Development of a classifier for gambling disorder based on functional connections between brain regions.

AIM: Recently, a machine-learning (ML) technique has been used to create generalizable classifiers for psychiatric disorders based on information of functional connections (FCs) between brain regions at resting state. These classifiers predict diagnostic labels by a weighted linear sum (WLS) of the correlation values of a small number of selected FCs. We aimed to develop a generalizable classifier for gambling disorder (GD) from the information of FCs using the ML technique and examine relationships between WLS and clinical data. METHODS: As a training dataset for ML, data from 71 GD patients and 90 healthy controls (HCs) were obtained from two magnetic resonance imaging sites. We used an ML algorithm consisting of a cascade of an L1-regularized sparse canonical correlation analysis and a sparse logistic regression to create the classifier. The generalizability of the classifier was verified using an external dataset. This external dataset consisted of six GD patients and 14 HCs, and was collected at a different site from the sites of the training dataset. Correlations between WLS and South Oaks Gambling Screen (SOGS) and duration of illness were examined. RESULTS: The classifier distinguished between the GD patients and HCs with high accuracy in leave-one-out cross-validation (area under curve (AUC = 0.89)). This performance was confirmed in the external dataset (AUC = 0.81). There was no correlation between WLS, and SOGS and duration of illness in the GD patients. CONCLUSION: We developed a generalizable classifier for GD based on information of functional connections between brain regions at resting state.

Framing effects on financial and health problems in gambling disorder.

Gambling disorder (GD) patients show excessively risky decision-making in the financial domain. We aimed to clarify whether GD patients show risky decision-making in domain-general or in domain-specific. Furthermore, we also investigated the effect of the well-known cognitive bias, the framing effect on GD's decision-making under risk. Sixty-two male GD patients and 74 age-matched healthy male controls (HC) conducted a risky choice task in which they should choose solutions for difficult situations between a sure and a risky option that had the same expectations. Six situations were prepared for each financial and health domain. For each domain, three situations were presented with options using positive frames, and the other three situations were presented with options using negative frames. The results showed that GD chose more risky options in the financial domain with positive frames than HC, but chose comparably in the financial domain with negative frames, whereas GD and HC chose comparably in the health domain regardless of the frames. Thus, GD showed risky decision-making in domain-specific. In addition, the results indicate the importance of considering the influence of the framing effect for assessment of risky decision-making by GD. Domains and the influence of the framing effect should be considered when decision-making patterns of neuropsychiatric disorders are studied.

Amygdala volume is associated with risky probability cognition in gambling disorder.

Gambling disorder (GD) is characterized by continual gambling despite negative consequences. Risky decision-making is a hallmark of the disorder. We applied a tool from behavioral economics for assessing probability cognition in both gain and loss domains to GD. We aimed to examine the alteration of probability cognition and its relationship with brain structure in GD. Forty-six GD patients and 52 age-matched healthy controls (HCs) conducted a risky choice task in which subjects should choose between a sure and a risky option in both loss and gain domains. The distortion and elevation parameters of the probability weighting function were estimated. We compared the parameters between GD and HC and examined their relationships with the striatum and amygdala volumes in GD. GD showed greater elevation parameter in the gain domain and smaller regional gray matter volume in the left amygdala than HC. The elevation parameter in the gain domain showed a negative correlation with the left amygdala volume in GD. Altered probability cognition in the gain domain but not in the loss domain might be more relevant to risky decision-making in GD. Our findings indicate that alteration in the amygdala might play a significant role in risky decision-making of GD. Longitudinal studies are recommended to examine the causal relationship between brain abnormalities and risky decision-making in GD.

An fMRI study of decision-making under sunk costs in gambling disorder.

The sunk cost effect is the tendency to continue an investment, or take an action, even though it has higher future costs than benefits, if costs of time, money, or effort were previously incurred. This type of decision bias is pervasive in real life and has been studied in various disciplines. Previous studies and clinical observations suggest that decision-making under sunk costs is altered in gambling disorder (GD). However, the neural mechanisms of decision-making under sunk costs in GD remain largely unknown, and so is their association with the clinical characteristics of this patient group. Here, by combining functional magnetic resonance imaging and the task that demonstrated a clear example of the sunk cost effect, we investigated the neural correlates during decision-making under sunk costs in GD. We found no significant differences in the strength of the sunk cost effect between the GD and healthy control (HC) groups. However, the strength of the sunk cost effect in patients with GD showed a significant negative correlation with abstinence period and a marginally significant positive correlation with the duration of illness. We also found a reduction in the neural activation in the dorsal medial prefrontal cortex during decision-making under sunk costs for the GD group compared with the HC group. Furthermore, in patients with GD, the levels of activation in this area negatively correlated with the duration of illness. These findings have important clinical implications. This study will contribute to a better understanding of the mechanisms underlying altered decision-making abilities in GD.

Common and differential brain abnormalities in gambling disorder subtypes based on risk attitude.

Studying brain abnormalities in behavioral addiction including GD enables us to exclude possible confounding effects of exposure to neurotoxic substances, which should provide important insight that can lead to a better understanding of addiction per se. There have been a few brain structural magnetic resonance imaging studies for GD, although the results have been inconsistent. On the other hand, GD was suggested to be a heterogeneous disorder in terms of risk attitude. We aimed to examine the heterogeneity of GD by combining a behavioral economics task and voxel-based morphometry. Thirty-six male GD patients and 36 healthy male control subjects underwent a task for estimation of loss aversion, which can assess risk attitude in real-life decision-making. The GD patients were divided into two groups based on their level of loss aversion, low and high. While both groups showed common gray matter volume reduction in the left supramarginal gyrus and bilateral posterior cerebellum, high loss-aversion GD showed pronounced reduction in the left posterior cerebellum and additional reduction in the bilateral medial orbitofrontal cortex. Our study suggests that the heterogeneity of GD is underpinned at the brain structural level. This result might be useful for understanding neurobiological mechanisms and for the establishment of precise treatment strategies for GD.

Heterogeneity of Loss Aversion in Pathological Gambling.

Pathological gambling (PG) is characterized by continual repeated gambling behavior despite negative consequences. PG is considered to be a disorder of altered decision-making under risk, and behavioral economics tools were utilized by studies on decision-making under risk. At the same time, PG was suggested to be a heterogeneous disorder in terms of personality traits as well as risk attitude. We aimed to examine the heterogeneity of PG in terms of loss aversion, which means that a loss is subjectively felt to be larger than the same amount of gain. Thirty-one male PG subjects and 26 male healthy control (HC) subjects underwent a behavioral economics task for estimation of loss aversion and personality traits assessment. Although loss aversion in PG subjects was not significantly different from that in HC subjects, distributions of loss aversion differed between PG and HC subjects. HC subjects were uniformly classified into three levels (low, middle, high) of loss aversion, whereas PG subjects were mostly classified into the two extremes, and few PG subjects were classified into the middle range. PG subjects with low and high loss aversion showed a significant difference in anxiety, excitement-seeking and craving intensity. Our study suggested that PG was a heterogeneous disorder in terms of loss aversion. This result might be useful for understanding cognitive and neurobiological mechanisms and the establishment of treatment strategies for PG.