Introduction

Methamphetamine (METH) is a widely misused psychostimulant, and methamphetamine use disorder (MUD) poses significant public health challenges, contributing to substantial morbidity and mortality worldwide [1]. METH abuse results in a range of severe neurological and physical consequences, including sleep disturbances, aggression, emotional dysregulation, and psychosis characterized by delusions and hallucinations [2]. According to a 2023 provisional estimate from the Centers for Disease Control and Prevention (CDC) in the United States, over 36,000 deaths were linked to psychostimulant-related causes [3]. Despite the severity of these outcomes, effective treatment strategies remain lacking. A major barrier to therapeutic advancement is the limited understanding of the neurobiological mechanisms underlying METH addiction, highlighting the urgent need to clarify its pathophysiological basis.

The pathological mechanisms underlying drug addiction are exceedingly complex, with deficits in impulse control recognized as a central phenotype that significantly contributes to the development of addiction [4]. Neurobehavioral studies indicate that dysfunction within the “decision-action execution” circuitry contributes to the progression of addiction, with heightened impulsivity emerging as a primary behavioral phenotype associated with substance use disorders [5, 6]. Compelling evidence points to the hyperdirect pathway, an essential inhibitory control network comprising the right inferior frontal gyrus (rIFG), pre-supplementary motor area (pre-SMA), thalamus, and subthalamic nucleus (STN), among other regions, as neurophysiologically vital for the regulation of impulsive behaviors [7]. However, the specific mechanisms through which METH addiction disrupts the functional integrity of this inhibitory control circuit, thereby influencing addiction trajectories, remain inadequately elucidated.

The relationship between abnormal brain structure and impulse control has been investigated in several studies. In particular, METH users with reduced gray matter volume (GMV) in the right superior frontal gyrus have been found to exhibit higher impulsivity scores [8]. Similarly, decreased gray matter intensity in the left superior frontal gyrus, coupled with increased gray matter density in the posterior cingulate cortex and ventral striatum, has been correlated with poorer performance on a delay-discounting task [9]. However, findings have not always been consistent. For example, one study reported no significance association between reduced gray matter intensity in the rIFG and inhibitory control performance in METH users [10]. In addition, enlarged striatal structures have been linked to higher levels of impulsivity in both stimulant users and their non-using siblings, suggesting a possible genetic predisposition to these phenotypes [11]. Given the inconsistencies in previous findings regarding brain morphometry and impulsive behavior, as well as the limited exploration of underlying genetic mechanisms, more research is warranted to clarify the neural and genetic correlates of impulsivity in METH users.

Therefore, the present study aimed to investigate the effects of METH use on structural brain alterations, their association with impulsivity, and the underlying genetic mechanism, using magnetic resonance imaging (MRI) scanning and whole exon sequencing. Based on prior morphometric findings, we hypothesized that METH users would exhibit structural abnormalities within the hyperdirect pathway compared to age- and sex-matched healthy controls. Furthermore, we anticipated that METH users with morphometric alterations in the inhibitory control network would display higher levels of impulsivity. Lastly, we sought to identify specific genes and biological pathways associated with these structural abnormalities, providing insight into the genetic underpinnings of brain morphometric changes in METH users.

Patients and methods

Study design and participants

This was a cross-sectional case-control study. Participants were recruited between January 2018 and January 2021, including patients from a Compulsory Detoxification Center in Chengdu and healthy controls (HCs) from the local community. HCs were matched to patients by gender, age, and ethnicity through community advertisements. Inclusion criteria for patients were as follows: aged 18–55 years, Han Chinese, right-handed, and meeting the diagnostic criteria for MUD as defined by the Chinese version of the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). Inclusion criteria for HCs were: matched with patients in terms of gender, age, and ethnicity; right-handed; and without any comorbid psychiatric or physical disorders. Exclusion criteria for all participants included: current or past diagnosis of a substance use disorder other than METH or nicotine; a history of psychiatric disorders unrelated to substance use; major medical conditions (e.g., cardiovascular disease, severe hepatic or renal dysfunction); history of brain injury or loss of consciousness exceeding 10 min; left-handedness; and contraindications for MRI scanning. A total of 173 participants were initially recruited. Of these, 31 were excluded: 10 for polysubstance use, 4 for major psychiatric comorbidities, and 17 for incomplete data (e.g., failed MRI scans, missing genetic or clinical data). The final sample comprised 142 male participants.

