The present exploratory pilot study investigated genetic variants potentially associated with EA among elite Turkish athletes using a GWAS approach. To the best of the authors’ knowledge, the present study is among the first attempts to explore the potential genetic basis of EA among this population. Although no SNP reached genome-wide significance, 11 SNPs exceeded the suggestive significance threshold and were identified as candidates for further investigation. Among these, two SNPs (rs117523538 and rs7312447) were negatively associated with higher EAI scores, and nine (rs79233502, rs13318101, rs7699799, rs4690145, rs3799055, rs7789550, rs73682313, rs6577987, rs113672848) were positively associated. Moreover, several variants were observed at relatively low frequencies in the reference population, highlighting their potential relevance for future research. Additionally, four SNPs (rs4690145, rs7789550, rs117523538, and rs113672848) were associated with altered gene expression in brain and nerve tissues, suggesting potential biological pathways warranting further investigation. However, these findings should be interpreted with caution, given the study’s exploratory nature, limited sample size, and the absence of an independent replication cohort.
EA is considered a complex phenotype, influenced by the interaction of multiple genetic and environmental factors. In this context, traditional candidate gene association studies may remain limited in reproducibility and in their ability to reliably identify genetic susceptibility28. Therefore, GWAS approaches have become increasingly important for investigating the genetic basis of complex phenotypes due to their broader and more reproducible analytical framework37. However, despite their advantages, GWAS requires large sample sizes to achieve sufficient statistical power, which is challenging to achieve among elite athlete cohorts. To minimize the impact of this limitation, individuals from four different sport disciplines were combined, resulting in a total of 168 participants. Nevertheless, the demographic characteristics of these groups (number of participants, sex, age, and sport experience) differed considerably (Table 1), which may have influenced the results of the association. This heterogeneity should therefore be considered when interpreting the findings.
Using EAI scores, levels of exercise addiction symptoms of the participants were evaluated within each sport discipline (Table 2). The scores differed across groups, with rowers exhibiting the highest mean EAI scores and cyclists the lowest. Previous studies have suggested that the demanding environmental and training conditions experienced by rowers may influence athletes’ mood and psychological well-being38. In a study with a large cohort, the ratio of EA among cyclists was reported to be significantly lower independent of age, sex, and training39, which is consistent with the descriptive findings of the present study. However, given the heterogeneity of the participant groups in sample size, sex, age, and sports experience, these observations should be interpreted with caution. Further studies involving larger and more homogeneous cohorts are needed to confirm these findings.
In the present exploratory study, 11 SNPs exceeded the suggestive significance threshold and were identified as potential candidates for further investigation. Among these, rs7789550 has previously been reported in relation to alcohol dependence, suggesting that it may warrant further examination in the context of compulsive behaviors more generally40. Similarly, rs3799055, located within the ADGRB3 gene, was significantly associated with EA. ADGRB3 belongs to the adhesion G protein-coupled receptor (GPCR) family and is involved in several central nervous system processes, including axon guidance, myelination, and synaptic organization. Previous studies have linked SNPs and copy number variations in ADGRB3 to various psychiatric disorders41, suggesting that rs3799055 may contribute to neurobiological mechanisms underlying addictive behaviors. In addition, a GWAS approach has identified a genome-wide significant variant in the ADGRB3 gene associated with anxious temperament42. Although these observations suggest the potential relevance of this gene to behavioral traits, the findings of the present study should be interpreted with caution, given the exploratory nature of the analysis, the absence of genome-wide significant associations, and the lack of an independent replication cohort.
