Methamphetamine use disorder (MUD) is a severe global health burden with a lifetime prevalence of over 0.7%. MUD could profoundly reduce life quality by disrupting both social and physical functions. Previous studies have found that MUD could lead to various abnormalities, including immunological activation and microglia abnormality, stress response, and metabolomic disruptions. A comprehensive understanding of these abnormalities is necessary for the intervention of MUD as well as its complications and provides the opportunity for the discovery of MUD biomarkers. Cross-sectional studies have suggested that such abnormalities could be detected in the peripheral blood of patients with MUD, which suggested that analyzing blood molecule profiles of MUD is beneficial for understanding MUD pathology. For example, the elevated levels of malondialdehyde in MUD were considered evidence of oxidation stress activation in methamphetamine (MA) exposure.

However, the current studies of MUD (as well as other substance abuse) blood molecule profiles are still at a simple and premature stage, with many critical questions left unexplored. First, substance dosage and withdrawal time could have a significant impact on the blood molecules, whereas only a few studies have considered this temporal pattern. This temporal pattern is specifically important for the development of MUD biomarkers, as early alterations may reflect addiction risk, while late alterations may reflect sequential impairs of other organs. Second, the molecular disruption of neuroactive substance involves multiple biological pathways, which may impact each other in a network structure. For example, the cardiac injury caused by MUD is mediated by various molecular mechanisms, including elevated oxygen consumption and neurotransmitter release, hyperoxidation, and inflammation. Thus, a cross-sectional study focusing on isolated cardiac injury markers may not fully capture the underlying mechanisms, which are central to the currently available results. Third, how peripheral signals reflect alterations in clinical assessment and the central nervous system is a valuable yet unanswered question. It has been shown that MUD is associated with multiple brain structural alterations such as periventricular, subcortical, and deep white matter lesions, but how these alterations linked with the blood molecular profile is unknown. In other fields of psychiatry, researchers have found that peripheral C-reactive protein was associated with the microstructure of the human brain. It is reasonable to infer that the blood molecule profile of MUD could also be linked to specific alterations in central nervous system.

Until now, research on molecular markers of MA addiction has been fragmented with inconsistent results. The mechanism by which these markers affect brain function and further impact MA prognosis remains unclear. In light of these caveats, we recruited a cohort of patients with MUD as well as healthy controls. After a systematic literature review of molecular markers related to addiction, we conducted a multimodal analysis that integrated the expression data of 52 blood molecules from nine categories (neuropeptides, oxidation, stress, vitamin, cytokine, neurotransmitter, cardiac injury markers, liver markers, and metabolite) (Fig. 2, Table S1 and Reference in Supplement), clinical assessments and characteristics (such as drug exposure dosage and withdrawal time), and brain magnetic resonance imaging (MRI) data. This study design allowed us to elucidate the following questions: (i) what is the overall molecular alterations of MUD? and (ii) how are these molecular alterations associated with other molecules and brain structural alterations? Therefore, this study hypothesized that peripheral molecular biomarkers of MA users influence the clinical phenotype through interactions with brain function, potentially offering insights for treatment and intervention.

Method

Cohort and ethical approval

This study was approved and supervised by the ethics committee of Shanghai Mental Health Center (approval number: 2016KY-49) and was in accordance with the Declaration of Helsinki. All participants provided signed written consent for participation.

Sample recruitment

We recruited 271 participants according to the following criteria:

1. A total of 231 patients with MUD who: (i) met the diagnosis criteria of severe MA use according to DSM-5; (ii) had ≥9 years of education; (iii) were aged 18 to 50 years; (iv) were not currently receiving any drug treatment; and (v) did not have a co-occurrence of a physical or axis-I disorder according to DSM-5 including depression, anxiety, or schizophrenia, or have serious medical illnesses that required pharmacological treatment or drug use disorders other than MA. Among these patients, we further classified them according to the withdrawal time (<3 months: acute group; >3 months and <6 months: medium group; >6 months: chronic group).

2. A total of 40 healthy controls who: (i) did not have any severe physical or neurological conditions; (ii) had no history of drug use; (iii) had no previous head injury; (iv) did not meet the diagnosis criteria of any psychiatric disorders according to DSM-5; and (v) had no family history of psychiatric disorders.

Data collection

The flowchart of the study is shown in Fig. 1. A self-administrated case report form included sociodemographic characteristics (e.g. age, education, marriage, weight, and height) and drug use history (e.g. age of onset, total duration of drug use, and dose) was administrated to each participant by one psychiatrist in a separate room. We also obtained the demographic characteristics of the controls.

Fig. 1

_{Flowchart of the study. Data from 231 methamphetamine users and 40 healthy controls were collected, including clinical assessments, peripheral blood metabolites, and magnetic resonance imaging (MRI) data.}

For a subset of patients, we performed a 1-year follow-up study and collected the same demographic, clinical assessment, and blood samples. Patients with follow-up data did not show significant differences in demographic characteristics (Tables S5 and S6).

Sample processing and ELISA

We collected fasting blood samples of each participant between 8 am and 9 am in 10-mL EDTA standard tubes and centrifuged for 15 min at 3000 rpm and 4°C. Fifty-two molecules were selected by a comprehensive literature review on biomarkers related to addiction. Levels of each blood marker were measured via sandwich enzyme-linked immunosorbent assay (ELISA) by a commercially available kit (Beijing Rongxin Zhihe Biotechnology Co. Ltd., Beijing, China). The ELISA was performed separately for each biomarker. All assays were performed in duplicate and expressed as picogram per milliliter. The detection range of this assay was 20 to 4000 pg/mL. The intra-assay and interassay coefficients were <5% and <10%, respectively. All procedures were performed by one technician who was blinded to the sample group to minimize technical variance.

Global analysis of metabolites

We scaled expression levels of each molecule of all samples into zero-mean and SD equal one. We applied linear discriminant analysis on all baseline samples by lda function in the MASS 7.3 R package. The samples were labeled as “MA” and “Control.” We first applied linear discriminant analysis (LDA) using all 52 molecules, then using molecules in each of the categories (such as cytokine and neurotransmitter) (Table S1). The discriminant accuracy was evaluated by the area under the curve (AUC) calculated by the ROCR 1.0 R package. Within each patient group, we used linear regression to evaluate the relationship between linear discriminant (LD) score and substance withdrawal time, controlling for age, sex, and education. For a subset of patients with follow-up data, we calculated the LD score at follow-up using the LD loadings in the baseline sample and used a paired t test to evaluate the difference between baseline and follow-up.

Association test of each molecule

We applied linear mixed regression (the sample recruitment site was denoted as a random effect) using the lmer function from lme4 1.1 R package to test the association between the expression level of each molecule and the following variables:

1. Demographic variables, including age, sex, and education. No further covariate was included in these analyses.

2. Clinical assessments: we assessed clinical impulsiveness by the Barratt Impulsiveness Scale (BIS-11), anxiety symptoms by the General Anxiety Disorder-7 (GAD-7), general health conditions by the Patient Health Questionnaire-9 (PHQ-9),

   sleep quality by the Pittsburgh Sleep Quality Index (PSQI), nicotine usage by the Fagerstrom Test for Nicotine Dependence (FTND), and withdrawal time. When calculating the association between each molecule and these assessments, all demographic variables were included as covariates.

3. Substance abuse status, including all MA users versus controls, and acute MA users versus controls. All demographic variables were included as covariates.

We set the significance threshold as the P value of coefficient <0.05/52.

Brain MRI procedure

All brain imaging for the present study was performed on a Siemens Tim Trio 3T scanner (Erlangen, Germany). High-resolution T1-weighted anatomical scans were acquired using a magnetization-prepared rapid gradient-echo sequence (repetition time = 2300 ms, TE = 3 ms, TI = 1000 ms, flip angle = 9°, and voxel size = 1.0 × 1.0 × 1.0 mm³). The voxel-wise cortical thickness (CT) of each participant were estimated using the diffeomorphic registration-based CT (DiReCT) method implemented in Advanced Normalization Tools (ANTs) (http://stnava.github.io/ANTs/), which exploits tissue probabilistic maps to identify a maximum likelihood correspondence between the white matter (WM) surface and the outer gray matter (GM) surface. DiReCT is a reliable volume-based technique for estimating the CT of both human and nonhuman primates, which yields similar results to using surface-based algorithms. In addition, GM volume (GMV) and WM volume (WMV) were obtained using the VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm/) executed in MATLAB (Mathworks, Sherborn, Massachusetts). All images were visually and automatically checked by using covariance analysis on the sample homogeneity of segmented GM images. T1-weighted images were corrected for field inhomogeneities and registered using a DARTEL (diffeomorphic anatomical registration through exponentiated lie algebra) template. Spatially normalized images were then tissue-segmented into GM, WM, and cerebrospinal fluid and modulated for different tissue segments to preserve the regional volumetric information of a particular tissue within a voxel. All images were then smoothed with an isotropic Gaussian kernel (8 mm full width at half maximum). The GMV, WMV, and CT of each brain region were then extracted using Automated Anatomical Labelling Atlas 3 (AAL3), which included the nucleus accumbens. Finally, linear mixed regression was performed to investigate the association between each molecule-brain region pairs, with age, sex, and comorbidity as covariates.

Result

Characteristics of patients with MUD and the control group

We recruited 231 patients with MUD with different withdrawal times and 40 healthy controls. The clinical demographic characteristics are presented in Table 1. There were significant differences in age and sex (P < 0.001) between patients and controls, which were adjusted in the following analyses. Among different patient subgroups, the only demographic difference is that the acute group had a significantly higher age. Other demographic characteristics, including body mass index, smoking, and drinking behaviors, and somatic comorbidities showed no significant groupwise difference.

Table 1. Characteristics of participants

Global analysis of metabolite expression

We quantified the blood expression levels of 52 molecules related to different biological functions (Table S1). To gain an overview of the molecular alteration, we applied LDA to construct linear combinations of tested molecules that could maximally capture differences among sample groups. The first combination LD1 could distinguish patients with MUD from controls (AUC = 0.95, sensitivity = 0.80, specificity = 0.75, McNemar test P = 8.30 × 10⁻⁸) (Figs 2a and S1a–c). We separately ran LDA using only cytokines, oxidation markers, neuropeptides, or stress response markers (Fig. S1c–f), and they were unable to fully distinguish patients (AUC <0.75), which suggested that MA caused wide-range disturbances not limited to a single system. LD2 showed no association with patient groups or withdrawal time (P > 0.05).

Fig. 2

_{Overview of blood molecular alteration in MUD. (a) LDA of all samples at baseline. LD1 and LD2: linear discriminant score. Color represents patient group; line represents contour line corresponding to the sample density within each group. (b) Each dot represents an MA user. The x-axis represents the withdrawal time at baseline and the y-axis represents the LD1 score. (c) Boxplot of LD1 score of MA users at baseline (left) and after treatment (middle) and healthy controls (right). Red lines link the baseline and follow-up data of the same patient. (d) Left: heatmap showing the relationship between each molecule and the withdrawal time as well as patient groups. Middle: heatmap showing the contribution of each molecule to LD. The full names of all tested molecules can be found in Table S1. The LD1 score was rescaled by Z-score. MA, methamphetamine; MUD, methamphetamine use disorder; LD, linear discriminant; LDA, linear discriminant analysis.}

We then tested the association between LD combinations and clinical characteristics. We found that LD1 was significantly associated with MA withdrawal time (longer withdraw time was linked to higher LD1 score and was closer to healthy controls, regression P = 0.0008) (Fig. 2b), suggesting that the global metabolomic disturbance could partially be recovered after the termination of MA exposure. The acute, mid, and chronic groups of patients had significantly lower LD1 than controls, where the acute group had the lowest (t test P = 1.07 × 10⁻¹², Fig. S2). We further performed a longitudinal analysis on a subset of patients with MUD (n = 59) and, as expected, found that LD1 scores recovered close to control after 1-year follow-up (paired t test P = 2.20 × 10⁻⁵) (Fig. 2c). In addition, we found that a lower LD1 score was nominally associated with higher PHQ-9 and GAD-7 scores (P = 0.01 and 0.03, respectively).