This study was approved by the West China Hospital of Sichuan University Biomedical Research Ethics Committee (Date: Sep 1st, 2017/No: 2017-449), and it adhered to the tenets of the Declaration of Helsinki. All the participants who were willing to participate in this study were informed of the study purpose, methods, and possible risks and benefits they could receive from this study in advance, and also informed to that they have liberty to refuse and this would not influence their further treatment. Those who agreed to participate signed an informed consent form. To minimize coercion risks, data collection was conducted by independent researchers who were unaffiliated with the facility staff, and participants were explicitly assured that their responses would not influence treatment outcomes or privileges.

Assessments

General questionnaire

All participants underwent a structured interview conducted by a certificated psychiatrist. The interview collected detailed information, including age, ethnicity, educational level, smoking and drinking status, co-use of other substances, age at first METH use, duration of METH use, average dose per use, frequency of use, and duration of abstinence. A history of METH use was confirmed through urine toxicology testing, which was administered immediately after the participants entered the agencies.

Visual analogue scale

The visual analogue scale (VAS) is a widely used tool for assessing current psychological craving. In this study, it was employed to evaluate participants’ craving for METH following exposure to drug-related cues. Drug-related images depicting METH-related paraphernalia and drug-use contexts were compiled from publicly available sources and prior cue-reactivity studies. All images were screened by experienced clinicians and researchers to ensure ecological validity and relevance to METH use. Representative examples of the stimuli are provided in the Supplementary Materials (Figure S1). Participants were shown a line marked from 0 to 10, with higher values indicating stronger craving. After being presented with drug-related stimuli, they were instructed to select a number on the scale that best represented the intensity of their current craving for METH.

Barratt impulsiveness scale, version 11

The Chinese version of the Barratt Impulsiveness Scale, Version 11 (BIS-11), one of the most widely used self-report questionnaires for assessing impulsivity, was utilized in this study [12, 13]. The BIS-11 consists of 30 items that evaluate three dimensions of impulsivity: attentional impulsivity (AI; i.e., difficulties in focusing or completing tasks), motor impulsivity (MI; i.e., acting without considering the consequences), and nonplanning impulsivity (NPI; i.e., a lack of foresight and a tendency toward an unstructured lifestyle). Each item is rated on a 5-point Likert scale ranging from 1 (never) to 5 (always), with higher scores indicating greater impulsivity [14].

MUD severity

The severity of MUD was assessed using the latest edition of DSM-5, which provides standardized criteria for diagnosing substance use disorders. The DSM-5 adopts a dimensional approach, focusing on the severity of the disorder rather than distinguishing between substance abuse and dependence. Severity is classified based on the number of diagnostic criteria met: no diagnosis (0-1 symptoms), mild (2-3 symptoms), moderate (4-5 symptoms), and severe (6 or more symptoms).

MRI data acquisition and preprocessing

All participants underwent MRI scanning using a 3.0 T scanner (Achieva; Philips, Amsterdam, the Netherlands) equipped with an eight-channel phased-array head coil in the Department of Radiology. High-resolution T1-weighted images were acquired using a three-dimensional magnetization-prepared rapid gradient-echo (3D MPRAGE) sequence. The imaging parameters were as follows: repetition time (TR) = 8.37 ms, echo time (TE) = 3.88 ms, flip angle = 7°, in-plane matrix resolution = 256 × 256, field of view (FOV) = 240 mm × 240 mm, and number of slices = 188. The quality of each brain scan was assessed immediately after acquisition, and scans were repeated if any substantial artifacts or distortions were detected.

