist's critique of a recent study prompted additional blog posts from YHN, Coturnix and drdrA (so far). At the root of much of the bloviating is a comment the original article's authors posted at Isis' blog which appeared to recommend that the most appropriate venue for critical discussion of a scientific paper is the Letters-to-Editors section of the journal in question. Reading additional commentary from the original authors on various of the blogs I am not convinced they meant this point to be as absolute as it came across but it does get me thinking about such discussions.
And I thought I would overview one such prior discussion that bears on some of my posts on MDMA.


Vollenweider and colleagues published a paper in 1998 on the "Psychological and cardiovascular effects and short-term sequelae of MDMA ("ecstasy") in MDMA-naive healthy volunteers."

Vollenweider FX, Gamma A, Liechti M, Huber T.Neuropsychopharmacology. 1998 Oct;19(4):241-51 [publisher link; I think this and the following letters are freely available, if not check the MAPS database linked in the sidebar]. In this study they administered a 1.7 mg/kg oral dose of MDMA to human volunteers.

A letter to the editors of the journal was submitted by Gijsman and colleagues in which they objected to the study:

...we think this study should not have been performed because of the risk ... the authors state that "animal research strongly suggests that a single recreational dose of MDMA is unlikely to produce long-term serotonergic deficits in humans". We disagree with this assertion for the following reason: Repeated administration of MDMA to animals leads to damage of serotonergic axons and terminals which regenerate only to a certain extent and in a very abnormal manner (Fischer et al. 1995). This damage is associated with decreased concentrations of serotonin in the brain. Single dose MDMA also causes a rapid, biphasic decrease in concentration of serotonin in the brains of animals: concentration drops within 1-3 hours restores within 24 hours and drops again after 24-36 hours, lasting for months or even a year (Steele et al. 1994). Therefore it cannot be excluded and even seems likely that administration of a single dose of MDMA to humans causes damage of serotonergic neurons. Even more because primates seem to be more sensitive to both acute and chronic effects of MDMA: in rats, 10 mg/kg (Colado et al. 1995) and in monkeys, 5 mg/kg causes these effects (Ricaurte et al. 1988), a dose that closely approaches the usual recreational dose.If MDMA were a newly developed drug it would almost certainly not be allowed in clinical phase I studies on these grounds.

Readers will recall that I have an interest in these very issues in large part because of ongoing clinical trials (which must certainly surprise the letter authors!)
This letter resulted in a defense from the original study's authors which is extensive, citation heavy and freely available so I won't quote it extensively. Suffice it to say they argued that there was little evidence from either human or animal studies that a single 1.7 mg/kg dose of MDMA was likely to lead to lasting damage. There was one pertinent hook to the continuing story:

In this respect, Ricaurte and colleagues' study (cited by the authors), where a single dose of 5 mg/kg MDMA was given orally to monkeys (Ricaurte et al. 1988), is probably the closest approach to the dose regimen used in our human study. This study found a reduction in 5-HT and 5-HIAA content of about 20% in the thalamus and hypothalamus 2 weeks postdrug.However, as shown above, this does not permit any conclusions as to a possible loss of 5-HT terminals.

Presumably since the Gijsman et al letter suggested there were potential ethical issues at hand with respect to the treatment of human subjects, the editors also addressed the issue. Lieberman and Aghajanian conclude:

Thus, it would appear that while Gjisman et al. (1999) raise valid concerns about this type of research in general and this study in particular, the data do not support the view that single oral doses at 1.7 mg/kg of MDMA (which was one third of the doses used in monkeys by Ricaurte et al. (1988)) are likely to produce damage to serotonin terminals.

These comments would count as trolling a colleague in the blogosphere and sure enough, McCann and Ricaurte soon contributed a viewpoint. Their argument relied on some reasonably accepted* intra-species dose scaling principles.

Unfortunately, when extrapolating the animal data to humans, Dr. Vollenweider et al. (1999) and the Neuropsychopharmacology editorial omitted a critical and fundamental factor in their calculations: the principle of interspecies drug dose scaling (see Mordenti and Chappell 1989). This principle, which is based upon the underlying anatomical, physiological, and biological similarities among mammals, permits researchers to extrapolate animal data to human beings under a variety of experimental conditions. Put simply, smaller animals require higher dosages of drug, on a mg/kg basis, to achieve the same effect. Stated mathematically: Dhuman = Danimal(Whuman / Wanimal)0.7 where D = dose of drug in milligrams (mg) and W = weight in kilograms (Kg). If the known single oral neurotoxic dose of MDMA in a monkey (5.0 mg/kg in a 1 kg monkey) is substituted into the equation, the equivalent dose in a human being weighing 70 kg is 1.4 mg/kg, slightly lower than the 1.7 mg/kg dose used by Vollenweider and colleagues. When identical calculations are carried out using rodent neurotoxicity data (O'Shea et al. 1998), the equivalent MDMA dose in humans is approximately 1.7 mg/kg

