Defining “risk” in NRXN1
When I first received my son’s diagnosis, I was provided with a copy of the Unique pamphlet on 2p16.3 deletions. While I found it helpful in providing a general overview of characteristics that some children experience, I struggled to wrap my head around what this meant for my specific child. A comprehensive Google search was also scary, especially seeing so many studies linking NRXN1 deletions to risk for schizophrenia and other major mental illnesses that are often not diagnosed until adulthood. I wondered what the actual risk was for my son in developing these conditions as he grew older.
One thing that’s been clear to researchers for a while now is that the first cases of genetic syndromes described in the literature are often on the more severe end of what is typical for a condition. These early case reports are just not very representative of the group of individuals as a whole. This is due to something called “ascertainment bias.” Simply put, children who present to their doctor with significant delays or behavioral difficulties are the ones who end up getting genetic testing. Kids who exhibit mild symptoms often don’t come to the attention of doctors and never receive clinical genetic testing. If they never receive genetic testing, they’re never diagnosed with the genetic condition and so aren’t represented in published research studies.
Take the case of intellectual disability (which used to be called “mental retardation” before it was considered pejorative). In the Simons Searchlight handout, they cite a study that reported intellectual disability in 90% of individuals with NRXN1 disorder. With what we know about ascertainment bias, however, this certainly does NOT mean that 90% of people with NRXN1 deletions have intellectual disability. It just means that out of the 15 children who came to the attention of the geneticist who ordered testing, 14 met diagnostic criteria for intellectual disability. It tells us nothing about individual risk.
To get around the problem of ascertainment bias, researchers use a study design called “case control.” In a case control study, a large group of “normal” individuals gets genetic testing. The number of NRXN1 deletions in the control population is compared to the number of NRXN1 deletions in a patient population of interest (e.g., a group of people diagnosed with autism, schizophrenia, or what have you). The difference between these numbers is called the odds ratio.
Let’s take a made up example to demonstrate this point. There are 2 groups of 100 people each. One group is comprised of healthy volunteers and the other group is comprised of individuals who have been diagnosed with a serious mental illness. Both groups are given genetic testing. In the healthy group, 2 people were found to have a specific genetic mutation, while in the mental illness group, 4 people were found to have the same genetic mutation. In this case, those with mental illness are twice as likely to have the genetic change, resulting in an odds ratio of 2 (an odds ratio of “1” would mean there is no difference between the groups).
Taking the idea of “risk” a step further, let’s continue with our previous example. Let’s say that this serious mental illness occurs in 1/100 (1%) of the general population (this is the prevalence). We could make a rough estimate of relative risk for individuals with the genetic change. With an odds ratio of 2, we might estimate the relative risk for individuals with that particular genetic syndrome to be 1/50 (2%). I’ll be honest that the relationship between odds ratio and relative risk is more complicated than this, but for our purposes I think it provides an OK rough estimate.
So what do actual case control studies say about risk for developmental disorders and psychiatric illness in NRXN1-related disorder? I’ve made a graph to summarize.
You’ll notice that the largest odds ratios occur for autism and for Tourette syndrome (approximately 20 for both). However, autism occurs much more frequently in the general population compared to Tourette’s (2.3% vs 0.6%) so the actual risk of individuals with NRXN1-related disorder is much higher for autism than for Tourette’s (46% vs 12%).
You might also notice that none of the diagnoses occur in more than half of individuals with NRXN1-related disorder. This is in line with population studies, which demonstrate that NRXN1 deletions are not actually very rare in the general population (perhaps 1 in ). Most people who carry the deletion have probably never had clinical genetic testing and likely have no idea they have the deletion. About 60-70% of children who are diagnosed with NRXN1 deletion inherit it from a parent with few, if any symptoms. All of these findings support the idea that risk for serious developmental disorders and neuropsychiatric illness is increased for those with NRXN1 deletions; however, most individuals will not experience any serious symptoms.