T1-weighted images were processed using the Diffeomorphic Anatomical Registration Through Exponentiated Lie algebra (DARTEL) toolbox implemented in Statistical Parametric Mapping 12 (SPM12), following the standard procedure described by Ashburner [15]. The MRI data processing steps were as follows: First, T1-weighted images were manually realigned according to the commissure-posterior commissure (AC-PC) line and the midsagittal plane. Second, all T1-weighted images were segmented into probability maps of gray matter (GM), white matter (WM), and cerebrospinal fluid using SPM12. The resulting GM and WM maps were then rigidly aligned (three rotations and three translations) to the Montreal Neurological Institute (MNI) space and resampled to 1 mm isotropic voxel resolution. Third, flow fields and a series of template images were generated using the ‘DARTEL’ routine, based on the previously segmented GM and WM images. Fourth, the flow fields and final DARTEL template were used to generate smoothed (6 mm full-width at half-maximum [FWHM] isotropic Gaussian kernel), modulated, and spatially normalized GM images in MNI space. Finally, GMV maps were extracted for subsequent statistical analyses.

Genetic data acquisition and preprocessing

Sequence processing

Genomic DNA was collected from peripheral blood and extracted using established Illumina paired-end protocols. Agilent SureSelect Human All Exon V6 Kit (Agilent Technologies, Santa Clara, CA, USA) was used for exome capture for all cases and controls according to the manufacturer’s instructions. GATK Best Practices were adopted for data processing [16].

Quality control

FastQC was employed to assess the overall quality of the raw sequencing reads. VeryfyBamID [17] was then utilized to detect contamination across all subjects. To further ensure data integrity, PLINK [18] was applied to evaluate population stratification, relatedness between subjects, and sex consistency. Quality control (QC) at the variant and genotype levels was performed using GATK [16] and KGGSeq [19]. Detailed parameters for these QC steps are provided in the online Supplementary Materials.

Annotation

Variant annotation was performed using KGGSeq [19], with classification into 17 functional categories based on the Human Genome Variation Society recommendations. For subsequent analyses, only non-synonymous exonic and splicing variants were retained. Minor allele frequencies (MAF) for each variant were obtained from three public databases: the Exome Aggregation Consortium (ExAC r0.3), the East Asian panel of the 1000 Genomes Project (Aug 2015 release; 1kgeas), and the South Asian panel of the 1000 Genomes Project (Aug 2015 release; 1kgsas). The potential functional impact of each variant was further evaluated using the Combined Annotation Dependent Depletion (CADD) score [20].

Statistical analyses

Independent samples t tests were used to compare normally distributed continuous variables between groups, while nonparametric tests were applied for variables that did not follow a normal distribution. Categorical variables were analyzed using the chi-square test. For neuroimaging analyses, whole-brain voxel-wise comparisons of GMV between groups were conducted using analysis of covariance (ANCOVA) in SPM12, with age, educational level, current smoking status, history of alcohol use, and total intracranial volume (TIV) included as covariates. Statistical maps were thresholded at a voxel-wise level of p < 0.001 (uncorrected), followed by cluster-level family-wise error (FWE) correction at p < 0.05. Significant clusters were identified based on this cluster-wise correction and reported in the Results section. Effect sizes were calculated at the cluster level by extracting mean modulated GMV values from each significant cluster for each participant, followed by computation of Cohen’s d to quantify group differences.

Partial correlation analyses were conducted to examine the relationships among GMV of brain regions exhibiting intergroup differences, drug use parameters, and trait impulsivity. Mediation analyses were further performed to investigate whether trait impulsivity mediated the relationship between GMV in specific brain regions and drug use characteristics in individuals with MUD (MUDs), adjusting for age, educational level, current smoking status, history of alcohol use, TIV, and duration of METH abstinence. General statistical analyses were performed using SPSS software (version 26.0). Mediation models were estimated using the PROCESS macro for SPSS (version 3.2), with 5000 bootstrap samples to assess the significance of indirect effects [21, 22]. A two-tailed p-value < 0.05 was considered as statistically significant. FWE correction and false discovery rate (FDR) correction were applied as appropriate to control for multiple comparisons.