Substance abuse–activated stress response, immune response, and hyperoxidation

Having described the global metabolomic disturbance, we further analyzed the disturbance on each molecule. As shown in Fig. 2d and Table S2, the expression level of 26 molecules was significantly altered between controls and patients with MUD. The top molecules included soluble intercellular adhesion molecule 1 (P = 3.20 × 10⁻⁶ downregulated in patients), folic acid (P = 4.59 × 10⁻¹², downregulated in patients), as well as interleukin (IL) 6 (P = 4.40 × 10⁻⁶, upregulated in patients) and myoglobin (P = 3.51 × 10⁻⁹, upregulated in patients).

Taken together, 100% of myocardial injury markers (6/6), vitamins (3/3), and neurotransmitters (3/3) were altered by substance abuse, followed by oxidation markers (6/7) and stress markers (5/8). The direction of these alterations consistently suggested the activation of: (i) stress response (upregulation of stress markers and hormones), (ii) immune response (upregulation of proinflammatory markers such as C-reactive protein, IL-6, and tumor necrosis factor, alongside downregulation of the anti-inflammatory cytokine IL-10), and (iii) hyperoxidation (indicated by the upregulation of oxidation products and downregulation of antioxidant enzymes).

We further analyzed the potential heterogeneity within the MUD group. We compared the acute MUD group with the controls and found that 24 of 26 molecules disturbed in the MUD group were disturbed in the acute MUD group in the same direction (Fig. 2d). Exceptions were vitamin B12 and norepinephrine. On the other hand, brain natriuretic peptides (P_acute = 0.0005), ghrelin (P_acute = 0.0005), and melatonin (P_acute = 0.0002) were significantly altered in the acute MA group but not in the overall MUD group. Consistently, the moderate and chronic MUD groups showed similar but less significant disturbance: all 26 significant molecules had the same direction of disturbance in both the moderate and chronic groups, but only 24 and 19 reached a significance threshold, respectively (Table S2). Generally speaking, although withdrawal time was significantly associated with 14 molecules (Fig. 2d), the differences were far less pronounced than the differences between MA users and healthy controls. We illustrate several typical examples (glutathione reductase, epinephrine, gamma-aminobutyric acid [GABA], and tumor necrosis factor) in Figure S2.

Last, we analyzed whether the cumulative dosage of MA usage has an impact on blood molecules. As shown in Table S2, we only found one suggestive negative association between MA dosage and epinephrine (P = 0.005).

5-HIAA, melatonin, and SOD reflected clinical symptoms

As expected, patients with MUD had significantly higher BIS, PHQ-9, GAD-7, and FTND scores than controls. These scores partially alleviated as withdrawal time prolonged, but did not fully return to normal (linear regression slope >0, P < 0.001, Fig. S3). We further analyzed whether such symptoms were associated with peripheral molecular disruption. As shown in Figure 3, the blood level of 5-hydroxyindoleacetic acid (5-HIAA) had a significant reverse association with PHQ-9 (P = 0.0001, Fig. 3a) and GAD-7 (P = 0.0001, Fig. 3b). Melatonin was also reversely associated with two assessments: GAD-7 (P = 0.0003, Fig. 3c) and BIS (P = 0.0009, Fig. 3d). Last, superoxide dismutase (SOD) was associated with GAD-7 (P = 0.0003, Fig. 3e) and vitamin B 12 was associated with PHQ-9 (P = 0.0006, Fig. 3f). In summary, the expression level of peripheral molecules is associated with clinical symptoms in MUD.

Fig. 3

_{Associations between peripheral molecules and clinical assessment in patients with MUD. Each point represents a patient, and the x- and y-axes represent the molecule levels and clinical assessments. BIS, Barratt Impulsiveness Scale; FDR, false discovery rate; GAD-7, General Anxiety Disorder-7; MUD, methamphetamine use disorder; PHQ-9, Patient Health Questionnaire-9. Color dots corresponded to sample groups as shown by the x-axis.}

Strong association between BDNF and brain structures of patients with MUD

Since we have identified the association between blood molecules and symptoms, we then asked whether they were also associated with brain structure. In a subset of patients with MUD (n = 127, including 20, 44, and 63 from the acute, mid, and chronic groups, respectively), there was no significant difference between patients with and without MRI data (Table S6). We conducted MRI and associated MRI-derived features with withdrawal time and blood molecules. As shown in Fig. 4a,b and Table S3, after P value adjustment, the CT of bilateral calcarine sulcus as well as right putamen were significantly different among MUD subgroups with different withdrawal times (false discovery rate [FDR] <0.05). We also found nine significant associations between blood molecules and brain structures (Table S4). The brain-derived neurotrophic factor (BDNF) was significantly associated with the GMV of the right fusiform (FDR = 0.009, Fig. 4c), subregions in the right cerebellum (FDR = 0.02) and vermis (FDR = 0.04), left superior temporal pole (P_FDR = 0.03), and right caudate (P_FDR = 0.04, Fig. 4e,f). Significant associations were found between the brain volumes of the orbitofrontal cortex and GABA subunit A5 (FDR = 0.01, Fig. 4d) and orexin (FDR = 0.03).

Fig. 4

_{Associations between peripheral molecules and brain morphology in patients with MUD. (a,b) Group difference of cingulate gyrus volume. (c,d) Association between blood molecules and brain structure. (e) Structure of caudate. (f) Association between blood brain-derived neurotrophic factor and caudate volume. BDNF, brain-derived neurotrophic factor; GABA, gamma-aminobutyric acid; GMV, gray matter volume; MUD, methamphetamine use disorder; OFC, orbitofrontal cortex. Color dots corresponded to sample groups as shown by the x-axis.}

Discussion

In the current study, we systematically screened the blood molecular alterations associated with MUD and analyzed their association with withdrawal time, clinical characteristics, and brain structural alteration. We found widespread activation of inflammation and oxidative stress in patients with MUD and highlighted molecules such as BDNF and melatonin, which mediated other molecular or brain structural alterations.

We used the LDA method to distinguish MUD from healthy controls. Generally speaking, a model with an AUC exceeding 0.8 is considered to have good discriminative ability. In this study, the AUC was 0.95, indicating that the model had good discriminative ability and could distinguish between two groups. Specificity refers to the proportion of samples that are actually controls and correctly identified as controls. The value of 0.75, although not very high, is still acceptable in many studies. Sensitivity (also known as true positive rate) refers to the proportion of samples that are actually patients and correctly identified as patients. The value of 0.8 indicates that the model could correctly identify 80% of patients with MUD. This is also a relatively high value.

The inflammatory and oxidative activation caused by MUD, such as IL-10, IL-9, glutathione peroxidase, and melatonin, is consistent with the existing knowledge of MUD complications. Previous studies have found that MA could alter a wide variety of immune cells including T and B, dendritic, and natural killer cells, and could promote the polarization of macrophage. MA could also trigger hyperoxidation via neurotransmitter releases, which, in turn, lead to demyelination, neuron damage, and cardiac injury. The neurotoxicity of MA on neuron and glia cells is found to be mediated by redox-sensitivity transcription factors such as nuclear factor E2–related factor 2, which further highlights the role of hyperoxidation in MA complication. Similarly, alcohol also impairs the oxygen reduction system, in particular NADH and the cytoP450 system, which disrupts mitochondria functions and eventually leads to liver injury. Taken together, both our results and previous findings suggest that inflammation and oxidative stress are the key processes in MUD complications, which may be ideal therapeutic targets for the treatment of these complications. It is recommended that antioxidants be included in drug regimens prescribed for MA abusers who are referred to physicians to seek medical care for any reason.

In this study, we found that patients with MUD have higher levels of anxiety symptoms, depression symptoms, impulsivity, and nicotine use, which is not difficult to understand. In addition, we also found a correlation between peripheral blood indicators and these behaviors. 5-HIAA (5-hydroxyindoleacetic acid) is a metabolite of 5-hydroxytryptamine (5-HT) and belongs to the serotonin system. Its changes often occur in pathological processes such as anxiety, depression, and stress. A decrease in 5-HIAA has also been found in animal studies of MA administration, indicating that changes in 5-HIAA may suggest anxiety and depression symptoms in MA users. Melatonin is an indoleamine produced in the pineal gland and released into the blood every night. In addition to its early discovered effects on sleep, it has also been proven in animal and clinical experiments to have antianxiety effects and participate in addiction. This study demonstrated that melatonin was associated with anxiety symptoms in patients with MUD, and there were studies suggesting its association with impulsivity, which is consistent with our results. SOD is an important component of the antioxidant enzyme system, and there was a significant decrease in anxiety disorders and stress. A previous study also showed that SOD2 genetic variants could predict GMV reduction in chronic alcohol users, suggesting that it may also be involved in the development of MA addiction. The relationship between vitamin B12 and depression has been confirmed by numerous studies, and MA users have been shown to have lower vitamin B12 levels than healthy controls. Therefore, it is supposed that early supplementation of vitamin B could improve depressive symptoms in patients with MUD.

A reduction in CT was observed in the bilateral calcarine sulcus in patients undergoing drug withdrawal. To date, we have found only one paper on heroin-dependent patients that showed similar results in reduced CT (left calcarine sulcus). We speculate that long-term use of MA would have irreversible effects on the calcarine sulcus and gradually worsen over time, indicating that the damage caused by MA is persistent. Our previous study also found significantly decreased GMV of MA users during the 6-month withdrawal period in the precontral gyrus, caudate, fusiform, and cerebellum. At 12 months, the cerebellum had recovered, but the cingulate gyrus showed continued reduction. Moreover, the decrease in GMV correlated with the cumulative use of MA, which is consistent with the results of this paper.

Another interesting finding of the current study is that blood molecule levels of patients with MUD were associated with their structural alterations of the central nervous system. One notable result is the significant association between BDNF and GMV of nine brain regions. BDNF is a critical neurotrophin that plays a pivotal role in neuronal survival, growth, and differentiation. Reduced BDNF levels have previously been implicated in the pathophysiology of various psychiatric and neurodegenerative disorders and MA early withdrawal. However, reports on the changes in BDNF levels over time during MA withdrawal are not particularly consistent; for example, one study showed that BDNF levels were higher during the initial stages of withdrawal, but subsequently declined after 1 month. In the context of our study, we did not find significant differences between BDNF and healthy individuals, which is inconsistent with previous research results. This may be attributable to differences in sample size and withdrawal time, but a significant correlation was established between BDNF concentrations and the GMV across nine specific brain regions, including the fusiform gyrus, orbitofrontal cortex, temporal pole, caudate, cerebellum crus, and vermis. This might suggest that alterations in BDNF levels, possibly as a result of chronic MA use, could either be a consequence of, or a contributing factor to, the observed volumetric changes in these brain regions. Numerous studies have been conducted on the effects of BDNF on brain function within the context of substance use disorders, mainly focusing on the prefrontal-striatal circuit, as well as the cerebellum. BDNF, its genes, mRNA, and metabolites are involved in the pathogenesis of addiction. Given that BDNF is involved in neuroplasticity and can modulate synaptic strength and structure, it is plausible to infer that the neurotoxic effects of MA on the brain might be modulated by BDNF. Furthermore, these findings align with previous research that has shown MA to decrease BDNF expression in specific brain regions, leading to a potential avenue for therapeutic intervention targeting BDNF signaling pathways.

Moreover, blood levels of GABA were also found to correlate with orbital frontal cortex. Previous studies have also found that both alcohol addiction and cocaine addiction could cause changes in the orbital frontal cortex GABA system, manifested as an increase in GABA(A) subunit mRNA levels. The use of GABA receptor antagonists could reduce addictive behavior, indicating that GABA receptors are potential effective targets for drug intervention in addiction.

There are several limitations that should be noted in this study. First, the number of participants (patients and controls) with longitudinal data or brain MRI data is still relatively small, and the sample size disparity between each group is also relatively large. We also failed to collect MRI data from healthy individuals, but we found that statistical power was still sufficient after calculation. In addition, the extent to which peripheral blood indicators could reflect the changes of brain structure and function is an old-fashioned problem, so we need to be cautious in making conclusions. Furthermore, since individuals in the preaddiction phase rarely seek medical support, our cohort recruited only patients who already exhibited significant symptoms. Each patient's stage of addiction is not identical to that of others. Therefore, the temporal profiles of blood molecules at different stages still need to be analyzed, and the explanation of causality by state indicators remains to be discussed.

