Transcranial Magnetic Stimulation for Methamphetamine Use Disorder: A Scoping Review within the Neurocircuitry Model of Addiction

Abstract

The use of methamphetamine in the United States is increasing, contributing now to the “fourth wave” in the national opioid epidemic crisis. People who suffer from methamphetamine use disorder (MUD) have a higher risk of death. No pharmacological interventions are approved by the FDA and psychosocial interventions are only moderately effective. Transcranial Magnetic Stimulation (TMS) is a relatively novel FDA-cleared intervention for the treatment of Major Depressive Disorder (MDD) and other neuropsychiatric conditions. Several lines of research suggest that TMS could be useful for the treatment of addictive disorders, including MUD. We will review those published clinical trials that show potential effects on craving reduction of TMS when applied over the dorsolateral prefrontal cortex (DLPFC) also highlighting some limitations that affect their generalizability and applicability. We propose the use of the Koob and Volkow’s neurocircuitry model of addiction as a frame to explain the brain effects of TMS in patients with MUD. We will finally discuss new venues that could lead to a more individualized and effective treatment of this complex disorder including the use of neuroimaging, the exploration of different areas of the brain such as the frontopolar cortex or the salience network and the use of biomarkers.

1. Introduction.

Amphetamines (amphetamine and methamphetamine) are among the most widely used illicit drugs worldwide after cannabis (Drugs and Crime, 2019), with methamphetamine being the most frequently consumed illicit stimulant, especially in south-east Asia. In 2020, 0.6 % of the US population over 12 years (or 1.4 million people) reported suffering from methamphetamine use disorder (MUD) (Center for Behavioral Health Statistics and Quality, 2021).

People with MUD have an increased risk of death, with a recent study estimating an all-cause standardized mortality ratio of 6.8 (Stockings et al., 2019). Drug overdose deaths involving methamphetamine increased from 547 in 1999 to more than 20,000 in 2020 (CDC, 2020), particularly in minoritized groups and people with opioid use disorder (Friedman and Shover, 2023).

Despite decades of research, there is no clear evidence for specific pharmacological interventions for MUD (Paulus and Stewart, 2020), although several studies using medications with different mechanisms of action are underway (Ballester et al., 2017). A recent meta-analysis of psychosocial interventions in people with cocaine and amphetamine use disorders (De Crescenzo et al., 2018) showed that a combination of contingency management (CM) with community reinforcement was the most effective intervention for promoting abstinence. However, limitations of psychosocial interventions, include lack of availability, high drop-out rates (in some cases, up to 50%), (Lappan et al., 2020) and modest long-term efficacy (McKetin et al., 2012). Thus, there is a need to develop and evaluate novel MUD treatments. One such potential intervention is Transcranial Magnetic Stimulation (TMS). TMS constitutes an emerging family of neuromodulation techniques that combine focal brain effects, favorable profile of side effects (tolerability) and relatively few contraindications or systemic side effects. The physical principles of TMS can be explained by Faraday’s Law of electromagnetic induction that is based on the ability of a magnetic field to induce an electrical current in a closed circuit. In this case a “coil” that produces a magnetic field is applied to the surface of the scalp and the resulting perpendicular electrical field propagates downward. This electrical field depolarizes neurons according to their action potential threshold and/or modify their excitability. Some TMS devices are now FDA-cleared treatments for major depressive disorder (MDD) (O’Reardon et al., 2007), obsessive compulsive disorder (OCD) (Carmi et al., 2019), and most recently for tobacco use disorder (TUD) (Zangen et al., 2021).