Statistical analyses of rare variants

A total of 35,950 rare variants were analyzed using a gene-based burden approach because individual rare variants (MAF ≤ 1%) typically lack sufficient statistical power to be tested independently in samples of this size. The Zeggini burden test collapses multiple rare variants within the same gene into a single genetic burden score, thereby increasing statistical power under the assumption that rare variants within a gene influence the phenotype in the same direction. Accordingly, we adopted gene-based Zeggini burden tests implemented in RvTest [23] to examine the association between aggregated rare variants within each gene and thalamic GMV. Specifically, rare variants (MAF ≤ 1%) within a given gene were collapsed into a single burden score, which was then entered into a linear regression model with thalamic GMV as the dependent variable, adjusting for age and the first two principal components (PCs) to account for population stratification. All curated gene sets from the Molecular Signatures Database (MSigDB) [24] were evaluated.

Statistical analyses of common variants

For common variants analyses, exonic variants passing quality control were further filtered based on a MAF at least 5%, as defined in the ExAC database. These variants were first tested using logistic regression models comparing MUDs and HCs, adjusted for age, sex, and the top two PCs. Variants showing nominal associations with diagnostic status (p ≤ 0.1) were retained as candidate loci, resulting in a total of 362 common variants. Genes mapped from these variants were subsequently subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using the ClusterProfiler package (version 4.16.0) in R software (version 4.5.1) to identify gene pathways associated with these common variants [25]. Furthermore, selected common variants were further examined in gene-environment interaction analyses to assess the interaction effects between individual single nucleotide polymorphisms (SNPs) and drug use characteristics on reginal GMV, while controlling for age, educational levels, current smoking status, history of alcohol consumption, and TIV. A two-tailed p-value < 0.05 was considered as statistically significant and FDR correction was applied as necessary to account for multiple comparisons.

Results

Demographic data, drug use, and trait impulsivity

A total of 142 participants were included in the study, comprising 51 HCs and 91 MUDs, as diagnosed according to DSM-5 criteria. As shown in Table 1, there was no significant difference in age between the two groups (p > 0.05). However, significant group differences were observed in educational level, current smoking status, and history of alcohol consumption (all p < 0.001). Table 1 also summarizes the METH use characteristics of the MUD group. Compared to HCs, MUDs exhibited significantly higher scores in MI, AI, NPI, and BIS-11 scores (all p < 0.01).

Table 1 Demographic information of healthy controls and patients with methamphetamine use disorder.
Full size table

Group differences in GMV across the brain

After controlling for age, educational level, current smoking status, history of alcohol consumption, and TIV in the group comparison analyses, MUDs exhibited significantly reduced GMV in the left thalamus compared to HCs (cluster-wise p < 0.0001, FWE corrected; clusters size: 1771 mm3; Fig. 1). The mean GMV extracted from this cluster was significantly lower in the MUD group, with a moderate effect size (Cohen’s d = 0.44). No other brain regions exhibited significant group differences after correction for multiple comparisons.

Fig. 1: Differences in gray matter volume between healthy controls and individuals with methamphetamine use disorder.
Fig. 1: Differences in gray matter volume between healthy controls and individuals with methamphetamine use disorder.
Full size image

HC vs. MUD; p < 0.0001, FWE corrected.

Relationships between METH use parameters, trait impulsivity and GMV of specific brain in MUDs

Partial correlation analyses were conducted to explore the potential relationships among METH use parameters, trait impulsivity and GMV of the left thalamus within the MUD group. After controlling for age, educational levels, current smoking status, history of alcohol consumption, TIV and duration of abstinence, a significantly negative correlation was found between GMV of the left thalamus and MI (r = −0.250, p < 0.05, Fig. 2). Additionally, MI was positively associated with MUD severity (r = 0.283, p < 0.01, Fig. 2). Furthermore, simple mediation analyses revealed that MI significantly mediated the relationship between GMV of the left thalamus and both the duration of METH use (indirect effect = 40.00%, 95%CI: [−103.55, −4.544], Fig. 3a) and MUD severity (indirect effect = 52.82%, 95%CI: [−2.064, −0.0.016], Fig. 3b).

