Some monkey business in autism research (or, why it is not a good idea to provoke Orac)

Believe it or not, even I, Orac, sometimes get tired of blogging about antivaccination idiocy. Indeed, this week was just such a time. I hope you can’t blame me. After all, the last few months have been so chock-full of some of the most bizarre and annoying antics of antivaccinationists at such a frequent clip that there was just no way I could even keep up with it, and trying was starting to burn me out. (I guess there’s only so much that the stupid can burn before even Orac’s nearly indestructible clear plastic case can handle before he needs a break.) Truth be told, not wanting Respectful Insolence to become an “all antivaccinationist all the time” blog, I had been planning on taking a break for a bit and keeping my powder dry for when Jenny McCarthy’s merry band of antivaccinationists descend on Washington, DC on June 4.

But then yesterday afternoon, my old irritant, J.B. Handley, that bull-in-a-china-shop mercury- and vaccine-obsessed founder of Generation Rescue, showed me some love in an e-mail:

From: “Generation Rescue” <info@generationrescue.org>
To: “‘Orac'”
Subject: FW: SICK MONKEYS: RESEARCH LINKS VACCINE LOAD, AUTISM SIGNS
Date: Fri, 16 May 2008 12:33:25 -0700

Orac:

You are a complete jack-ass.

– Generation Rescue

Appended to the message was the text from this link.

Having felt the love, I have to admit that J.B. sure does know how to charm a blogger. When he draws my attention to some abstracts so politely, how can I refuse to return to him manyfold what he obviously craves so badly, a dose of my characteristic and inimitable not-so-Respectful Insolence™, a dose that, given who he is, I am more than happy–nay, eager–to administer to him with loving care, the way a vaccine should be administered. (Too bad there’s no vaccine against bull-headedness.) Indeed, when an antivaccinationist asks me so nicely, I cannot refuse, prior vows to chill out for a while notwithstanding. I even view it as my moral duty to give J.B. what he wants, a task facilitated by the fact that my wife was working last night, that I didn’t feel like working more on a talk I have to give on Monday (making writing this little screed an excellent excuse for procrastination when it comes to finishing the slides for that talk), and that, other than watching Battlestar Galactica, I didn’t have anything better to do. I normally don’t do long, substantive, Orac-length posts on the weekend, but for J.B., I’ll make an exception. I like him just that much.

No need to thank me, J.B.; writing this was reward enough for me.

Besides, I was curious; after I had asked around about this e-mail, I found that other antivaccinationists had been sending this same link to other bloggers who frequently write about autism and vaccines and who share my conclusion that the science just doesn’t support the antivaccinationist myth that vaccines somehow cause autism. Also, the article linked to was written by Dan Olmsted, that credulous reporter who claims that the Amish don’t vaccinate and don’t get autism when they do both and who conveniently couldn’t find the special needs clinic in the heart of Amish country that treats Amish with some forms of autism (not to mention who takes the word of an antivaccination-leaning crunchy physician in Chicago as apparently scientific evidence that the unvaccinated don’t get autism), among other developmental conditions. So credulously eager to believe any bit of trash that supports his need to blame vaccines for autism, Olmsted’s articles are almost always good for a chuckle, if not an outright belly laugh. Although not quite as amusing as Kent Heckenlively‘s overblown and ignorant screeds on AoA, they’re usually still fairly entertaining in a “gawk at the stupid” sort of way. So onward, we go! Here’s Danny-boy:

The first research project to examine effects of the total vaccine load received by children in the 1990s has found autism-like signs and symptoms in infant monkeys vaccinated the same way. The study’s principal investigator, Laura Hewitson from the University of Pittsburgh, reports developmental delays, behavior problems and brain changes in macaque monkeys that mimic “certain neurological abnormalities of autism.”

The findings are being reported Friday and Saturday at a major international autism conference in London.

Although couched in scientific language, Hewitson’s findings are explosive. They suggest, for the first time, that our closest animal cousins develop characteristics of autism when subjected to the same immunizations – such as the MMR shot — and vaccine formulations – such as the mercury preservative thimerosal — that American children received when autism diagnoses exploded in the 1990s.