There are different TMS protocols depending on the type of coil used (figure of eight, H coils), the frequency of the pulses delivered (single pulses, repetitive TMS which can be delivered in high or low frequency) or even the pattern of the delivery of the electromagnetic pulses (intermittent or continuous theta burst stimulation or TBS). For clarity, when talking about TMS in general, we will refer to therapeutic TMS (often abbreviated rTMS but hereafter simply TMS) to minimize confusion and to differentiate from single pulse TMS studies. As we will show below, the most general protocol investigated in MUD consists on a daily session of repetitive TMS (high frequency, low frequency or in form of TBS) during 4–6 weeks (total of 20–30 sessions). Similarly to the protocol used in MDD, this is done in an outpatient basis and besides the time commitment, patients continue with their normal daily life activities (no sedation is needed and normally there is no significant changes in patient’s pharmacological treatments).

There is significant promise for the use of TMS for MUD. We will first provide an overview of the current understanding of MUD within the three-stage of Koob and Volkow’s neurocircuitry model of addiction. Describing MUD within this model provides a theoretical framework that could help designing future TMS studies. This perspective will then be used to explain brain effects of TMS in patients with MUD. We will then summarize the available published studies about TMS for MUD, and end by suggesting new venues that could lead to a more individualized treatment, in the hope that precision approaches yield superior efficacy. Of note, other neuromodulatory techniques, such as transcranial direct-current stimulation, have demonstrated some promise in this area (Ekhtiari et al., 2022; Guaiana et al., 2023). However, this review will focus on TMS due to the amount of available evidence and its FDA-clearance for the above indications with the hopes that it might be more rapidly translated into clinical trials.

2. Conceptualization of Methamphetamine Use Disorder within the Addiction Neurocircuitry perspective.

Addiction represents a complex multi-stage process where different neurocircuitries might play specific role(s) at different times of disease progression (Bickel et al., 2018). The Koob and Volkow’s neurocircuitry model of addiction (Koob and Volkow, 2010) explains this process as formed by three different stages: in the binge/intoxication stage the use of a drug produces increases in the concentration of dopamine (DA) in the basal ganglia which reinforces subsequent drug use (positive reinforcement). With ulterior drug use, and due to compensatory changes in the extended amygdala among other mechanisms, withdrawal and negative affective symptoms might emerge which would lead to continue drug use to avoid these symptoms (negative reinforcement). Finally, or at the same time, changes in the prefrontal cortex would be responsible for the preoccupation/anticipation stage where addiction is maintained due to a failure in inhibitory mechanisms in behavior (impulsivity) or in the attribution of importance (salience) to drug cues. We will now review more in detail the neurocircuitry model of addiction and describe some evidence that places MUD within this framework (Figure 1).:

Figure 1: Neurocircuitry schematic illustrating the potential impact of TMS in the brain circuitry for the three stages of the addiction cycle according to Koob and Volkow’s model.

The preoccupation/anticipation stage is represented in green. The frontal cortex system is compromised, producing deficits in executive function (mediated by DLPFC) and contributing to the incentive salience of drugs compared to natural reinforcers (mediated among other effectors by vmPFC). HF-rTMS/iTBS is supposed to exert stimulatory effects in the DLPFC and LF-rTMS/cTBS is supposed to exert inhibitory effects in the vmPFC. dTMS is supposed to exert stimulatory effects in the insula and the mPFC. The binge/intoxication stage is represented in blue. In this case, the dopamine system (represented by discontinuous lines) is compromised. The activation of the ventral striatum/dorsal striatum is driven by cues through the hippocampus and basolateral amygdala. Finally, the withdrawal/negative affect stage is represented in red. Brain stress systems such as NE or CRF are activated by cues through the hippocampus and basolateral amygdala and stress through the insula. It is not known how TMS modulates this stage, e.g. improving the hypodopaminergic tone or though between-systems adaptations (figure modified with permission from Koob, G.F. and Volkow, N.D, 2010). AC: anterior cingulate; Acb, nucleus accumbens; AMG: amygdala; BLA, basolateral amygdala; BNST, bed nucleus of the stria terminalis; CeA, central nucleus of the amygdala; CRF, corticotropin-releasing factor; dTMS, deep TAM; DA, dopamine; DGP, dorsal globus pallidus; DLPFC: dorsolateral prefrontal cortex; DS: dorsal striatum; GP: globus pallidus; HF-rTMS/iTBS: high frequency repetitive transcranial magnetic stimulation/intermittent theta burst stimulation; Hippo: hippocampus; LF-rTMS/cTBS: low frequency repetitive transcranial magnetic stimulation/continuous theta burst stimulation; mPFC: medial prefrontal cortex; NE, norepinephrine; OFC: orbitofrontal cortex; SNc, substantia nigra pars compacta; Thal: thalamus; VGP, ventral globus pallidus; VS: ventral striatum; VTA, ventral tegmental area.

