A Note From Hyman on the Genetic Complexity of Psychiatric Disorders
“Our brains are not like Mendel’s peas.”
Steven Hyman, MD, is the Director of the Stanley Center for Psychiatric Research at the Broad Institute of MIT and Harvard in Cambridge, Massachusetts. He is also the Harvard University Distinguished Service Professor of Stem Cell and Regenerative Biology. Hyman was the director of the U.S. National Institute of Mental Health (NIMH) from 1996 to 2001. The following is personal correspondence from Hyman in response to my recent posts on genetics and schizophrenia (see here and here). It is being shared on Psychiatry at the Margins for the benefit of the readers with his permission and has been slightly edited for clarity.
Dear Awais,
I enjoy reading your blogs, and I fear that you are getting yourself into an intellectual labyrinth as you respond to interlocutors focused on whether psychiatric disorders are genetic or not. I think that it is most helpful to eschew narrow Mendelian views of genetics in thinking about common psychiatric disorders. While severe neurodevelopmental disorders (NDDs) caused by penetrant de novo mutations approximate Mendelian models of causality, Mendelian causal logic has been seen as inapplicable to common mental disorders for decades. I sometimes think that it is criminal that Mendel is taught in school without even a whiff of quantitative genetics to balance it.
To my mind, Mendel’s genius lay in his ability to ignore almost all traits of his pea plants except those that fit a deterministic dominant or recessive model of transmission. Heritabilities across all populations studied tell us that psychiatric disorders such as schizophrenia, bipolar disorder, ADHD, and anorexia nervosa are highly influenced by genes, but the high heritabilities are partitioned within populations into hundreds or perhaps thousands of alleles of small effect. Each person then receives a larger or smaller grab bag of risk alleles from their parents. That is not the whole story, however. In addition to loading for genetic risk, the etiology also reflects developmental bad luck (perhaps somatic mutations during early cell divisions in the brain, perhaps the leakiness of transcriptional regulation at just the wrong time and place, perhaps infelicitous brain wiring), and the person’s exposome1. These are all causal, just not by themselves. The ultimate state of one’s brain circuits, neural computations, and symptoms might be considered an emergent property resulting from many causes—and chance. In the context of such complexity, the Mendelian monogenic framework is not a reasonable method to investigate the small causal contribution of any particular allele (see the Visscher papers).
I am not sure why anyone who understands science would be hung up about whether to call a mental disorder “genetic” or not — high heritabilities tell us that genes matter a lot, but for complex disorders, genes by themselves are never fate. Some psychiatrist colleagues feel that genetics is just a new high-tech way of blaming families. The reality is that in every human genome, there are risk alleles for essentially all polygenic traits, including psychiatric disorders. The dark dreams of eugenicists are not only depraved but also scientifically untutored, as there is no human genome without many alleles that confer risks and many that confer possible benefits, and sometimes the same allele confers both. I know several people who feel guilty about familial mental illness, but given the complexity and probabilistic nature of risk, it would be churlish in the extreme to blame them. In addition, genetic contributions to illness should not encourage fatalism because genetic influences do not connote immutability. For the social scientists and psychiatrists still worried about genetic determinism my usual response is, “Our brains are not like Mendel’s peas.”
Heritabilities across all populations studied tell us that psychiatric disorders such as schizophrenia, bipolar disorder, ADHD, and anorexia nervosa are highly influenced by genes, but the high heritabilities are partitioned within populations into hundreds or perhaps thousands of alleles of small effect. In addition to loading for genetic risk, the etiology also reflects developmental bad luck (perhaps somatic mutations during early cell divisions in the brain, perhaps the leakiness of transcriptional regulation at just the wrong time and place, perhaps infelicitous brain wiring), and the person’s exposome. These are all causal, just not by themselves.
Even though, as NIMH director in the 1990’s, I vastly underestimated the level of genetic complexity of psychiatric disorders by a few orders of magnitude (as did almost everyone), Gottesman and Shields had already written a very forward-looking paper arguing for a polygenic model of schizophrenia in 1967. Had Robins and Guze read and understood it, they might not have foisted their categorical abominations on us in their short, seemingly reasonable, but ultimately disastrous 1970 paper that the undisciplined adventurer Spitzer turned into the DSM-III. They would have known, for example, that psychiatric disorders could not possibly be natural kinds, that they would not breed true, that they would be intrinsically dimensional with no bright lines between ill and well, and highly heterogeneous.
Finally, there is a reason why alleles associated with early-onset psychiatric disorders (ranging from autism spectrum to schizophrenia) must have a very small effect size if they are common or else be ultra-rare. The reason is negative selection. For example, the unfortunate kids with severe de novo NDD mutations (i.e., representing new mutations not carried by the parents) that are associated with significant intellectual disability, seizure risk, and autism inter alia, rarely have offspring themselves. And so these Mendelian or near-Mendelian mutations typically start and end with them. Because of the properties of our genomes, however, these same de novo mutations do recur to plague other unfortunate families. The ultra-rare mutations that increase schizophrenia risk by 20-fold or more also don’t last very many generations in the gene pool. The Power et al. (2013) paper on fecundity shows that a male with schizophrenia has approximately 20% as many children as his unaffected siblings. Alleles of small effect that are discovered by GWAS can become common in the human gene pool because they slip by under natural selection’s radar. This is because they do not exert significant individual effects on phenotype and contribute to mental illness only in infelicitous combinations—bad luck at meiosis—plus stochastic developmental effects and environmental exposures. There is a large literature—see Chen et al. (2024)—that identifies the fraction of human genes that do not tolerate loss of function (LoF) mutations. Some may be embryonically lethal, for example, but many LoF intolerant genes encode proteins involved in synaptic structure, function, and plasticity. Damaging mutations in these genes result in neuropsychiatric disorders that begin early in life, thus decreasing reproductive success and selecting against transmission of that allele. Once we are well past reproductive age, evolution stops loving us, and thus there can be common alleles with large effects like APOE4, associated with risk of Alzheimer’s disease. This thought of evolutionary abandonment pains me since I am old.
Steve
References and Suggested Readings
Chen, S., Francioli, L. C., Goodrich, J. K., Collins, R. L., Kanai, M., Wang, Q., ... & Karczewski, K. J. (2024). A genomic mutational constraint map using variation in 76,156 human genomes. Nature, 625, 92–100.
Hyman, S. E. (2018). The daunting polygenicity of mental illness: making a new map. Philosophical Transactions of the Royal Society B: Biological Sciences, 373(1742), 20170031.
Gottesman, I. I., & Shields, J. (1967). A polygenic theory of schizophrenia. Proc Natl Acad Sci USA, 58(1), 199-205.
Power, R. A., Kyaga, S., Uher, R., MacCabe, J. H., Långström, N., Landen, M., ... & Svensson, A. C. (2013). Fecundity of patients with schizophrenia, autism, bipolar disorder, depression, anorexia nervosa, or substance abuse vs their unaffected siblings. JAMA Psychiatry, 70(1), 22-30.
Robins, E., & Guze, S. B. (1970). Establishment of Diagnostic Validity in Psychiatric Illness: Its Application to Schizophrenia. Am J Psychiatry, 126, 983–87.
Visscher, P. M., Hill, W. G., & Wray, N. R. (2008). Heritability in the genomics era—concepts and misconceptions. Nature reviews genetics, 9(4), 255-266.
Visscher, P. M., Yengo, L., Cox, N. J., & Wray, N. R. (2021). Discovery and implications of polygenicity of common diseases. Science, 373(6562), 1468-1473.