Readers will recall that I've tended to ignore this species-scaling argument when discussing the MDMA doses used in the human clinical trials. Instead, I limited my discussion to the size of the typical subject range and the supplemental dosing practices. It is worth recognizing here that most comparisons of dose / exposure to a drug are estimates. Differences in human body size means that human clinical prescriptions (including OTC recommendations) for a fixed mg dose of drug only get in the ballpark. Differences in individual metabolic and other within-species factors mean that using a mg dose per kg bodyweight approach is still only an approximation. Differences in the way two species metabolize different classes of compounds may mean that a dose-scaling equation such as described above holds true for certain types of drugs but not others!
Ultimately, additional data are required to resolve a tight threshold question such as "Is 1.7 mg/kg MDMA likely to cause lasting damage?".
One more response letter from the senior author on the study in question, Vollenweider:

Allometric interspecies scaling is based on the fact that the regression of the logarithm of a pharmacokinetic parameter and the logarithm of species weight is generally linear. As a result, pharmacokinetic (and therefore pharmacodynamic, including toxicity) parameters for a given drug can be estimated in any species if this linear relationship is determined (Ings 1990). Yates and Kugler (1986) and others have criticized the potential inaccuracy of interspecies scaling, noting the 10-fold range of estimates that may be derived depending on which pharmacokinetic and corrective factors are thought relevant. Accordingly, the accuracy of the technique is dependent on the availability of sufficient data.

Vollenweider then throws up some additional smoke screen regarding mechanism of toxicity and potential toxic metabolites but I think the point was made with the general observation. We simply did not then, and still do not now, have very tight estimates about the likely threshold for MDMA-induced damage of a lasting nature. I think the fact that MDMA itself is capable of producing essentially permanent alterations in brain serotonin function is not debatable but the question of threshold is certainly not well resolved.
Getting back to the meta-point of this post, I return to the question of the venue for discussion. Very occasionally a really good discussion of a scientific paper or finding breaks out via exchange of Letters-to-the-Editor in a scientific journal. In my experience, however, this is quite a rare event. The one I detail here was very useful although admittedly many people in the field were already discussing these dose-scaling issues. Still, it brings the discussion out into the open where any first-year graduate student getting up to speed on the field would run across it. A GoodThing.
This is the sort of discussion of a paper that I believe the journals which put effort into such things (such as Nature and PLoS) desire. Their online discussion formats seem to beg for more discussion and wax disappointed with the current state of affairs. The anonymous-comment blog format, however, seems to have no difficulty supporting interesting and fruitful discussion of scientific papers.
Hmm.
__
*as always I'm not really a pharmacologist and would welcome any necessary correction from experts such as Abel.
Mordenti J, Chappell W (1989): The use of interspecies scaling in toxicokinetics In Yacobi A, Kelly J, Batra V (eds), Toxicokinetics and New Drug Development. New York, Pergamon Press, pp 42-96

No responses yet

  • The anonymous-comment blog format, however, seems to have no difficulty supporting interesting and fruitful discussion of scientific papers.
    I'm curious how much of this is the blog format, and how much the anonymous-comment bit? For blogs that discuss general academic and funding issues, such as this one, of course anonymity is very important to the author. But I also read several blogs that describe "just the science" that are signed.
    A disadvantage of the blog-discussion-format is that there is not a direct link from the paper to the blogs that discuss it.

  • anonymous says:

    On this notes, maybe major journal should create a page with the published study for the comments from readers. This way journals will provide platform to discuss the study and it will be easy to find it for anyone. For example, if there is a discussion on this particular NEJM article, wouldn't it be great if all the comments of critiques and authors are available in a page linked with the article? This would be great service for scientific community by the journals.

  • DrugMonkey says:

    I'm curious how much of this is the blog format, and how much the anonymous-comment bit?
    It's the anonymous bit, no question. The journals that seem interested in online commentary have mechanisms that are almost identical to blog comment mechanisms. Until the aforementioned journals "get" this, they are doomed to fail IMNSHO. The occasional incivility is simply the cost of doing business.
    A disadvantage of the blog-discussion-format is that there is not a direct link from the paper to the blogs that discuss it.
    True although I see no reason this couldn't become integrated into PubMed. You know that little sidebar for "related articles"? Why not also have a "related blogs"? Technically this is already quite possible.

