Our Diagnoses Are Categories. Our Genes Are Not.
What a landmark Nature study means for families and clinicians navigating autism and neurodevelopmental conditions
I have spent nearly four decades in paediatric practice, the last several years focused on autism diagnostic services. In that time I have sat across the desk from hundreds of families in which neurodevelopmental and psychiatric conditions cluster across siblings, across generations, and within the same child. A paper now published in Nature provides the most compelling genomic evidence to date for what those families, and many of their clinicians, have long observed. What follows is my attempt to explain what it means in practice.
A paper published in Nature in December 2025, the largest cross-disorder psychiatric genetics study ever conducted, has confirmed something many clinicians have long suspected: the diagnostic categories we use in clinical practice bear little resemblance to the way genetic risk is distributed across the human genome.
The study, led by the Psychiatric Genomics Consortium Cross-Disorder working group, analysed data from over one million people across 14 psychiatric and neurodevelopmental conditions, including autism, ADHD, OCD, schizophrenia, bipolar disorder, major depression, PTSD, anorexia nervosa, and substance use disorders. Using sophisticated genomic statistical methods, the researchers identified five underlying genomic factors that together account for approximately two thirds of the genetic variance across all 14 conditions, and 238 specific genetic locations where individual variants influence multiple conditions simultaneously.
What is a pleiotropic locus, and why does it matter?
Pleiotropy refers to the phenomenon whereby a single genetic variant influences more than one trait or condition at the same time. Rather than one gene for autism, another for ADHD, and another for anxiety, many variants influence risk across several conditions at once. The 238 pleiotropic loci identified in this study are the genomic locations where that cross-condition influence is particularly strong and statistically robust. The clinical implications are considerable: they bear directly on why conditions co-occur within individuals and families, and how we should approach both diagnosis and support.
The five genomic factors — and where autism sits
The five-factor model maps the landscape of shared genetic risk more accurately than any diagnostic manual. The factors identified were: a Compulsive factor, encompassing OCD, anorexia nervosa, and partly Tourette’s syndrome; a Schizophrenia-Bipolar factor; a Neurodevelopmental factor defined primarily by autism and ADHD; an Internalising factor covering depression, anxiety, and PTSD; and a Substance Use factor. A higher-order general factor, the so-called p-factor, captured genetic liability cutting across all five.
For those of us working with autistic children and their families, the Neurodevelopmental factor is particularly significant. Autism and ADHD load on the same genomic factor. Tourette’s syndrome also has a partial loading here. This is not a statistical curiosity — it reflects a shared biological architecture that explains why ADHD co-occurs in autism at rates that no model of independent conditions could account for, and why tic disorders are substantially over-represented in neurodevelopmental clinics.
What this means in practice for families
Families frequently ask why multiple neurodevelopmental or psychiatric conditions seem to cluster in the same person or run through multiple branches of their family tree. The answer, now supported by the most comprehensive genetic evidence ever assembled, is that this clustering is not coincidence and not bad luck. It is shared biology.
The genetic variants that increase risk for autism overlap substantially with those that increase risk for ADHD, and to a lesser but meaningful degree with variants relevant to OCD and anxiety. When a parent tells me they have ADHD and anxiety, that a sibling has OCD, and that their autistic child also has features of all three, the current data support exactly what that family is observing: these conditions share underlying genomic foundations.
There is reassurance in this. The clustering is not caused by parenting, by diet, by vaccines, or by any other environmental factor families are sometimes made to feel responsible for. It reflects inherited biology that predisposes the nervous system to develop in certain ways.
What this means for clinical assessment
When evaluating an autistic child, we should be actively looking for co-occurring neurodevelopmental and psychiatric conditions rather than treating them as unexpected extras. The genetic architecture makes comorbidity the rule rather than the exception. ADHD is the most obvious example, but anxiety disorders, OCD, and Tourette’s syndrome all warrant systematic attention.
When a sibling or parent presents with ADHD, anxiety, or depression, that family history should be taken seriously as contextual information. A first-degree relative with any condition on the Neurodevelopmental or Internalising genomic factors tells us something meaningful about the genetic landscape of that family.
The study also has implications for genetic counselling. As whole genome sequencing becomes embedded in NHS practice through the Newborn Genomes Programme and expanding paediatric pathways, families will increasingly receive genetic results framed around a single diagnosis, usually autism. Clinicians will need to explain that variants identified may have implications not only for autism specifically, but for a broader range of neurodevelopmental and psychiatric traits across the family.
The question of treatment
One of the more optimistic implications of this research concerns treatment development. The authors note that several pharmacological interventions already show efficacy across multiple psychiatric conditions — selective serotonin reuptake inhibitors being the obvious example — and suggest that the shared genomic signal captured by the five factors could point towards new transdiagnostic treatment targets. Rather than developing separate medications for each diagnostic category, future therapeutics might address the shared biological pathways underlying multiple conditions simultaneously. This is a longer-term prospect, but one grounded in coherent logic: if the genetic architecture is shared, the molecular biology is likely, at least in part, to be shared too.
A note on diagnostic categories
None of this renders diagnostic categories useless. They remain practically important for accessing services, communicating with schools, directing reasonable adjustments, and guiding families towards appropriate support. What the genetic data do make clear is that the boundaries between diagnostic categories are considerably more porous than our classification systems imply. Autism is not a wholly separate biological entity from ADHD. OCD and autism share more genetic architecture than either shares with schizophrenia.
The authors suggest this evidence should inform ongoing debates about nosology, the science of classification, and that a neurobiologically valid psychiatric classification system might look quite different from DSM-5 or ICD-11. That is probably true, though the transition from genomic research findings to clinical reclassification is measured in decades rather than years.
For families, the message is more immediate: the co-occurrence of neurodevelopmental and psychiatric conditions in the same person or the same family has a genetic explanation, and that explanation is increasingly well understood. It is shared biology, not misfortune, and it points towards more comprehensive assessment, more honest counselling, and, in time, more effective treatment.
Reference: Grotzinger AD et al. Mapping the genetic landscape across 14 psychiatric disorders. Nature 649, 406–415 (2026). https://www.nature.com/articles/s41586-025-09820-3