Psychotropic Base Rates: The Argument From Antibiotics

The obscure antiprotozoal drug suramin has the prettiest molecular structure I’ve ever seen. It also has some evidence as a potential treatment for autism based on a cell danger response model of the condition.

I’m not going to get too excited here, because there are a lot of things that can cure stuff in mice. Still, one has to ask – an antiprotozoal? Really?

Protozoa are primitive little parasites kind of like bacteria. The most famous is plasmodium, which causes malaria. There’s not much reason an antiprotozoal drug should cross-react with the brain, so if it does treat autism it’s probably just a coincidence.

But it’s a pretty common one. I’ve already noted how an antibiotic used to treat acne, minocycline, is a promising schizophrenia treatment. The news article on such is called “Scientists Shocked To Find Antibiotics Alleviate Symptoms Of Schizophrenia”, but maybe by this point they should start being less shocked.

Off the top of my head, I can think of two other antibiotics with a significant psychiatric role. Iproniazid, the first antidepressant ever discovered, was originally used as an anti-tuberculosis drug – and just by eyeballing the chemical structure it’s pretty easy to see the relationship to current-mainstay-of-tuberculosis-treatment isoniazid. There are records – amusing in retrospect – of doctors remarking on how unusually happy and excited patients were to finally be getting treatment for their tuberculosis. Eventually someone put two and two together. realized the drug itself was a mood-lifting agent, and the antidepressants were born.

Cycloserine is a totally different antitubercular drug that doesn’t even share a chemical structure or mode of action with iproniazid. Nevertheless, it seems to affect classical conditioning in interesting ways, which makes it of use to psychiatrists. The most exciting possibility is that it speeds up the extinction response, meaning it could theoretically help someone “unlearn” behavior. There’s a common use – which the evidence only ambiguously supports – where you use it to treat something like social anxiety disorder by having a patient take it in relatively safe social situations. When nothing bad happens (hopefully), the cycloserine speeds up the usual process of “unlearning” the social situation-fear link and the patient gets better more quickly than if they had to become comfortable with crowds the old fashioned way. There are also some proposed uses regarding cocaine and other drug addictions.

And these are just the ones with good psychiatric effects. Less positive psychiatric effects are a dime a dozen in antibacterials and antiprotozoals – for example, people on mefloquine do some pretty weird stuff.

There’s no really good reason why antibiotics should have psychiatric effects. As mentioned before, beyond the fact that we’re selecting for bioactive chemicals here, it’s probably just coincidence. But that itself is a very interesting finding. If we think of antibiotics as chemicals chosen at random – as far as psychiatry is concerned – that means that random chemicals will often change mental processes around in important ways.

This shouldn’t be surprising – the brain is full of stuff and pretty easy to chemically disrupt. But it’s worth remembering. A lot of skepticism about new drugs – or new toxicity claims – comes from low base rates: the expectation that most chemicals are not active medications. But in psychiatry, the base rates might be higher than we think.

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

31 Responses to Psychotropic Base Rates: The Argument From Antibiotics

  1. Nisan says:

    If the base rate for psychoactivity is high, we should expect to find several psychoactive chemicals in the kitchen. Nutmeg and poppies come to mind. Alcohol, coffee, and tea are psychoactive food, but they may have been selected more for their psychoactive properties than their taste or nutrition. I’m not sure which category chocolate falls into. In your depression post you mentioned saffron and curcumin. Various nutrient deficiencies and ordinary meals can affect the mind in surprising ways (as in the Israeli judges study), but those don’t really count as an accident of pharmacology.

    • Douglas Knight says:

      There are two selection pressures. In addition to the humans selecting for the kitchen, why did the plant produce the spice in the first place? They are generally antimicrobial. They are selected to have biological effects, so it should be no surprise that they have additional biological effects. Caffeine and nicotine are insect neurotoxins. They work on the brain because that is their job.

      Conditioning on biological effect is the same situation as in the post.

      • Douglas Knight says:

        Actually, there is a difference between spices and oral antibiotics, which is that oral drugs have been selected for getting past the stomach. Spices aimed at animals are also selected for this, but most spices are aimed at microbes.

  2. Brett says:

    The other obvious possibility is that antibiotics can (sometimes) treat schizophrenia because infection is part of the etiology of the disease.

    • Scott Alexander says:

      We know exactly why iproniazid works – it’s a MAO inhibitor aside from its antibiotic properties. And we’re pretty sure we know why cycloserine works – it’s an NDMA partial agonist.

      There only a theory for why suramin works, but even if it’s wrong, it’s a good bet autism isn’t a protozoal disease – especially in the mouse model used in the experiment where it was deliberately induced with an in utero viral infection.

      • CaptainBooshi says:

        Thank you for clarifying this! I was wondering the exact same thing after reading that part of the article, and I wasn’t looking forward to researching to find out if this was likely, since I don’t know enough in this area to know what’s good research or bad.

