Friday, July 29, 2005

Dr. Hornig's Autistic Mice

The September 2004 issue of Molecular Psychiatry (may not have been available at your local news-stand) contained an article by Dr. M. Hornig et al titled, "Neurotoxic effects of postnatal thimerosal are mouse strain dependent". Athough the title doesn't give any hints (other than the mention of thimerosal), the article is all about autism. This struck me as strange, since I had a hard time imagining how you would tell if a mouse was autistic. But, I digress.

The Autism Diva has already addressed the topic of autistic mice and Dr. Hornig in some detail in her delightful post, "Rain Mouse?", so I'll try not to cover the same ground. Instead, I would like to address some of the technical aspects of the study.

In a nutshell, Dr. Hornig and her colleagues gave intramuscular (i.m.) injections of either thimerosal or a placebo to newborn mice. The mice were from three different strains - one that was susceptible to autoimmune disorders (SJL/J) and two that were not (C57BL6/J and BALB/cJ).

In order to simulate the immunization schedule given to human infants (see where they're going with this?), they gave the injections on the 7th, 9th, 11th and 15th days of life. For reference, human infants receive their immunizations at ages 2 months, 4 months, 6 months and 12 months (one year). However, mouse infancy is much more accelerated than human infancy, requiring the injections to be spaced closer together in order to happen during the same developmental stages.

A plan of overdosage:

The timing of the injections is the first flaw that I would like to address. While it is true that this dosing schedule allows the thimerosal to be administered in the same neurodevelopmental stage as in human infants, it overlooks a big fact of biology.

The plasma half-life (time that it takes for the plasma concentration to decrease 50%) of methylmercury in humans is variously reported as being between 40 and 56 days. There are no published studies of the half-life of thimerosal in humans, but animal studies make it likely that it will be longer. The half-life of methylmercury in mice is much shorter - 158 hours.

So, a dose of mercury given to a human will be half gone in about 50 days and, in a mouse, 6 1/2 days. Let's look at what this means in the context of this study:

If an infant is given a mercury dose at time zero (0) that produces a blood mercury level of 1.0 (arbitrary units), that blood level will be down to 0.44 after 60 days. A second dose at 60 days (see where I'm headed?) will bring the blood level to 1.44. In another 60 days, the level will be down to 0.63 and another dose will bring it up to 1.63. After waiting 180 days (about three half-lives), the blood level is down to 0.13 and a fourth dose will raise it to 1.13.

Now let's look at it from the perspective of the mice in this study. If they receive a dose at time zero that produces a blood level of 1.0 (arbitrary units), after two days, the blood level will only be down to 0.81. Another dose raises the blood level to 1.81 and, after another two days, it is down to only 1.46. Another dose raises the blood level to 2.46, which is down to 1.61 when the final dose raises it to 2.61.

So, the human experiences a maximum blood level of 1.63 (arbitrary units) and the mouse - since it is being dosed at a smaller fraction of its half-life - sees a maximum blood level of 2.61. In short, the mouse gets to a blood level 60% higher than the human.

The case of the underweight child:

Now, this all assumes that the mouse and the infant are not growing - which is not true - and that the doses they are getting are equivalent on a microgram per kilogram body weight basis, which also is not true. The study dosed the mice on a per kilogram basis equivalent to what a human infant would have received had they received their vaccinations on schedule. However - and this is a curious thing - they used the 10th percentile (10% of children that age weigh less, 90% weigh more) body weights for children at ages 2, 4, 6 and 12 months.

I found myself wondering, "Why didn't they use the 50th percentile (50% weigh more than this weight, 50% weigh less - sort of an 'average weight')?" I have no answer - but I have an idea. By using the 10th percentile, they were able to give the baby mice an even bigger dose of mercury. Let's run the numbers, as they say on NPR:

Dosing is based on the child receiving 62.5 micrograms of thimerosal at 2,4 and 6 months and 50 micrograms at 12 months. Using the numbers for 10th percentile weights and 50th percentile weights, this works out to doses of:

2 months
10th - 4.4 kg - 14.2 ug/kg
50th - 5.3 kg - 11.8 ug/kg (17% less)

4 months
10th - 5.7 kg - 10.8 ug/kg
50th - 6.8 kg - 9.2 ug/kg (15% less)

6 months
10th - 6.8 kg - 9.2 ug/kg
50th - 7.9 kg - 7.9 ug/kg (14% less)

12 months
10th - 9.0 kg - 5.6 ug/kg
50th - 10.3 kg - 4.9 ug/kg (13% less)

So, by using the 10th percentile weights, the authors were able to give the mice about 15% more thimerosal. This goes nicely with the dosing schedule to significantly raise the dose the mice receive.