The first thing I notice here before even reading more is just how much the mercury militia has morphed its “hypothesis” (if you can dignify it with that term). No longer do they talk about “it’s the mercury, stupid” or how “autism and autism spectrum disorders are all all ‘misdiagnoses for mercury poisoning.'” They’ve now conveniently pivoted to blaming vaccines and the “vaccination schedule” for….well, it’s not entirely clear what. That brings us to the second thing that I noticed, which is that they are no longer claiming that vaccines “cause” autism. They are touting this study as showing that vaccines somehow cause behavioral and brain changes that mimic “certain neurological abnormalities of autism.” Truly, antivaccinationists are fluid in their use of language, as long as it can be somehow twisted into implying that vaccines cause autism.

I’m not going to discuss Olmsted’s commentary on these abstracts any more because Olmsted has proven time and time again that he doesn’t know what science is, how it works, or what he is talking about. Besides, in response to tweaking over at the Autism Blog, the merry band of antivaccinationists over at AoA kindly posted all three of the actual abstracts that were presented at the International Meeting for Autism Research (IMFAR). Why bother with Olmsted’s uninformed rants and mischaracterizations when I can cut through his obfuscation and go straight to the source?

But before I do that, one should understand that these abstracts were poster presentations. In the biomedical field, poster presentations are the lowest form of “publication” of one’s data. Indeed, several meetings that I go to fairly regularly accept nearly every abstract that is submitted as a poster. It is from this pool that reviewers decide which abstracts will be granted oral presentations. That’s not to say that some posters, especially at AACR, aren’t excellent. Many are. because the ratio of posters to presentations is quite high. In any case, the standards IMFAR appears to subject posters to do not seem to be anything resembling rigorous peer-review. Thus, given that these are only abstracts being submitted as posters, I consider them much less seriously. Posters are a dime a dozen. If these are published in a decent peer-reviewed journal, I’ll take more notice. Their publication in such a journal would also allow me to examine the methodology, which is only sketchily described in these abstracts. Remember, lots of work is presented at meetings as abstracts but either never makes the cut for publication in a peer-reviewed journal or, when it finally does appear in print, has a radically different form.

So, on to the studies themselves. The first thing that became apparent when I read the abstracts is that this is really only one study, the results of which have been split into three different abstracts. In the biz, this is known as divvying up one’s data into MPUs (minimal publishable units). It’s an unfortunately common practice, but not particularly savory. No justification is given why these shouldn’t just be one abstract or one paper. Oh, well. Let’s move on and look at what is common to the three abstracts and see what we can glean from what is there.

What first leaps to mind in looking at the study is that there are 13 monkeys in the “vaccine” group and only three in the control group. No explanation is given for why there are such unequal numbers. Similarly, there is no mention of how the monkeys were assigned to one group or the other (randomization, anyone?), whether the experimenters were blinded to experimental group and which shots were vaccine or placebo, whether the monkeys were weight- and age-matched, or any of a number of other controls that careful researchers would do. Right off the bat, from the small numbers (particularly with only three monkeys in the control group), I can say that the study almost certainly doesn’t have the statistical power to find much of anything with confidence. Let’s put it this way. I do experiments with mouse tumor models, and if I put such a large mismatch in terms of the number of controls relative to the experimental group, I would be highly unlikely to get results I could have any confidence in.

On the other hand, maybe it’s a good thing that there weren’t more monkeys in the study, given the questionable ethics of subjecting monkeys to so many repeated procedures, not even counting the number of injections given to the monkeys as vaccines or placebos. Let’s take a look to see what I mean:

