Friday, January 18, 2008

Microbes and Chronic Disease

In the US, most deaths are attributable to chronic afflictions, such as heart disease and cancer. Typically the medical community has attributed these diseases to accumulated damage, such as plaque formation in arteries or mutations in genes controlling cellular replication.

This view is changing.

Scientists are now beginning to recognize that many of these chronic illnesses are due to microbial infections. A recent report in the American Journal of Psychiatry suggests that schizophrenia, a mental illness leading to errors in perception, is associated with the pathogen, Toxoplasma gondii.

"Our findings reveal the strongest association we've seen yet between infection with this very common parasite and the subsequent development of schizophrenia," study investigator Dr. Robert Yolken of John Hopkins Children's Center in Baltimore said in a statement.

Toxoplasma gondii
, a parasitic protozoa usually carried by cats, is an interesting microbe, having been associated with behavioral changes in humans and rats.

Other diseases are also showing microbial connections. The most famous example is Helicobacter pylori, the unequivocal cause of peptic ulcers and suspected agent of gastric cancer. Human papillomavirus is associated with 90% of cervical cancers. Hepatitis B is linked with 60% of liver cancers.

Some diseases look suspiciously like infections, such as Hodgkin's disease, multiple sclerosis, juvenile onset diabetes and Crohn's disease.

In an evolutionary terms, these speculations make sense. Microbes exist to pass on their genes, and they may have evolved ways to "cryptic" rather than "conspicuous". Paul Ewald divides diseases into three categories: environmental, genetic and infectious. Environmental diseases are acquired from toxins such as those in cigarettes and pollution. Genetic diseases are those caused by errors in replication and development. Ewald reasons that, if diseases are too common to have arisen by random mutation and too ruinous to have survived natural selection, and if it is not environmental, then it must be infectious. By being cryptic, these infectious diseases enhance their spread by increasing the odds of transmission.

Toxoplasma is one of Carl Zimmer's favorite parasites.

Photo: Toxoplasma gondii from Wikicommons.


  1. Thanks for freaking me out. Evolution=awesome.

  2. I think there's an important tension in Ewald's brilliant arguments in favor of microbial etiologies - a tension which he points out himself, but does not exactly highlight. Namely, the fact that the deep, evolutionary cause of a disease may be a microbe that never has to actually touch the diseased individual. The classic example is sickle cell disease, an illness that's ultimately "due" to malaria, since the causitive allele exists only because it gives heterozygotes superior net fitness in malaria endemic zones. Analogous ideas exist regarding at least some other common-ish recessive monogenic diseases, including cystic fibrosis (the microbe there being cholera).

    Going further, the same sort of thinking could suggest autoimmune pathogenesis of many immune diseases, with microbes as a deep cause which might not ever touch the diseased individual, but instead select for autoimmunogenic alleles in the host population. Over the long run of the host-parasite arms race, hosts should evolve alleles that suppress parasites with minimal (almost insignifcant) cost to the host. But parasites always evolve faster than hosts, due to having far shorter generation times, so the arms race always continues. It strikes me that over the short run, each time parasite pressure rises due to a parasite evolving a new protein tool (which is often), the host population might in turn experience, as a "quick and dirty fix," an increase in the frequency of alleles favoring "hair-triggered" B and T lymphocytes (which would tend to promote both autoimmunity and anti-parasite immunity). This could happen again and again because the quick and dirty fix is often simpler and easier than the "clean fix" in terms of fitness landscapes and "proximity in genome sequence space" and such. (Y'all probably have a better way of putting this in the evo field, and better thoughts about it as well, but I am jes' a country patho-biologist). The clean fix might often (not always) be very, very slow to evolve - especially when the host faces not merely a parasite with a new protein tool, but one with a radically new tool acquired by horizontal gene transfer. Or, for that matter, when the host faces a whole new parasite taxon that has not infected this host lineage at all for the past 10,000 to 1,000,000 years or more. Eventually, a clean fix does evolve in the host, and then frequency of "hair-trigger" B and T lymphocyte alleles declines (because their their advantages have become redundant, while their downside remains the same). But their frequency will always be higher than what it "should" be - what it would be if fully safeguarding against autoimmunity were far, far more important than suppressing parasites.

    I don't actually know of any empirical evidence that autoimmune-prone mammals have better xenoimmunity against parasites. So, pardon my speculation. But the sickle cell example is solid, anyway; the speculation about autoimmunity shows that it's at least possible there could be other classes of "microbe-sponsored" disease where microbes need not touch the host. The fact that plasmodia can "give" you sickle cell disease in this way attenuates Ewald's microbial etiology arguments a bit - the predictions that come out of his thought are rendered significantly less bold and clean. For example, should we feel almost sure that microbes must be hidden somewhere in a disease lesion - and therefore assign high confidence to B Balin's results (chlamydiae in the Alzheimer's brain), despite the failure of other groups to replicate those results? No, probably not. (Although there are other reasons to maintain robust interest in Balin's apparantly disconfirmed result - such as the relative difficulty of explaining how Balin's result could be falsely positive, in contrast to the relative ease of obtaining inexplicably negative results when using most kinds of sophisticated labwork.)

    Nevertheless, I am a slavish Ewald admirer - the value of his viewpoint remains overwhelming despite this "tension." Also, he has worked on several other subjects (such as virulence evolution) just as much he has on theoretics of microbial etiology of chronic disease. I'd be proud to shake the man's hand, buy him a drink, shovel snow off his driveway for free, etc. Actually, he spoke at my university a while ago, and I somehow missed finding out about it until 15 minutes after his talk ended; I totally flipped out.