In conclusion, we found that MUD was associated with widespread peripheral molecular disruption, which was partially recoverable after withdrawal but could not fully recover to healthy status. Major disruption was associated with inflammatory and oxidative activation as well as stress response. The disruption of melatonin, SOD, and BDNF was further linked to clinical and brain structural abnormalities. These results could guide future studies aimed at identifying biomarkers for MUD and therapy targets for MUD complications.

Methamphetamine Use Disorder and Molecular Changes

Methamphetamine use disorder (MUD) represents a significant global health concern. It can significantly reduce a person's quality of life by affecting both social interactions and physical well-being. Prior research has identified various issues linked to MUD, including increased immune system activity, abnormalities in brain immune cells (microglia), heightened stress responses, and disruptions in metabolism. A thorough understanding of these changes is essential for developing effective treatments for MUD and its related problems, and for identifying biological markers (biomarkers) of the disorder. Studies that examine data from a single point in time suggest that these changes can be detected in the blood of individuals with MUD, indicating that analyzing blood molecule profiles is valuable for understanding the disease process. For instance, higher levels of malondialdehyde in MUD have been seen as evidence of increased oxidative stress due to methamphetamine exposure.

However, current research on blood molecule profiles in MUD, and other substance use disorders, remains basic and incomplete, with many key questions still unaddressed. First, the amount of substance used and the time since last use (withdrawal time) can greatly affect blood molecules. Few studies have considered these time-related patterns. Such patterns are crucial for developing MUD biomarkers, as early changes might signal addiction risk, while later changes could indicate damage to other organs. Second, the molecular changes caused by psychoactive substances involve many biological pathways that may interact as a network. For example, heart damage from MUD involves several molecular processes, including increased oxygen use, neurotransmitter release, excessive oxidation, and inflammation. Therefore, a study that only focuses on specific heart injury markers may not fully capture the complex underlying mechanisms. Third, how signals from the bloodstream reflect changes in clinical assessments and the central nervous system is an important, yet unanswered question. MUD has been linked to various brain structural changes, such as lesions in areas near the ventricles, deep within the brain, and in the white matter, but the connection between these changes and blood molecule profiles is unknown. In other areas of psychiatry, researchers have found that C-reactive protein in the blood is associated with brain microstructure. It is reasonable to suggest that blood molecule profiles in MUD could also be linked to specific changes in the central nervous system.

Previously, research on molecular markers for methamphetamine addiction has been inconsistent and disconnected. The precise ways these markers affect brain function and influence the outcome of methamphetamine use remain unclear. To address these gaps, a group of patients with MUD and healthy individuals were recruited for this study. Following a thorough review of existing literature on addiction-related molecular markers, a comprehensive analysis was performed. This analysis combined data from 52 blood molecules across nine categories (including neuropeptides, oxidation markers, stress indicators, vitamins, cytokines, neurotransmitters, heart and liver injury markers, and metabolites), along with clinical assessments and individual characteristics (such as drug exposure and withdrawal time), and brain magnetic resonance imaging (MRI) data. This study design aimed to clarify two main questions: (i) what are the overall molecular changes associated with MUD? and (ii) how are these molecular changes connected to other molecules and brain structural alterations? Therefore, this study proposed that peripheral molecular biomarkers in methamphetamine users affect their clinical presentation through interactions with brain function, potentially offering valuable information for treatment and intervention strategies.

Method

Cohort and ethical approval

This study received approval and oversight from the ethics committee of Shanghai Mental Health Center (approval number: 2016KY-49) and followed the principles of the Declaration of Helsinki. All participants provided written informed consent before participating.

Sample recruitment

A total of 271 participants were recruited based on specific criteria. This included 231 patients diagnosed with severe methamphetamine use disorder according to DSM-5 criteria. These patients also met criteria for at least nine years of education, were between 18 and 50 years old, were not currently receiving drug treatment, and did not have co-occurring physical or Axis-I psychiatric disorders (such as depression, anxiety, or schizophrenia) or other serious medical illnesses requiring treatment, or other drug use disorders beyond methamphetamine. Patients were further categorized into acute (less than 3 months), medium (3 to 6 months), and chronic (more than 6 months) withdrawal groups based on the time since last use.

The study also recruited 40 healthy control individuals. These controls had no severe physical or neurological conditions, no history of drug use or previous head injury, no diagnosis of any psychiatric disorders according to DSM-5, and no family history of psychiatric disorders.

Data collection

A flowchart illustrating the study design is presented in Fig. 1. A self-administered case report form, which included demographic details (e.g., age, education, marital status, weight, and height) and drug use history (e.g., age of onset, total duration of drug use, and dosage), was given to each participant by a psychiatrist in a private room. Demographic information for the control group was also collected.

Fig. 1

Flowchart of the study. Data from 231 methamphetamine users and 40 healthy controls were collected, including clinical assessments, peripheral blood metabolites, and magnetic resonance imaging (MRI) data.

For a portion of the patients, a one-year follow-up study was conducted, and the same demographic, clinical assessment, and blood sample data were collected again. Patients who participated in the follow-up did not show significant differences in demographic characteristics compared to those who did not (Tables S5 and S6).

Sample processing and ELISA

Fasting blood samples were collected from each participant between 8 am and 9 am using 10-mL EDTA tubes. Samples were then centrifuged for 15 minutes at 3000 revolutions per minute and 4°C. Fifty-two molecules, chosen based on a comprehensive literature review of addiction-related biomarkers, had their levels measured. The concentration of each blood marker was determined using a commercially available sandwich enzyme-linked immunosorbent assay (ELISA) kit. Each ELISA was performed individually for each biomarker. All tests were conducted in duplicate, and results were expressed in picograms per milliliter. The detection range for the assay was 20 to 4000 pg/mL. The variations within a single assay (intra-assay) and between different assays (interassay) were less than 5% and 10%, respectively. To minimize technical variability, all procedures were carried out by a single technician who was unaware of the sample grouping.

Global analysis of metabolites

The expression levels of each molecule from all samples were standardized to have a mean of zero and a standard deviation of one. Linear discriminant analysis (LDA) was applied to all baseline samples using the lda function in the MASS 7.3 R package. Samples were labeled as either “MA” (methamphetamine users) or “Control.” LDA was first performed using all 52 molecules, then separately for molecules within each category (such as cytokines and neurotransmitters) (Table S1). The accuracy of discrimination was assessed by calculating the area under the curve (AUC) using the ROCR 1.0 R package. Within each patient group, linear regression was used to evaluate the relationship between the linear discriminant (LD) score and substance withdrawal time, adjusting for age, sex, and education. For patients with follow-up data, the LD score at follow-up was calculated using the LD loadings from the baseline sample, and a paired t test was used to evaluate changes between baseline and follow-up.

Association test of each molecule

Linear mixed regression was applied, using the lmer function from the lme4 1.1 R package, to investigate the association between the expression level of each molecule and various factors. The sample recruitment site was included as a random effect in these analyses. The tested variables included:

1. Demographic variables, such as age, sex, and education. No additional covariates were included in these specific analyses.

2. Clinical assessments, including impulsiveness (measured by the Barratt Impulsiveness Scale [BIS-11]), anxiety symptoms (General Anxiety Disorder-7 [GAD-7]), general health (Patient Health Questionnaire-9 [PHQ-9]), sleep quality (Pittsburgh Sleep Quality Index [PSQI]), nicotine use (Fagerstrom Test for Nicotine Dependence [FTND]), and withdrawal time. When examining the association between each molecule and these assessments, all demographic variables were included as covariates.

3. Substance use status, specifically comparing all methamphetamine users versus controls, and acute methamphetamine users versus controls. All demographic variables were included as covariates in these comparisons.

The threshold for statistical significance was set at a P value of less than 0.05/52.

Brain MRI procedure

All brain imaging for this study was conducted using a Siemens Tim Trio 3T scanner. High-resolution T1-weighted anatomical scans were acquired using a magnetization-prepared rapid gradient-echo sequence with specific parameters (repetition time = 2300 ms, TE = 3 ms, TI = 1000 ms, flip angle = 9°, and voxel size = 1.0 × 1.0 × 1.0 mm3). Voxel-wise cortical thickness (CT) for each participant was estimated using the diffeomorphic registration-based CT (DiReCT) method, which is part of the Advanced Normalization Tools (ANTs). This method uses tissue probability maps to find the most likely match between the white matter surface and the outer gray matter surface. DiReCT is a reliable volume-based method for estimating CT in both human and nonhuman primates, yielding results similar to surface-based algorithms.

Additionally, gray matter volume (GMV) and white matter volume (WMV) were obtained using the VBM8 toolbox, operated within MATLAB. All images underwent visual and automatic quality checks using covariance analysis to ensure homogeneity of segmented gray matter images. T1-weighted images were corrected for field inhomogeneities and registered using a DARTEL template. Spatially normalized images were then segmented into gray matter, white matter, and cerebrospinal fluid, and adjusted to preserve regional volumetric information within each voxel. All images were then smoothed with an isotropic Gaussian kernel (8 mm full width at half maximum). The GMV, WMV, and CT of each brain region were then extracted using the Automated Anatomical Labelling Atlas 3 (AAL3), which includes the nucleus accumbens. Finally, linear mixed regression was performed to investigate the association between each molecule and specific brain regions, with age, sex, and co-existing medical conditions included as covariates.

Result

Characteristics of patients with MUD and the control group

The study included 231 patients with methamphetamine use disorder (MUD) exhibiting various withdrawal times, and 40 healthy control individuals. The demographic characteristics of these groups are detailed in Table 1. Significant differences in age and sex were observed between patients and controls (P < 0.001); these differences were statistically adjusted for in subsequent analyses. Among the different patient subgroups, the only demographic distinction was that the acute withdrawal group was significantly older. Other demographic factors, such as body mass index, smoking, drinking habits, and co-existing physical illnesses, showed no significant differences between the groups.

**Table 1. **Characteristics of participants

Global analysis of metabolite expression

Blood expression levels for 52 molecules, linked to various biological functions, were quantified (Table S1). To gain a broad understanding of the molecular changes, linear discriminant analysis (LDA) was applied to create linear combinations of the tested molecules that best highlighted differences among sample groups. The first combination, Linear Discriminant 1 (LD1), successfully distinguished patients with MUD from healthy controls, showing high discriminative ability (AUC = 0.95, sensitivity = 0.80, specificity = 0.75, McNemar test P = 8.30 × 10−8) (Figs 2a and S1a–c). When LDA was run separately using only specific categories like cytokines, oxidation markers, neuropeptides, or stress response markers, these individual categories were less effective at distinguishing patients (AUC <0.75). This suggested that methamphetamine use leads to widespread disturbances affecting multiple bodily systems, not just a single one. The second linear discriminant (LD2) showed no association with patient groups or withdrawal time (P > 0.05).

The association between LD combinations and clinical characteristics was then examined. LD1 was significantly associated with methamphetamine withdrawal time; longer withdrawal periods were linked to higher LD1 scores, bringing them closer to those of healthy controls (regression P = 0.0008) (Fig. 2b). This suggests that the overall metabolic disruption can partially recover after methamphetamine use stops. Patients in the acute, medium, and chronic withdrawal groups all showed significantly lower LD1 scores than controls, with the acute group having the lowest scores (t test P = 1.07 × 10−12, Fig. S2). A longitudinal analysis of a subset of patients with MUD (n = 59) further demonstrated that LD1 scores moved closer to control levels after a one-year follow-up period (paired t test P = 2.20 × 10−5) (Fig. 2c). Additionally, a lower LD1 score was found to be nominally associated with higher scores on the PHQ-9 and GAD-7, indicating more severe depression and anxiety symptoms (P = 0.01 and 0.03, respectively).

Fig. 2

Overview of blood molecular alteration in MUD. (a) LDA of all samples at baseline. LD1 and LD2: linear discriminant score. Color represents patient group; line represents contour line corresponding to the sample density within each group. (b) Each dot represents an MA user. The x-axis represents the withdrawal time at baseline and the y-axis represents the LD1 score. (c) Boxplot of LD1 score of MA users at baseline (left) and after treatment (middle) and healthy controls (right). Red lines link the baseline and follow-up data of the same patient. (d) Left: heatmap showing the relationship between each molecule and the withdrawal time as well as patient groups. Middle: heatmap showing the contribution of each molecule to LD. The full names of all tested molecules can be found in Table S1. The LD1 score was rescaled by Z-score. MA, methamphetamine; MUD, methamphetamine use disorder; LD, linear discriminant; LDA, linear discriminant analysis.