Fig. 2: Results of partial correlation analysis among gray matter volume of the specific brain region, trait impulsivity, and drug use characteristics.
Fig. 2: Results of partial correlation analysis among gray matter volume of the specific brain region, trait impulsivity, and drug use characteristics.
Full size image

GMV, gray matter volume; MI, motor impulsivity; AI, attention impulsivity; NPI, nonplanning impulsivity; METH, methamphetamine; VAS, visual analogue scale; MUD, methamphetamine use disorder. ***: p < 0.001, **: p < 0.01, *: p < 0.05.

Fig. 3: Mediating effects of trait impulsivity in the relationship between gray matter volume of the specific brain region and drug use characteristics.
Fig. 3: Mediating effects of trait impulsivity in the relationship between gray matter volume of the specific brain region and drug use characteristics.
Full size image

a, mediating role of motor impulsivity between thalamic volume and drug use duration; (b), mediating role of motor impulsivity between thalamic volume and clinical severity. GMV, gray matter volume; MI, motor impulsivity; METH, methamphetamine; MUD, methamphetamine use disorder.

Correlation between GMV of left thalamus and rare variants

In the gene-based Zeggini burden test, 72 genes (online supplementary table S1) were identified as significantly associated with GMV of the left thalamus. Furthermore, as shown in Table 2, three pathways from the curated gene sets in the MSigDB were significantly associated with thalamic GMV: Cytidine Analogs Cytotoxicity (M19805; p = 1.99*10−6, FDR corrected), Muscle Structure Development (M12572; p = 2.97*10−6, FDR corrected), and Nuclear Speck (M17120; p = 6.18*10−6, FDR corrected).

Table 2 Gene set of rave variants associated with gray matter volume of left thalamus.
Full size table

Enriched biological pathways associated with common genetic variants in MUDs

A total of 362 common variants showing evidence of association in the regression analyses were selected and subsequently included in downstream functional annotation and GO enrichment analyses (online supplementary table S2). GO enrichment analyses revealed that enriched biological process (BP) were primarily related to ciliary and microtubule-based motility, including cilium movement (GO:0003341, adjusted p = 0.021), microtubule-based movement (GO:0007018, adjusted p = 0.021), and cilium-dependent cell motility (GO:0060285, adjusted p = 0.021). In the cellular component (CC) category, enriched terms were mainly associated with motile ciliary structures and nuclear envelope-associated complexes, such as motile cilium (GO:0031514, adjusted p = 0.001), microtubule organizing center attachment site (GO:0034992, adjusted p = 0.001), and nuclear membrane microtubule tethering complex (GO:0106094, adjusted p = 0.001). In the molecular function (MF) domain, significant enrichment was identified in terms related to cytoskeletal motor and structural activity, such as minus-end-directed microtubule motor activity (GO:0008569, adjusted p = 0.003), actin binding (GO:0003779, adjusted p = 0.006), and cytoskeletal motor activity (GO:0003774, adjusted p = 0.006) (Fig. 4a).

Fig. 4: Pathways enrichment analyses for common genetic variants in individuals with methamphetamine use disorder.
Fig. 4: Pathways enrichment analyses for common genetic variants in individuals with methamphetamine use disorder.
Full size image

a, Gene Ontology (GO) enrichment results for common variants in methamphetamine users; (b), Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment results for common variants in methamphetamine users. BP, biological process; CC, cellular component; MF, molecular function; ECM, extracellular matrix.

KEGG pathways enrichment analysis further supported the involvement of genes in cell motility and intercellular signaling (Fig. 4b). The most significantly enriched pathway was Cytoskeleton in muscle cells (hsa04820; adjusted p = 1.24 × 10⁻⁶), involving genes such as HSPG2, OBSCN, TTN, and ACTN3. Another enriched pathway in the same functional cluster was Motor proteins (hsa04814; adjusted p = 0.007), which encompassed several kinesin and dynein family members, including KIF15, DYNC2H1, DNAH14. In addition, genes were significantly enriched in the ECM-receptor interaction pathway (hsa04512; adjusted p = 0.007), a key signaling pathway involved in extracellular matrix binding and cellular communication.