2.1. The binge/intoxication stage and MUD.

As stated earlier, every drug with addiction potential produces an increase in the concentrations of DA (Di Chiara and Imperato, 1988), that binds to D1-like and/or D2-like family receptors (among other actions). It is thought that these effects in the dopaminergic meso-cortico-limbic neurocircuitry are fundamental in the reinforcing capacities of the different drugs with addiction potential (Volkow et al., 1999). Importantly, substance-related DA levels surpass the physiological elevations produced by natural rewards (Schultz et al., 2000). This supraphysiological exposure constitutes the biochemical basis of the so called “incentive sensitization” theory (Robinson and Berridge, 1993), where drugs and associated stimuli are incentivized (wanted) and the brain becomes hyper-reactive (i.e., sensitized) to them. DA functions, under this perspective, as a tag that informs the brain which events/stimuli are important for survival.

Numerous studies have found DA dysregulation in people with MUD. A recent meta-analysis of 14 studies using nuclear imaging (Single-Photon Emission Computed Tomography or SPECT, Positron Emission Tomography or PET) in patients with MUD (Proebstl et al., 2019) in different stages of abstinence, found a reduction of D2/D3 receptors and dopamine transporters in the striatum, and specifically in the caudate and the putamen. This was interpreted as compensatory changes in the brain secondary to the increased levels of DA in the synaptic cleft. This is not surprising as methamphetamine works among different actions by promoting the release of the storaged DA from synaptic vesicles, inhibiting its reuptake from the pre-synaptic neuron and even inhibiting its catabolism by the enzymes monoamine oxidase (MAO) (Moszczynska and Callan, 2017). A small fMRI study of n=25 patients with MUD indicated an increase in resting state functional connectivity within the meso-cortico-limbic circuit (main circuit of the DA system in the brain), more specifically between the midbrain and the amygdala, hippocampus, putamen, and prefrontal cortex, compared to healthy controls (n=27) (Kohno et al., 2014). Thus, neuroimaging studies in MUD point towards a dopaminergic circuit that becomes hyperconnected (with an increase in functional connectivity) with concurrent compensatory changes (reduction in D2/D3 DA receptors) within a cellular and molecular level. These compensatory changes produce a state of low DA that might explain some of the dysphoria experienced by patients specially but not only in early stages of recovery (see next).

2.2. The withdrawal/negative affect stage and MUD.

Koob and LeMoal (Koob and Le Moal, 2001) have described how the developing process of addiction entails a progressive dysregulation of emotional states by the recruitment of stress systems in the brain. Clinically, this stage explains how patients with addiction would continue using drugs not only due to the reinforcing effects of them but ultimately motivated to alleviate dysphoria and negative states. The extended amygdala and its connections to major autonomic centers, serves as the major effector of these changes. The extended amygdala releases neurotransmitters and hormonal regulators such as corticotropin releasing factor (CRF) that will influence other modulators central in the stress process such as orexin, vasopressin or norepinephrine (NE) (Koob, 2008). These increases in the stress process (i.e., between-system changes) add to the previous dopaminergic dysregulation explained in the previous stage (within-system changes) and together will produce processes of negative reinforcement that can facilitate the continuous use of drugs.