  • DSK Samways says:

    "The anonymous-comment blog format, however, seems to have no difficulty supporting interesting and fruitful discussion of scientific papers."
    Agreed. I'd like to see some basic improvements to the comment section mechanics for most blog providers, though, to better cope with active discussions taking place in comments sections. e.g. Scienceblogs should at least implement the basic "reply to comment" function that is common to blogs like Daily Kos, and serves to make it easier to navigate the comments and follow a specific discussion thread. Particularly for high profile papers that would generate a lot of interest and feedback.

  • becca says:

    "You know that little sidebar for "related articles"? Why not also have a "related blogs"? Technically this is already quite possible."
    I long for the day I can get a "related messageboards and chats"...
    Hey, that's a neat protein, but their methods section (as always) SUCKS! Anybody know what antibody from Santa Cruz they're talking about here?
    Then, INSTANT ANSWER!
    Generation.com. We

  • becca says:

    Grrr. I hate when arrows get interperted as HTML tags I don't want.
    The last part was supposed to read- "Generation.com We (heart) our instant informational gratification."
    Yeah, the comments for Scienceblogs should totally be reformated to be like the PhD Phorums. Then I wouldn't have to keep switching my style back and forth, and I could use fake html tags for editorial comments.

  • Matthew Baggott says:

    Vollenweider then throws up some additional smoke screen regarding mechanism of toxicity and potential toxic metabolites but I think the point was made with the general observation.
    Ok, I'll bite.
    I emphatically agree that the question of threshold is not resolved and is worth worrying about.
    However, I don't think it was a "smoke screen" for us to point out, for example, that if there's a neurotoxic metabolite (as is often hypothesized for MDMA), scaling is just not going to work. In such a case, it's not simply that the estimate has wide confidence intervals, but that it is a mistake to attempt scaling at all. Because Mordenti & Chappell clearly give this and other caveats about allometric scaling, it's weird to me that their chapter became the one to cite when people want to support scaling MDMA doses. Scaling makes significant (& controversial) assumptions about the mechanisms of toxicity.
    Rafael de la Torre and Magi Farre (I'm not certain how to put the accents on in html) subsequently made similar points in their 2004 Trends in Pharmacological Sciences article. [Newcomers to the MDMA literature with an interest in this topic may enjoy that article and the resulting exchange with Jon Sprague and Dave Nichols, as well as Mike Baumann, Xiaoying Wang, and Richard Rothman's 2007 Psychopharmacologia paper.]
    It strikes me as odd when people treat interspecies scaling as anything other than a rough rule-of-thumb. As I see it, it gives guesstimates for thinking about drug exposures before pharmacokinetics are known. But once we know pharmacokinetics in the relevant species, we don't need the guesstimates anymore. For example, once we have human and rat PK, it isn't hard to determine dosing regimens for rats that give good approximations of common human exposures (ignoring possibly important differences in metabolite formation). A benefit of doing it this way is that we are (hopefully) forced to make our assumptions about mechanism more explicit.
    I may be biased (I switched from rats to humans long ago), but I think that is the direction in which any dose conversions should flow: If you want to study mechanisms, use any dosing regimen that produces reliable effects. But if you want to argue the exposures are relevant to drug users, then you should imitate exposures seen in users -- don't rely on a magic scaling formula to justify a dosing regimen that was originally developed to get big monoamine depletions.
    As a side comment, I find it interesting that half of the four authors of the Gijsman et al. letter have gone on to conduct MDMA administration studies in humans themselves. So they seem to have made their peace with the risks and ethics of such research.
    That's my two bits. Thanks for the interesting & thoughtful posting!
    regards,
    Matt Baggott

  • DrugMonkey says:

    One fascinating note from the SfN conference. Ricaurte had a poster describing non-linear pharmacokinetics in squirrel monkeys with closely-spaced dosing. A nice comparison to the sort of data that the de la Torre group has been reporting in humans.
    The fascinating thing is that they represented the animal doses as human scaled doses using (I assume) the equation published in the letter described in this post. The actual mg/kg doses had to be extracted from the author who was presenting the poster.
    It is a fascinating way to try to insist that the argument has been settled in your favor. I wonder if anyone will follow suit or pointedly contest this practice? I wonder if they'll try to pull this in a publication?

Leave a Reply