Skip to Main Content

Far be it from us to tell 23andMe how to run its business, but if it or any other DNA company wants to give customers a better read of disease risk, they might start asking how mom’s pregnancy went.

That’s the key message of a schizophrenia study, published Monday, which showed that 108 regions of the genome previously identified as raising the risk of schizophrenia do so only slightly if the mother experiences no complications during pregnancy — but by some twelvefold if she does.


Taking into account prenatal conditions gives a starkly different read of how risky schizophrenia risk genes are: Taken together, they raise the chance of developing the disease to 12 in 1,000 if there are no complications — or to 80 in 1,000 if there are.

“This should be an eye-opening study, especially for anyone who thinks disease risk is all genetic,” said Janine LaSalle, of the University of California, Davis, who studies the genetics of autism and was not involved in the schizophrenia research. “Genes don’t exist in a lock-box away from everything else that happens to you.”

The idea that both genes and environment affect risk of disease is conventional scientific wisdom. But that usually means that while both genes and environment are involved, they act via separate biological pathways. BRCA mutations can raise the risk of breast cancer, for instance, and so can not bearing children, but the two don’t work by the same molecular mechanism.


The schizophrenia study, in Nature Medicine, finds something more intriguing. Whether or not risk genes live up to their billing depends on what environment they find themselves in. For schizophrenia, that means the placenta: It can act like either a bump stock on a rifle, increasing the power of schizophrenia genes, or a broken firing pin, rendering the genes non-functional.

To figure this out, Dr. Daniel Weinberger of Johns Hopkins University and his colleagues started with the 108 genes linked to schizophrenia. They constructed what’s called a polygenic risk score, which calculates the extra risk of a disease from numerous genetic variants. In thousands of healthy people and people with schizophrenia, the scientists found that what is thought to be a high risk score raised the risk of schizophrenia substantially only if in utero complications such as a mother’s diabetes, obesity, pre-eclampsia, or smoking were also present.

For instance, a high score raised the risk more than 800 percent if any of these conditions was present, but just 50 percent if the pregnancy had been healthy. A 50 percent higher risk translates to 12 rather than 8 per 1,000 people developing schizophrenia.

The 108 genes that make up the risk score are expressed at particularly high levels in the placenta, the study also found, and at higher levels in the placentas of male fetuses than female ones. That fits with the fact that more men than women develop schizophrenia, usually in young adulthood.

The activity of the schizophrenia genes is dialed up especially if the mother has pre-eclampsia or another pregnancy complication. Presumably, “this reflects the placenta’s response to stress,” Weinberger said, such that when it’s flooded with inflammatory molecules (the standard response to stress), “it turns on genes related to immune response.” Recent discoveries connect schizophrenia to the immune system.

Though specifics of how placental health and immune status affect fetal brain health are poorly understood, there is strong circumstantial evidence that it does. The chance of developing the disease is three to seven times greater in people whose mother had a respiratory infection during pregnancy, which can cause inflammation in the placenta, and up to twice as great if the mother had pre-eclampsia, a difficult delivery, or diabetes, obesity, alcohol use, vaginal bleeding, maternal smoking, or preterm birth, all of which can also affect the placenta.

“We need to create a new risk score for schizophrenia, incorporating not only genes but also placental health,” Weinberger said. “The odds of becoming schizophrenic based on your polygenic risk score is more than 10 times greater with these early-life complications than without them.” But the genes-only risk score, “in the absence of a serious complicated pregnancy, explains very little risk for schizophrenia — less that 2 percent” of the chance the individual will develop the disease.

UC Davis’ LaSalle, who studies the expression of autism-related genes in the placenta, also thinks it important to include the environment in genetic risk scores, since “if you don’t have the environmental risk you don’t have the genetic risk.” But more work needs to be done to identify which specific environmental influences on the placenta matter, she said.

23andMe doesn’t disagree. The company’s disease-risk analyses emphasize that, with the exception of single-gene disorders such as Huntington’s, you can have disease genes yet not develop the disease, or develop a disease despite not having the disease genes.

None of the analyses currently incorporate genotype-environment interactions like those found in the schizophrenia study. But at some point, said a 23andMe spokesman, if the technology and research identify more “GxE”’s for more disorders, and if the Food and Drug Administration signs off, that could well change.

  • Thought of posting these two excellent references (regarding genetics of psychiatric conditions) that I happened to come across:

    Zubenko, G. S., Sommer, B. R., & Cohen, B. M. (2018). On the Marketing and Use of Pharmacogenetic Tests for Psychiatric Treatment. JAMA psychiatry.

    Ross, C. A. (2013). Biology and Genetics in DSM-5. Ethical Human Psychology and Psychiatry, 15(3), 195-198.