  • From Pediatric Vaccines Influence Primate Behavior, and Amygdala Growth and Opioid Ligand Binding: “Amygdala growth and binding were measured serially by MRI and by the binding of the non-selective opioid antagonist [11C]diprenorphine, measured by PET, respectively, before (T1) and after (T2) the administration of the measles-mumps-rubella vaccine (MMR).” In other words, these monkeys were subjected to repeated MRI studies. That means they almost certainly had to be restrained and anesthetized each time.
  • From Microarray Analysis of GI Tissue in a Macaque Model of the Effects of Infant Vaccination: “Infant male macaques were vaccinated (or given saline placebo) using the human vaccination schedule. Dosages and times of administration were adjusted for differences between macaques and humans.  Biopsy tissue was collected from the animals at three time points: (1) 10 weeks [pre-MMR1], (2) 14 weeks [post-MMR1] and, (3) 12-15 months [at necropsy].  Whole genome microarray analysis was performed on RNA extracted from the GI tissue from 7 vaccinated and 2 unvaccinated animals at each of these 3 time points (27 samples total).” So these same monkeys were also subjected to at least two colonoscopies. At least, I assume it was colonoscopies; they don’t say.

That’s a lot of procedures and injections for these poor infant monkeys to undergo for what appears to be a poorly-designed experiment. Where on earth was the University of Pittsburgh’s IACUC? Of course, antivaccinationists dismiss concerns about the treatment of the animals in this study thusly, as a commenter on AoA says in response to complaints about just such issues:

The compassion for the monkeys expressed here is pure irony. An entire generation of baby humans has already been abused by the mandatory but un-tested vaccine schedule. Nearly 20 years later, the full schedule is finally being tested animals?!? The irony, the absurdity, the horror of it all…

I love Animallover’s idea about getting PETA involved. Let them raise hell and generate public outrage for these poor monkeys. Only morally corrupt people like Offit would fail to appreciate the irony and ask, “yea, but what about the children?”

And:

Was this experiment cruel to these primates? Yes it was. But it was only as cruel as the larger version that has been forced upon their human cousins.

The present vaccination schedule is simply unsafe for any primate.

At least the second commenter admitted that this experiment was unnecessarily cruel. You know, if I had said something like this, I’d deserve having PETA demonize me in an ad campaign. No responsible researcher takes this sort of attitude towards animals. I realize that these commenters are almost certainly not researchers, but they are dumb. These same two clowns, I’m guessing, will probably be at Jenny’s rally in DC with sandwich boards bearing some slogan like “Vaccines are weapons of mass destruction” or variations thereof. Yes, we should all believe these dim bulbs when they piously assure us that they really, truly are “not antivaccine.”

Yeah, right.

But I digress; back to the studies. After all, you never know. They may actually be OK; we have to judge them on their methodology, data analysis and conclusions. True, this is hard to do from just abstracts (and not particularly informative or quantitative abstracts at that!), but fairness demands that we give the investigators the benefit of the doubt for the moment. Once again, perhaps the most critical variable that isn’t discussed is whether proper blinding was used. Someone named Kelli Ann Davis, who acts as though she has inside information, claims that it was. I have my doubts. Of course, proper blinding of investigators is particularly important for behavioral studies, but it’s also important for any sort of imaging study or examination of histopathology of biopsies. For behavioral studies, the investigators absolutely needed to be blinded to which monkeys had received placebo and which had received vaccine, both when administering the injections and especially when observing and measuring behavior. The radiologists who interpreted the MRI scans had to be blinded as to which scans came from which group, as did the pathologists who interpreted the biopsy results; otherwise subtle (and not-so-subtle) biases can affect the results. Moreover, it’s unclear whether a saline injection is an adequate placebo. Saline injections don’t hurt very much; some vaccines can. This alone could potentially account for differences in behavior. If the monkeys receiving vaccines hurt more, they might become more afraid and withdrawn.

Here’s another question: How long is the life expectancy and time to maturity of these monkeys? In other words, were the investigators scaling down the time between injections proportionally to the difference in time to maturity between humans and these monkeys? That could end up being a lot of shots in a short period of time. So I looked it up. Rhesus Macaque monkeys live around 25 years and males reach sexual maturity by around four years of age, approximately 1/4 of the time it takes humans males to reach sexual maturity. That means, if I interpret correctly the methodology claiming to “adjust for age” that these monkeys could have received a lot of shots in a really short period of time.

