Regular readers of this blog may have noticed that it’s been a while since I’ve written a substantive post on the fear mongering and bad science that are used by activists to support the claim that mercury in the thimerosal used as preservatives in vaccines is the cause of an “autism epidemic.” The closest I’ve come is using Robert F. Kennedy, Jr.’s credulous reporting and conspiracy-mongering, in which he uncritically parroted the claims of the worst of the mercury militia and arguing that his recent article in Rolling Stone uses the same sort of dubious and fallacious techniques, showing that, when it comes to bad arguments, highly selective quoting of data to argue a point, and all around misrepresentation of the evidence, RFK, Jr. is the gift that keeps on giving to skeptics.
The reason I haven’t written on this topic much in a while was mainly because not much was happening. At least at first that was true. However, over the last couple of weeks, there have been a few developments that are worth commenting on. I had been waiting a bit to do so, because I didn’t want to steal the thunder of a blogger who has done a lot to reveal the depths of bad science that some will stoop to to support a “link” between mercury and autism.
This particular blogger, Kathleen Seidel, has finally started to publish the results of her investigation, and I can now break my self-imposed silence. And, boy oh boy, has she dug up some dirt.
The last time I left that father-son tag team of Don Quixotes tilting at mercury windmills, Mark and David Geier, they were using truly bad statistics to argue that autism rates have been falling in the Vaccine Adverse Events Reporting System (VAERS) database and the Californial Department of Developmental Services (CDDS) database. I wrote a rather prolonged fisking of them (later mentioning their apparently foreordained conclusions), only to have my position reinforced by Mark at Good Math, Bad Math. In addition, recently, the Geiers really made a name for themselves by claiming that testosterone somehow “potentiates” the toxicity of mercury in autism and prevents it from being removed by chelation therapy. They even went so far as to propose treating autism with Lupron, which shuts down the synthesis of sex hormones, both testosterone and estrogen. Worse, as Kathleen reported, Mark and David Geier are trying to patent this use of what is in essence chemical castration to treat autism.
Now Kathleen has done some more digging. I hadn’t really thought much of this at first, but on the paper that the Geiers published in that right-wing antivaccination conspiracy “journal,” The Journal of American Physicians and Surgeons (JAP&S) claiming that autism rates were falling based on extremely dubious statistical “analyses” of the VAERS and CDDS databases, David Geier had listed himself thusly:
David A. Geier, B.A. is a graduate student in biochemistry, George Washington University.
I thought it curious, given that most biomedical journals merely list affiliation and don’t list whether someone is a graduate student or faculty. I also know that David Geier had formed a company known as Medcon, whose main purpose is to help parents bring legal proceedings before the National Vaccine Injury Compensation Panel (NVICP). But it didn’t really catch my attention much more than as something rather odd that somehow rubbed me the wrong way and made me suspicious that David Geier was listing something that most people wouldn’t list, but I thought no more of it.
Kathleen has found out that it’s not so odd, but rather a misrepresentation. She noted that the Geiers had published another paper in Hormone Research, entitled A Clinical and Laboratory Evaluation of Methionine Cycle-Transsulfuration and Androgen Pathway Markers in Children with Autistic Disorders. It was published online as an E-pub ahead of print. In it, David Geier once again listed himself as being affiliated with the Department of Biochemistry, George Washington University, Washington, D.C. That struck me as quite odd as well. First off, by the conventions of academic publishing, an author usually lists his or her affiliation as the institution where he or she is either faculty or where the work was done. For graduate students, it is assumed that the affiliation listed represents the department and institution at which the work was done, usually as a part of a graduate student’s thesis work with a faculty advisor from that particular department. Moreover, David Geier had listed his affiliation with George Washington University in a number of other articles and conference proceedings, including his biographical sketch for various conferences at which he presented recently, including Autism One, Defeat Autism Now!, and the National Autism Association.
There’s only one problem. David Geier is not a graduate student at George Washington University. Indeed, as far as Kathleen has been able to tell, he is no longer affiliated with George Washington University in any way, nor did he do any of the work reported in the papers listed above during his brief stint as a graduate student at GWU. Kathleen explains:
A search of public areas on the GWU website yielded no instances of David Geier’s name, the “Institute for Chronic Illnesses,” or any indication that the work described in this article took place under the auspices of the GWU Biochemistry Department. Since the GWU website does not allow unaffiliated individuals access to their faculty and student directory, I contacted Dr. Allen Goldstein, Chairman of the GWU Department of Biochemistry and Molecular Biology, to inquire about David Geier and the “Institute for Chronic Illnesses.”
Dr. Goldstein stated that David Geier enrolled in GWU’s graduate biochemistry program in 2003, presenting an an application supported with an impressive bibliography of academic journal articles co-authored with his father. According to Dr. Goldstein, Mr. Geier took two courses in biochemistry during the 2003-2004 school year and none thereafter; he took the last of three public health courses during the Spring 2005 semester.
