The Genetics of Schizophrenia
A Traitwell crosspost
Every now and again I’ll post this sort of stuff about genetics.
This was written by Gavan Tredoux, the chief science officer of Traitwell.
E. Fuller Torrey liked to ask his schizophrenic patients to explain the meaning of proverbs, which they almost invariably cannot fathom. For instance: "People who live in glass houses should not throw stones." One of them replied "I used to live in a glass house, but all I did was wave."
Schizophrenia is a thought disorder, as opposed to a mood disorder like depression. The term literally means “fragmented mind” but it has nothing to do with multiple or split personalities. Sufferers are unable to properly integrate their thoughts and sensory inputs. Often they hear voices. They may even see visions. They may be catatonic, unable to experience pleasure or achieve goals. They may experience paranoia, convinced that people are scheming against them, or surveilling them. They may suffer from delusions of grandeur.
Schizophrenia is an extremely debilitating mental disorder, often requiring hospitalization for prolonged periods. It imposes a tremendous cost on society and on those who suffer from it. Available treatments seek to control the symptoms of schizophrenia through medications. A fraction of sufferers may recover, but unfortunately most do not.
The disease typically manifests itself in early adulthood, but in less common cases may occur in childhood or even in late adulthood. On average about 1 in 100 people will develop schizophrenia, but this varies sharply by society and region. For example, much higher rates of 7% have been observed in the West of Ireland in the past. Men experience somewhat more severe effects than women, responding less readily to treatment. Some have speculated that it is a relatively modern disease, since few seem to have described anything like it before the year 1750, but this is disputed. Be that as it may, the modern description of schizophrenia dates only from the late 1800s and early 1900s.
Early Investigations: Twins and Schizophrenia
The far-reaching consequences of schizophrenia have resulted in over a century of intensive efforts to find out more about its causes and development. From the beginning many suspected that it has hereditary dimensions. This can be clarified by investigating relatives who share the condition. In 1875, Francis Galton described a case of (probably) identical French twins who exhibited clear signs of shared schizophrenia, though the condition was not yet named that. In 1916, Ernst Rüdin gathered data showing that the relatives of schizophrenics are more likely to have the condition than one would expect by chance. By that time Mendel’s laws of genetics had been rediscovered, and much effort was expended speculating whether the condition might be caused by a single dominant or recessive gene. Interestingly, Rüdin’s statistical collaborators thought it more likely to be polygenic, caused by multiple genes with additive effects, but were short of evidence.
Subsequent investigations accumulated data from twin pairs, contrasting identical with fraternal twins, and other relations. This early data suggested that identical twins were much more likely to match for schizophrenia than fraternal ones, but sample sizes were small and skeptics abounded. Very rare cases of schizophrenia among twins separated at birth and raised entirely apart were also examined, starting with Eliot Slater in the England (Slater had met Rüdin and his collaborators in Germany in the mid-1930s). By the 1960s this evidence became conclusive, bolstered by new data from Scandinavian countries, where twin registers and excellent record keeping made tracking down cases feasible, with additional data from the UK and USA. This led David Rosenthal to conclude in 1970 that it was no longer plausible to deny that schizophrenia has genetic dimensions.
Non-Genetic Explanations?
Resistance to genetic explanations in psychiatry has old roots and the fever dreams of the 1960s diverted attention---explanations blaming either family rearing practices, the economic system as a whole, or more vaguely “society”, dominated. Popular authors like the Scottish “existential” psychiatrist R. D. Laing promoted the idea that schizophrenics are distorted by social pressures, and may actually be showing healthy reactions to that distortion. At least that is how many interpreted his theories, which never attained a state that could be unambiguously understood—this appears to have enhanced rather than retarded his influence. Others blamed “schizogenic” mothers for warping their children through coldness and other pathologies, often without pausing to consider that many of these mothers may have been schizophrenic or otherwise mentally ill themselves, to a greater or less degree.
Today a mountain of evidence has accumulated showing that schizophrenia has a strong genetic component. This data also allows us to estimate its strength with confidence. One way to do this is to compare sufficiently large samples of (genetically) identical twins with fraternal twins, which like ordinary siblings only share half their genes on average. One of the most striking pieces of evidence this yields is the fact that the offspring of discordant identical twins have the same sharply elevated risk for developing schizophrenia. It doesn’t matter whether or not their parent actually developed the condition. The risk is passed on genetically.
