Partitioning the Heritability of Tourette Syndrome and Obsessive Compulsive Disorder Reveals Differences in Genetic Architecture

Twin studies have made a confident claim about Tourette syndrome: it is roughly seventy-seven percent heritable. Genes do most of the work. However, decade after decade of genome-wide association studies — large scans that seek individual risk variants — kept returning almost nothing. There were too few significant hits to explain that family signal. The gap between what twins tell us and what genome-wide association studies deliver is known as the missing heritability problem, and for Tourette syndrome and obsessive-compulsive disorder, it was glaring. Davis and colleagues set out to find where that heritability actually went. What they discovered reshaped how we think about two of psychiatry's most persistently misunderstood disorders. The missing heritability problem has a specific shape. Family and twin studies estimate what's called broad-sense heritability — the proportion of variation in a trait that is explained by genetic differences between people. The narrower version counts only additive effects from common variants, the kind genome-wide association studies are designed to detect. For obsessive-compulsive disorder, published estimates from family studies range from about twenty-seven to sixty-five percent, depending on age of onset. For Tourette syndrome, these estimates cluster around sixty percent. Meanwhile, the handful of genome-wide significant hits produced by those studies explain only a small portion of that signal. The question posed by Davis and colleagues was direct: what if we stop looking for individual significant loci one at a time and, instead, ask what all common variants together can explain? The tool they used is called GCTA — Genome-wide Complex Trait Analysis — which implements a method known as restricted maximum likelihood, or REML. Rather than hunting for top hits, GCTA builds a genetic relationship matrix across all individuals using genome-wide single nucleotide polymorphism data, then estimates how much variance in disease liability those variants collectively account for. It has already been applied to height, schizophrenia, autism, and type two diabetes. Davis and colleagues applied it to Tourette syndrome and obsessive-compulsive disorder simultaneously, aiming to estimate total single nucleotide polymorphism heritability and carve that heritability into biologically meaningful pieces. The dataset required substantial assembly. Cases and controls were drawn from multi-site international cohorts: six hundred seventeen Tourette syndrome cases paired with four thousand one hundred sixteen controls, and one thousand sixty-one obsessive-compulsive disorder cases paired with four thousand two hundred thirty-six controls. All subjects were restricted to genetically defined European ancestry, established through principal components clustering. Quality control was stringent: single nucleotide polymorphisms with differential missingness, deviation from Hardy-Weinberg equilibrium, or platform effects were removed, and individuals with relatedness above a pi-hat threshold of zero point zero five were excluded. Control-control dummy analyses and ten phenotype permutations confirmed no systematic inflation. The final analyses used three hundred ninety-three thousand three hundred eighty-seven genotyped single nucleotide polymorphisms for Tourette syndrome and three hundred seventy-three thousand eight hundred forty-six for obsessive-compulsive disorder, with imputed datasets extending to more than seven point six million single nucleotide polymorphisms. The top twenty principal components were included as covariates throughout. The headline numbers are striking. GCTA yielded a heritability point estimate of zero point fifty-eight for Tourette syndrome and zero point thirty-seven for obsessive-compulsive disorder, both on the liability scale, with p-values below one in a trillion and one in ten million, respectively. These estimates are close to — and statistically consistent with — the heritability values from twin and family studies. That alignment is the core finding. It means very little heritability is truly missing from genome-wide association data for either disorder. The signal was never absent; it was spread across hundreds or thousands of common variants, each too small to cross a genome-wide significance threshold on its own, but collectively accounting for most of the familial risk. The permutation tests confirmed that the estimates were not artifacts of population structure or platform differences. Now comes the part where Tourette syndrome and obsessive-compulsive disorder start to diverge. Davis and colleagues ran three partitioning analyses — by chromosome, by allele frequency, and by functional annotation — and the picture that emerged was of two disorders that share a genetic neighborhood but live in different houses. Start with chromosome length. For Tourette syndrome, heritability correlates significantly with chromosome length — the longer the chromosome, the more heritability it contributes, with a correlation of zero point forty-six and a p-value of zero point zero three. That's the hallmark of a broadly polygenic architecture, with risk scattered more or less uniformly across the genome. For obsessive-compulsive disorder, the correlation is weaker and doesn't reach significance at zero point thirty-five. Chromosome fifteen is a dramatic outlier for obsessive-compulsive disorder — contributing far more heritability than its