      • Ialdabaoth says:

        Query: what changes did the viral infection have? Were they genetic, epigenetic or more downstream?

  3. Ialdabaoth says:

    Theory: Toxoplasmosis infection replicates some of the symptoms of schizophrenia sufficiently to be labeled ‘schizophrenia’; antiprotozoals stress the toxoplasma gondii enough that they are no longer attempting to influence the human’s behavior into being eaten by a cat, on the erroneous assumption that the human is in fact a mouse.

    • Scott Alexander says:

      Suramin seems to work for autism, not schizophrenia.

      And it works against a mouse model of autism produced by infecting the mice’s mothers with a virus during pregnancy – in other words, we already know why the mice are faux-autistic, and it’s not toxo.

      As far as I know minocycline, which does work against schizophrenia, isn’t antiprotozoal.

      • Levi Aul says:

        There’s a table at the end of this linking to a bunch of suramin studies. I don’t have time to go through them all, but my prior would be—given that so many studies were done—that there was a promising response in more than just the autism trial.

        Edit: but schizophrenia is nowhere to be found on that table! And all the results on googling “schizophrenia suramin” are just articles that compare schizophrenia to autism and then talk about suramin’s effect on autism. Have no schizophrenia suramin studies been done, or is there a null result paper everyone’s conspicuously ignoring?

      • Anonymous says:

        Wikipedia says that minocycline is sometimes used to treat toxoplasmosis:

        https://en.wikipedia.org/wiki/Toxoplasmosis

        …which suggests that it does have an antiprotozoal effect.

        Toxo is so weird. I am not sure whether to be worried about the link between latent toxoplasmosis and increased risk of traffic accidents. There is at least a 30% chance that I am infected.

        As for minocycline, I am also reminded of the thing about minocycline improving decision-making ability:

        http://lesswrong.com/lw/hdn/link_antibiotic_seemingly_improves_decisionmaking/

  4. suntzuanime says:

    This post gave me the urge to try quaff-IDing every liquid in the house. I dunno if that sort of response is sufficiently rare that you don’t need a “don’t experiment with the psychoactive effects of random chemicals, dipshits” disclaimer.

  5. Steve Johnson says:

    “There’s no really good reason why antibiotics should have psychiatric effects.”

    Again, unless your base assumption that parasites that effect behavior are rare is incorrect.

    I wonder when they’ll find Cochran’s hypothesized gay germ.

    • Scott Alexander says:

      See above comments like here – we know why these work and it’s not by antiparasitic means.

    • Paul Torek says:

      Bingo. Only, for “parasites” substitute “bacteria”. Isn’t there substantial evidence that intestinal bacteria populations affect mental states? I admit I was half-dreaming when I heard that on the radio, but I think that’s what I heard.

      To Scott’s reply: do we know all the reasons why selected antibiotics work, or do we know at least some reasons?

  6. jrayhawk says:

    Re: antibiotics and psychosis: UTIs come to mind.

    http://www.drtimothylau.com/documents/UTI%20and%20SCZ%20JCP%202013.pdf
    http://europepmc.org/abstract/MED/24499998

    http://evolutionarypsychiatry.blogspot.com/2013/06/infection-and-psychosis-in-schizophrenia.html

    I expect will be getting more interesting symptomology connections out of the exploding research interest in intestinal flora, too, in the next few years.

  7. This is interesting (I take it as speculation at this point).

    Would higher base rates imply higher base-rates for things like geographically linked environmental exposure? I.e., you move between places, are exposed to a different mix of environmental chemicals and, boom, you become more religiously oriented. A very 19th century idea, with a molecular twist.

    • Levi Aul says:

      AFAIK, this happens, but only in the same ways that inflammatory response can produce long-lasting effects: either by affecting foetal development, or by methylating your gamete DNA to produce epigenetic changes in the following generation.

      Which is to say, if you’re a person with high openness-to-experience who happens to stay in an area with high parasite load right around puberty, don’t be surprised if your kids come out with low openness-to-experience, even if you’ve long since moved away.

  8. Quixote says:

    Would higher base rates of activity be downthink that should make us more inclined to take “pesticides cause peanut allergies” or other theories that the auto immune disease / allergy increase is enviormentaly caused more seriously?

  9. Sarah says:

    Yeah, my heuristic is “damn near everything is a brain drug.” Birth control. Aspirin. Really almost everything affects your state of mind.

    In my foolish youth I took thermogenic pills (meant to cause weight loss) and they made me absurdly moody. Turns out they’re full of MAO-inhibitors! When you take random shit, kids, wiki it first.

  10. zslastman says:

    Your general point stands but you are mistaken in treating anti-protozoals as being drawn from the space of random small molecules. Rather they are being drawn from the space of molecules which bind protozoal proteins, and in particular surface receptors. This increases the likelihood of them being psychotropically active because neuronal proteins may be homologous to the surface receptors of simpler organisms (they had to evolve from something).