Substance use–activated stress response, immune response, and hyperoxidation

Following the global analysis of metabolic disruption, specific changes in individual molecules were further analyzed. As presented in Fig. 2d and Table S2, the expression levels of 26 molecules were significantly different between control individuals and patients with MUD. Key molecules showing alterations included soluble intercellular adhesion molecule 1 (downregulated in patients, P = 3.20 × 10−6), folic acid (downregulated in patients, P = 4.59 × 10−12), interleukin (IL) 6 (upregulated in patients, P = 4.40 × 10−6), and myoglobin (upregulated in patients, P = 3.51 × 10−9).

In total, all myocardial injury markers (6 out of 6), vitamins (3 out of 3), and neurotransmitters (3 out of 3) were altered by substance use. This was followed by oxidation markers (6 out of 7) and stress markers (5 out of 8). The consistent direction of these changes pointed to the activation of three main processes: (i) stress response, indicated by increased stress markers and hormones; (ii) immune response, characterized by higher levels of pro-inflammatory markers like C-reactive protein, IL-6, and tumor necrosis factor, along with reduced levels of the anti-inflammatory cytokine IL-10; and (iii) hyperoxidation, evidenced by increased oxidation products and decreased antioxidant enzymes.

Further analysis explored variations within the MUD group. When comparing the acute MUD group to controls, 24 of the 26 molecules found to be altered in the overall MUD group showed changes in the same direction in the acute group (Fig. 2d). Exceptions were vitamin B12 and norepinephrine. Additionally, brain natriuretic peptides (Pacute = 0.0005), ghrelin (Pacute = 0.0005), and melatonin (Pacute = 0.0002) were significantly altered in the acute MUD group but not in the overall MUD group. Consistently, both the moderate and chronic MUD groups showed similar, but less significant, disturbances; all 26 significant molecules were altered in the same direction in both groups, but only 24 and 19, respectively, met the significance threshold (Table S2). While withdrawal time was significantly associated with 14 molecules (Fig. 2d), these differences were generally less pronounced than the differences between methamphetamine users and healthy controls. Several typical examples, such as glutathione reductase, epinephrine, gamma-aminobutyric acid (GABA), and tumor necrosis factor, are illustrated in Figure S2. Lastly, the impact of cumulative methamphetamine dosage on blood molecules was investigated. Only one suggestive negative association was found between methamphetamine dosage and epinephrine (P = 0.005) (Table S2).

5-HIAA, melatonin, and SOD reflected clinical symptoms

As anticipated, patients with MUD exhibited significantly higher scores on measures of impulsiveness (BIS), depression (PHQ-9), anxiety (GAD-7), and nicotine dependence (FTND) compared to control individuals. These scores partially improved as the withdrawal period lengthened but did not fully return to normal levels (linear regression slope >0, P < 0.001, Fig. S3). The potential association between these clinical symptoms and peripheral molecular disruptions was further analyzed. As shown in Figure 3, blood levels of 5-hydroxyindoleacetic acid (5-HIAA) displayed a significant inverse relationship with both PHQ-9 (P = 0.0001, Fig. 3a) and GAD-7 scores (P = 0.0001, Fig. 3b). Melatonin also showed an inverse association with two assessments: GAD-7 (P = 0.0003, Fig. 3c) and BIS (P = 0.0009, Fig. 3d). Furthermore, superoxide dismutase (SOD) was associated with GAD-7 scores (P = 0.0003, Fig. 3e), and vitamin B12 was linked to PHQ-9 scores (P = 0.0006, Fig. 3f). In summary, the expression levels of certain molecules in the blood are associated with clinical symptoms in individuals with MUD.

Fig. 3

Associations between peripheral molecules and clinical assessment in patients with MUD. Each point represents a patient, and the x- and y-axes represent the molecule levels and clinical assessments. BIS, Barratt Impulsiveness Scale; FDR, false discovery rate; GAD-7, General Anxiety Disorder-7; MUD, methamphetamine use disorder; PHQ-9, Patient Health Questionnaire-9. Color dots corresponded to sample groups as shown by the x-axis.

Strong association between BDNF and brain structures of patients with MUD

After identifying associations between blood molecules and symptoms, the potential link between these molecules and brain structure was explored. In a subset of patients with MUD (n = 127, comprising 20 from the acute group, 44 from the medium group, and 63 from the chronic group), no significant differences were observed between patients with and without MRI data (Table S6). MRI data were collected, and MRI-derived features were associated with withdrawal time and blood molecules. As shown in Fig. 4a,b and Table S3, after adjusting for statistical significance (P value adjustment), the cortical thickness (CT) of both the bilateral calcarine sulcus and the right putamen significantly differed among MUD subgroups based on withdrawal time (false discovery rate [FDR] <0.05).

Nine significant associations between blood molecules and brain structures were also found (Table S4). Brain-derived neurotrophic factor (BDNF) was significantly associated with the gray matter volume (GMV) of the right fusiform (FDR = 0.009, Fig. 4c), subregions in the right cerebellum (FDR = 0.02) and vermis (FDR = 0.04), left superior temporal pole (PFDR = 0.03), and right caudate (PFDR = 0.04, Fig. 4e,f). Significant associations were also identified between the brain volumes of the orbitofrontal cortex and levels of GABA subunit A5 (FDR = 0.01, Fig. 4d) and orexin (FDR = 0.03).

Fig. 4

Associations between peripheral molecules and brain morphology in patients with MUD. (a,b) Group difference of cingulate gyrus volume. (c,d) Association between blood molecules and brain structure. (e) Structure of caudate. (f) Association between blood brain-derived neurotrophic factor and caudate volume. BDNF, brain-derived neurotrophic factor; GABA, gamma-aminobutyric acid; GMV, gray matter volume; MUD, methamphetamine use disorder; OFC, orbitofrontal cortex. Color dots corresponded to sample groups as shown by the x-axis.

Discussion

In the current study, the blood molecular changes associated with methamphetamine use disorder (MUD) were systematically screened, and their connections to withdrawal time, clinical characteristics, and brain structural alterations were analyzed. The findings revealed widespread activation of inflammation and oxidative stress in patients with MUD. Furthermore, specific molecules like brain-derived neurotrophic factor (BDNF) and melatonin were highlighted for their role in mediating other molecular or brain structural changes.

The linear discriminant analysis (LDA) method was employed to differentiate individuals with MUD from healthy controls. Generally, a model with an Area Under the Curve (AUC) exceeding 0.8 is considered to have good discriminatory power. In this study, the AUC of 0.95 indicated strong discriminatory ability, effectively distinguishing between the two groups. Specificity, which refers to the proportion of true controls correctly identified as controls, was 0.75. While not exceptionally high, this value is acceptable in many research contexts. Sensitivity (also known as the true positive rate), representing the proportion of true patients correctly identified, was 0.8, indicating that the model accurately identified 80% of patients with MUD, which is a relatively high rate.

The inflammatory and oxidative processes activated by MUD, involving molecules such as IL-10, IL-9, glutathione peroxidase, and melatonin, align with existing understanding of MUD complications. Prior research has shown that methamphetamine can alter various immune cells, including T cells, B cells, dendritic cells, and natural killer cells, and can promote macrophage polarization. Methamphetamine can also trigger excessive oxidation through neurotransmitter release, leading to demyelination, neuron damage, and heart injury. The toxicity of methamphetamine to neurons and glial cells is known to be mediated by redox-sensitive transcription factors like nuclear factor E2–related factor 2, further emphasizing the role of hyperoxidation in MUD complications. Similarly, alcohol impairs the oxygen reduction system, particularly NADH and the cytoP450 system, which disrupts mitochondrial functions and eventually causes liver damage. Together, the findings from this study and previous research suggest that inflammation and oxidative stress are central processes in MUD complications, making them potential targets for therapeutic interventions. It is therefore recommended that antioxidants be considered as part of treatment regimens for methamphetamine users seeking medical care.

In this study, individuals with MUD were found to exhibit higher levels of anxiety, depression, impulsivity, and nicotine use compared to controls, which is consistent with clinical observations. Additionally, a correlation was identified between peripheral blood indicators and these behavioral symptoms. 5-hydroxyindoleacetic acid (5-HIAA), a metabolite of 5-hydroxytryptamine (5-HT) within the serotonin system, often shows changes in conditions like anxiety, depression, and stress. Decreased 5-HIAA levels have also been observed in animal studies of methamphetamine administration, suggesting a link to anxiety and depressive symptoms in methamphetamine users. Melatonin, an indoleamine produced by the pineal gland and released at night, is known for its effects on sleep and has demonstrated anti-anxiety properties and involvement in addiction in both animal and human studies. This study demonstrated melatonin's association with anxiety symptoms in patients with MUD, and other research has linked it to impulsivity, aligning with these findings. Superoxide dismutase (SOD), a key component of the antioxidant enzyme system, shows significant decreases in anxiety disorders and stress. A previous study also indicated that genetic variations in SOD2 could predict gray matter volume reduction in chronic alcohol users, suggesting its potential role in the development of methamphetamine addiction. The link between vitamin B12 and depression is well-established, and methamphetamine users have been found to have lower vitamin B12 levels than healthy controls. Thus, early vitamin B supplementation might potentially alleviate depressive symptoms in patients with MUD.

A reduction in cortical thickness (CT) was observed in the bilateral calcarine sulcus in patients undergoing drug withdrawal. To date, only one other study, involving heroin-dependent patients, has reported similar results with reduced CT in the left calcarine sulcus. It is speculated that long-term methamphetamine use may lead to irreversible and progressively worsening damage to the calcarine sulcus, indicating persistent harm from the substance. Previous research also noted a significant decrease in gray matter volume (GMV) in methamphetamine users during a six-month withdrawal period, specifically in the precentral gyrus, caudate, fusiform, and cerebellum. After 12 months, the cerebellum showed recovery, but the cingulate gyrus continued to exhibit reduced volume. Furthermore, the decrease in GMV correlated with the cumulative amount of methamphetamine used, which is consistent with the findings of this study.

Another significant finding of the current study is the association between blood molecule levels in patients with MUD and their structural changes in the central nervous system. A notable result is the strong association between brain-derived neurotrophic factor (BDNF) and the gray matter volume (GMV) of nine brain regions. BDNF is a crucial neurotrophin essential for neuronal survival, growth, and differentiation. Reduced BDNF levels have previously been implicated in various psychiatric and neurodegenerative disorders, as well as during early methamphetamine withdrawal. However, reports on BDNF level changes during methamphetamine withdrawal have not always been consistent; for instance, one study indicated higher BDNF levels during initial withdrawal stages, followed by a decline after one month. This study did not find significant differences in BDNF levels between patients with MUD and healthy individuals, which is inconsistent with some prior research. This discrepancy may be due to differences in sample size and withdrawal times. Nonetheless, a significant correlation was established between BDNF concentrations and GMV across nine specific brain regions, including the fusiform gyrus, orbitofrontal cortex, temporal pole, caudate, cerebellum crus, and vermis. This suggests that altered BDNF levels, potentially resulting from chronic methamphetamine use, could either be a consequence of, or contribute to, the observed volumetric changes in these brain regions.

Extensive research has explored BDNF's role in brain function within the context of substance use disorders, often focusing on the prefrontal-striatal circuit and the cerebellum. BDNF, along with its genes, messenger RNA, and metabolites, is involved in the development of addiction. Given BDNF's role in neuroplasticity, where it can modify synaptic strength and structure, it is plausible to infer that the neurotoxic effects of methamphetamine on the brain might be influenced by BDNF. These findings align with previous research indicating that methamphetamine decreases BDNF expression in specific brain regions, suggesting a potential pathway for therapeutic intervention targeting BDNF signaling. Moreover, blood levels of gamma-aminobutyric acid (GABA) were also found to correlate with the orbitofrontal cortex. Previous studies have similarly linked both alcohol and cocaine addiction to changes in the orbitofrontal cortex GABA system, characterized by increased GABA(A) subunit mRNA levels. The use of GABA receptor antagonists has been shown to reduce addictive behaviors, indicating that GABA receptors are potentially effective targets for drug interventions in addiction.