While the negative affect stage has been studied more thoroughly in alcohol and opioid use disorders (Koob, 2021), where withdrawal symptoms are more evident, there is growing evidence that it is as well important in MUD (May et al., 2020). A functional magnetic resonance imaging (fMRI) study using a feedback expectancy task with monetary gain/loss, showed that subjects with MUD (n=17) vs controls (n=23) exhibited higher blood oxygenation level dependent (BOLD) responses in the caudate to monetary loss, suggesting increased reactivity to negatively valenced stimuli (Bischoff-Grethe et al., 2017). A recent resting state-fMRI (rs-MRI) study in MUD patients (n=48) with a variable number of abstinent days, compared to controls (n=48), indicated an inverse relationship between anxiety and depressive symptoms with the strength of functional connectivity between the rostral ACC and the medial PFC. In other words, weaker connectivity of higher cognitive and emotional cortical centers (rostral ACC and medial PFC) was associated to higher symptoms of depression and anxiety. Taken together, these studies indicate abnormal functioning of different areas of the brain implicated in emotional regulation in MUD (Jiang et al., 2021).

2.3. The preoccupation/anticipation stage and MUD.

This stage focuses on how the progressive use of drugs translates to reduced inhibitory control from higher cortical/prefrontal structures: the use of drugs becomes progressively not only impulsive but eventually compulsive. Several PET studies have established an association between the already described downregulation of the D2 receptors of the indirect basal ganglia pathway (Hikida et al., 2010) and decreased metabolic activity within the prefrontal cortex (PFC) (Volkow et al., 2001). This suggests that those changes explained by the neurocircuitry model of addiction might not occur linearly but in parallel. Goldstein and Volkow (Goldstein and Volkow, 2011) described a syndrome of impaired response inhibition and salience attribution (iRISA) produced by a dysfunctional PFC: a ventral sub-region, formed by the medial orbitofrontal cortex (mOFC), the ventromedial prefrontal cortex (vmPFC) and the ventral anterior cingulate cortex (vACC) would become more responsive to drugs or drug stimuli; a dorsal sub-region formed by the dorsolateral prefrontal cortex (DLPFC), the dorsal anterior cingulate cortex (dACC) and the inferior frontal gyrus (IFG) with a role in conflict resolution and regulation of the ventral subregion would instead become hypofunctional to drug-related stimuli and broader cognitive tasks. The net effect is the development of higher attribution of salience to drug-related stimuli alongside reduced inhibitory control which yields a more automatic response during decision-making processes involving drugs.

An fMRI study in n=65 male patients with MUD in early abstinence showed higher activation by drug>neutral cues in the PFC, amygdala, striatum, insula, and secondary visual processing regions (Ekhtiari et al., 2021). Another fMRI study of n=75 individuals with MUD assessed how visual cue-induced craving and negative emotion visual stimuli disrupts response inhibition measured with a Go/No-Go task (Dakhili et al., 2022) suggesting a potential shifting of neural resources to processing addiction related cues due to their higher salience.

In summary, the Koob and Volkow’s neurocircuitry model provides a heuristic explanation of the pathophysiology of addiction where the basal ganglia-dopaminergic system becomes hypersensitive to drug cues, the prefrontal cortex loses its ability to regulate the use of drugs and becomes hyper-reactive to drug-related cues and hypo-responsive when the control from higher order cognitive nodes is expected and finally the extended amygdala gradually pushes the individual to a state of dysphoria that potentiates negative reinforcement processes. Numerous neuroimaging studies in people with MUD, including cue-reactivity and resting state, provide support to this model (Bischoff-Grethe et al., 2017; Dakhili et al., 2022; Ekhtiari et al., 2021; Grodin et al., 2019; Jiang et al., 2021; Kohno et al., 2014; May et al., 2020; Proebstl et al., 2019; Weafer et al., 2020).