length would predict — and when chromosome fifteen is removed, the length correlation for obsessive-compulsive disorder becomes significant. Several specific chromosomes also stand out for Tourette syndrome: chromosomes two, five, eleven, twelve, sixteen, and twenty all contribute more heritability than expected. The allele frequency results are where the two disorders look most different. Using directly genotyped single nucleotide polymorphisms, variants with a minor allele frequency below five percent — the rarer end of the common variant spectrum — accounted for twenty-one percent of Tourette syndrome heritability, with a heritability estimate of zero point thirteen and a standard error of zero point zero four. For obsessive-compulsive disorder, that same bin explained essentially zero. Imputed data sharpened the contrast: those low-frequency imputed variants captured about thirty percent of Tourette syndrome heritability and still zero percent of obsessive-compulsive disorder heritability. The authors tested whether population stratification could be driving the Tourette syndrome rare-variant signal by comparing summed per-chromosome estimates to the single-bin estimate and found no significant difference, ruling out substructure as an explanation. The result stands: Tourette syndrome has a meaningful rare-variant contribution; obsessive-compulsive disorder does not. The third partition used expression quantitative trait loci, or eQTLs — genetic variants known to influence gene expression levels in a specific tissue. Davis and colleagues annotated eQTLs from parietal cortex, cerebellum, and muscle, then used those annotations as partitions in joint REML models. Parietal cortex eQTLs accounted for twenty-eight percent of Tourette syndrome heritability and twenty-nine percent of obsessive-compulsive disorder heritability. Cerebellar eQTLs explained thirty-five percent of obsessive-compulsive disorder heritability but only nineteen percent of Tourette syndrome heritability. When parietal and cerebellar eQTLs were combined into a single brain partition, brain eQTLs accounted for thirty-three percent of Tourette syndrome heritability — approaching but not reaching significance at a p-value of zero point zero six — and fifty-nine percent of obsessive-compulsive disorder heritability, which was significant at a p-value of zero point zero zero nine. Genic variants separately accounted for fifty-three percent of Tourette syndrome heritability and forty percent of obsessive-compulsive disorder heritability, both significant. Muscle eQTLs showed no enrichment for either disorder, which makes the brain-specific signal more convincing. With the architecture established for each disorder separately, Davis and colleagues then asked how genetically related the two disorders are to each other. The genetic correlation between Tourette syndrome and obsessive-compulsive disorder came out at zero point forty-one, with a standard error of zero point fifteen and a p-value of zero point zero zero two. A genetic correlation measures the degree to which the same variants, weighted by their effects, drive both traits simultaneously. A value of zero point forty-one means moderate overlap — these disorders share genetic signals, but they are not the same condition genetically. In a sensitivity analysis that removed cases with documented comorbidity of both disorders, the genetic correlation rose to zero point fifty, though with considerably more uncertainty given the reduced sample size. That correlation provides a genomic account for something clinicians have long observed: Tourette syndrome and obsessive-compulsive disorder co-occur far more often than chance would predict. People with Tourette syndrome frequently develop obsessive-compulsive disorder symptoms, and the two conditions are often treated as related. A genetic correlation of zero point forty-one puts a number on that relationship. However, the architectural differences — especially the rare-variant contribution to Tourette syndrome and the concentration of obsessive-compulsive disorder heritability in common variants — mean the two disorders require different genetic strategies going forward. The practical implications follow directly from the architecture. For obsessive-compulsive disorder, heritability resides almost entirely in common variants, and brain expression quantitative trait loci from the parietal cortex and cerebellum concentrate a substantial fraction of that signal. That profile is suited to large-scale genome-wide association studies — larger samples should uncover the individual common variants that we can now confirm are present. For Tourette syndrome, about twenty-one to thirty percent of heritability exists in lower-frequency variants that standard genome-wide association studies are not designed to capture efficiently. That advocates for whole-genome or whole-exome sequencing studies. The enrichment of heritability in brain expression quantitative trait loci for both disorders points toward the parietal cortex and cerebellum as biological territories worth investigating mechanistically. The larger point is this: the missing heritability for Tourette syndrome and obsessive-compulsive disorder was never truly missing. It was distributed across many common variants too small to find individually, and in Tourette syndrome's case, into lower-frequency variants that standard analyses had been skimming past. Davis and colleagues didn't just quantify heritability. They provided researchers with a map. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.