    By the way, IMHO the following is the single most amazing of all the many obscure-ish papers on concrete detection of a microbe in a chronic disease: PMID 11787831.

    The text is free here - unfortunately without the high-quality immunoelectron micrographs which are pretty much the whole point:

    Many non-bacteriologists are prone to think that any cryptic bacterial cause of chronic disease ought to have already been uncovered rather easily, by accident, because it should respond rapidly to antibacterials (which most people take occasionally for one reason or another). So I'll just mention that HIV-negative M. avium complex infection in man is an example of a bacteriosis which is often very poorly poorly responsive to even long-term combination antibacterial treatment. Free text:

    Demonstrations and theoretics of why bacteria may sometimes enjoy nonspecific, nongenetic resistance to antibacterials (and other insults) are reviewed in PMIDs 17215163, 15807669, 17143318, and numerous others.

  3. Fascinating comments. A common theme in evolutionary biology is the trade-off, and the cost of resistance is certainly one. For example, a bacterium that is resistant to, say, ampicillin, is less fit in ampicillin-free media than a wildtype bacterium of the same species.

    Sickle-cell anemia is a perfect example of this trade-off. I debated whether to include this in the article, but decided to focus on diseases directly linked to human ailments. Its fascinating to speculate that many other human pathologies may have indirect microbial causes, but it is pretty difficult to tease these associations out to the standards of scientific evidence. Sickle cell is so obvious, but others, such as those you mention, are pretty cryptic. Thanks for the wonderful comments.

  4. I liked Ewald's book, but if there are many different mutations that cause the "same" genetic disease, would that disease still have to be rare?

  5. Nice blog. Interesting comment above on "trade-off" in pathogenesis. When I was a surgical resident in the 1960s, my professor had a theory that non-resistant bacteria would displace the antibiotic resistant strain. For this purpose, he kept a culture of E-Coli that was sensitive to all antibiotics in the hospital lab. If a patient came in for elective surgery who had been on broad spectrum antibiotics (the 1965 variety), we were instructed to culture the stool and check for resistant strains. The bacteriologists were always annoyed by this but he had enough sway to overcome the purist objections. If we found a resistant strain of E. Coli, the patient was given the sensitive strain in a malted milkshake. A few days later, the culture would always show replacement of the resistant strain. Of course, we had no knowledge of B. fragilis in those days so I don't know how effective our practice was in practical terms but it did support the "trade-off" theory.

  6. Good point, Ford. Multiple mutations can lead to the same disease.

    Fascinating story, Mike. I see more and more evidence for the role of microbes in health and disease, particularly in ways that we don't expect. In the 1/12/08 issue of New Scientist (, an article mentions that dairy farmers may have 5x less lung cancer, possibly due to exposure to dried manure.

    I wonder however if you could give someone an E. coli milkshake these days?

  7. That's a neat experiment. I don't know if it's been tried, but at least one author has suggested that probiotics could have a role in keeping both hands and wounds "clean" - well, healthy (clean of MRSA etc), if not literally clean:

    The generation time of E. coli in the GI is often quoted at 12-24 hours, although I don't personally know what the evidence is. I wonder if your strain might have had something more on its side, in addition to the wild type specs that confer higher fitness in an abx-free environment. I'm thinking of colicins here, though maybe lysogenic phage could be another possibility. Different strains vary in their colicin production capacities and colicin sensitivities; I wouldn't know whether strains exist which are able to lyse almost all other strains.

    Or maybe the wild type is indeed so much fitter that it can "peacefully" displace an abx-resistant strain in just a few generations. After all there is only ~1 gram of E coli in the gut. I can't quite think out how it all works. I guess major factors are the time needed for intestinal epithelial cells to slough off (I don't have a number), and the amount of probiotic innoculum that survives the stomach.

    I wonder if there would be any chance of displacing MRSA from the nares? Perhaps one could even use an engineered S. aureus that's been shorn of most of its toxins? - I'm not familiar with whether any/all of the toxins are needed to begin or sustain quiescent colonization of the nares. I bet someone has thought of this, but it might be hard to google out.

    Ever heard of "stool transfusion?" Hopefully no one's having a sandwich or something while reading. Anyway, it's actually been studied for ulcerative colitis and for relapsing C diff. I don't know whether it's ever been broadly adopted.

  8. I wonder why p53 doesn't evolve a little, considering that, as most biologists have probably heard, it is lesioned in over 50% of tumors - and is conserved all the way back to the mouse.

    In doing a little googling, I'm mainly seeing the opinion that human p53 is functional solely as a homotetramer, and is thus suceptible to dominant-negative effects. Surely(?) the whole thing could be rearranged somehow so that it wouldn't be like this - so that both copies would need to be lesioned for loss of function, and any somatic mutation would prove recessive.

    I once asked a postdoc this question; she suggested cancer might not have been a serious selection pressure on man during pre-agricultural history (ie prior to 8000 BC). This link offers total cancer incidence by age in the UK:

    The incidence doesn't look all that low to me, even at "reasonably" young ages. While I haven't read any of the material, I am aware that there is a bit of disagreement in the field of figuring out the primordial human lifespan (and that archaeological methods have been employed as well as examination of present-day tribes in New Guinea leading reasonably pristine nonagricultral lives.)