    Also, specific psychotropic effects do not imply specific neuronal biochemical activity because neurons are precious little snowflakes and are often the first cells to manifest an effect from things which have a very general activity. E.g. alcohol, or faulty mitochondria.

  11. Interesting facts, and I like that you start from a simple theory: drugs selected for an effect on bacteria or protozoa might have some random effect on the brain as well. A more elaborate description of random effect for antibiotics should probably include the indirect impact of changing our bacterial biome on the prevailing chemical soup of bacterial excretions (that is, you shouldn’t only make guesses about the shape of the drug itself because direct activity on the brain is, for a “random” chemical shape (it’s not random; we should expect all evolved life to have similarities), less likely than indirect through an already known antibiotic effect).

    Anyway, the above was just to quibble about the nature of the “random effect on the brain”. I agree w/ the conclusion that the baseline likelihood for subjectively psych-drug-like effects is pretty high (look at how many things people figured out how to get high off).

    • pxib says:

      We are also, on a very superficial level, integrated colonies of protists. Those eukaryote cells are, in turn, walled and gated factories employing what are functionally prokaryote workers and mechanisms. Chemistry which influences protists will, naturally, influence the operation of several of our colonies, and chemistry which influences bacteria may occasionally sneak through the factory gates.

      At a deep level, all (non-viral) life operates like other life. Many critical and widespread adaptations occurred only once and a few of the creatures that managed them (famously the chloroplasts and mitochondria) have simply been installed as symbiotes.

      It’s uniquely difficult to notice brain effects in the same way that we notice, for example, liver effects… but it’s not surprise that they should occur.

  12. A440 says:

    In general, I would expect a good fraction of chemicals that can cross the blood-brain barrier to have some sort of psychoactive effect. But, of course, many can’t.

  13. Sarah says:

    By the way, let me be the first to cast GIANT DOUBTS on mouse models of autism.

    There are theories, but no consensus, on the physiological causes of autism in humans. In fact, we don’t know that there’s one cause rather than many.

    The diagnosis of autism in humans is behavioral, and very vague, and relies heavily on “high-level” skills like language and socializing, which mice, of course, don’t have in the same way humans do. I suspect that lots of different kinds of neural abnormalities can break language, for the same reason that lots of different kinds of damage can cause a car not to start. Not to mention that observable autistic traits vary a whole lot.

    People say that “cancer’s not one disease.” But cancer at least has a common *form* — uncontrollably dividing, undifferentiated cells. And you can attack many kinds of cancer at once through mechanisms that deal with that form — hit it in the glucose supply, for instance. I think that, to our current knowledge, autism is a lot less of a unified phenomenon than cancer is.

    Some mouse models of autism (or cousin conditions) are relatively sane. For example, the FMR1 gene, loss-of-function mutations in which cause Fragile X syndrome, causes some social/cognitive stuff in knockout mice. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3396120/

    But this “maternal immune activation” thing is just a “face value” model. That is, when you stimulate the mouse mother’s immune system, you get babies who vocalize oddly, self-groom more, and approach other mice less. Is that autism? I’d say the answer is “mu” or “there’s no way to tell.” Do similar mechanisms cause autism in humans? Well, there’s conflicting evidence in the epidemiological studies whether mothers exposed to viruses during pregnancy are more likely to have autistic children.

    IL-6 is today’s favorite Big Bad, and my gut says stuff like this (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3068755/) is probably buzzword-salad when you look at it closely. I’m not sure it’s not true, I just don’t trust these fuckers. They have no incentive at all not to fool themselves spinning these conjunctive webs of regulatory pathways.

    My heuristic is to look at what actual differences (genetic & physiological) show up between autistics and nonautistics, and maybe even describe behavioral traits observationally again, because people see what they want to see. The mouse thing is on the shady side, even for an animal model.

  14. Anthony says:

    If you want more pretty molecular structures, look at Derek Lowe’s blog. I particularly like Things I won’t work with. Hexanitrohexaazaisowurtzitane is lovely! But he’s got chemical structures on all sorts of posts.

    • Nornagest says:

      That series of posts is a special kind of hair-raising. I usually only hear the like from soldiers and security researchers.

      • Anthony says:

        If you’re in the field, his jobs posts might be as hair-raising, but yeah – that’s some scary stuff.

  15. Alrenous says:

    Makes sense. It is the nature of complex structures to be delicate and sensitive. Most food is psychoactive in some subtle way.

    Is there any plausibility to presuming there’s symbiotic brain bacteria? I was going to say that the lethal properties of the chemical can’t possibly be an explanation when the effect is negative. Then I realized I thought the brain was sterile of foreign life and I have zero evidence for that. Time to accept my ignorance.