Several limitations of this study should be acknowledged. First, the number of participants with longitudinal data or brain MRI data, for both patients and controls, is relatively small, and there is a considerable disparity in sample size between the groups. Additionally, MRI data from healthy individuals were not collected, although statistical power was found to be sufficient after calculations. Furthermore, the extent to which peripheral blood indicators can accurately reflect changes in brain structure and function is a long-standing question, necessitating caution when drawing conclusions. Moreover, since individuals in the pre-addiction phase rarely seek medical help, this study's cohort only included patients who already displayed significant symptoms. The stage of addiction for each patient was not uniform. Therefore, temporal profiles of blood molecules at different stages still require further analysis, and the explanation of causality based on current state indicators remains an area for discussion.

In conclusion, this study found that methamphetamine use disorder (MUD) is associated with widespread disruptions in peripheral molecules. These disruptions were observed to partially recover after withdrawal but did not fully return to a healthy state. Major disruptions were linked to inflammatory and oxidative activation, as well as stress responses. The alterations in melatonin, superoxide dismutase (SOD), and brain-derived neurotrophic factor (BDNF) were further connected to clinical symptoms and brain structural abnormalities. These results can guide future research aimed at identifying biomarkers for MUD and developing therapeutic targets for MUD-related complications.

Introduction

Methamphetamine Use Disorder (MUD) is a significant health issue worldwide, affecting over 0.7% of people in their lifetime. It can greatly lower a person's quality of life by impacting social interactions and physical health. Earlier research shows that MUD can lead to various problems, such as changes in the immune system, abnormalities in certain brain cells (microglia), increased stress responses, and disruptions in the body's metabolism. A thorough understanding of these changes is essential for treating MUD and its related complications, and for finding markers to help diagnose the disorder. Studies examining people at a single point in time suggest that these abnormalities can be found in the blood of individuals with MUD, indicating that analyzing blood molecule patterns is useful for understanding the disease. For instance, higher levels of malondialdehyde in MUD have been seen as evidence of increased oxidative stress from methamphetamine exposure.

However, current research on blood molecule patterns in MUD and other substance use disorders is still in its early stages, leaving many important questions unanswered. First, the amount of the substance used and the time spent in withdrawal can significantly affect blood molecules, but few studies have considered these changes over time. Understanding these temporal patterns is crucial for developing MUD markers, as early changes might indicate addiction risk, while later changes could point to damage in other organs. Second, the way neuroactive substances disrupt molecules involves multiple biological pathways that are interconnected. For example, heart damage from MUD is caused by various molecular processes, including increased oxygen use, neurotransmitter release, excessive oxidation, and inflammation. Therefore, a study focusing only on isolated heart injury markers at one point in time might not fully capture the underlying mechanisms. Third, how blood markers relate to changes observed in clinical evaluations and the central nervous system is a valuable yet unanswered question. MUD has been shown to be linked to multiple brain structural alterations, such as damage in specific brain regions, but how these changes connect to blood molecule profiles is unknown. In other areas of psychiatry, researchers have found a link between C-reactive protein in the blood and the brain's microscopic structure, suggesting that MUD blood molecule profiles could also be linked to specific brain alterations.

Until now, research on molecular markers of methamphetamine addiction has been scattered and produced inconsistent results. How these markers affect brain function and impact the long-term outlook for methamphetamine users remains unclear. Given these challenges, a group of patients with MUD and healthy control participants were recruited. Following a thorough review of existing research on addiction-related molecular markers, a comprehensive analysis was conducted. This analysis combined data on the levels of 52 blood molecules from nine categories (including neuropeptides, oxidation markers, and neurotransmitters), clinical information (such as drug dosage and withdrawal time), and brain magnetic resonance imaging (MRI) data. This study design aimed to answer two main questions: (i) what are the overall molecular changes in MUD? and (ii) how are these molecular changes related to other molecules and alterations in brain structure? Therefore, this study proposed that blood-based molecular markers in methamphetamine users influence observable symptoms and characteristics by interacting with brain function, potentially providing insights for treatment and intervention.

Method

Cohort and Ethical Approval

This study received approval and oversight from the ethics committee of Shanghai Mental Health Center and followed the principles of the Declaration of Helsinki. All participants gave informed written consent to participate.

Sample Recruitment

A total of 271 participants were recruited based on the following criteria:

1. A total of 231 patients with MUD who: (i) met the diagnostic criteria for severe methamphetamine use disorder as defined by DSM-5; (ii) had at least 9 years of education; (iii) were aged 18 to 50 years; (iv) were not currently undergoing any drug treatment; and (v) did not also have a physical illness or a major mental health disorder (such as depression, anxiety, or schizophrenia) as defined by DSM-5, or serious medical conditions requiring medication, or other drug use disorders besides methamphetamine. Among these patients, they were further categorized by their withdrawal time (<3 months: acute group; >3 months and <6 months: medium group; >6 months: chronic group).

2. A total of 40 healthy control participants who: (i) had no severe physical or neurological health conditions; (ii) had no past history of drug use; (iii) had not experienced previous head injuries; (iv) did not meet the diagnostic criteria for any psychiatric disorders as defined by DSM-5; and (v) had no family history of psychiatric disorders.

Data Collection

A case report form, completed by the participant, gathered social and demographic information (e.g., age, education, marital status, weight, and height) and drug use history (e.g., age of onset, total duration of drug use, and dose). This form was given to each participant by a psychiatrist in a separate room. The demographic characteristics of the control group were also collected.

For a subset of patients, a 1-year follow-up study was conducted where the same demographic, clinical assessment, and blood samples were collected. Patients with follow-up data did not show significant differences in demographic characteristics compared to those without.

Sample Processing and ELISA

Fasting blood samples were collected from each participant between 8 am and 9 am in 10-mL EDTA tubes and then centrifuged for 15 minutes at 3000 revolutions per minute and 4 degrees Celsius. Fifty-two molecules related to addiction biomarkers were chosen based on a thorough review of existing literature. The levels of each blood marker were measured using a sandwich enzyme-linked immunosorbent assay (ELISA) with a commercially available kit. Each biomarker's ELISA was performed separately. All tests were done twice, and results were expressed as picograms per milliliter. The assay's detection range was 20 to 4000 pg/mL. The within-assay and between-assay variation coefficients were less than 5% and 10%, respectively. A single technician performed all procedures, unaware of the sample groups, to minimize technical differences.

Global Analysis of Metabolites

The expression levels of each molecule for all samples were normalized to have a mean of zero and a standard deviation of one. Linear discriminant analysis (LDA) was applied to all baseline samples. Samples were categorized as "MA" (methamphetamine users) or "Control." LDA was first applied using all 52 molecules, and then separately for molecules within each category (e.g., cytokines and neurotransmitters). The accuracy of discrimination was assessed using the area under the curve (AUC). Within each patient group, linear regression was used to evaluate the relationship between the linear discriminant (LD) score and substance withdrawal time, while controlling for age, sex, and education. For a subset of patients with follow-up data, the LD score at follow-up was calculated using the LD loadings from the baseline sample, and a paired t-test was used to evaluate the difference between baseline and follow-up.

Association Test of Each Molecule

Linear mixed regression was applied to test the association between the expression level of each molecule and the following variables, with the sample recruitment site included as a random effect:

1. Demographic variables, including age, sex, and education. No further controlling variables were included in these analyses.

2. Clinical assessments: clinical impulsiveness was assessed using the Barratt Impulsiveness Scale (BIS-11); anxiety symptoms using the General Anxiety Disorder-7 (GAD-7); general health conditions using the Patient Health Questionnaire-9 (PHQ-9); sleep quality using the Pittsburgh Sleep Quality Index (PSQI); and nicotine usage using the Fagerstrom Test for Nicotine Dependence (FTND), along with withdrawal time. All demographic variables were included as controlling variables when calculating the association between each molecule and these assessments.

3. Substance use status, including comparisons between all methamphetamine users versus controls, and acute methamphetamine users versus controls. All demographic variables were included as controlling variables.

The significance threshold was set at a p-value of less than 0.05/52.

Brain MRI Procedure

All brain imaging for this study was conducted using a Siemens Tim Trio 3T scanner. High-resolution T1-weighted anatomical scans were acquired using a specific MRI sequence with detailed parameters. The cortical thickness (CT) for each small 3D volume (voxel) of each participant's brain was estimated using the DiReCT method, part of Advanced Normalization Tools (ANTs). This method uses tissue probability maps to find the most likely match between the brain's white matter surface and the outer gray matter surface. DiReCT is a dependable volume-based technique for measuring CT in both humans and animals, yielding results similar to surface-based algorithms. Additionally, gray matter volume (GMV) and white matter volume (WMV) were obtained using the VBM8 toolbox, which runs in MATLAB. All images were visually and automatically checked for consistency using covariance analysis on the segmented gray matter images. T1-weighted images were corrected for magnetic field variations and aligned using a DARTEL template. Spatially normalized images were then divided into gray matter, white matter, and cerebrospinal fluid, and adjusted to preserve the regional volumetric information of specific tissues within a voxel. All images were then smoothed using an isotropic Gaussian kernel to reduce noise. The GMV, WMV, and CT of each brain region were then extracted using the AAL3 brain atlas, which includes the nucleus accumbens. Finally, linear mixed regression was performed to examine the association between each molecule and brain region pair, with age, sex, and other health conditions included as controlling variables.

Result

Characteristics of Patients with MUD and the Control Group

The study recruited 231 patients with MUD, who had varying withdrawal times, and 40 healthy control participants. Clinical demographic characteristics showed significant differences in age and sex between patients and controls, which were accounted for in subsequent analyses. Among the different patient subgroups, the only demographic difference observed was that the acute group was significantly older. Other demographic characteristics, including body mass index, smoking, drinking habits, and physical health conditions, did not show significant differences between groups.

Global Analysis of Metabolite Expression

The blood expression levels of 52 molecules related to different biological functions were quantified. To get an overall view of the molecular changes, Linear Discriminant Analysis (LDA) was applied to create linear combinations of the tested molecules that could best capture differences between the sample groups. The first combination, LD1, successfully distinguished patients with MUD from control participants (with an AUC of 0.95, sensitivity of 0.80, and specificity of 0.75). The McNemar test p-value was very low. When LDA was run separately using only cytokines, oxidation markers, neuropeptides, or stress response markers, these individual categories were not able to fully distinguish patients (with AUC values below 0.75). This suggested that methamphetamine caused widespread disruptions, not limited to a single system. LD2 did not show any association with patient groups or withdrawal time.

The association between LD combinations and clinical characteristics was then tested. It was found that LD1 was significantly associated with methamphetamine withdrawal time; longer withdrawal times were linked to higher LD1 scores, bringing them closer to those of healthy controls. This suggested that the overall metabolic disruption could partially reverse after stopping methamphetamine use. The acute, medium, and chronic groups of patients had significantly lower LD1 scores than controls, with the acute group showing the lowest scores. A longitudinal analysis was performed on a subset of 59 patients with MUD, and as anticipated, LD1 scores recovered to near control levels after a 1-year follow-up. Additionally, a lower LD1 score was broadly associated with higher scores on the PHQ-9 (depression symptoms) and GAD-7 (anxiety symptoms).

Substance Use-Activated Stress Response, Immune Response, and Hyperoxidation

After describing the overall metabolic disruption, the disruption of each individual molecule was further analyzed. The expression levels of 26 molecules were significantly different between control participants and patients with MUD. The most notable molecules included soluble intercellular adhesion molecule 1 (lower in patients), folic acid (lower in patients), as well as interleukin (IL)-6 (higher in patients) and myoglobin (higher in patients).

Overall, all heart injury markers (6 out of 6), vitamins (3 out of 3), and neurotransmitters (3 out of 3) were altered by substance use, followed by most oxidation markers (6 out of 7) and stress markers (5 out of 8). The direction of these changes consistently indicated the activation of: (i) the stress response (increased stress markers and hormones), (ii) the immune response (increased markers that promote inflammation, such as C-reactive protein, IL-6, and tumor necrosis factor, alongside a decrease in the anti-inflammatory cytokine IL-10), and (iii) excessive oxidation (shown by increased oxidation products and decreased antioxidant enzymes).

