Altered local intrinsic neural activity and molecular architecture in internet use disorders

,


Introduction
Internet gaming disorder (IGD) is a prototypical behavioral addiction characterized by compulsive gaming, inhibitory control deficits, and emotional dysregulation, which may cause various psychological and health-related issues, including depression, social anxiety, and insomnia (Macur and Pontes, 2021;Jorgenson et al., 2016;Paulus et al., 2018).In Western countries, the prevalence of pathological gaming in adolescents ranges from 1 % to 10 %.In contrast, the prevalence of IGD in Asian countries ranges from 10 % to 15 % (Saunders et al., 2017).This phenomenon has imposed tremendous financial and psychological burdens on the younger generation.Consequently, the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-V) incorporated IGD as a non-substance addiction for the first time (Petry et al., 2014).However, the underlying pathophysiology of IGD remains unknown despite numerous efforts.Therefore, investigating the neurobiological mechanisms of inhibitory control early in the disease course is critical for the treatment of IGD.
Recently, the triple network model (executive control network, default mode network, and salience network) has been proposed not only as the basis of the pathophysiology of mental disorders but also described as a typical feature of addiction (Wang et al., 2020a;Zhang et al., 2017a;Mestre-Bach et al., 2023).A group-level independent component analysis (ICA) reveals impaired regulation between networks in individuals with IGD (IGDs) and provides a theoretical basis for the triple network model of IGDs (Zhang et al., 2017a).However, merely focusing studies on large-scale alterations in the brain function network is not sufficient.The correlation between local blood oxygen fluctuations and global functional connectivity has been analyzed in schizophrenia and suggests that alterations in large-scale functional connectivity are associated closely with self-imposed local neural activity (Fu et al., 2018).Previous studies have demonstrated the significant value of local INA features in understanding the underlying pathology of diseases (Rice, 2006).And aberrant energy metabolism in brain regions may indicate an increase or decrease in resting-state metabolic rate (Fox and Raichle, 2007).Currently, amplitude of low-frequency fluctuation (ALFF) and regional homogeneity (ReHo) are frequently used to elucidate the features of local INA (Han et al., 2018;Ye et al., 2022).Compared with healthy controls (HCs), Sun and his colleagues found that IGDs revealed decreased ALFF in the left superior frontal gyrus (SFG) and the right middle frontal gyrus (MFG) (Sun et al., 2018).Dong and his colleagues revealed that ReHo decreased in the left temporal lobes and occipital lobes while increasing in the bilateral inferior parietal lobules and left cerebellum (Dong and Du, 2012).Therefore, focusing on the local intrinsic neural activity of brain regions holds the promise of providing information for understanding the neuropathological mechanisms in IGDs.Especially abnormal local neural activity may be an indication of adolescent-onset IGDs, which could facilitate early identification of individuals at higher risk of developing IGD and intervention treatments.
However, we suggest that the current study remains deficient in the following points: First, ALFF is sensitive to physiological and pathological noise.Therefore, the researchers proposed another indicator to describe local INA based on ALFF calculations-the fractional amplitude of low-frequency fluctuation (fALFF) (Zang et al., 2007).This index improves the sensitivity and specificity of neural activity detection (Zou et al., 2008).Yet, it is relatively less used in IGD.Second, the previous studies demonstrated that significantly different results were observed using ReHo and ALFF, respectively.A study suggests that ALFF provides a significant benefit in measuring neural activity intensity, while ReHo could complement the inadequacy of ALFF in local homogeneity (Zou et al., 2008).However, the effects of functional segregation and functional integration in IGDs were poorly studied, and whether fALFF combined with ReHo reflects vulnerability markers for IGDs is unclear.Third, the specific molecular architecture behind these changes in local neural activity is also unclear.Fourth, whether the combination of ReHo and fALFF improves the predictive probability of an IGD diagnosis.
Therefore, our study has the following objectives: 1) The fALFF and ReHo methods were used to explore the local INA alternations in IGDs in order to provide more comprehensive information; 2) The correlation between different brain regions and clinical characteristics was analyzed; 3) The JuSpace (https://github.com/juryxy/JuSpace)toolbox is used to calculate specific neurotransmitter system alterations in IGDs; 4) Receiver operating characteristic curve (ROC) analysis was used to examine the ability of fALFF and ReHo in distinguishing IGDs from HCs.

