Disorders such as autism, schizophrenia and epilepsy each affect about 1% of the population and are therefore defined as “common disorders”. But are they really? I mean, they are clearly really that common, but are they really “disorders”? Are they natural categories that reflect some shared underlying etiology or are they simply groupings based on sets of shared symptoms? Genetics is providing an answer to that question and demonstrating that so-called “common disorders” are really collections of rare disorders with similar symptoms. This represents a complete paradigm shift in psychiatry, the full ramifications of which have yet to be appreciated.
We have known for decades of examples of rare genetic syndromes that can include symptoms of autism spectrum disorder (such as Fragile X syndrome or Rett syndrome) or of schizophrenia (such as velo-cardio facial syndrome, now called 22q11 deletion syndrome), while epilepsy is a known symptom of many genomic disorders. But such examples were typically thought of as exceptional and distinct from the much larger group of idiopathic cases of ASD, SZ or epilepsy. (Idiopathic simply means of currently unknown cause). Such conditions were often dismissed as not “real autism” or “real schizophrenia”, despite the fact that clinicians could not make any such assessment based on symptoms alone.
Instead, it was widely held to be a proven fact that the genetics of ASD and SZ generally followed a very different mode – rather than being caused by single mutations, as with the syndromes mentioned above, the idea was that the idiopathic cases were caused by combinations of tens or hundreds (or even thousands) of minor genetic differences, each with only a tiny effect on its own, but collectively sufficient to result in disease if enough of them were inherited.
Modern genomic technologies are revealing that this supposed dichotomy between rare and common disorders is artificial – merely a reflection of our current state of knowledge (or, more correctly, our current state of ignorance). Over the past five years, researchers have discovered many more rare genetic conditions that manifest with psychiatric symptoms, and which collectively can account for an ever-growing percentage of patients presenting with ASD or SZ. These include deletions or duplications of whole chunks of chromosomes, often affecting many genes, as well as mutations that affect only one gene.
[The DNA sequence of each gene codes for production of a specific protein. Genes are strung along chromosomes, like the information encoding successive songs on a cassette tape. (I may be showing my age with this reference!) Localised damage to the tape at one specific point can affect just one song, but cutting out a whole section could remove or disrupt multiple songs at the same time. Similarly, changing one letter of the DNA sequence can alter the code for a single protein, while deleting a whole section of chromosome can remove multiple genes and thereby affect production of multiple proteins at once].
Some regions of the genome are particularly prone to errors in DNA replication that result in deletions or duplications. While still rare, these recur at a high enough frequency that many cases with effectively the same genetic lesion can be identified. This has enabled researchers to recognise and characterise a growing number of genomic disorders that carry a high risk of psychiatric or neurological symptoms. In addition to previously known conditions such as 22q11 deletion syndrome, Williams, Angelman and Prader-Willi syndromes, new conditions have been defined involving deletions or duplication at 1q21.1, 3q29, 7q36.2, 15q11.2, 16p11.2, 22q13 and many others, with more being recognised all the time.
All of these mutations have variable effects, sometimes presenting as ASD, sometimes as SZ or epilepsy – often, but not always, with developmental delay or intellectual disability. Because their clinical manifestations are so variable, there was no way to detect or recognise these patients prior to genetic screening (except for conditions with other characteristic symptoms, such as distinct facial morphology). But once a genetic diagnosis can be made, it becomes possible to group patients with the same mutation together and determine whether there are any patterns to their symptoms, their course of illness, how they respond to medications, and other clinical parameters. This is useful information for clinicians and also for patients and their families – indeed, international support groups have been formed for many of these rare genomic conditions.
New conditions caused by mutations in specific genes are also being defined. Rett syndrome is a classic example – a form of autism and intellectual disability in girls that is caused by mutations in a gene called MeCP2. New genomic sequencing technologies are now revealing many more such conditions, although the pace of discovery here has been slower, for two reasons.