Interactive effects of genetic variants and drug use on brain structure

In the analyses of gene-environment interactions, individuals with the GG genotype at the CR1L rs12729569 locus exhibited a more substantial decrease in thalamic volume among drug users compared to non-users and those with other genotypes after controlling for age, educational levels, current smoking status, history of alcohol consumption, and TIV (p = 0.002, uncorrected, Fig. 5a). Similarly, MUDs with the AG genotype at the SDC3 rs2282440 locus demonstrated significantly reduced thalamic volume relative to individuals with the GG or AA genotypes (p = 0.002, uncorrected, Fig. 5b).

Fig. 5: Interaction effects between common genetic variants and drug use characteristics on gray matter volume in the thalamus.
Fig. 5: Interaction effects between common genetic variants and drug use characteristics on gray matter volume in the thalamus.
Full size image

a, interaction between rs12729569 and methamphetamine use status on thalamic gray matter volume; (b), interaction between rs2282440 and duration of methamphetamine use on thalamic gray matter volume. GMV, gray matter volume; METH, methamphetamine; Non-use, individuals with no history of methamphetamine use; Use, methamphetamine user.

Discussion

Inconsistent with previous reports [8,9,10], this study identified reduced GMV in the left thalamus in MUDs compared to HCs. Moreover, thalamic atrophy was associated with higher MI, which significantly mediated the relationship between thalamic volume and both the duration of METH use and MUD severity. Additionally, GMV in the left thalamus was linked to rare genetic variants and biological pathways involved in cytidine analog cytotoxicity, muscle structure development, and nuclear speckle organization. GO and KEGG enrichment analyses further revealed a significant overrepresentation of ciliary and microtubule-based biological processes in MUDs. Interaction analyses indicated notable gene-environment interactions, particularly between the rs12729569 locus and drug use, as well as between the rs2282440 locus and the duration of METH use, both influencing thalamic volume. Collectively, these findings provide novel insights into the neuroanatomical and genetic underpinnings of MUD.

Using VBM, we observed significant GMV reductions in the left thalamus of MUDs, compared to healthy volunteers. To our knowledge, thalamic atrophy has not been previously reported in METH users; however, similar structural deficits have been documented in users of alcohol [26], tobacco [27], cocaine [28], and synthetic cannabinoids users [29]. The absence of thalamic findings in earlier METH studies may be attributable to limited sample sizes and methodological variability [8, 30]. Several potential mechanisms may underlie the observed thalamic volume reductions in METH users. One prominent hypothesis involves the neurotoxic effects of METH. Specifically, METH enters presynaptic neurons via dopamine transporters, promoting dopamine efflux and elevating dopamine concentrations within the synaptic cleft [31]. This cascade induces oxidative stress and mitochondrial dysfunction, which can ultimately lead to neuronal apoptosis and synaptic damage within the thalamus [32]. Another plausible mechanism is METH-induced neuroinflammation: chronic METH exposure activates microglia and astrocytes in the brain, triggering neuroinflammatory responses that contribute to thalamic neurodegeneration [33, 34]. Furthermore, the genetic background of an individual may affect the susceptibility to METH-induced neurotoxicity. While no dopamine receptor gene (DRD2) variants were detected in our cohort, previous studies suggest that variations in the dopamine receptor gene (DRD2) may increase vulnerability to neuronal damage by dysregulating dopaminergic signaling in the thalamus [35]. METH exposure may also alter the expression of thalamus-related genes via epigenetic modification [36]. Taken together, these findings suggest that thalamic GMV reductions in METH users likely result from a convergence of neurotoxic, inflammatory, and genetic mechanisms. The observed left-lateralized effect may reflect functional hemispheric specialization, although subtle structural alterations in the right thalamus cannot be excluded and warrant further investigation.