Dr. Goldstein stated unequivocally that David Geier has never served on the faculty of the Department of Biochemistry and Molecular Biology at GWU; that neither the “Institute for Chronic Illnesses” or its Institutional Review Board (that is, the committee that approves and supervises research on human subjects) are in any way associated with GWU; and that none of the research described in the article was sponsored by GWU or conducted in the GWU laboratories. He described the affiliation with the Department of Biochemistry in the Hormone Research article as “fallacious,” and stated that it conveyed a “significant misrepresentation” of Mr. Geier’s position in the field of biochemistry.
Ouch. That one’s gotta leave a mark.
Worse for the Geiers, as soon as the editors of Hormone Research learned about the misrepresentation of David Geier’s affiliation, they yanked their paper from online publication, leaving it in limbo.
To me, far more disturbing than than David Geier’s misrepresentation is the question of just what the Institute for Chronic Illnesses is. Is it a real Institute? Is its IRB qualified? Kathleen promises more about this, and, in the meantime, I reserve judgment. I do find it telling, however, that Dr. Goldstein emphasized that this IRB was not affiliated with GWU. In any case, lest I be accused of nothing more than ad hominem attacks, what about the actual Hormone Research paper that was yanked? What was in it? What did it conclude?
Well, not much.
The first thing you have to remember when considering this paper is that the Geiers are using it as a justification for their “Lupron protocol,” which, as I described before, involves treating autistic children with Lupron to block the production of testosterone (and, although the Geiers don’t mention this, all of the other steroid sex hormones). They’ve been pushing their protocol through all the usual venues that advocate biomedical “treatments” for autism. For example, there are videos of Mark and David Geier pontificating on their concept of “testosterone sheets” at F.A.I.R. Autism Media, based on an X-ray crystollography study from 1968 that crystallized testosterone with mercury using hot benzene (very physiological). Oddly enough, no mention of this proposed therapy is made in the paper. This is particularly odd since the methods sections states that “consecutive” children with autism who presented to Dr. Mark Geier’s Genetic Centers of North America from November 2004 to November 2005 and only 16 patients with autistic spectrum disorders (ASDs) were identified. Only 16? That’s not very many for a whole year. Another thing: Since the Geiers were actively recruiting patients for their “Lupron” protocol at the time, one has to wonder whether that skewed the patient population because parents whose children showed signs of elevated testosterone may have preferentially brought their children to the Geiers, given their evangelism for their “protocol.”
Be that as it may, though, the rest of the paper is pretty lame and provides only the weakest of support for the hypothesis that excess testosterone coupled with defects in oxidative metabolism has anything to do with autism. For one thing, other than the total testosterone levels, the dataset for the children reported upon is maddeningly incomplete. For example, only 11/16 had a serum/plasma DHEA ratio. Only 14/16 had a serum FSH level reported. Levels of serum testosterone were elevated above normal ranges for 8/16, but most were only slightly above the top end of the normal range. However, the Geiers somewhat deceptively represented the aggregate from all 16 patients as being 256% of normal (no standard deviation or standard error of mean reported, no 95% confidence interval). The only way I can figure out their coming up with this figure was to use the average of the normal range as the “normal” testosterone concentration and then doing a ratio. The problem is, patients with testosterone levels within the normal range are considered to have normal testosterone levels. If the normal range is 0-20 and the testosterone is reported as 19, that’s within the normal range. It is not 1.9 times normal. Or, if they considered only patients with testosterone elevated above normal ranges, in the previous example, a testosterone of 25 would not represent a testosterone of 2.5 times normal. The Geiers appear to pull the same thing with plasma FSH values. All of the values reported for the 16 patients are within the normal range. True, many of them are at the low end of the normal range, but they are within the normal range. Yet, in aggregate, they are reported as being 35% of “normal.” As for the level of molecules related to oxidative metabolism, a complete set of levels of glutathione, cysteine, cystathione, homocysteine, and methionine were reported for only 7/16 patients. And–surprise, surprise–they did the same thing with the levels of all of these molecules. Most of the values for them were within the normal range. Again, they were on the low end of the normal range, but they were within the normal range. Yet, again, the Geiers manage to report them as being anywhere from 56% to 81% of “normal.
Finally, for all of these lab values, I have my doubts about the way that the Geiers calculate their p-values using Student’s t-test. In essence, they are trying to compare their values to the mean and standard deviation for normals from the laboratory. The problem is, with such a small number of patients, most of whose lab values actually fall within the laboratory’s reference range, there’s a problem. Given that the Geiers don’t report the mean and standard deviation of normal specific to the laboratories that they used, it’s impossible to judge if they did their statistics correctly. Worse, Student’s t-test is only designed to compare one value measured in two different groups. If you’re going to compare multiple variables, as the Geiers did, your odds of finding a spurious result go up if you don’t use statistical tests designed to deal with multiple comparisons, such as, for example, Hotelling’s T2 statistic. Why?Statistical significance is arbitrarily defined as there being only a 1 in 20 (0.05) chance that the two “different” values could in fact be statistically indistinguishable. Consequently, for two values reported to be statistically significantly different, there is a 5% chance that they are in fact not different. Thus, the more variables you look at, the greater the chance of picking up “statistically significant” differences that in reality aren’t. That’s why tests for that account for testing multiple variables between two groups are needed to account for this. The Geiers tested for eight different markers in the two groups, not one. Of course, the Geiers aren’t the only investigators guilty of this particular statistical misstep; it’s sadly all too common in the biomedical literature. Nonetheless, they could be finding statiistical significance where none exists.