Schizophrenia and Heritability
A formal heritability statistic can be calculated from this twin data. For schizophrenia this estimate is on the higher end for all known traits, about as high as for general intelligence, at about 0.7-0.8. This means that differences in genes can account for at least 70% of the differences in diagnosed schizophrenia.
Other methods using different relatives and more complex calculations produce more or less the same result. It is worth pausing over this heritability estimate because it illustrates an important conundrum.
To calculate heritability from twins, one doubles the difference between the correlations. If identical twins correlate at 0.5, and fraternal twins correlate at 0.15, then heritability is 2x(0.5-0.15)=0.7. The difference is doubled because identical twins share twice the genes that, on average, fraternal twins do. This should be intuitively obvious, but it can also be proved easily from the first principles of genetics. However, stop right there to reflect on the fact that heritability can be as high as 0.7, but the correlation (“concordance”) between twins can be as modest as 0.5. Why? The easiest way to understand this is to consider what would happen if we made the diagnosis of schizophrenia more strict, using higher thresholds. This would increase the correlation between twins for schizophrenia, whether fraternal or identical. Similarly, making the diagnosis weaker or more permissive would decrease the correlation. Because heritability is expressed in terms of differences, it already corrects to some extent for arbitrary changes in diagnosis. It is tricky to use heritability with a sure grasp, and an entire book has been written detailing myriad failings, even among experts, in its use (see Sesardić 2005).
What is interesting here is that a concordance rate of 0.5 between identical twins suggests that there is room for other influences on the development of schizophrenia.
Even if your identical twin is schizophrenic, you need not be. We know from other evidence, also gathered from twin and adoption studies, that the other factors are not derived from the shared environment that people are raised in. This shared environment is what people typically mean when they talk about environment at all, and includes all family rearing influences, education and similar factors. The influence of shared environment can be estimated in the same way that heritability can be estimated, and it is negligible.
More or less zero.
That leaves us with some work to do in understanding what it is, exactly, that causes identical twins to be discordant for schizophrenia. The discordance must come from the “non-shared environment” peculiar to each individual. A popular hypothesis, investigated extensively by E. Fuller Torrey and Irving Gottesman for several decades, but already described by Paul Meehle in the early 1960s, is known as the “diathesis-stress” theory. According to this, genes provide a risk for schizophrenia, but stressors are needed for the condition to develop. Discordant twin pairs have been extensively studied to try and identify what those stressors might be, in the hope that progression to the condition may be prevented or at least ameliorated. But to date all these efforts have been fruitless and no hard evidence has been found for any stressors that can be rigorously understood.
This is a familiar pattern in behavior genetics. “Non-shared environment” has been investigated for many traits, and never satisfactorily understood. Part of the problem stems from the use of the word “environment”. This may evoke warm and fuzzy feelings on the part of psychologists and sociologists, but strictly speaking it means nothing more than non-genetic here. It includes all sources of purely random error. It may even stop at purely random error for all we know. Consider again the idea of “stressors”. They may well exist, as discrete events that can be detected and understood. But the evidence from twin studies does not directly support any such idea. All it tells us is that something non-genetic is happening.
Consider the following alternative scenario. We may think of genes as a program attempting to build a complex product. Suppose it operates in a space in which there is random noise present at all times. A program with perfect fidelity is able to overcome this noise, say through redundancy and error correction steps, nearly all the time. Less successful programs may only do so part of the time. Perhaps only 50% of the time. Mutations in DNA may degrade a high fidelity program to one with lower fidelity. A degraded program may be repeated multiple times and produce different results each time. Think of the differences as random developmental errors, in which, try as it may, the program is often not able to get all the “ducks in a row” that it requires. Sometimes it gets more “ducks”, sometimes less. Here the outcomes will be as we observe for conditions like schizophrenia, only partially concordant. But it is not sensible to chase after the background noise as a target for intervention. It is unstructured and cannot be addressed directly.
Given that twin and pedigree studies strongly implicate genetics in schizophrenia, it is natural to suppose that specific genes will be found if we look hard enough. The distribution of the disease strongly suggests that it is polygenic, so multiple genes should be found. This is exactly what has happened in the last decade. Hypothesis-free methods trawl through very large data sets in order to find which differences in specific locations of the genome are associated with schizophrenia. By using an extreme statistical significance threshold, false discoveries are protected against. Replication raises confidence in the reality of the findings.