Another issue is that it is not clear how many different behaviors were examined in total. It’s possible that the investigators looked at a much wider variety of behaviors and then “cherry-picked” the ones that were positive, especially since many of the behaviors reported as being different between the vaccinated and unvaccinated monkeys are not particularly “autistic”-seeming. In any case, given the extremely small sample size, it is not at all surprising that there were some positive findings just by chance alone. Indeed, just for yucks, to look at a simplistic “yes-no” question of whether a monkey exhibited a particular trait or not, I ran a Fisher’s Exact Probability test for a control group of 3 in which zero exhibit the trait and the experimental group of 13, modeling varying numbers of monkeys exhibiting the trait. To achieve statistical significance at the p=0.05 level, 10/13 of the vaccinated monkeys, or 77%, would have to exhibit a given trait if it assumed that zero out of the three controls exhibit that trait. That’s a huge number, and that’s for a trait that is a noncontinuous variable with two possible values; i.e., a “yes-no” question, as in “yes, the monkey exhibits that behavior” or “no, the monkey does not exhibit that behavior.” If we start looking at traits that rely on measurements of a behavior over time, in other words a continuous variable, then a control group of three is totally inadequate; even a relatively small variance would make it virtually impossible to achieve statistical significance.

Finally, let’s look at the microarray. Sending this particular abstract to me was a big mistake, as I’ve actually done gene expression profiling before using microarrays and am experienced with PCR and quantitative real time PCR. I can’t say I’m an expert at cDNA microarrays, but I’ve picked up a few principles. Let’s look at what the investigators say in the last abstract:

Whole genome microarray analysis was performed on RNA extracted from the GI tissue from 7 vaccinated and 2 unvaccinated animals at each of these 3 time points (27 samples total).

Results: Histopathological examination revealed that vaccinated animals exhibited progressively severe chronic active inflammation, whereas unexposed animals did not. Gene expression comparisons between the groups (vaccinated versus unvaccinated) revealed only 120 genes differentially expressed (fc >1.5; log ratio p<0.001) at 10 weeks, whereas there were 450 genes differentially expressed at 14 weeks, and 324 differentially expressed genes between the 2 groups at necropsy.

One thing that leaps right out at me is the question of why on earth specimens from only slightly more than half of the vaccinated monkeys and only two out of the three unvaccinated monkeys were evaluated. What happened to the other specimens? No explanation is given for why specimens from all the monkeys weren’t studied. This alone makes me highly suspicious of the all the results.

The second thing that leaps right out is the cutoff they were using. They appear to have used a cutoff for “differential expression” of a 1.5 times increase or decrease in the level of a given gene’s transcript. In other words, a gene qualifies as being “differentially expressed” if its messenger RNA (mRNA) level in the vaccinated group is 1.5 times control or 1/1.5, or 0.67 times its level in the unvaccinated group. If I’m interpreting correctly how they did this, that’s a very loose standard for deciding on whether a gene is differentially expressed in the vaccinated group or not, especially in a first pass experiment with low numbers like this one. (Let’s put it this way: In one of the microarray experiments I did, all the genes of interest that I looked at had ratios of over 6, and one had a ratio of over 200, which, not surprisingly, really got our attention.) Although sometimes we will accept 1.5-fold differences, in general in doing a microarray experiment, on the first try we ignore any gene with less than a two-fold change, and we prefer to see three-fold or greater changes in expression levels of the messenger RNA. This is especially true when one uses a log ratio to calculate each gene expression level, given that the log ratio is prone to large changes due to error at low expression levels. This would be particularly true in a dataset that includes only two control samples, which is the absolute minimum number that statistics can be done on and totally inadequate for an experiment like this. The selection of this liberal cutoff strongly suggests to me that the investigators were trying to pad the number of differentially expressed genes. Even if they were referring to a true log ratio of 1.5 (i.e., a 21.5-fold change, or 2.8-fold change, which they may very well have been doing), this would still be fairly liberal for an experiment with only two samples in the control group. Remember, cDNA microarray expression profiling looks at thousands of genes at the same time; without truly rigorous statistics, there will be dozens, if not hundreds of false positives. That’s one reason why In microarray experiments it is absolutely critical to verify any “positive” findings for genes that are up- or downregulated by doing:

  • Reverse transcriptase quantitative real time PCR
  • Western blot or immunoprecipitation to verify that the difference observed in the mRNA level is also seen at the protein level (assuming, of course, a suitable antibody is available)

Did the authors validate any of the genes they found with differential expression levels using these techniques? I see no mention that they did. If they didn’t, their results mean very little, except perhaps for genes with differential expression with a ratio of greater than 10 (a very small number, I’d bet). Once again, cDNA microarray experiments are tricky and often prone to producing false positives in terms of finding genes that are differentially expressed between a control and a test group. Any findings must–and I can’t repeat this enough–must be validated, or at the very least a subset of the genes must be validated by other techniques, particularly when sample sizes are so small. Overall, the abstract suggests to me that the investigators are probably newbies doing expression profiling work or at least in reporting it. One thing that reinforces that impression in my mind is that they only report in the abstract raw numbers of transcripts that were up- or downregulated; usually investigators report a few of the specific genes that were differentially regulated and the ratio by which they were different. All they say is that they are genes consistent with “inflammation,” but that could mean a lot of things, and we don’t diagnose inflammation through the use of cDNA microarrays. Also, in a time course experiment like this, it would be of great interest to know if it was the same genes that were elevated, whether they were continuing to increase or whether they peaked and came back down. No mention is made of this. Raw numbers of genes going up or down mean little. The identities of the genes, how much they go up and down, and the pattern are what matter. That’s why there is software to look at expression profiling data and identify whether changes observed are consistent with the activation of various signaling pathways or intracellular programs. I speculate that the investigators probably just gave their samples to a cDNA microarray service at one of the investigators’ institutions without much consultation with anyone about how such experiments should be designed and carried out, and the service dutifully ran the samples.

Finally (and I saved this for last because I wanted to address what little substance thre was in these three abstracts first), who did this experiment? One name leaps stands out: Andrew Wakefield. Yes, it’s the same Andrew Wakefield whose incompetent science (particularly in the area of dealing with RNA and doing PCR!) done with huge conflicts of interest led to a scare that caused MMR vaccination rates to plummet in the U.K. ten years ago. Kev informs us a bit about some of the other authors:

The primary author seems to be Laura Hewiston of Pittsburgh University. She is registered on that page as a DAN! Doctor. She (I think its the same person) also appears here (see 953) and here.

Also listed as an author according to AoA is one AJ Wakefield. Enough said about that!

Lastly, is Steve Walker who did a poster presentation at an IMFAR in the past (can’t recall which one) which also appeared to offer support for the MMR hypothesis. Oddly, that poster presentation never made it into any kind of peer reviewed journal.

Oddly enough, Dr. Hewiston appears to have a background in primate research and has presented multiple times at the meeting of the American Society of Primatologists, an observation that makes me wonder how she got roped into these studies. Apparently she has an autistic son, and that may be coloring her decisions. Unfortunately, Dr. Hewitson wouldn’t be the first researcher whose personal brush with autism led her down the path of questionable science; I hope she doesn’t descent too far into antivaccination-related research to get out before doing permanent damage to her career.

Judging from the abstracts so helpfully provided by AoA, I am completely underwhelmed. This appears to be nothing more than Hornig v.2.0, except this time with monkeys. I suppose it’s possible, albeit unlikely, that the science in the actual study is better than what is represented in the abstracts. For that, we will have to wait for the actual papers to be published–if they ever are published, which is by no means certain, given what I can glean of the quality (or, more correctly, the lack thereof) of the science presented in these abstracts. (If anyone is actually attending IMFAR now and saw the posters, I’d love to hear your account of what was actually reported.) Of course, if all else fails, and Hewitt and Wakefield can’t get this published in a real journal, even a low impact, bottom-feeding journal, there’s always the Journal of American Physicians and Surgeons. I’m sure the editors there would eat it up and publish all three abstracts as full papers in a very special “monkeys and autism” edition of their revered journal.