Participant selection
The present study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (2019-KY-297).After explaining the aims of the current study, all subjects signed informed consent forms.All examinations were carried out under the guidance of the 1975 Declaration of Helsinki (Shephard, 1976).
One hundred and three right-handed subjects, including 47 IGDs (age: 14.63 ± 0.33) and 56 HCs (age: 15.88 ± 0.81), were involved in this study.This study used the DSM-V and Young's Internet Addiction Test (IAT) as the diagnostic criteria for IGDs.The DSM-V consists of nine items and could be summarized into two core features: 1) the inability to undertake previous social roles and no longer participate in the life of the community; and 2) the loss of inhibitory control of self-behavior.The IAT scale consists of 20 items associated with online internet use, including psychological dependence, compulsive use, and withdrawal, as well as related problems in school or work, sleep, family, and time management.For each item, a graded response is selected from 1 = "rarely" to 5 = "always" or "does not apply" (Widyanto and McMurran, 2004).All subjects were first assessed for the diagnosis of IGD by an experienced psychiatrist against DSM-5 diagnostic criteria.Psychiatric medical disorders were screened by the Mini International Neuropsychiatric Interview.Subjects with an IAT score of 50 and more who meet more than five DSM-5-based criteria for the diagnosis of IGD are enrolled in the IGD group (Widyanto and McMurran, 2004;Battle, 2013).Those who scored at least 50 points on the IAT and did not meet the DSM-5 diagnostic criteria were enrolled in the control group.IAT has proven to be a reliable tool for IGD classification.All subjects did not have any drug abuse, dependence, or intellectual disability.Two questionnaires were used to access the subjects' clinical features, namely the Hamilton Anxiety Scale (HAMA) and the Hamilton Depression Scale (HAMD).After all interventions were completed, all subjects were instructed not to use any drugs of abuse, including coffee, alcohol, etc.

Imaging acquisition
All images were acquired on a 3 T magnetic resonance imaging scanner (Siemens Healthcare, Erlangen, Germany) with a standard head coil.The head motion and scanner noise were minimized by foam pads.During the scanning process, participants were required to close their eyes, not sleep, not think about anything, and not move around.The scanning parameters were repetition time (TR) = 1000 ms, echo time (TE) = 30 ms, flip angle = 70 • , field of view (FOV) = 220 × 220 mm 2 , voxel size = 2 × 2 × 2.2 mm 3 , slice thickness = 2.2 mm, slice = 52, and 400 volumes.At the end of scanning, participants were immediately asked whether they remained awake during scanning.

Data analysis
The functional magnetic resonance imaging data were preprocessed by the Resting-State fMRI Analysis Toolkit (DPARSF, v4.3; www.res tfmri.net) in MATLAB.The main steps and parameters are: 1) conversion of data formats (DICOM to NIFTI); 2) the first 10 volumes of each functional time series were removed; 3) slice-timing; 4) realignment (excluding subjects with a maximum head motion > 2.0 mm or rotation > 2.0 • ); 5) all the images were normalized to the EPI template and resampled to 3 mm 3 × 3 mm 3 × 3 mm 3 ; 6) removing linear trends and regression out nuisance covariates of 24 head motion parameters, global signals, white matter, and cerebrospinal fluid signals.There are two subtle details that we need to clarify here: For fALFF calculations, the data were not further filtered.However, for ReHo calculations, the data were further filtered using a bandpass temporal filter of 0.01-0.08Hz to reduce noise and low-frequency drift.And, for ReHo calculation, smooth was not used to avoid artificially inflating regional homogeneity because smoothing enforces a degree of autocorrelation or synchrony among spatially proximal voxels.Eventually, 7 IGDs and 6 healthy controls were excluded because of excessive head motions.40 IGDs and 50 healthy subjects were included in further analysis.
After being preprocessed, fALFF and ReHo were calculated based on DPABI software.For the fALFF calculation, the time series of each voxel was transformed to the frequency domain without bandpass filtering to obtain the power spectrum.Then the average square root of the power spectrum at each voxel across 0.01-0.1 Hz was taken as the ALFF value.
The fALFF was computed as the ratio of the power spectrum of the lowfrequency range divided by the entire frequency range (Zang et al., 2007).The ReHo was computed by calculating Kendall's coefficient of concordance (KCC) of the time series of a given voxel with its nearest 26 neighbors (Zang et al., 2004).Then all images were also Z-transformed and smoothed with an 8 mm full-width half maximum Gaussian kernel (FWHM) for statistical analysis.