The potential differences within the MUD group were further analyzed. When comparing the acute MUD group with the controls, it was found that 24 of the 26 molecules disrupted in the overall MUD group were similarly disrupted in the acute MUD group. Exceptions were vitamin B12 and norepinephrine. Conversely, brain natriuretic peptides, ghrelin, and melatonin were significantly altered in the acute methamphetamine group but not in the overall MUD group. Consistent with this, the moderate and chronic MUD groups showed similar but less pronounced disruption: all 26 significant molecules showed the same direction of change in both the moderate and chronic groups, but only 24 and 19, respectively, reached statistical significance. Generally, although withdrawal time was significantly associated with 14 molecules, these differences were much less noticeable than the differences observed between methamphetamine users and healthy control participants.

Finally, an analysis was conducted to determine whether the cumulative dosage of methamphetamine use impacted blood molecules. Only one suggestive negative association was found between methamphetamine dosage and epinephrine.

5-HIAA, Melatonin, and SOD Reflected Clinical Symptoms

As expected, patients with MUD had significantly higher scores on the Barratt Impulsiveness Scale (BIS), Patient Health Questionnaire-9 (PHQ-9), General Anxiety Disorder-7 (GAD-7), and Fagerstrom Test for Nicotine Dependence (FTND) compared to control participants. These scores partially eased as withdrawal time increased, but they did not fully return to normal levels. A further analysis investigated whether these symptoms were associated with disruptions in blood molecules. The blood level of 5-hydroxyindoleacetic acid (5-HIAA) showed a significant inverse relationship with PHQ-9 and GAD-7 scores. Melatonin was also inversely associated with two assessments: GAD-7 and BIS. Lastly, superoxide dismutase (SOD) was associated with GAD-7, and vitamin B12 was associated with PHQ-9. In summary, the expression levels of blood molecules are associated with clinical symptoms in MUD.

Strong Association Between BDNF and Brain Structures of Patients with MUD

Given the identified association between blood molecules and clinical symptoms, the question then arose whether these molecules were also associated with brain structure. In a subset of 127 patients with MUD (including those from acute, medium, and chronic withdrawal groups), there was no significant difference between patients with and without MRI data. MRI scans were conducted, and brain characteristics derived from these scans were associated with withdrawal time and blood molecules. After statistical adjustment, the cortical thickness (CT) of both the left and right calcarine sulcus, as well as the right putamen, differed significantly among MUD subgroups with different withdrawal times. Nine significant associations were also found between blood molecules and brain structures. Specifically, brain-derived neurotrophic factor (BDNF) showed a significant association with the gray matter volume (GMV) of the right fusiform gyrus, specific subregions in the right cerebellum and vermis, the left superior temporal pole, and the right caudate. Significant associations were also found between the brain volumes of the orbitofrontal cortex and a subunit of GABA, as well as orexin.

Discussion

In this study, the blood molecular alterations associated with MUD were systematically screened, and their association with withdrawal time, clinical characteristics, and brain structural alterations was analyzed. The study found widespread activation of inflammation and oxidative stress in patients with MUD. It also highlighted molecules such as brain-derived neurotrophic factor (BDNF) and melatonin, which played a role in other molecular or brain structural changes.

The LDA method was used to distinguish MUD from healthy control participants. Generally, a model with an Area Under the Curve (AUC) greater than 0.8 is considered to have good discriminative ability. In this study, the AUC was 0.95, which indicates that the model performed well in distinguishing between the two groups. Specificity refers to the proportion of actual control samples that are correctly identified as controls; a value of 0.75, though not extremely high, is often acceptable in many studies. Sensitivity, also known as the true positive rate, refers to the proportion of actual patient samples that are correctly identified as patients. A sensitivity of 0.8 indicates that the model could correctly identify 80% of patients with MUD, which is considered a relatively high value.

The observed inflammatory and oxidative activation in MUD, involving molecules like IL-10, IL-9, glutathione peroxidase, and melatonin, aligns with existing knowledge about MUD complications. Previous studies have shown that methamphetamine can alter various immune cells (such as T and B cells, dendritic cells, and natural killer cells) and encourage macrophages to change their function. Methamphetamine can also trigger excessive oxidation through the release of brain chemicals, which then leads to damage to nerve coatings (demyelination), neuron damage, and heart injury. The harmful effects of methamphetamine on nerve cells (neurons) and support cells (glia) are mediated by certain transcription factors (like nuclear factor E2–related factor 2) sensitive to oxidative changes, further emphasizing the role of excessive oxidation in MUD complications. Similarly, alcohol impairs the oxygen reduction system, specifically NADH and the cytoP450 system, disrupting mitochondrial functions and ultimately leading to liver injury. Taken together, both the current findings and previous research suggest that inflammation and oxidative stress are central processes in MUD complications, potentially serving as good targets for treatment. This suggests that antioxidants might be beneficial additions to treatment plans for individuals with methamphetamine use disorder seeking medical care.

In this study, patients with MUD were found to have higher levels of anxiety, depression, impulsivity, and nicotine use. Additionally, a correlation was found between blood markers and these behaviors. 5-HIAA (5-hydroxyindoleacetic acid) is a breakdown product of serotonin and part of the serotonin system. Changes in its levels often occur during conditions like anxiety, depression, and stress. A decrease in 5-HIAA has also been observed in animal studies of methamphetamine administration, suggesting that changes in 5-HIAA might indicate anxiety and depression symptoms in methamphetamine users. Melatonin, a hormone produced in the pineal gland, is released into the blood every night. Beyond its well-known effects on sleep, it has been shown in animal and human studies to have anti-anxiety effects and play a role in addiction. This study demonstrated that melatonin was associated with anxiety symptoms in patients with MUD, and other studies suggest its association with impulsivity, which is consistent with the current findings. Superoxide dismutase (SOD), an important component of the antioxidant enzyme system, shows significantly decreased levels in anxiety disorders and stress. Previous research also indicated that genetic variations in SOD2 could predict gray matter volume reduction in long-term alcohol users, suggesting its potential involvement in the development of methamphetamine addiction. The link between vitamin B12 and depression has been confirmed by numerous studies, and methamphetamine users have shown lower vitamin B12 levels than healthy controls. Therefore, it is hypothesized that early vitamin B supplementation could improve depressive symptoms in patients with MUD.

A reduction in cortical thickness (CT) was observed in both the left and right calcarine sulcus in patients undergoing drug withdrawal. To date, only one paper on heroin-dependent patients has been found showing similar results in reduced CT (specifically in the left calcarine sulcus). It is speculated that long-term methamphetamine use could have lasting effects on the calcarine sulcus, gradually worsening over time, indicating that the damage caused by methamphetamine is persistent. A previous study also found significantly decreased gray matter volume (GMV) in methamphetamine users during the 6-month withdrawal period in areas such as the precentral gyrus, caudate, fusiform, and cerebellum. At 12 months, the cerebellum had recovered, but the cingulate gyrus continued to show reduction. Furthermore, the decrease in GMV correlated with the cumulative use of methamphetamine, which is consistent with the current findings.

Another interesting finding of this study is that the blood molecule levels of patients with MUD were associated with structural alterations in their central nervous system. A notable result is the significant association between brain-derived neurotrophic factor (BDNF) and the gray matter volume (GMV) of nine brain regions. BDNF is a vital protein that plays a key role in the survival, growth, and differentiation of neurons. Reduced BDNF levels have previously been linked to the development of various psychiatric and neurodegenerative disorders, as well as early methamphetamine withdrawal. However, reports on how BDNF levels change over time during methamphetamine withdrawal are not always consistent; for example, one study showed that BDNF levels were higher during the initial stages of withdrawal but then declined after one month. In this study, no significant differences were found in BDNF levels between patients and healthy individuals, which contradicts some previous research. This discrepancy may be due to differences in sample size and withdrawal time, but a significant correlation was found between BDNF concentrations and GMV across nine specific brain regions, including the fusiform gyrus, orbitofrontal cortex, temporal pole, caudate, cerebellum crus, and vermis. This could suggest that changes in BDNF levels, potentially resulting from chronic methamphetamine use, might either be an outcome of or contribute to the observed volume changes in these brain regions. Numerous studies have investigated the effects of BDNF on brain function in the context of substance use disorders, mainly focusing on the brain circuit involving the prefrontal cortex and striatum, as well as the cerebellum. BDNF, its genes, mRNA, and metabolites play a role in how addiction develops. Given that BDNF is involved in the brain's ability to change and can adjust the strength and structure of connections between neurons, it is reasonable to conclude that the harmful effects of methamphetamine on the brain might be influenced by BDNF. Furthermore, these findings align with previous research showing that methamphetamine can decrease BDNF expression in specific brain regions, suggesting a potential direction for treatment interventions that target BDNF signaling pathways.

Furthermore, blood levels of GABA were found to correlate with the orbitofrontal cortex. Previous studies have also shown that both alcohol addiction and cocaine addiction can cause changes in the orbitofrontal cortex's GABA system, indicated by an increase in GABA(A) subunit mRNA levels. Using GABA receptor blockers (antagonists) could reduce addictive behavior, suggesting that GABA receptors are potentially effective targets for drug interventions in addiction.

Several limitations to this study should be noted. First, the number of participants (both patients and controls) with longitudinal data or brain MRI data was relatively small, and the sample sizes differed considerably between groups. MRI data from healthy individuals were not collected, although statistical power was still deemed sufficient after calculation. Additionally, the degree to which blood markers reflect changes in brain structure and function is a long-standing challenge, requiring caution in drawing conclusions. Furthermore, because individuals in the pre-addiction phase rarely seek medical support, the study cohort included only patients who already showed significant symptoms. Each patient's addiction stage was not identical. Therefore, the changes in blood molecules at different addiction stages still require further analysis, and attributing causality based on single-time-point indicators needs further discussion.

In conclusion, it was found that MUD was associated with widespread disruptions in blood molecules. These disruptions were partially reversible after withdrawal but did not fully return to a healthy state. The main disruptions were linked to inflammatory and oxidative activation, as well as stress responses. Disruptions involving melatonin, superoxide dismutase (SOD), and brain-derived neurotrophic factor (BDNF) were further connected to clinical symptoms and brain structural abnormalities. These results can guide future studies aimed at identifying biomarkers for MUD and therapeutic targets for MUD complications.

Understanding Methamphetamine Use Disorder

Methamphetamine use disorder (MUD) is a serious worldwide health problem that affects more than 0.7% of people in their lifetime. MUD can greatly reduce a person's quality of life by harming social and physical abilities. Past research has shown that MUD can lead to different problems, such as changes in the immune system, brain cells (microglia), stress responses, and metabolism. A full understanding of these problems is needed to treat MUD and its related complications, and to find ways to detect MUD early. Studies have shown these problems can be found in the blood of patients with MUD, suggesting that studying blood molecules in MUD helps understand the disease. For instance, higher levels of malondialdehyde in MUD pointed to increased oxidative stress from methamphetamine exposure.

However, current studies of blood molecule profiles in MUD (and other substance use) are still basic, with many important questions unanswered. First, the amount of drug used and how long someone has been off it can greatly affect blood molecules, but few studies have looked at this timing. This timing is especially important for finding MUD markers because early changes might show addiction risk, while later changes could show harm to other organs. Second, drug-related changes in molecules involve many biological processes that affect each other. For example, heart damage from MUD happens through different molecular processes, like increased oxygen use, release of brain chemicals, high oxidative stress, and inflammation. So, a single study looking only at heart damage markers might not fully show the main reasons for the damage, which are key to current findings. Third, how blood signals show changes in a person's health and brain is an important but unanswered question. MUD has been linked to changes in brain structure, like damage around the ventricles, under the cortex, and in deep white matter, but how these brain changes connect to blood molecules is unknown. In other areas of psychiatry, researchers have found that C-reactive protein in the blood was linked to the detailed structure of the human brain. It is likely that blood molecules in MUD are also linked to specific changes in the brain.

Until now, research on molecular markers of methamphetamine addiction has been scattered with different results. The way these markers affect brain function and influence the future course of MA addiction remains unclear. Because of these challenges, a group of patients with MUD and healthy individuals were recruited. After a thorough review of research on addiction-related molecular markers, a combined analysis was done. This analysis looked at data from 52 blood molecules from nine groups (neuropeptides, oxidation, stress, vitamin, cytokine, neurotransmitter, cardiac injury markers, liver markers, and metabolites), along with health evaluations and features (like drug exposure amount and time since last use), and brain MRI scans. This study design allowed researchers to answer what the overall molecular changes in MUD are, and how these molecular changes are connected to other molecules and brain structure changes. Therefore, this study suggested that blood markers in MA users affect their symptoms by interacting with brain function, which could help in finding treatments.