    • There are no exact genes responsible for “schizophrenia.” The 108 genes mentioned are ASSOCIATED with a higher risk of schizophrenia, but also with the development of other “mental disorders.” More significantly, these genes altogether explained at best something like 15% of the total risk, so that means a lot of people who are diagnosed with schizophrenia don’t have these genes. Also worth noting is that not all of these same genes are present in those who DO have this supposed genetic risk factor – they just have SOME of this set of genes. So all in all, the idea that these are “schizophrenia genes” is largely propaganda.

    • Thank you for this information Steve. Most articles nowadays present information in a way that implies to the reader that ‘all mental disorders have a genetic origin.’ If you happen to have an academic reference for this helpful information you have provided, could you please cite it? Thank you!

    • You have to read carefully to suss out what they are saying. I’m still looking for a direct analysis from a critical viewpoint, but here is a sample of the kind of quote we see:

      “A genetic analysis of more than 65,000 people with and without schizophrenia identified that those who are carrying a particular variation of the gene – one that leads to a higher expression of it in the immune system – had a higher risk of developing schizophrenia in adolescence.”

      Notice how they carefully say “a higher RISK of developing schizophrenia in adolescence” but fail to mention how much higher that risk is. This phrasing clearly suggests that not everyone with this variation develops schizophrenia. It also fails to address the very important question of what percentage of people who develop schizophrenia do NOT have this gene. For genetic causality to be shown, we’d need to see that MOST of the people who have schizophrenia DO have this gene (or collection of genes) and MOST (if not all) people who don’t have schizophrenia do NOT have this gene.

      What we are really looking at in these “Genome Wide Association Studies” (GWAS) is taking a whole bunch of genes and a very large sample size and trying to find any CORRELATION between schizophrenia and one or more combinations of these genes. These correlations can be as low as 1% or less of the total diagnosed population in some cases, because very large samples can give a positive correlation in even a very small overlap. Unless these genes are always or almost always found to exist in people diagnosed and rarely if ever found in the non-diagnosed population, they can not be called “schizophrenia genes.” Moreover, even if a higher correlation is found, the genes in question may be coding for sensitivity, or high creativity, or high visual discrimination, or any of a ton of other qualities, and schizophrenia may just be an occasional event that happens when under high stress. With child abuse correlations (“The weighted average prevalence of either sexual or physical abuse was claimed to be 59% in males and 69% in females, and the prevalence of both types 19% in males and 36% in females.” being far higher than those for any combination of genes, it should be obvious to anyone that whatever risk factors may come from genetics, abuse and other forms of stress (migration to another country and urban living situations in particular) contribute at least half and probably a whole lot more to schizophrenia diagnoses than genes.

      The other puzzling question is: since genes can’t be changed, and environments can, and since more than half of the variation appears to be environmental, why are we spending so much on researching the part we can’t change, and pretending the part we can change has no impact?

    • Thank you Steve! You make some very good points – yes, many top universities still appear to focus on researching genetic contributions to schizophrenia – perhaps they are suffering from confirmation bias! Thanks also for the links!

  • The Australian branch of The International Society for Psychological and Social Approaches to Psychosis has decided to drop the label for Schizophrenia (because the label lacks a scientific basis, leads to stigmatization and loss of hope). Check out the article titled “Schizophrenia Awareness Week – Drop the Label!”

    • You asked for a reference. I just found an article by coincidence today. I don’t know the quality.
      “The emperors of the schizophrenia polygene have no clothes.” Psychol Med. 2008 Dec;38(12):1681-5. doi: 10.1017/S0033291708003395. Epub 2008 Apr 21.

      “candidate genes…. have proven unreplicated in large sibling pair linkage surveys and a targeted association study.”

  • The ability to diagnose schizophrenia is arbitrary. With this fact the science of linking the behaviour to the DNA is also not scientific.
    If you want to read a criticism of the science of behaviour originating in the DNA , look for Jay Joseph, author of The Gene Illusion .

    • Below are some references that further point out how LOW the genetic risk for psychiatric disorders actually are – thought someone may find them to be useful:

      Joyner, M. J., Paneth, N., & Ioannidis, J. P. (2016). What happens when underperforming big ideas in research become entrenched?. JAMA, 316(13), 1355-1356.

      Leo, J. (2018). Memo to the newest generation of gene hunters: read Jay Joseph. Psychosis, 10(1), 58-61.

      Kendler, K. S., & O’donovan, M. C. (2014). A breakthrough in schizophrenia genetics. JAMA psychiatry, 71(12), 1319-1320.

      Joseph, J. (2017). Schizophrenia and genetics: The end of an illusion. BookBaby.

      Leo, J., & Joseph, J. (2002). Schizophrenia: Medical students are taught it’s all in the genes: But are they hearing the whole story? Ethical Human Psychology and Psychiatry, 4, 17–30.

  • Changes in genetic expression happen throughout life – not just in the prenatal period – i.e., stressful life events (including social stresses) as well as how one deals with these stresses (various coping skills) constantly alter genetic expression throughout life.

Comments are closed.