3. Potential Neurocircuitry effects of therapeutic TMS in Methamphetamine Use Disorder.

From the previous section, it follows that if therapeutic TMS is expected to be beneficial in MUD or in other substance use disorders, then it would need to correct a hyper-reactive dopamine system to internal or external drug related cues (stage 1), a basal hypodopaminergic tone responsible for dysphoria and hypo-hedonia (stage 2), especially on post-acute withdrawal symptoms, a sensitized stress system (stage 2) and/or an unbalance in higher cognitive centers, by activating the dorsal network responsible of inhibitory process or by inhibiting the ventral network reactive to drug-related cues (stage 3).

Most of the available studies in MUD have focused on the neurocircuitry effects induced by the PFC stimulation, either the dorsal or the ventral networks. This is in part due to the ample knowledge acquired using TMS in the field of depression, where the FDA-cleared protocol includes the stimulation of the left DLPFC with emerging data supporting its use in other disorders such as in PTSD (Philip et al., 2019). The following are the potential mechanisms that could explain some of the benefits of using TMS in MUD.

3.1. Reverting an imbalanced PFC functioning: effects on the anticipation/preoccupation stage

Functional MRI studies are giving us a deeper and more complete understanding of some of the changes in the brain resulting from processes of addiction. Resting-state functional connectivity (RSFC) studies identify those areas in the brain, such as the default mode network (DMN) that are functionally connected during a task-free state (resting). Different neuroimaging studies have found alterations in the DMN (within and between this and other networks) in patients with SUD (Zhang and Volkow, 2019). Other functional MRI studies include task-oriented imaging where different areas and circuits of the brain will activate depending on the cognitive or emotional task demands. The theoretical stimulation of the PFC with TMS, in line with the alterations described earlier in the preoccupation/anticipation stage, could theoretically lead to an activation of the dorsal executive control network (ECN), formed by the DLPFC and the posterior parietal cortex (PCC). This could revert what it is believed to be a general hypofunction or a reduced connectivity between this neurocircuitry and the DMN in processes of addiction (Su et al., 2020c; Zhang and Volkow, 2019). A recent study (Su et al., 2020c) in 60 individuals with MUD evaluated the effects of intermittent Theta Burst Stimulation (iTBS), a newer form of TMS (Huang et al., 2005), over the left DLPFC in resting state after 20 daily sessions. Compared to pre-treatment, there was an increase in functional connectivity between the left DLPFC and inferior parietal lobe and this correlated with craving reduction. The authors attributed this change to better top-down regulatory effects of the ECN in craving control.

Not only the DLPFC has attracted attention as a therapeutic target. As stated earlier when describing the iRISA syndrome (Goldstein and Volkow, 2011), the ventral PFC subregion becomes hyper-responsive to drug cues (increased salience) so another attempt to curve down methamphetamine use could consist on inhibiting this area when stimuli are present. A recent study (Chen et al., 2020b) in 74 individuals with MUD compared iTBS (putatively excitatory) over the L-DLPFC, continuous TBS (putatively inhibitory) over the left vmPFC, a combination of both, or sham, for a total of 10 sessions and found that either active approach was effective in craving reduction compared to sham, and the combined group responded faster than the individual approaches.

3.2. Improving the withdrawal/negative affect stage.

Pioneering studies in healthy controls have shown a regulatory function of the PFC in dopamine release. High frequency TMS (HF-TMS) over the left DLPFC produced a decrease in binding potential with raclopride in the ipsilateral caudate, an effect that can be explained by an increase in dopamine concentration (Strafella et al., 2001). These DA effects have also been studied in extra striatal regions such as the anterior cingulate cortex and the orbitofrontal cortex with the use of a different radioligand (Cho and Strafella, 2009). At present, it is unknown if PFC stimulation in MUD yields dopaminergic effects, but this would be of special interest considering the described hypodopaminergic state (within systems adaptations) that could contribute to a state of dysphoria and particularly relevant to early withdrawal from methamphetamine use.