In actuality, although the Geiers’ study reports on a very small number of patients and doesn’t report complete datasets for most of even this small number of patients, the conclusions of the study are on the surface actually fairly narrowly stated, which is probably why the paper was accepted. It is the subtext of the study that the reviewers almost certainly didn’t know about, namely that the Geiers were already treating autistic patients with a combination of Lupron to suppress their testosterone production and the quackery known as chelation therapy to “get rid of mercury,” which, according to the Geiers, is a major cause of autism. Never mind that there is no good scientific evidence to provide a rationale for such a treatment. The real hypothesis behind the paper remains unwritten, namely that many autistic children are undergoing “precocious puberty.” Thus, although some of the children are listed as having either body hair or facial hair or as masturbating or exhibiting “aggressive” behavior, no real clinical information is presented to show that these children truly were undergoing precocious puberty, even though the Geiers are basing the use of very powerful drugs on the belief that many autistic children are, in fact, undergoing precocious puberty. Unfortunately, they don’t seem to know what they are doing. Kathleen Seidel has discussed the diagnositic criteria for precocious puberty, as have I. Kevin Leitch has described what the Geiers are doing from the accounts of parents of autistic children:
It was clear that the ‘scientists’ advising these people had not informed them of basic facts about the condition that was allegedly affecting their kids autism. Neither of them had had their childrens hand and wrist radiographed which is the standard way of determining if a child is undergoing Precocious Puberty or not. Basically, If bone age is within 1 year of chronological age, puberty has not started. If bone age is advanced by 2 or more years, puberty likely has been present for a year or more or is progressing more rapidly.
The single most basic fact about Precocious Puberty is that it is immediately subdivided into Central Precocious Puberty (CPP) or Pseudo Precocious Puberty (PPP). It is vital to make this difference as the treatment is different in each division. The division can only be made by testing for premature activation of the hypothalamic-pituitary-gonadal axis. When I asked one of these people if the Geiers (yes, it was they) had subcategorised into CPP or PPP they did not know what I was talking about. They were entirely ignorant of these terms. It was clear neither of the two people I had spoken to had undergone this sub-categorisation.
They claimed it was ‘enough’ to ‘know’ that their children had excess testosterone. One of these children is female. This child’s parent was utterly ignorant of the fact that excess testosterone in females was not called ‘precocious puberty’ but indicative of ‘Androgen excess’. Lupron is not mentioned as a treatment for Androgen Excess.
One other interesting fact about increased testosterone is that in patients diagnosed with PPP, this can result from an excess of vitamins and other dietary supplements. Its common knowledge that this is a common part of DAN! and DAN! style treatment regimes. Yet again, the Geier’s patients parents were entirely unaware of this fact.
Surprise, surprise. One has to wonder how many of the patients reported in the Hormone Research paper were on supplements or, more importantly, had actually undergone a real diagnostic workup for precocious puberty. (Of course, if the autism discussion boards are any indication, children older than 9, who by definition can’t have precocious puberty, are being treated with this protocol.) I’m surprised and disappointed that the reviewers didn’t pick up on that lack, but perhaps I shouldn’t be. After all, the Geiers never really stated their real hypothesis. On the other hand, I’m justifiably surprised and disappointed that the reviewers didn’t question the other shortcomings of the article, such as the lack of complete datasets for half the patients, the question of whether Student’s t-test was the most appropriate statistical test for these multiple comparisons, the problem that the lab values reported for most of the patients were actually within the normal range according to laboratory reference values, and a very unclear figure in which they lay out their hypothesis. Finally, the biggest question of all, a question that you wouldn’t know to ask about if you didn’t know the context of the Geiers’ “research” described above, is whether this group of patients were truly representative of a population of autistic children. Chances are pretty good that they are not and that they represent a group of patients brought in because their parents had heard about the Geiers’ Lupron protocol and who therefore may have been selected because they were perceived to have signs of elevated androgens.
The bottom line is that, if there is indeed a relationship between testosterone, methionine trans-sulfuration, and autism, this paper sure as heck doesn’t provide any convincing evidence to show it it. More importantly, although it almost certainly would have been used for such a purpose (just as the Geiers Medical Hypotheses paper was), had it not been pulled by Hormone Research, it doesn’t provide justification for “therapies” involving heavy duty pharmacological interventions such as shutting down the production of testosterone and other steroid hormones using Lupron.
But that’s what the Geiers are selling. They’re even trying to get a patent on the treatment. And David Geier, at least, seems willing to claim an affiliation with George Washington University that he no longer has in order to cover their work with a patina of “credibility” by implying that it was done at GWU.
As Kathleen says, “To be continued.” Personally, I wonder if this “Institute for Chronic Diseases” IRB is also the one that supposedly approved the Lupron “clinical trial” that the Geiers are promoting.