Cluster analysis in gambling disorder based on sociodemographic, neuropsychological, and neuroendocrine features regulating energy homeostasis

Background: The heterogeneity of gambling disorder (GD) has led to the identification of different subtypes, mostly including phenotypic features, with distinctive implications on the GD severity and treatment outcome. However, clustering analyses based on potential endophenotypic features, such as neuropsychological and neuroendocrine factors, are scarce so far. Aims: This study firstly aimed to identify empirical clusters in individuals with GD based on sociodemographic (i.e., age and sex), neuropsychological (i.e., cognitive flexibility, inhibitory control, decision making, working memory, attention, and set-shifting), and neuroendocrine factors regulating energy homeostasis (i.e., leptin, ghrelin, adiponectin, and liver-expressed antimicrobial peptide 2, LEAP-2). The second objective was to compare the profiles between clusters, considering the variables used for the clustering procedure and other different sociodemographic, clinical, and psychological features. Methods: 297 seeking-treatment adult outpatients with GD (93.6% males, mean age of 39.58 years old) were evaluated through a semi-structured clinical interview, self-reported psychometric assessments, and a protocolized neuropsychological battery. Plasma concentrations of neuroendocrine factors were assessed in peripheral blood after an overnight fast. Agglomerative hierarchical clustering was applied using sociodemographic, neuropsychological, and neuroendocrine variables as indicators for the grouping procedure. Comparisons between the empirical groups were performed using Chi-square tests (χ2) for categorical variables, and analysis of variance (ANOVA) for quantitative measures. Results: Three-mutually-exclusive groups were obtained, being neuropsychological features those with the greatest weight in differentiating groups. The largest cluster (Cluster 1, 65.3%) was composed by younger males with strategic and online gambling preferences, scoring higher on self-reported impulsivity traits, but with a lower cognitive impairment. Cluster 2 (18.2%) and 3 (16.5%) were characterized by a significantly higher proportion of females and older patients with non-strategic gambling preferences and a worse neuropsychological performance. Particularly, Cluster 3 had the poorest neuropsychological performance, especially in cognitive flexibility, while Cluster 2 reported the poorest inhibitory control. This latter cluster was also distinguished by a poorer self-reported emotion regulation, the highest prevalence of food addiction, as well as a metabolic profile characterized by the highest mean concentrations of leptin, adiponectin, and LEAP-2. Conclusions: To the best of our knowledge, this is the first study to identify well-differentiated GD clusters using neuropsychological and neuroendocrine features. Our findings reinforce the heterogeneous nature of the disorder and emphasize a role of potential endophenotypic features in GD subtyping. This more comprehensive characterization of GD profiles could contribute to optimize therapeutic interventions based on a medicine of precision.