Different studies have shown associations of methamphetamine use with elements that impact recovery such as sleep problems (Mahoney et al., 2014; Sun-Suslow et al., 2020) and depressive symptoms (Luo et al., 2022) (Glasner-Edwards et al., 2009). TMS performed over the left DLPFC was done in 50 subjects with MUD (active vs sham) in early abstinence from methamphetamine use (average 6 days, SD 4.4 days). This study showed benefits in craving reduction, quality sleep, depressive symptoms, and a trend towards withdrawal symptoms reduction in the active group (Liang et al., 2018b).

Changes in depression and sleep quality might thus be explained by between-systems adaptations following the neurocircuitry model of addiction (Koob and Volkow, 2010).

4. Potential efficacy of TMS treatment in Methamphetamine Use Disorder.

Limitations.

For this review, the PubMed database was searched with the terms “methamphetamine” and “transcranial magnetic” or “TMS” or “theta burst” (see Figure 2 for PRISMA (Page et al., 2021) representation). The initial search identified 38 articles. Only randomized clinical trials in methamphetamine use disorder were selected. Review articles, protocols, pilot data, commentaries or TMS indications for other than MUD were not included. The subsequent search identified 15 articles. Three of those studies did not use a sham group (Liu et al., 2019; Liu et al., 2020; Zhao et al., 2020), two of the studies used the same sample to assess different outcomes (Chen et al., 2020a; Su et al., 2020c), and two of them used a single TMS session to study cognitive changes (Liang et al., 2018a; Zhang et al., 2018). Thus, a total of 8 studies met our quality criteria (see Table 1).

Figure 2: Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram depicting study identification and selection process.

Table 1:

Randomized placebo-controlled clinical trials of transcranial magnetic stimulation studies for the treatment of methamphetamine use disorder.

Author	Sample	Intervention	Tools	Findings	Conclusions
Wang L-J et al. (2021)	N=131 Active:33 Sham:33v HC:65	iTBS 20 sessions (4 weeks) 900 pulses/day L-DLPFC	-Cue-induced VAS -BRIEF-A	-VAS effective -iTBS reduced BRIEF-A subscales -GEC, inhibition and WM related to craving changes. -GEC and WM predictive value.	iTBS had a higher effect on executive function of patients with MUD. ExFu could become a predictor of treatment response.
Su H et al. (2020a)	N=126 (20 W) Active:70 Sham:56 Multicenter	iTBS 20 sessions (4 weeks) 900 pulses/day L-DLPFC	-Cue-induced VAS -Sleep Quality PSQI (1 month) -CogState Battery (1 and 12 months): ISL and GML	-VAS effective. -PSQI effective -CogState: Effective month 1 not month 12.	iTBS improves cue-induced cravings. iTBS improves sleep quality. iTBS improves some cognitive domains only in the first month.
Su H et al. (2020b)	N=50 (20 W) Active:25 Sham:25	iTBS 20 sessions (4 weeks) 900 pulses/day 100%MT L-DLPFC	-Cue-induced VAS -CogState Battery (baseline and post):ISL and GML	-GABA/NAA decreased in iTBS. -Glx/NAA decreased in sham. -VAS effective. -CogState effective. -GML and GABA changes correlated.	iTBS might induce changes in GABA. Decreases in GABA correlated with problem solving function.
Yuan J et al. (2020)	N=73 MUD Active:37 Sham:36 N=33 HC	rTMS: 1Hz 10 sessions (2weeks) 100% MT L-DLPFC	-Cue-induced VAS -2-choice oddball task (baseline, post)	Improvements in VAS. Improvements in impulsivity, more accuracy and more reaction time delay.	Low frequency rTMS improved impulsive behavior in MUD.
Chen T et al. (2020b)	N=74 MUD iTBS:18 cTBS:18 iTBS+cTBS:18 sham:18	iTBS (L-DLPFC): 900 pulses/day cTBS (L-vmPFC): 900 pulses/day iTBS+cTBS: 1800 pulses/day Treatment for 2 weeks, 10 sessions.	-Cue-induced VAS -HAMD-17 -HAMA-14 -PSQI -AWQ -CogState Battery	-Time and TimeXgroup effects in VAS. -Faster response in combined group -Active groups: Lower HAMA, AWQ	iTBS (DLPFC), cTBS (vmPFC) and combined are effective in craving reduction. Combined group responded faster and better sleep indexes.
Liang Y et al. (2018)	N=50 men Active:24 Sham:24	rTMS 10 Hz 100% MT DLPFC 10 sessions	-MWSS -Cue-induced VAS -PSQI -Self-rating depression and anxiety scales.	Improvement in MWSS in both groups (rTMS>placebo) Improvement in VAS (rTMS) Improvement in PSQI (rTMS) Improvement in depression and anxiety (rTMS)	HF-rTMS over the DLPFC for 10 days is effective for reducing withdrawal symptoms, craving, sleep and depression and anxiety symptoms in MUD.
Liu Q et al. (2017)	N=50 Active:40 (n=10x4 groups) Sham:10	rTMS 10 Hz or 1 Hz 100% MT 5 Sessions L or R DLPFC	Cue-induced VAS	-VAS effective in either 1Hz or 10Hz treatment vs placebo in L-DLPFC or R-DLPFC.	rTMS in high or low frequency applied over the left or right DLPFC effective for cue-induced craving
Su H et al. (2017)	N=30 Active:15 Sham:15	rTMS 10 Hz 5 Sessions 80% MT L-DLPFC	-Cue-induced VAS -CogState Battery -PSQI -HAMD-17 -HAMA-14	-VAS effective -Increase verbal learning and social cognition.	rTMS improves cue-induced craving and some cognitive domains.

