Nonlinear association between gambling severity and retrospective gambling-related emotions: exploratory evidence from resting-state degree centrality

Objective Existing research on gambling disorder has primarily focused on emotional states following gambling episodes, difficulties in emotion regulation, or abnormalities in reward-based decision-making, whereas relatively less attention has been paid to patterns of change in individuals’ subjective emotional experiences in gambling-related contexts. It should be noted that emotional experiences in real-life gambling situations are immediate and fluctuating in nature. In contrast, the present study used questionnaires to assess participants’ retrospective, global evaluations of their emotional experiences in gambling-related contexts over the past year, which should not be equated with direct measurement of real-time emotional responses during gambling episodes. Therefore, this study aimed to examine the association pattern between gambling severity and retrospective gambling-related emotional experiences, and further to explore resting-state brain functional correlates that may be involved in this association. Methods A total of 111 participants who had gambling-related experiences within the past year were included in this study. Gambling severity was assessed using the Problem Gambling Severity Index (PGSI). The Positive and Negative Affect Schedule (PANAS) was used to assess participants’ retrospective evaluations of positive and negative emotional experiences in gambling-related contexts over the past year. A net affect score was calculated by subtracting the negative affect score from the positive affect score. Piecewise regression and restricted cubic spline regression were used to examine the nonlinear relationships between PGSI scores and net affect, positive affect, and negative affect scores. Paired-sample tests were conducted to compare positive and negative affect within different PGSI risk categories. In addition, 34 participants completed resting-state functional magnetic resonance imaging scans. Based on PGSI scores, the imaging sample was divided into a lower-PGSI group (PGSI ≤ 4, n = 15) and a higher-PGSI group (PGSI > 4, n = 19). Whole-brain degree centrality (DC) values were calculated, and two-sample t-tests were used to compare whole-brain DC differences between the two groups. DC values were then extracted from brain regions showing significant group differences, and moderation regression analysis was performed to explore whether PGSI group moderated the relationship between DC values in this region and retrospective net affect. Results Behavioral analyses showed a nonlinear association between PGSI scores and retrospective gambling-related net affect. Exploratory piecewise regression suggested that the vicinity of PGSI = 3 may represent a candidate turning point in changes in net affect. Within the PGSI > 3 range, net affect decreased significantly as PGSI scores increased (slope = -1.01, p < 0.001), and the direction of this slope differed from that observed in the PGSI ≤ 3 range, as indicated by a significant interaction term ( p = 0.005). Decomposition of affective components showed that the decline in net affect was mainly attributable to different trajectories of positive and negative affect as gambling severity increased: positive affect initially increased with higher PGSI scores and then tended to plateau, whereas negative affect continued to increase as PGSI scores increased. The restricted cubic spline regression results generally supported the findings from the piecewise regression analyses. Paired-sample tests showed that, at the problem gambling stage defined by PGSI ≥ 8, the predominance of positive affect over negative affect was attenuated and no longer reached statistical significance (median difference = 2, p = 0.0714). Exploratory neuroimaging analyses showed that, compared with the lower-PGSI group, the higher-PGSI group exhibited significantly lower DC values in the left medial superior frontal gyrus/paracingulate gyrus/BA10 region. Resting-state–retrospective affect association analysis showed that the interaction between PGSI group and DC values in this region reached statistical significance (β = -18.49, p = 0.045), suggesting that the association pattern between degree centrality in this region and retrospective net affect may differ across individuals with different levels of PGSI. Conclusion This study found a nonlinear association between gambling severity and retrospective gambling-related net affect, with the vicinity of PGSI = 3 emerging as a candidate interval in which net affect may shift from relative elevation to decline. This decline does not indicate that negative affect predominates at all levels of gambling severity; rather, it may reflect a gradual weakening of the predominance of positive affect, a continued increase in negative affect, and a tendency for the two affective components to become more balanced. Exploratory resting-state fMRI analyses showed that individuals with higher PGSI scores exhibited lower degree centrality in the left medial superior frontal gyrus/paracingulate gyrus/BA10 region, and that the association between DC values in this region and retrospective net affect may vary according to PGSI level.