The show must go on!:

What Dr. Hornig didn't show in her article, but is showing to concerned parents of autistic children, is the video of her autistic mice. The study article adresses such prosaic behavioral changes as decreased spontaneous movement, decreased exploration and decreased streotypic behaviors, but the video (CAUTION: large file) she shows includes such titillating tidbits as one mouse grooming another to death and another mouse biting its own tail. These, she says, are evidence that these mice have become autistic....

Hold the phone! Grooming is a social activity mice engage in, and autistic people are supposed to be averse to social interactions! And although self-injurious behavior is seen in autism, there are other, better explanations that were overlooked.

In fact, I felt a strange hot flush come over me as I watched the video (maybe it was just a hot flash). I have seen exactly this type of behavior before - but not associated with mercury.

Many years ago, I was associated with a group studying treatments for chronic neuropathic pain. The model we used was a rat model in which we injured the sciatic nerve and let it heal. A number of the rats would develop what was assumed to be neuropathic pain - evidenced (so we thought) by biting and chewing on the affected hind limb. It was a grisly sight and, what was worse, it wasn't really neuropathic pain.

Another group discovered that anything that caused abnormal sensation would cause rats (and other animals) to bite and gnaw on the affected limb. What I saw in these mice was exactly what we saw in the partially denervated rats. And - oh, by the way - mercury causes parasthesias (numbness and tingling). Dr. Hornig has probably managed to make these mice so neurotoxic that they are experiencing parasthesias - a common sign of mercury poisoning.

And now, a word from our sponsors!:

These days, no discussion of mercury and autism is complete without a thorough exposition of who paid for the study and who might have a conflict of interest. So, you might ask, who paid for this study?

The UC Davis M.I.N.D. Institute
The Coalition for Safe Minds

To remind you, the Coalition for Safe Minds' mission statement is:

"Our mission is to end the health and personal devastations caused by the needless use of mercury in medicines. "

But I'm sure that didn't influence the outcome of the study.

Happy reading!



Anonymous Anonymous said...

Thank you.

Explains a lot.

29 July, 2005 22:18  
Anonymous Anonymous said...

I wonder if anyone has ever injected this strain with mercury in the past?
Ahhhh, yes. I see they are even referred to as mercury-susceptible SJL mice. Seems like a safe choice for a Safe Mind.

These mice are so prone to autoimmunity that they develop autoimmune and neuroimmune disorders spontaneously. Of course Mady also knows that a viral infection works in this strain too.

30 July, 2005 11:18  
Blogger Autism Diva said...

Thank you Prometheus.

One can only hope that the ever so helpful Bobby Kennedy Jr. and David Kirby will get a chance to read this.


30 July, 2005 14:27  
Blogger jaws said...

Once again I'm left puzzled and suprised that a study this flawed could make it into (what I assume) is a peer-reviewed journal.

With all that being said, I've heard rumors from reliable soures that there are reputable research groups trying to create a mouse model of autism (through genetics, not Hg injections). One of the biggest difficulties they face though is determining what kind of mouse behavior is equivlent to that of Autistic humans.

30 July, 2005 16:36  
Anonymous Anonymous said...

I saw a presentation at the 2004 IMFAR conference by a Japanese researcher who was working on a "rat model" of autism. They injected pregnant rats with valproic acid, because prenatal exposure of Valproic acid (VPA?) can cause brain changes leading to autism or autistic like behavior anyway, in children. It has the same effect as thalidomide and another chemical (Dr Patricial Rodier's work) on embryos.

Anyway, the rats had short attention spans/short memories and got confused in the star shaped maze.

One of those in the audience said it looked like the Japanese researchers had created a rat model of Attention Deficit Disorder.

I personally wasn't convinced that it had any behaviors that could be stretched to be "autistic".

Is there a rat model for schizophrenia? I doubt it.

So much of the neocortex is involved in autism, as I understand it, that you couldn't really create an animal with a similar brain as autistics have.