Statistics analysis
Demographic and clinical characteristics were evaluated between the two groups.Two-sample t-tests were performed for the demographic characteristics (age and education level) and clinical scores (IAT scores, HAMD scores, and HAMA scores).Significant differences were considered at p < 0.05.
Group analysis of fALFF and ReHo maps was computed using the general linear model (GLM) in SPM12 with age, education level, and mean framewise displacement (FD) as nuisance covariates (Gaussian random field theory GRF corrected, p voxel < 0.005, p cluster < 0.05, twotails).The ROC curves were realized through Statistical Product and Service Solutions (SPSS).Predictive probability was first calculated using SPSS's embedded binary logistic regression separately for the brain regions that fALFF and ReHo altered and for the brain regions that fALFF and ReHo co-altered.Further, the ROC curve was construed with IGD diagnosis as the outcome variable and predicted probability as a covariate.

Correlation analysis
In order to determine the associations between the altered ReHo/ fALFF and clinical characteristics (including the severity of internet gaming addiction, depression, and anxiety), Pearson correlation analyses were performed between differential regions and IAT scores, HAMD scores, and HAMA scores with mean FD as a covariate.

Demographics and clinical date
No significant differences were found in age, education years, or mean FD.The IGDs had significantly higher IAT scores, HAMD scores, and HAMA scores compared with the HCs.The detailed demographic information is shown in Table 1.

Correlation analysis
The results showed that the ReHo values in the cerebellum_7b_R were positively correlated with IAT scores and HAMA scores (r = 0.476, p = 0.005; r = 0.377, p = 0.025).See Fig. 3.The fALFF value in the right SFG and bilateral SMA had no significant effect on clinical characteristics (IAT scores, HAMD scores, and HAMA scores).