The reduction of GMV observed in the thalamus may reflect functional insufficiency in this region, potentially contributing to the maladaptive behavioral patterns observed in MUDs. In the present study, MUDs with reduced thalamic GMV exhibited significantly higher MI scores. While most prior neuroimaging studies probing the neural basis of METH addiction related-impulse control in METH addiction have focused on the prefrontal cortex (PFC) and basal ganglia [8, 9], few have explored the association of thalamus with impulsive behavior, despite its central role in regulating cortico-subcortical communication. As a key relay station, the thalamus mediates bidirectional signal flow between cortical and subcortical regions [37] and serves as a critical node within the frontostriatal-thalamo-cortical circuit, which mediates goal-directed and automatic aspects of inhibitory control [7]. The inverse association between thalamic GMV and MI observed in our study suggests that thalamic atrophy may disrupt the dynamic interplay between top-down cognitive control (mediated by prefrontal circuits) and bottom-up reward-driven processes (modulated by striatal pathways) [38]. Specifically, neurodegeneration of the thalamus may impair its function as a regulatory hub within this circuit [39], potentially leading to two maladaptive consequences: diminished relay of inhibitory signals from the PFC to subcortical structures, weakening the ability to suppress impulsive responses, and heightened transmission of striatal reward signals via thalamocortical projections, reinforcing compulsive drug-seeking behaviors. Our findings are also consistent with recent functional neuroimaging studies reporting thalamic hyperactivity during cue-reactivity tasks in stimulant users [40], highlighting a possible dissociation between structural atrophy and functional hyperactivation. This paradox may reflect compensatory neural recruitment in early stages of addiction or reflect maladaptive plasticity resulting from chronic METH exposure. Future multimodal imaging studies integrating structural MRI with resting-state or task-based functional connectivity analyses are warranted to further elucidate how regional thalamic volume loss impacts network-level communication within impulse control circuits. Moreover, our mediation analysis revealed that MI significantly mediated the relationship between thalamic GMV and clinical indicators of METH use severity, underscoring the potential of impulsivity as both a behavioral marker and a therapeutic target. Together, these findings highlight thalamic structural deficits as a previously underrecognized neurobiological substrate of impulsive dysregulation in METH addiction and support the development of interventions targeting thalamocortical circuitry to mitigate disease progression.

The identification of 72 genes significantly associated with left thalamic GMV through gene-based burden testing provides novel insights into the molecular mechanisms underlying thalamic structural vulnerability in METH addiction. Gene set enrichment analysis further identified three biologically distinct pathways: Cytidine Analogs Cytotoxicity, Muscle Structure Development, and Nuclear Speck, each offering unique mechanistic hypotheses for the observed thalamic volume reduction. The strongest association with the Cytidine Analogs Cytotoxicity pathway suggests a convergence between the neurotoxic effects of METH and disruptions in nucleoside metabolism. Enrichment in this pathway implies that genetic variants influencing cellular responses to cytidine analogs may exacerbate METH-induced thalamic damage via shared mechanisms such as impaired DNA repair or mitochondrial dysfunction [41]. The unexpected enrichment of the Muscle Structure Development pathway may reflect the thalamic critical role in sensorimotor integration [42]. Genes in this pathway could be involved in thalamocortical axonal guidance or neuroglial interactions essential for maintaining structural stability within thalamic motor relay nuclei. Finally, the Nuclear Speck pathway, commonly associated with RNA splicing and transcriptional regulation [43], suggests that alterations in post-transcriptional control of neurodevelopmental genes may contribute to thalamic atrophy in the context of chronic METH exposure.