First, if you sequence the entire genome of any individual you will find many serious mutations – severely affecting production or function of a couple hundred proteins (out of ~20,000 in total). Recognising which one of those is causing disease in a particular patient is impossible, unless you have some prior information. That information can come from seeing the same gene mutated in multiple patients with a particular condition. That brings up the second problem – the number of genes in which mutations can cause ASD or SZ or epilepsy is very large, probably on the order of a thousand. So the likelihood that any two patients will have a mutation in the same gene is very low. This means we will need to sequence very large samples of patients to start to see the signal of meaningful repeat hits amongst the background noise of repeats that arise by chance, simply because we all carry many mutations.
Those efforts are underway and are beginning to pay off, with new conditions being defined at an ever-increasing rate. One recent example involves mutations in the gene CHD8. Mutations that disrupt this gene have been observed in multiple patients with diagnoses of developmental delay or ASD (15 independent mutations in 3,730 cases), but never in a sample of 8,792 clinically unaffected controls. You can see how rare these mutations are – accounting for only 4 of every 1000 cases – but the fact that you don’t see such mutations in controls provides strong evidence that they are in fact the cause of disease in those patients. (See here for a much more nuanced discussion of causality in genetic disorders – the phenotypic effects of any single mutation will always be modified, sometimes strongly, by additional genetic variants in the background).
By finding multiple patients with mutations in the same gene, clinicians were able to define a new syndrome that was previously unrecognisable. In this case, patients with CHD8 mutations display macrocephaly (increased head size), distinct faces and gastrointestinal problems (the CHD8 protein has independent functions in both the brain and the nervous system innervating the gut). The genetic information is thus directly and immediately relevant to the clinical management and treatment of these cases.
Now, one might say that such mutations are so rare that they don’t really tell us anything about the generality of conditions like ASD or SZ. But the point is, there is no reason to think such a thing exists. As more and more mutations causing high risk of psychiatric conditions are discovered, the percentage of cases remaining idiopathic decreases. Those diagnostic categories are not founded on knowledge but on ignorance of underlying cause, by definition.
Known, high-risk mutations can now be identified in >10% of cases of SZ, 25-30% of cases of ASD, and over 60% of cases of severe intellectual disability. Those numbers represent a vast increase from even a few years ago and are sure to increase rapidly in the very near future. Even if the genetic effects in many cases are more complicated (involving more than one mutation at a time, with contributions from common variants), the major message remains the same: these conditions are incredibly genetically heterogeneous. It is probably far more appropriate to think of “autistic symptoms” or “schizophrenic symptoms” as a common consequence of many distinct genetic conditions, than to think of “autism” or “schizophrenia” as monolithic disorders.
That has hugely important implications not just for clinical practice but also for research. If you take a hundred patients with ASD, you might have 70-80 distinct genetic causes. That’s something to consider in the context of, say, neuroimaging studies that look for commonalities across groups of ASD or SZ patients. Any time I see a study reporting some difference in brain structure “in autism” or “in schizophrenia”, I replace that phrase with “in intellectual disability” and see if it still makes any sense. (It doesn’t, give the well-accepted heterogeneity of ID). Of course, there may be some commonalities in the final outcome in these patients, given they end up with similar symptoms, but research purporting to look at causes should bear the genetic heterogeneity in mind.
Genetics is increasingly providing the means to distinguish the underlying causes in different patients and hopefully develop a far more personalised approach to care. Fortunately, new technologies of genome editing are making it much easier to recapitulate disease-causing mutations in animals so that pathogenic mechanisms can be elucidated. Just in the past couple weeks, very exciting results have been published that help localise the primary effects of particular mutations (in the genes SYNGAP1 and NLGN3) to specific cell types in specific regions of the developing brain in mouse models.
The recognition that these common diagnostic categories are really collections of very rare conditions will necessitate a shift in approaches aimed at developing new treatments. The economics of drug development for rare conditions are obviously very different from the search for the new blockbuster. The next big challenge is to elucidate the biological mechanisms leading to disease across many different mutations to determine if there are any shared pathways or common pathophysiological endpoints that might be targeted in large groups of patients or if individualised treatments can be (or need to be) developed for very small and specific sets of patients, as is happening in other areas of medicine.