Finally, the question comes up: Why now? Why are these abstracts being touted now by antivaccinationists? In other words: Cui bono? I think the answer is probably two-fold. First, the. the Autism Omnibus is cranking back into high gear again. The plaintiffs have been having their backsides handed to them lately and probably need more “ammunition,” however poor quality it is. Second, Jenny McCarthy’s little shindig with Generation Rescue and antivaccinationists everywhere in Washington, DC is coming up in less than three weeks. True, abstracts had to be submitted to IMFAR months ago, but Jenny had been talking about a protest at the CDC late last year. In any case, even if the original intent wasn’t to boost the efforts to blame “toxins” in vaccines for all sorts of evils, the way these abstracts are being distributed now, even to those of us most likely to be skeptical about it, suggests, besides an “in your face,” “see, I told you so” gloating attitude that really isn’t warranted, plans for a publicity campaign based on these “studies.” Look for them to be featured prominently on signs and in speeches and literature at Jenny’s little convention of know-nothings on June 4. Also look for a new post touting this study at that repository of everything antivaccine, The Huffington Post, by that useful idiot for the mercury militia David Kirby. Any bets that it’ll appear there on Monday?

You’re smart not to take that bet.

Oh, and before I forget, J.B: Right back at ya! I hope this little exercise shows you the truth of this song, adapted for my own nefarious purposes:

You don’t tug on Superman’s cape
You don’t spit into the wind
You don’t pull the mask off the old Lone Ranger
And you don’t mess around with Orac

At least, it’s not wise to mess with Orac when it comes to shoving in my face dubious scientific studies that haven’t even been published in a decent peer-reviewed journal as though they were some great refutation of everything I’ve written about vaccine pseudoscience, that is.

ADDENDUM: It appears that perhaps I was too easy on Dr. Hewitson. I really, truly wanted to give her the benefit of the doubt. However, it turns out that she appears to have a very serious and seemingly undisclosed conflict of interest, as a commenter has informed me. Not only is she married to Dan Hollenbeck, a regular contributor to the Age of Autism website, but she and her husband are listed as litigants in the Autism Omnibus proceedings. She did not, as far as I can tell, disclose that conflict of interest. At least, it is not appended to the abstract. See #437:

437. Laura Hewiston and Dan Hollenbeck on behalf of Joshua Hollenbeck, Dallas, Texas, Court of Federal Claims Number 03-1166V

Here’s what INSAR says about presenters at IMFAR and conflicts of interest:

INSAR requires authors to disclose their sources of contributed support (commercial, public, or private foundation grants, and off-label use of drugs, if any). INSAR also requires authors to signify whether there may be a real or perceived conflict of interest. Any potential for financial gain that may be derived from reported work may constitute a potential conflict of interest.

“Any” potential financial gain, and “…real or perceived conflict of interest”! The authors should have listed some. There are several. Dr. Hewitson has a child in the Omnibus, and Dr. Wakefield, a hearing about his medical license in process where the charges are based on his unethical behavior in doing the MMR study in 1998. These studies, if decent, are almost certainly going to be cited as exculpatory evidence that his research was correct after all. That’s on top of his medical practice at Thoughtful House, where his treatments of children are based on the idea that vaccines somehow cause autism and that the MMR vaccine causes “autistic enterocolitis.” Neither of these conflicts of interest are listed on the AoA posting of the abstracts, which means either that AoA left them out or neither Dr. Hewitson nor Dr. Wakefield reported them to INSAR when submitting or finalizing the abstract.

Funny, I wonder if J.B. knew that when he sent me this. In fact, this might not have come out so quickly were it not for J.B. e-mailing me and annoying me enough to write about these studies. Good going, J.B.! Keep it up!

ADDENDUM #2: Kev spells Dr. Hewitson’s conflicts of interest out in detail. It’s more than just being a litigant in the Autism Omnibus:

So, here we are with three poster presentations from a woman who has an autistic son, affiliated with DAN!, is married to the Thoughtful House IT guy (who also happens to be on the Board of Directors of SafeMinds) and these afore-mentioned poster presentations are also co-authored by Andrew Wakefield.

I wonder just how impartial this science can be?

Not very, I’m guessing.