Cohort and ethical approval

This study was approved and overseen by the ethics committee of Shanghai Mental Health Center and followed the Declaration of Helsinki. All participants gave their written consent to take part.

Sample recruitment

A total of 271 participants were recruited based on specific criteria.

This included 231 patients with MUD who met the diagnosis for severe MA use according to DSM-5, had at least 9 years of education, were aged 18 to 50, were not currently taking any medication, and did not have other serious medical illnesses or mental health conditions like depression, anxiety, or schizophrenia, or other drug use disorders besides MA. These patients were further divided into groups based on how long they had been off the drug: less than 3 months (acute group), between 3 and 6 months (medium group), or more than 6 months (chronic group). The study also included 40 healthy individuals who had no severe physical or neurological problems, no history of drug use or head injury, no mental health diagnoses, and no family history of mental health disorders.

Data collection

A flowchart showed the study process. A psychiatrist in a private room gave each participant a self-administered form to collect information about their social characteristics (like age, education, marital status, weight, and height) and drug use history (like age of first use, total time using drugs, and dose). Information about the healthy individuals' social characteristics was also collected.

For some patients, a one-year follow-up study was done, collecting the same demographic, health evaluation, and blood samples. Patients who provided follow-up data did not show major differences in their social characteristics.

Sample processing and ELISA

Fasting blood samples were collected from each participant between 8 am and 9 am in 10-mL EDTA tubes and spun for 15 minutes at 3000 rpm and 4°C. Fifty-two molecules were chosen after a thorough review of research on addiction markers. The levels of each blood marker were measured using a standard lab test called sandwich enzyme-linked immunosorbent assay (ELISA) with a commercial kit. Each marker was tested separately. All tests were done twice and reported in picograms per milliliter. The detection range for this test was 20 to 4000 pg/mL. The accuracy within a test and between tests was less than 5% and less than 10%, respectively. All steps were performed by one technician who did not know which samples belonged to which group to reduce differences from the testing process.

Global analysis of metabolites

The levels of each molecule in all samples were adjusted to have a zero mean and a standard deviation of one. Linear discriminant analysis (LDA) was applied to all initial samples using an R package. Samples were labeled as "MA" or "Control." LDA was first performed using all 52 molecules, then using molecules from each category (like cytokines and neurotransmitters). The accuracy of distinguishing between groups was evaluated by the area under the curve (AUC). Within each patient group, linear regression was used to check the relationship between the linear discriminant (LD) score and how long someone had been off the substance, while also accounting for age, sex, and education. For a subset of patients with follow-up data, the LD score at follow-up was calculated using the original LD values from the initial sample, and a paired t-test was used to evaluate the difference between the initial and follow-up scores.

Association test of each molecule

Linear mixed regression was used to test the connection between the level of each molecule and the following factors, with the sample collection site considered as a random effect:

1. Social characteristics, including age, sex, and education. No other factors were included in these analyses.

2. Health evaluations: Clinical impulsiveness was measured by the Barratt Impulsiveness Scale (BIS-11), anxiety symptoms by the General Anxiety Disorder-7 (GAD-7), general health by the Patient Health Questionnaire-9 (PHQ-9), sleep quality by the Pittsburgh Sleep Quality Index (PSQI), nicotine use by the Fagerstrom Test for Nicotine Dependence (FTND), and time since last drug use. When calculating the connection between each molecule and these evaluations, all social characteristics were included as influencing factors.

3. Substance use status, including all MA users compared to controls, and acute MA users compared to controls. All social characteristics were included as influencing factors.

The significance level was set so that the P-value of the coefficient was less than 0.05/52.

Brain MRI procedure

All brain imaging for this study was performed using a Siemens Tim Trio 3T scanner. High-resolution T1-weighted anatomical scans were taken using a specific sequence. The cortical thickness (CT) of each participant was estimated using a method that uses tissue maps to find the best match between the white matter surface and the outer gray matter surface. This method is reliable for estimating CT in both humans and other primates and gives similar results to other techniques. In addition, gray matter volume (GMV) and white matter volume (WMV) were obtained using another standard software tool. All images were checked visually and automatically for consistency. T1-weighted images were corrected for issues with signal uniformity and aligned using a specific template. Spatially aligned images were then divided into gray matter, white matter, and cerebrospinal fluid and adjusted to keep the regional volume information of specific tissue within a small area. All images were then smoothed with a Gaussian filter. The GMV, WMV, and CT of each brain region were then extracted using a brain atlas, which included the nucleus accumbens. Finally, linear mixed regression was performed to investigate the connection between each molecule and brain region pair, with age, sex, and other medical conditions considered as influencing factors.

Characteristics of patients with MUD and the control group

The study recruited 231 patients with MUD, who had varying times since their last drug use, and 40 healthy individuals. The clinical and social characteristics of these groups were presented. There were significant differences in age and sex between patients and healthy individuals, and these differences were accounted for in the later analyses. Among the different patient subgroups, the only social difference was that the acute group was significantly older. Other social characteristics, including body mass index, smoking and drinking habits, and other medical conditions, did not show major differences between groups.

Global analysis of metabolite expression

The blood levels of 52 molecules related to different biological functions were measured. To get an overall view of the molecular changes, LDA was used to create combinations of the tested molecules that could best show differences between the sample groups. The first combination (LD1) could tell patients with MUD apart from healthy individuals (AUC = 0.95, sensitivity = 0.80, specificity = 0.75). This suggested that MA caused widespread disturbances not limited to a single body system. LD2 did not show a link to patient groups or how long someone had been off the drug.

Researchers then tested the connection between LD combinations and clinical features. It was found that LD1 was significantly linked to the time since MA use (longer time off the drug was linked to a higher LD1 score and being closer to healthy individuals), suggesting that the overall metabolic disruption could partly recover after stopping MA use. The acute, medium, and chronic groups of patients had significantly lower LD1 scores than healthy individuals, with the acute group having the lowest. A follow-up analysis was performed on a subset of patients with MUD (n=59), and as expected, LD1 scores recovered to be similar to healthy individuals after a one-year follow-up. In addition, a lower LD1 score was associated with higher scores for depression (PHQ-9) and anxiety (GAD-7).

Substance use–activated stress response, immune response, and hyperoxidation

After describing the overall metabolic disruption, the study further analyzed the changes in each molecule. The levels of 26 molecules were significantly different between healthy individuals and patients with MUD. Key molecules included soluble intercellular adhesion molecule 1 (levels were lower in patients), folic acid (lower in patients), as well as interleukin (IL) 6 (higher in patients), and myoglobin (higher in patients).

Overall, 100% of the markers for heart injury (6 out of 6), vitamins (3 out of 3), and neurotransmitters (3 out of 3) were changed by substance use, followed by oxidation markers (6 out of 7) and stress markers (5 out of 8). The direction of these changes consistently suggested the activation of: (i) stress response (increased stress markers and hormones), (ii) immune response (increased markers of inflammation like C-reactive protein, IL-6, and tumor necrosis factor, along with decreased anti-inflammatory cytokine IL-10), and (iii) high oxidative stress (shown by increased oxidation products and decreased antioxidant enzymes).

The study further analyzed differences within the MUD group. When comparing the acute MUD group with healthy individuals, 24 of the 26 molecules disrupted in the MUD group were also disrupted in the acute MUD group in the same way. Brain natriuretic peptides, ghrelin, and melatonin were significantly changed in the acute MA group but not in the overall MUD group. Similarly, the moderate and chronic MUD groups showed similar but less strong disruptions: all 26 significant molecules had the same direction of disruption in both the moderate and chronic groups, but only 24 and 19 reached a significant level, respectively. Generally, although the time since drug use was significantly linked to 14 molecules, the differences were far less pronounced than the differences between MA users and healthy individuals. Finally, it was analyzed whether the total amount of MA used affected blood molecules, and only one weak negative link was found between MA dosage and epinephrine.

5-HIAA, melatonin, and SOD reflected clinical symptoms

As expected, patients with MUD had significantly higher scores for impulsivity, depression, anxiety, and nicotine use than healthy individuals. These symptoms partly improved as the time since drug use increased, but they did not completely return to normal. The study further analyzed whether these symptoms were linked to changes in blood molecules. The blood level of 5-hydroxyindoleacetic acid (5-HIAA) had a significant inverse link with depression (PHQ-9) and anxiety (GAD-7). Melatonin was also inversely linked with two assessments: anxiety (GAD-7) and impulsivity (BIS). Lastly, superoxide dismutase (SOD) was linked with anxiety (GAD-7), and vitamin B12 was linked with depression (PHQ-9). In summary, the levels of blood molecules are linked to clinical symptoms in MUD.

Strong association between BDNF and brain structures of patients with MUD

Since the study identified a link between blood molecules and symptoms, the next question was whether they were also linked to brain structure. In a subset of patients with MUD (n=127), there was no significant difference between patients with and without MRI data. MRI scans were conducted, and the brain features derived from these scans were linked to the time since drug use and blood molecules. The cortical thickness of the bilateral calcarine sulcus and the right putamen were significantly different among MUD subgroups with different times since drug use. Nine significant links were also found between blood molecules and brain structures. Brain-derived neurotrophic factor (BDNF) was significantly linked to the gray matter volume of the right fusiform, subregions in the right cerebellum and vermis, left superior temporal pole, and right caudate. Significant links were found between the brain volumes of the orbitofrontal cortex and GABA subunit A5 and orexin.

Discussion

In this study, researchers systematically looked for changes in blood molecules linked to MUD and analyzed how these changes related to the time since drug use, clinical symptoms, and changes in brain structure. They found widespread activation of inflammation and oxidative stress in patients with MUD and highlighted molecules like BDNF and melatonin, which played a role in other molecular or brain structure changes.

The LDA method was used to tell patients with MUD apart from healthy individuals. Generally, a model with an AUC over 0.8 is considered good at distinguishing groups. In this study, the AUC was 0.95, showing the model was very good at distinguishing between the two groups. Specificity refers to the percentage of healthy individuals correctly identified as healthy. The value of 0.75, while not extremely high, is acceptable in many studies. Sensitivity refers to the percentage of patients correctly identified as patients. The value of 0.8 means the model correctly identified 80% of patients with MUD, which is also a relatively high value.

The inflammation and oxidative activation caused by MUD, involving substances like IL-10, IL-9, glutathione peroxidase, and melatonin, matches what is already known about MUD complications. Previous studies have shown that MA can change many immune cells and can increase oxidative stress, which then leads to nerve damage and heart problems. The harm MA does to nerve and glial cells is thought to be through certain factors, further showing the role of high oxidative stress in MA complications. Similarly, alcohol also harms the oxygen reduction system, which disrupts cell functions and eventually leads to liver injury. Overall, both these results and previous findings suggest that inflammation and oxidative stress are key processes in MUD complications, making them potential targets for treatment. It is recommended that antioxidants be included in medical plans for MA users who seek medical care for any reason.

In this study, it was found that patients with MUD had higher levels of anxiety, depression, impulsivity, and nicotine use, which is understandable. In addition, a link was found between blood markers and these behaviors. 5-HIAA is a breakdown product of serotonin and is often changed in conditions like anxiety, depression, and stress. A decrease in 5-HIAA has also been seen in animal studies of MA use, suggesting that changes in 5-HIAA might indicate anxiety and depression symptoms in MA users. Melatonin is a hormone produced in the pineal gland and released into the blood every night. Besides its known effects on sleep, it has been shown in animal and human studies to reduce anxiety and play a role in addiction. This study showed that melatonin was linked to anxiety symptoms in patients with MUD, and other studies suggest its link to impulsivity, which matches these findings. SOD is an important part of the body's antioxidant system, and a significant decrease has been found in anxiety disorders and stress. A previous study also showed that SOD2 gene variations could predict reduced gray matter volume in chronic alcohol users, suggesting it might also be involved in the development of MA addiction. The link between vitamin B12 and depression has been confirmed by many studies, and MA users have been shown to have lower vitamin B12 levels than healthy individuals. Therefore, it is thought that early vitamin B supplementation could improve depressive symptoms in patients with MUD.