Discussion
IGDs showed abnormal local neural activity within multiple regions of the brain, with a large consensus in the literature.The prefrontal cortex has been a crucial cortex for the neurological and pathophysiological underpinnings of IGDs (Jin et al., 2016).Its essential role in mediating objective measures of deficits in inhibitory control and cognitive impairment constitutes a neurobiological marker for IGDs (Wang et al., 2020b).Moreover, numerous studies have demonstrated that impulse inhibitory control is modulated by noradrenergic, GABAergic, and cholinergic informational inputs (Bari and Robbins, 2013).Thus, the specific neurotransmitter system abnormalities of local neural activity in our results are significant from a neuropathological and pharmacological perspective, which contributes to our understanding of how microstructural alterations drive macroscopic neuroimaging abnormalities in IGDs.
This study used ReHo combined with fALFF to comprehensively investigate the pattern of abnormal changes in local neural activity between IGD and HC and its potential associated neurotransmitter system.Compared to HCs, local functional impairments are shown in the audio-visual and inhibitory control circuits of IGD.More importantly, the main effects of IGDs are associated with the dopaminergic system (D1, D2, FDOPA), the µ-opioid receptor (MU), and the serotonergic system (5-HT2a, SERT), while they are also associated with the less studied GABAa system (GABAa) and the cholinergic system (VAChT).Our results revealed the potential neurobiological mechanisms of IGD.
PFC constitutes a critical component of the executive control network (ECN), which is associated with working memory, cognitive behavior, and inhibitory control (Kuss et al., 2018).The prefrontal L. Ma et al. Brain Research Bulletin 216 (2024) cortex is also involved in significant attribution, consciousness, and top-down higher-order executive functions by modulating limbic reward regions (Goldstein and Volkow, 2011).Recently, Matthias Brand proposed the Interaction of Person-Affect-Cognition-Execution (I-PACE) model, which suggests that addictive behavior develops as a consequence of the interaction between multiple factors (Brand et al., 2019).
And inhibitory control attenuation is considered to contribute to the development of habituated behaviors (Brand et al., 2019).Moreover, two essential features of IGD are inhibitory control and cognitive deficits (Brand et al., 2014).A Go/NoGo fMRI study suggests that the IGDs showed a significantly increased probability of making commission errors compared to HCs and exhibited increased activation in the PFC (Liu et al., 2014).These are consistent with our findings.We speculated that over-activation indicates inefficiency in inhibitory control and cognitive flexibility, which fit the Impaired Response Inhibition and Significant Attribution (IRSA) model.The IRSA model attributes significant reliance on drug-related rewards to reduced sensitivity to non-drug rewards (Canterberry et al., 2016).Therefore, with deficits in core functioning, we hypothesize that IGDs are accustomed to the thrill of the internet world and excessive pursuit of internet gaming at the expense of other activities.
The temporal lobe mainly possesses the functions of hearing, language, cognition, and emotion (Olson et al., 2007).MTG is recognized as a crucial brain region for language processing and expression (Kiehl et al., 2004).The occipital lobe is an important part of the visual network, which involves visual information processing and attention bias (Mechelli et al., 2000).An auditory oddball event-related potential (ERP) combined with an fMRI study found that local neural activity in the occipital regions was positively correlated with P3 amplitude (Park et al., 2023).Behavioral studies revealed that, compared to casual gamers, IGDs showed longer reaction times to computer-related stimuli (Heuer et al., 2021).This is consistent with our results.Therefore, we hypothesize that the decreased ReHo reflects the impairment of visual and auditory abilities caused by excessive computer games, prolonged screen time, and loud noise.A recent study contributes to our comprehension of the cerebellum's involvement in addiction and suggests that its intermediary function between motor processes and reward-motivation (Moulton et al., 2014).According to anatomic proof, extensive connections have been established between the cerebellum and the cerebral cortex (Middleton and Strick, 1997).More importantly, the cerebro-cerebellar model further explains how adaptive behaviors could be effectively performed (Boven and Cerminara, 2023).A recent study used calcium imaging to demonstrate at the molecular level that granule cells in the cerebellum engage in non-motor predictive activity: reward prediction error coding (Hull, 2020).Prediction errors are thought to be an essential phenotype of adaptive behavior (Mundorf et al., 2023).This study showed decreased ReHo in multiple regions of the cerebellum.Therefore, we speculated that reduced local neural activity in the cerebellum modifies the coding of reward prediction errors and disrupts brain top-down adaptive control of behavior, which further aggravates the individual's addiction to internet games.Notably, the present study showed ReHo values in the cerebellum_7b_R were positively correlated with IAT scores and HAMA scores.This suggests that ReHo values of the cerebellum could significantly increase during the development of the severity of IGDs, which may reflect an imbalance in local neural activity and further interfere with inhibitory control activity.Moreover, the present study provides robust evidence that the cerebellum is involved in emotion regulation in IGD.
Numerous substance and non-substance addictions are thought to be caused by dysfunction of the brain's monoamine neurotransmitter systems (dopamine, serotonin, and µ-opioid) (Zhang et al., 2017b).These neurotransmitter systems perform significant roles in normal inhibitory control and emotion regulation.This study calculated the spatial correlation between the mappings of 12 neurotransmitter systems and the mappings of ReHo and fALFF.We discovered that fALFF and ReHo alternations in IGDs were significantly correlated with the dopaminergic system (D1, D2, and FDOPA) and the µ-opioid receptor system (MU).The dopamine system was involved in the projection of specific brain regions with cognitive, emotional, and motor functions (Grace, 2016).Emerging evidence suggests that dysfunction of the dopamine system is closely associated with the pathophysiology of IGDs (Weinstein and Lejoyeux, 2020).A PET study indicates that the dysregulation of the D2 receptor-mediated frontal cortex may represent a potential mechanism for loss of control and compulsive behavior in IGDs (Tian et al., 2014).Moreover, in clinical pharmacology, bupropion can regulate dopamine release and inhibit dopamine reuptake, suggesting the neural mechanism of treating IGD (Stahl et al., 2004).Similar to the dopamine system, µ-opioid receptors subserve mood regulation, motivation, and cognitive function (Tejeda et al., 2012).The opioid receptor is a crucial player in the pathophysiology of addiction.An earlier study discovered that nicotine upregulates mu opioid receptors and increases the release of  endogenous opioids in smokers (Kuwabara et al., 2014).A study discovered that opiate could induce self-directed aggression in internet users (Siever, 2008), while naltrexone (the opioid receptor antagonist) significantly improved this abnormal behavior and successfully treated an individual with compulsive internet use (Bostwick and Bucci, 2008).Notably, dopamine and opioid receptors also perform important regulatory roles in the development of the central nervous system (Tejeda et al., 2012;Money and Stanwood, 2013).Therefore, we speculate that alterations in local neural activity and behavior may have a significant impact on the stabilization of dopamine and opioid receptors during the neurodevelopmental period in IGDs.
Moreover, our results also revealed that ReHo alternations were weakly correlated with the serotonergic system (5-HT2a, SERT), the GABAergic system (GABAa), and the cholinergic system (VAChT).Structurally, serotonin is involved in neurogenesis, development, and the construction of neural pathways (Dayer, 2014).Functionally, the serotonergic system is further engaged in regulating reward and preference (Wedekind et al., 2010).Elevated serotonergic transporter genes have been identified in IGDs (Lee et al., 2008).We hypothesized that persistent game-rewarding behaviors lead to increased reuptake of serotonin.And serotonin reuptake inhibitors (SSRI) have also been reported to effectively mitigate the symptoms of internet addiction (Sá et al., 2023).The cholinergic systems are widely distributed in the thalamus, striatum, and limbic system and perform an instrumental role in memory and attention (Hampel et al., 2018).GABAergic systems are involved in maintaining the excitation/inhibition balance in the reward circuit (Hosseinzadeh Sahafi et al., 2023).Numerous addictive substances (alcohol, nicotine, marijuana, and cocaine) achieve a reduction in inhibitory GABAergic inputs to DAergic neurons and potentiation of glutamatergic inputs through activating the α4β2nAChRs and α7nAChRs in DAergic neuronal processes of the VAT projections into the NAc (Hosseinzadeh Sahafi et al., 2023;Thorpe et al., 2020;Vallés and Barrantes, 2023).However, as far as we know, the current studies regarding the impact of the GABAergic and cholinergic systems in IGDs remain limited.On the one hand, a new candidate gene in internet addiction, the nicotinic acetylcholine receptor subunit alpha 4 (CHRNA4), was first reported in a 2012 case-control study (Montag et al., 2012).Through further investigation of the pathogenesis of IGD and 72 candidate genes, only CHRNA4 was discovered to be significantly associated with IGD (Jeong et al., 2017).On the other hand, a study measured microRNA expression levels in IGDs and, further combined with western blot analysis, revealed that the miRNA downstream target GABRB2 was significantly overexpressed in IGDs (Lee et al., 2018).Moreover, dysregulation of the GABAergic and cholinergic receptors is associated with negative emotions, including anxiety, depression, and mood disorders (Kumar et al., 2013;Benowitz, 2009), which are also characteristic of IGD (Markett et al., 2011).Presently, GABAergic receptor agonists (baclofen) and acetylcholinesterase inhibitors (galantamine) have shown promising therapeutic effects in the treatment of substance addiction (Logge et al., 2022;Donlon et al., 2024).However, bupropion, methylphenidate, and serotonin reuptake inhibitors stand as the prevailing pharmacological agents in IGD treatment.In the future, it may be possible to develop a drug targeting the cholinergic system and the GABAergic system for the treatment of IGDs.