Functional annotation of common variants revealed significant enrichment in biological pathways primarily related to ciliary and microtubule-based processes. These findings implicate disruptions in intracellular transport and structural cytoskeletal organization as potential contributors to methamphetamine-induced thalamic vulnerability [44,45,46]. Notably, although no direct SNP-level overlap was observed between the variants examined here and loci reported in previous genome-wide association studies (GWAS) of METH or stimulant use disorders [47,48,49], such divergence is not unexpected given differences in phenotype definition, study design, ancestry composition, and statistical power between large-scale GWAS and the present imaging-genetics study. The thalamus is a critical relay center that integrates signals between cortical and subcortical regions involved in inhibitory control. Structural integrity within this region relies heavily on dynamic cytoskeletal remodeling, which governs axonal stability, vesicular trafficking, and synaptic plasticity [50, 51]. Variants in cytoskeletal genes such as DYNC2H1, KIF15, and TTN, which regulate dynein/kinesin motor function and axonemal architecture, may impair neurodevelopmental scaffolding or postnatal circuit maintenance, particularly in high-demand relay nuclei like the thalamus [51]. Despite the lack of direct genetic concordance at the single-variant level, our findings converge with prior GWAS at a broader biological scale, implicating pathways related to synaptic function, neurotransmission, and neuroplasticity, mechanisms repeatedly highlighted in genetic studies of stimulant dependence. Moreover, emerging evidence suggests that cytoskeletal abnormalities are directly linked to impulsivity-related phenotypes by altering synaptic connectivity and dendritic arborization within fronto-striatal-thalamic circuits [52, 53]. In parallel, disruption of extracellular matrix signaling may compromise neuron-glia communication and alter the plasticity of brain regions implicated in reward processing and cognitive control, such as the hippocampus, amygdala, and thalamus [54,55,56]. Together, our findings elucidate a multilevel mechanistic cascade in which genetically induced dysfunction in the cytoskeleton and extracellular matrix predispose individuals to structural deficits in the thalamus and behavioral disinhibition. This provides a biologically plausible pathway that links molecular risk factors to addiction-relevant phenotypes in MUD.

Our analyses of gene-environment interactions have demonstrated that genetic variation plays a significant role in modulating METH-related structural brain alterations. Specifically, individuals possessing the GG genotype at rs12729569 within the CR1L gene exhibited thalamic atrophy among METH users, in contrast to non-users and those with alternative genotypes. This suggests that this genetic locus may increase susceptibility to METH-induced neurotoxicity. Similarly, MUDs who carry the AG genotype at rs2282440 in the SDC3 gene displayed significantly reduced thalamic volume compared to those with the GG or AA genotypes, further emphasizing the influence of genotype-dependent vulnerability in this critical subcortical region [57]. Considering the thalamic essential role in sensorimotor gating and corticostriatal communication, these genotype-specific reductions in thalamic integrity may contribute to the dysregulation of frontostriatal-thalamic circuits frequently observed in MUD [7]. These findings highlight the necessity of integrating both genetic and environmental factors when modeling individual differences in addiction-related brain changes.

This study has several important limitations. First, the relatively small sample size for genetic analyses and cross-sectional design limit the ability to draw causal inferences regarding the relationships among drug use, thalamic atrophy, and the development of impulsivity. As such, the findings should be interpreted with caution. Second, the sample comprised exclusively male participants, predominantly recruited from a compulsory isolation and rehabilitation facility. Because the number of female residents in these facilities was very limited, recruitment of women was not feasible. Although the stringent exclusion criteria enhanced internal validity, they may also reduce the external validity of the findings and limit their generalizability to female individuals and more clinically heterogeneous MUD populations. Third, the study did not include thalamic subnuclei segmentation, thereby limiting the capacity to discern subregion-specific structural alterations that may differentially influence impulse control or the severity of addiction. Finally, physiological factors such as hydration status and circadian variation, which may subtly influence brain volume measures, were not directly assessed and should be considered in future studies.

This study integrates neuroanatomical, behavioral, and genetic evidences to identify thalamic structural deficits as a central feature of disrupted frontostriatal-thalamo-cortical circuitry and impulse control in MUD. Our findings reveal genotype-dependent susceptibility involving cytoskeletal and ciliary pathways, which may exacerbate METH-induced neurotoxicity. Although causality cannot be inferred due to the cross-sectional nature of the study, the results highlight the thalamus as a critical target for pharmacogenomic interventions. Future studies with larger samples, longitudinal designs, and multimodal approaches, incorporating detailed mapping of thalamic subnuclei and neurotransmitter imaging, are imperative to clarify the progression and functional implications of thalamic pathology in METH addiction. These insights lay the groundwork for precision medicine approaches and biomarker development in substance use disorders.