AWQ, amphetamine withdrawal questionnaire; CogState-GML, Groton maze learning task; CogState-ISL, international shopping list; cTBS, continuous theta burst stimulation; GABA, gamma amino butyric acid; Glx, glutamate; HAMA, Hamilton anxiety rating scale; HAMD, Hamilton depression rating scale; HC, healthy controls; HF-rTMS, high frequency repetitive transcranial magnetic stimulation; Hz, hertz; iTBS, intermittent theta burst stimulation; L-DLPFC, left dorsolateral prefrontal cortex; MT, motor threshold; MUD, methamphetamine use disorder; MWSS, methamphetamine withdrawal symptom scale; NAA, n-acetyl aspartate; PSQI, Pittsburgh scale quality index; R-DLPFC, right dorsolateral prefrontal cortex; rTMS, repetitive transcranial magnetic stimulation; VAS, visual analogue scale; vmPFC, ventromedial prefrontal cortex; W, women.

Most, but not all (see Liu Q (Liu et al., 2017) and Yuan J (Yuan et al., 2020)) of the published studies have used a pattern consisting of applying TMS over the left DLPFC at either high frequency (10 Hz) or in the form of iTBS. The most consistent finding is a reduction in craving with some studies demonstrating other effects such as an improvement in psychiatric symptomatology like depressive and anxiety symptoms (Chen et al., 2020b; Liang et al., 2018b), withdrawal symptoms (Liang et al., 2018b), quality of sleep (Chen et al., 2020b; Liang et al., 2018b; Su et al., 2020a), neuropsychological domains like problem solving (Su et al., 2020a; Su et al., 2020b), working memory (Su et al., 2020a; Su et al., 2020b), executive functions (Wang et al., 2021), verbal and social cognition (Su et al., 2020b; Su et al., 2017) and other constructs such as impulsivity (Yuan et al., 2020).