Limitation
The following limitations of this study cannot be ignored: First, the subjects in this study included both males and females.But previous studies have found gender differences in IGDs.Therefore, in the future, gender differences in the local functional activity of IGDs and their neurotransmitter changes need to be explored.Second, the sample size was relatively small, and future studies will need a larger sample to verify this result.Third, the questionnaire score depends on the subjective perception of different IGDs.Therefore, correlation analyses were not significant.Fourth, because current research is still in a cross-sectional design, we couldn't determine the causal relationship between abnormal INA and addiction.In the future, longitudinal research is still needed to complement the current findings.

Conclusions
In this study, our results revealed widespread local dysfunction in IGDs, which may be associated with impairments in audiovisual processing and inhibitory control.The combination of ReHo and fALFF results in a greater diagnosis of IGD.More importantly, the local intrinsic neural activity alternations are associated with the monoamine neurotransmitter system and the less-studied GABAergic system/ cholinergic system.This study builds a link between neuroimaging and neurotransmitter information, contributing to a comprehensive understanding of IGD.

Fig. 4 .
Fig. 4. The ReHo and fALFF separately and combine to distinguish the ROC curves of IGDs.

Table 1
Demographic and clinical characteristic of subjects.

Table 2
Significant Group Differences in fALFF and ReHo.