The Psychiatric Genetics Consortium has conducted a series of these Genome Wide Association Studies (GWAS) with ever increasing sample sizes, now up to of 76,755 positive cases and 243,649 controls (Trubetskoy et al 2022). Enough genomic variants have now been found (287 loci) to explain 7% of the variation in the disease, examining only the most common sites within the genome where variations are found. These are the strongest available results for the specific genetics of any psychiatric condition. We know that in theory, using these common variants alone (obtained from inexpensive DNA chips) at most 24% of the heritability can be explained with a sufficiently large sample size, much larger than obtained so far. Examination of more uncommon sites of variation within the genome, through whole-genome sequencing, using 24,248 cases and 97,322 controls, has already revealed a number of copy-number variants in up to 32 genes with large effects (Singh et al 2022). In the future a combination of these techniques will explain more and more of the heritability established by the study of relatives.
Open Questions and the Need for Further Research
It is curious that despite concerted efforts to separate the diagnosis of the thought disorder schizophrenia from the mooddisorder manic depression (bipolar), genetic evidence tends to equate them at a deeper level. In truth the disorders were never clearly separated because even though an individual might be clearly diagnosable with one rather than the other, in practice a person could progressively drift and be diagnosed with the other condition some years later, perhaps to revert once again. The diseases are not considerate enough to read the textbooks. GWAS studies of schizophrenia and manic-depression now clearly implicate many of the same genes in both disorders, strongly supporting the idea of a continuum between the two conditions. Schizophrenia itself is no longer properly thought of as a binary condition (where either you have it or you don’t) but rather as a spectrum, on which extreme individuals at the tail may end up with a traditional diagnosis of schizophrenia, but milder conditions may exist which do not cross that threshold. In the future, GWAS studies may attempt to address the spectrum itself. For example, there is strong evidence that between 6 and 15% of the otherwise normal population experience auditory, visual, olfactory and other delusions, yet live otherwise normal lives (Linszen et al, 2022). Some of the same genes may be implicated in such cases.
There are many other unresolved questions related to the genetics of schizophrenia. Consider the fact that schizophrenics have much reduced fertility. Over time this should result in the related genes being eliminated from the population, as they are replaced by genes associated with higher fertility. Why then do we still have schizophrenia genes? One possible explanation is that the mutations involved constantly recur with a low probability, replenishing the copies that were selected out. Similar explanations have been given for the persistence of hemophilia, which also leads to lower fertility (bleeding compromises survival) but still persists in populations: the hemophilia mutation must constantly recur. Though schizophrenia is polygenic the same principle applies. However other explanations are also available—the genes may be linked to compensating traits which make up for the loss of fertility—so the question is far from settled.
A Self-Assessment for Schizophrenia Based on DNA
The Traitwell Schizophrenia app (https://traitwell.com/schizo/) is based on the latest 2022 results from the Psychiatric Genomics Consortium. It is free and allows you to upload your DNA to satisfy your curiosity about where you may lie on the distribution of known relatively common genes for the condition. It will be continuously improved as larger studies are completed and more genes are found.
References
2004. Di Lalla, Lisbeth ed. Behavior Genetics Principles. APA.
1960. Fuller, John L. and William R. Thomson. Behavior Genetics. Wiley.
1991. Gottesman, Irving. Schizophrenia Genesis. Freeman.
2022. Linszen et al. “Occurrence and phenomenology of hallucinations in the general population: A large online survey”. Schizophrenia (2022)8:41.
1973. Meehl, Paul E. Psychodiagnosis: Selected Papers. Minnesota.
2018. Mitchell, Kevin J. Innate. Princeton.
2018. Plomin, Robert. Blueprint: how DNA makes us who we are. MIT.
1970. Rosenthal, David. Genetic Theory and Abnormal Behavior. McGraw Hill.
2005. Sesardić, Neven. Making Sense of Heritability. Cambridge.
1971. Shields, James and Irving Gottesman ed. Man, Mind and Heredity: Selected Papers of Eliot Slater on Psychiatry and Genetics. Johns Hopkins.
2022. Singh et al. “Rare coding variants in ten genes confer substantial risk for schizophrenia”. Nature 604, 509–516.
1988. Torrey, E. Fuller. Surviving Schizophrenia. Harper.
1994. Torrey, E. Fuller et al. Schizophrenia and Manic Depressive Disorder. Basic Books.
2022. Trubetskoy et al. “Mapping genomic loci implicates genes and synaptic biology in schizophrenia”. Nature 604, 502–508.
Maybe next post on bipolar disorder?