Another notable finding involved the rs7312447 polymorphism located within the LINC01619 gene, a long non-coding RNA (lncRNA) implicated in neurological processes, cellular stress responses, and vascular injury-related mechanisms. Previous evidence has suggested that LINC01619 may serve as a biomarker in neurological conditions such as cerebral vasospasm following subarachnoid hemorrhage43. Given the emerging evidence linking neurobiological processes to behavioral addictions, rs7312447 may represent a potential candidate for future research investigating the biological mechanisms underlying EA. However, the relevance of this variant to EA remains unclear and requires further investigation among independent cohorts.
rs113672848, located within the ZNF480 gene, was also significantly associated with EA. Although variants within ZNF480 have rarely been reported among people with schizophrenia, these findings support the notion that this gene may be involved in neuropsychiatric mechanisms characterized by substantial genetic heterogeneity44. Moreover, prior studies have suggested that genes in the zinc finger protein family, including ZNF408, may be promising candidates for neurodevelopmental disorders such as intellectual disability45. Collectively, these findings suggest that rs113672848 may be a candidate for future research investigating biological mechanisms underlying neuropsychiatric and behavioral phenotypes. However, its potential relevance to exercise addiction remains unclear and requires further investigation among independent cohorts.
rs117523538 was also associated with downregulation of TSPAN32 expression in the hypothalamus according to GTEx data. The hypothalamus plays a central role in neuroendocrine regulation, motivation, stress response, and behavioral control, all of which are closely linked to addiction-related mechanisms. Although a direct relationship between the TSPAN32 gene and addictive behaviors has not yet been clearly established, the gene is known to be involved in immune regulation, T-cell activation, and neuroinflammatory processes. Given the growing recognition of neuroinflammation in the neurobiology of addiction, alterations in hypothalamic TSPAN32 expression may potentially contribute to neurobiological mechanisms underlying addictive behaviors41,46,47,48,49. To date, no direct evidence has linked the remaining identified variants to EA or athlete-related phenotypes. Therefore, the biological relevance of these variants remains unclear, and further research with larger independent cohorts is required to determine whether these findings can be replicated and to explore their potential relevance to EA.
The present study has several notable strengths, including the use of a relatively large cohort of elite athletes for an exploratory genomic investigation, and the use of a microarray-based, genome-wide approach rather than a candidate single-gene strategy. These methodological features broadened the exploratory scope of the analysis. However, several limitations should be considered when interpreting the findings. First, the heterogeneity of participants both within and across sport disciplines may increase the risk of model overfitting, potentially reducing the statistical power and limiting the reliability and reproducibility of the results. Second, the inclusion of athletes from only four sport disciplines restricts the generalizability of the findings to other athletic populations. Third, although stringent significance thresholds were applied (p < 1.00 × 10⁻⁵ and p < 1.00 × 10⁻⁷), the possibility of false-positive findings cannot entirely be excluded, particularly given the modest sample size and the absence of an independent replication cohort.
In addition, the cross-sectional design of the study precludes any causal inference regarding the relationship between genetic variation and EA-related symptoms. Because EAI scores were assessed at a single time point, it remains unclear how these relationships may evolve over time. Moreover, the use of self-report measures may introduce response bias, social desirability bias, and potential over- or under-reporting of symptoms. This limitation could particularly be relevant among elite athletes, for whom intensive training behaviors are often considered normative and essential for performance.
The lack of systematic control for behavioral and environmental factors, including training practices, coaching approaches, nutritional habits, and psychosocial stressors, could also influence the observed associations. Moreover, psychological characteristics and psychiatric conditions, such as anxiety, obsessive-compulsive tendencies, eating disorders, and attention-deficit/hyperactivity disorder (ADHD), may overlap with EA-related symptoms. Consequently, some of the identified genetic associations may reflect broader psychiatric vulnerability rather than EA specifically.
Finally, exercise addiction-related symptoms were assessed using the EAI, a screening instrument rather than a diagnostic tool. Therefore, the findings should be interpreted as reflecting variation in EA symptoms or risk rather than clinically confirmed EA. Furthermore, because exercise addiction currently lacks universally accepted clinical diagnostic criteria, the present findings should be interpreted as reflecting variation in self-reported exercise addiction symptoms rather than a clinically established phenotype. Future studies incorporating clinical interviews and longitudinal assessments may help establish more robust phenotypic definitions for genetic association research. Taken together, these limitations indicate that the present findings should be regarded as preliminary, exploratory, and hypothesis-generating rather than confirmatory evidence of genetic susceptibility to EA.