They might be able to create animals with the same kinds of Purkinje cells problems in the cerebellum (maybe they have) but what about the tightly packed minicolumns in the cortex? Though I don't know if either of those are seen in all autistics.

Not all autistics have the problems in the brain stem that Patricia Rodier describes, either.

30 July, 2005 18:14  
Anonymous The Dirt said...

Why would they need to read it Diva? At least they can count

31 July, 2005 09:12  
Anonymous donna said...

I have a cat with allergies who grooms herself like crazy and gets antisocial (well, more than cats are already) when her allergies flair up. She gets this way because she itches and hurts. She's certainly not autistic.

What a dumb study.

31 July, 2005 13:54  
Anonymous Anonymous said...

Thank you,
Perhaps if you have the time you can take a look at another example of how NOT to do a study.
Environmental mercury release, special education rates, and autism disorder: an ecological study of Texas-Raymond F. Palmer

03 August, 2005 14:51  
Blogger JP said...

I'm curious about one thing - the Pichichero study (which isn't referenced here) talks about ethylmercury having a much shorter half-life than methylmercury...approximately 7 days. Admittedly, we don't have a similar reference model for rodents - but could the blood levels for mice be lower than what you originally postulate?

Your points about the relative ludicrousness of "autistic mice" and other explanations for their behaviors are well taken. uestions of toxicity aside, the study is basically useless.

04 August, 2005 14:49  
Blogger Prometheus said...

To jp:

If ethylmercury has such a significantly shorter half-life (7 days vs 50 days) in humans, it might be reasonable to assume that it would have a proportionally shorter half-life in the mouse. In that case, "dose stacking" would not occur to the degree that the numbers for methylmercury suggest.

However, other studies have shown that the half-life of various organomercury compounds in mice varies dramatically with strain - including the strains used in this study. What Dr. Hornig may have studied - without her knowledge - is the different organomercury half-lives in the three strains she investigated.

Regardless, nothing in the Hornig et al study connected to autism, since the behaviors she observed could have myriad explanations that have nothing to do with the site of autism - wherever that might be!


04 August, 2005 23:35  
Blogger JP said...

However, other studies have shown that the half-life of various organomercury compounds in mice varies dramatically with strain - including the strains used in this study. What Dr. Hornig may have studied - without her knowledge - is the different organomercury half-lives in the three strains she investigated.

I agree with that assessment. I just know that the first argument one would hear from the "anti-vax" side (irrational or not) is that "didn't Pichichero prove that ethylmercury has a shorter half-life"?

Regardless, nothing in the Hornig et al study connected to autism, since the behaviors she observed could have myriad explanations that have nothing to do with the site of autism - wherever that might be!

On that point, we both agree. As I said above, questions of toxicity aside, the study is basically useless in determining a cause (or even potential cause) for autism. All it shows is that a particular strain of mouse acts much differently when exposed to mercury.

05 August, 2005 09:15  
Blogger Citizen Cain said...

I agree that Hornig's mouse study isn't persuasive evidence that thimerosal exposure is associated with autism, and I generally am very skeptical of the argument that thimerosal contributes to autism (see

However, I disagree with the criticism of Hornig's use of the 10th percentile weight rather than 50th percentile weight. There's nothing wrong with a toxicological study trying to model exposures that mirror a more vulnerable segment of the human populuation. It is reasonable to assume that the children who are most vulnerable to harm from thimerosal are those who receive the same injections as everbody else but who are smaller than the average.

08 August, 2005 13:53  
Blogger Prometheus said...

To: Citizen Cain

I have no opjection to the use of the 10th percentile and I suppose that it was chosen for exactly the reasons you stated. However, there was no discussion in the article about why the 10th percentile was used, and it was a point that might be easily missed. As long as everyone is aware that the per kilogram dose was calculated using a weight that gives doses 15% or so greater than the median weight, all is well.

This might also be a good point to mention that the C57BL6 mouse strain used in the study has been known, since at least 1986, to have significantly faster excretion of methylmercury (and presumably ethylmercury) than other strains. I would have included this in the posting, but the papers just arrived via interlibrary loan.


08 August, 2005 14:07  
Blogger Vicoprofen said...

I think this topic is very different and should be more pages like this,Your
comment says a lot of truth, and I'd like to invited you
to visit my page:10/325-
Vicoprofen - Lortab- Tylenol
- Ativan
All Major Medications are available right here at:

25 October, 2006 13:15  

<< Home