A reduction in cortical thickness was seen in the calcarine sulcus on both sides of the brain in patients undergoing drug withdrawal. So far, only one paper on heroin-dependent patients has shown similar results in reduced cortical thickness. Researchers speculate that long-term MA use could have permanent effects on the calcarine sulcus and gradually worsen over time, indicating that the damage caused by MA is ongoing. Previous research also found significantly decreased gray matter volume in MA users during a 6-month withdrawal period in specific brain areas. At 12 months, some areas had recovered, but others continued to show reduction. Moreover, the decrease in gray matter volume was linked to the total amount of MA used, which matches the results of this paper.

Another interesting finding was that blood molecule levels in patients with MUD were linked to changes in their brain structure. One notable result was the significant link between BDNF and gray matter volume in nine brain regions. BDNF is a crucial protein that helps nerve cells survive, grow, and develop. Reduced BDNF levels have been previously connected to various mental health and brain diseases, and to early MA withdrawal. However, reports on how BDNF levels change over time during MA withdrawal are not always consistent. In this study, no significant differences were found between BDNF and healthy individuals, which is different from previous research. This might be due to differences in sample size and withdrawal time. However, a significant link was established between BDNF levels and gray matter volume across nine specific brain regions. This might suggest that changes in BDNF levels, possibly from long-term MA use, could either be a result of or contribute to the observed volume changes in these brain regions. Many studies have looked at BDNF's effects on brain function in substance use disorders, mainly focusing on specific brain circuits and the cerebellum. BDNF, its genes, and its related substances are involved in the development of addiction. Since BDNF helps with brain plasticity and can adjust connections between nerve cells, it is likely that MA's harmful effects on the brain might be affected by BDNF. These findings also align with previous research showing that MA decreases BDNF levels in certain brain regions, pointing to a potential area for treatment targeting BDNF pathways.

Furthermore, blood levels of GABA were also found to be linked to the orbital frontal cortex. Previous studies have also found that both alcohol and cocaine addiction can cause changes in the GABA system of the orbital frontal cortex, shown as an increase in GABA(A) subunit mRNA levels. Using drugs that block GABA receptors could reduce addictive behavior, suggesting that GABA receptors are potential targets for drug treatment in addiction.

There are several limitations to this study. First, the number of participants with long-term data or brain MRI data is still relatively small, and the sample size difference between each group is also quite large. Also, MRI data from healthy individuals were not collected, but statistical power was still considered sufficient. In addition, the extent to which blood markers can truly show changes in brain structure and function is a long-standing issue, so conclusions need to be made carefully. Furthermore, since people in the pre-addiction phase rarely seek medical help, this study only recruited patients who already had clear symptoms. Each patient's stage of addiction is not the same as others'. Therefore, how blood molecules change at different stages still needs to be analyzed, and explaining cause and effect using current markers needs further discussion.

In conclusion, this study found that MUD was linked to widespread changes in blood molecules, which partly recovered after stopping drug use but did not fully return to a healthy state. Major changes were linked to inflammation, oxidative stress, and stress response. The changes in melatonin, SOD, and BDNF were further linked to clinical symptoms and brain structure problems. These results could guide future studies aimed at finding markers for MUD and treatment targets for MUD complications.

Methamphetamine Use Disorder

Methamphetamine use disorder, or MUD, is a serious problem around the world. It can greatly harm a person's life by affecting their social life and body. Past studies found that MUD can lead to many unusual changes in the body. These include the body's defense system working too hard, stress, and changes in body chemicals. Knowing more about these changes is key to helping people with MUD and its other health problems. It also helps find signs of MUD in the body. Some studies have shown these changes can be seen in a person's blood. This means looking at chemicals in the blood of people with MUD can help understand the illness.

However, studies on blood chemicals in people with MUD are still very new. Many important questions remain unanswered. For one, how much of the drug was used and for how long someone has stopped using it can change their blood chemicals. Most studies have not looked at these changes over time. Also, drug use changes many body processes, and these changes can affect each other. For example, heart problems from MUD involve many things, like too much oxygen use and inflammation. So, looking at just one part, like heart problems, may not show the whole picture. Lastly, it is not clear how blood changes relate to a person's health and brain. MUD is linked to brain changes, but how these changes connect to blood chemicals is not known.

Research on signs of methamphetamine addiction in the body has been unclear and not always the same. It is not well understood how these signs affect the brain and how a person with MUD will do in the future. Because of this, a study was done with people who have MUD and healthy people. After looking at past research, the study looked at many things at once. It studied levels of 52 blood chemicals in nine groups (like stress markers, vitamins, and heart markers), along with health checks (like drug dose and how long someone stopped using), and brain scans. This helped answer: (i) What are the overall chemical changes in MUD? and (ii) How do these changes relate to other chemicals and brain changes? The study thought that blood signs in people using methamphetamine affect their health by working with brain function. This could help find new ways to treat and help people.

Cohort and ethical approval

This study was approved and watched over by the ethics committee at Shanghai Mental Health Center. It followed rules for treating people in research fairly. Everyone who took part in the study gave their written permission.

Sample recruitment

The study included 271 people. Of these, 231 had methamphetamine use disorder (MUD). They met specific health rules, had at least 9 years of school, were 18 to 50 years old, were not on any drug treatment, and did not have other major health or mental problems. These patients were put into groups based on how long they had stopped using the drug: less than 3 months, 3 to 6 months, or more than 6 months. The study also included 40 healthy people. They did not have any serious health problems, drug use history, or mental health issues.

Data collection

A form was used to gather details about each person, like their age, schooling, and drug use history. This was done by a doctor in a private room. Healthy people also shared their basic information. For some patients, information and blood samples were collected again after one year.

Sample processing and ELISA

Blood samples were taken from each person in the morning. The blood was spun to separate it. Levels of 52 chemicals linked to addiction were measured using a special lab test. Each test was done carefully by one person who did not know if the sample was from a patient or a healthy person.

Global analysis of metabolites

The levels of each chemical were adjusted to make them easier to compare. A special math method was used to find differences in all 52 chemicals between MUD patients and healthy people. This method also looked at chemicals within each group (like stress markers). The study checked how well this method could tell the groups apart. For MUD patients, this method also looked at how blood chemicals changed over time as they stopped using the drug.

Association test of each molecule

The study also looked at how the level of each blood chemical was linked to other information. This included things like a person's age, sex, and schooling. It also looked at links to health checks for impulsivity, anxiety, sadness, sleep quality, and nicotine use. Finally, it checked how each chemical was linked to having MUD or being in an early stage of MUD.

Brain MRI procedure

For some patients, brain scans were taken. These scans measured the thickness of the outer part of the brain and the size of brain areas. Then, a math method was used to see if there was a link between the levels of blood chemicals and changes in brain parts.

Characteristics of patients with MUD and the control group

The study included 231 patients with MUD and 40 healthy people. The patients had stopped using the drug for different lengths of time. There were clear differences in age and sex between the patients and healthy people, which were adjusted for in the study. Among the patient groups, the group that had just stopped using the drug was older. Other health details were similar across all groups.

Global analysis of metabolite expression

The study measured the levels of 52 blood chemicals. A special math method showed a clear difference in these chemicals between people with MUD and healthy people. This meant that methamphetamine caused many changes across different body systems, not just one.

The study found that these overall chemical changes were linked to how long someone had stopped using the drug. The longer someone stopped, the more their chemical levels returned closer to those of healthy people. Patients who had just stopped using the drug showed the biggest changes. When some patients were checked again after one year, their chemical levels had gotten closer to normal. Also, bigger chemical changes were linked to more signs of sadness and anxiety.

Substance use–activated stress response, immune response, and hyperoxidation

The study looked closely at each chemical. It found that 26 chemicals were clearly changed in MUD patients compared to healthy people. For example, all markers for heart injury, vitamins, and brain signals were changed. Many chemicals linked to body damage and stress were also changed.

These changes showed higher stress, an overactive immune system, and more cell damage (called "hyperoxidation"). This meant more signs of swelling and less help against damage to cells. The changes were strongest in patients who had just stopped using the drug. While some chemical levels changed over time, the biggest differences were between drug users and healthy people. The total amount of drug used did not seem to have a strong link to blood chemicals.

5-HIAA, melatonin, and SOD reflected clinical symptoms

Patients with MUD had more signs of impulsivity, sadness, anxiety, and nicotine use than healthy people. These signs got better as patients stopped using the drug for longer, but they did not go back to normal.

The study found that levels of certain blood chemicals were linked to these health signs. For example, lower levels of 5-HIAA were linked to more sadness and anxiety. Melatonin was also linked to anxiety and impulsivity. SOD, a chemical that fights cell damage, was linked to anxiety. And vitamin B12 was linked to sadness. This shows that blood chemicals can be connected to health problems in people with MUD.

Strong association between BDNF and brain structures of patients with MUD

Since blood chemicals were linked to health signs, the study also looked at links to brain structure. For some MUD patients who had brain scans, certain parts of the brain were different depending on how long they had stopped using the drug.

There were also clear links between blood chemicals and brain structures. A chemical called BDNF was strongly linked to the size of many brain areas, including parts of the brain that help with seeing, movement, and thinking. Other chemicals like GABA and orexin were also linked to the size of certain brain areas. This shows that blood chemicals in MUD patients can be connected to changes in the brain.

Discussion

This study looked closely at how blood chemicals change in people with MUD. It also checked how these changes related to how long someone had stopped using the drug, their health signs, and their brain structure. The study found a lot of inflammation and cell damage in MUD patients. It also pointed out chemicals like BDNF and melatonin, which were linked to other chemical or brain changes.

Inflammation and Cell Damage

MUD causes inflammation and too much cell damage. This is known from other studies too. Methamphetamine can change how the body's defense cells work. It can also cause cell damage, which leads to harm in nerve coverings, nerve cells, and the heart. This kind of cell damage is also seen with alcohol use, leading to liver problems. Both this study and past findings show that inflammation and cell damage are key problems with MUD. They might be good targets for new treatments. Doctors might consider giving anti-damage medicines to people with MUD who are seeking medical help.

Blood Chemicals and Health Symptoms

People with MUD often have more anxiety, sadness, and impulsivity. This study found that certain blood chemicals were linked to these symptoms. For example, low levels of 5-HIAA were linked to more anxiety and sadness. Melatonin, a chemical known for sleep, was also linked to anxiety and impulsivity. SOD, which helps fight cell damage, was linked to anxiety. And vitamin B12 was linked to sadness. Many studies have shown that vitamin B12 can help with sadness. These findings suggest that helping with vitamin B levels might improve sadness in MUD patients.

Brain Structure Changes

The study found that the outer layer of the brain (calcarine sulcus) was thinner in patients who were stopping drug use. This harm may last for a long time and get worse over time. Past studies also found less brain gray matter in MUD users, and this loss was linked to how much drug was used. This fits with the findings of this study.

BDNF and Brain Health

An important finding was the strong link between blood levels of a chemical called BDNF and the size of many brain parts. BDNF is very important for how brain cells live, grow, and change. Lower levels of BDNF have been seen in many brain problems and in early stages of stopping methamphetamine. While this study did not find BDNF levels to be different from healthy people, it did find a clear link between BDNF and changes in the size of brain areas. This may mean that BDNF changes, possibly from long-term methamphetamine use, could be a cause or a part of the brain changes seen. Past research also shows BDNF is involved in addiction. These findings suggest that aiming treatments at BDNF might help the brain problems caused by methamphetamine.

GABA and Brain Function

Also, blood levels of GABA, another brain chemical, were linked to a part of the brain called the orbitofrontal cortex. Other studies have found that GABA systems in this brain area change with drug use. This suggests that targeting GABA might be a helpful way to treat addiction.

Study Limitations

This study had some limits. The number of people in certain groups, like those with repeat visits or brain scans, was small. Also, the study did not have brain scan data for healthy people. It can be hard to say exactly how blood chemicals reflect brain changes. Last, the study looked at people who already had clear signs of addiction, not those just starting. This means the study cannot fully show how blood chemicals change at different stages of addiction.

Conclusion

In closing, the study found that MUD causes many changes in a person's blood chemicals. These changes got better after stopping the drug, but did not fully return to normal. The main changes were linked to inflammation, cell damage, and stress. Changes in melatonin, SOD, and BDNF were also linked to health signs and brain changes. These findings can help guide future studies to find better ways to test for MUD and treat its problems.