Technology VarSome.com Research Brief

What the Genome Has to Say About Stuttering

Jason Armstrong | August 11, 2025

A large-scale genome-wide association study (GWAS) published in Nature Genetics¹ has identified 57 distinct loci associated with developmental stuttering. The study analysed genetic data from over 1.1 million individuals, making it the most extensive investigation of its kind to date.

Background

Affecting more than 400 million people, developmental stuttering is the most common fluency disorder worldwide. Onset usually occurs between the ages of 2 and 5, with up to 80% of affected children spontaneously recovering without the aid of speech therapy. There are stark differences in stuttering between sexes. At onset, developmental stuttering seems to affect boys and girls at a similar rate. However, in adolescent and adult populations, the male-to-female ratio is around 4:1.

Young people who experience stuttering report low self-esteem, increased bullying, decreased classroom participation, depression, and suicidal thoughts. Stuttering in adults can affect employability, socio-economic status, and mental and social wellbeing. Various therapies exist to help manage stuttering, including behaviour modifications, technology-based feedback, and speech and cognitive interventions. However, there is no known cure, meaning many of those who retain a stutter into adolescence and beyond may never fully recover.

Study Design & Scope

Polikowsky et al. stratified participants by both genetic ancestry (European, African, East Asian, Latin/Admixed American) and sex, running eight primary GWAS alongside several secondary meta-analyses. Stuttering status was determined using self-reported survey responses from 23andMe participants who had consented to participate in genetic research. In total, the primary analysis included 99,776 cases and 1,023,243 controls. The findings were validated in two independent cohorts: the International Stuttering Project (ISP) and Add Health.

Key Findings

This study provides strong evidence that stuttering is a common complex trait with a substantial genetic component. Across the eight GWAS and multiple secondary meta-analyses, the authors identified 63 genome-wide significant associations, mapping to 57 unique loci, none of which have previously been linked to stuttering. The genetic architecture appears highly polygenic, similar to traits like insomnia or type 2 diabetes. Estimated SNP-based heritability was around 9% in European males and females. However, despite high genetic correlation between sexes, the male and female analyses revealed non-overlapping loci, suggesting partially distinct biological mechanisms.

Polikowsky et al. also uncovered functional enrichment in enhancer regions, active chromatin marks (e.g., H3K4me1, H3K27ac, H3K27ac), and genes expressed in neurons, particularly in the brain. Polygenic risk scores (PRS) trained on the European male cohort showed the strongest predictive power in both external validation datasets, especially for male participants.

Genetic correlation and Mendelian randomization analyses revealed overlap and potential causal relationships with several traits:

Shared architecture with:
- Autism
- Depression
- ADHD
- Impaired rhythm
- Asthma, BMI, and hearing loss (female-specific)
- Slower walking pace and lower alcohol consumption frequency (female-specific)
Bi-directional causal links with:
- Depression
- BMI
- Impaired rhythm
- Autism (sex-combined)
- ADHD (female-specific)

One of the study’s most striking findings relates to rhythm processing. It provides the first genetic support for the Atypical Rhythm Risk Hypothesis, which proposes a link between impaired rhythm processing and developmental speech disorders. Mendelian randomization in this study suggests a bidirectional causal link between stuttering and impaired rhythm. Notably, one of the strongest loci identified in European males, VRK2, has previously been linked to musical beat synchronisation in other research².

Implications for the Field

This study moves beyond the traditional view of stuttering as an exclusively behavioral or psychological issue, providing evidence for a genetic contribution that is both common and complex. It highlights the need to consider sex- and ancestry-specific genetic effects when interpreting stutter risk. The authors also caution that differences in recovery may affect how reliably stuttering is captured in self-reported data. Because females are more likely to recover from stuttering during childhood, they may be less likely to recall or report it accurately in adulthood. This could lead to a less specific definition of “cases” in females, potentially weakening the genetic signal and contributing to the lower predictive performance of female-derived PRSs.

From a broader perspective, this work raises an important point: genetic testing is often associated with rare or life-threatening conditions. But studies like this hint at how genomics could one day support conditions that may not be considered clinically urgent but still have a significant impact on many individuals. Understanding that a speech disorder has a biological basis, particularly during childhood, might help reduce stigma, improve self-esteem, and support targeted interventions.

The lack of curative treatment has shaped how stuttering is perceived; often as something to be endured or managed, rather than understood. For those who continue to stutter, the effects can be long-lasting and deeply personal. In this context, identifying biological contributions helps change the narrative, providing evidence that stuttering is rooted in neurodevelopment, rather than personal failure, and supporting more informed and empathetic approaches to care. This shift could empower families and clinicians to recognize stuttering earlier, pursue more tailored interventions, and reduce stigma, even when fluency does not fully return.

Final Thoughts

Polokowsky et al. reinforce the value of sex- and ancestry-stratified analysis, an approach increasingly relevant to large-scale variant interpretation. Although not yet clinically actionable, the PRS model offers a glimpse into how polygenic insights could support diagnosis, especially in developmental neurocognitive traits. However, the lack of genome-wide significant findings in some ancestry groups (particularly East Asian and African females) highlights ongoing limitations in statistical power due to underrepresentation.

As with many complex traits, the findings underscore the need for greater diversity in genomic datasets. The strong overlap with neurodevelopmental and psychiatric traits could inform multi-trait risk assessments or phenotype clustering models.