However, several limitations are important to highlight. The primary outcome in all these studies has been mainly focused on a reduction in craving. Surprisingly, there is a paucity of research on the link between craving and actual methamphetamine use (Hartz et al., 2001) (Galloway et al., 2010). Whether and how a reduction in craving translates into an actual reduction in drug use remains unclear and an important area of future research. Furthermore, it remains to be evaluated if other outcomes such as self-reported methamphetamine use, urine detection of metabolites of methamphetamine or other drugs provide a more realistic clinical outcome. Of note, most of these studies were performed in currently hospitalized patients, perhaps indicating some of the real-world difficulties engaging this population. Longitudinal outpatient follow-up and naturalistic studies are clearly needed.

Another important limitation has to do with the ability to generalize observed findings. To date, these interventional studies have evaluated individuals with MUD but excluded patients with other psychiatric comorbidities. This creates an immediate concern about validity, particularly because of the very high rates of depressive and anxiety disorders in MUD patients. To underscore this point, one study found a lifetime prevalence of anxiety disorders of up to 39% before the onset of the methamphetamine addiction (Zweben et al., 2004) and an Australian group reported rates of 40% of MDD in those individuals entering treatment for MUD (McKetin et al., 2011). Unfortunately, by excluding these individuals the efficacy of these treatments is only translatable to a subset of the patients with this serious condition.

Other important considerations have to do with the designs of the studies. Technical and biological variability play a role in the effectiveness of this treatment. Technically, most but not all the studies use high frequency (typically 10 Hz) to deliver the magnetic stimulation and most but not all the studies use 100% MT. However, the number of sessions and hence the total number of pulses delivered is highly variable. Su H (Su et al., 2020a) and Wang et al (Wang et al., 2021) used iTBS 20 sessions over four weeks, Yuan J (Yuan et al., 2020) used 1Hz TMS 10 sessions for two weeks, Chen T (Chen et al., 2020b) used 10 sessions of iTBS or cTBS for 2 weeks and Liu Q (Liu et al., 2017) used 5 sessions of 1Hz or 10Hz TMS for five days. A recent meta-analysis (Ma et al., 2019) on the use of non-invasive brain stimulation on stimulant craving in stimulant use disorder (both cocaine and MUD), found that TMS in the treatment of methamphetamine addiction was effective (N=10, Hedges’ g=1.5, CI=[0.7–2.3], z=3.7, p,0.001), but the number of sessions and the total number of pulses in TMS were not significant when all the studies on stimulant use disorders were analyzed together.

5. Future directions.

The methodological limitations explained earlier, including proper outcome selection, inclusion or not of comorbidities and technical design considerations, are inherent to any clinical study in TMS. We now discuss important considerations to take into account in MUD and in general in the addiction field with the hope to inform future studies.

5.1. Imaging guided treatment in TMS

Different anatomic factors, such as the distance of the cortex to the skull, could theoretically influence the distribution of the generated electrical field compromising the intended area of stimulation. Age and biological sex differences, for example, play an important role in the localization of the DLPFC (Mylius et al., 2013) and these and other factors might add variability when traditional techniques, such as the International 10–20 system (Klem et al., 1999) or the 5-cm rule (Herbsman et al., 2009) are being used to place the coil (Fabregat-Sanjuan et al., 2022).

None of the already discussed studies have used neuro-navigation techniques to localize the DLPFC which could add more precision in the location of the different intended areas to be modulated (Tik et al., 2023). Nevertheless, it is unclear the relative advantage of incorporating these techniques or in a broader sense neuroimaging data into the field of TMS, at least at present. Pioneering studies in depression suggested that the efficacy of TMS could be related to the intrinsic negative functional connectivity between the DLPFC and the subgenual cingulate cortex (sgACC) (Fox et al., 2012).

Since then, different studies have investigated the potential benefit of incorporating rsMRI data to better individualize this treatment in depression and other comorbid conditions (Philip et al., 2018). A recent study, however, using data from the largest published clinical trial of TMS (Blumberger et al., 2018), found that only 3% of clinical outcome variability could be modifiable by using rsMRI data from the sgACC (Elbau et al., 2023). This has led many authors (Siddiqi and Philip, 2023) to question the utility of using neuroimaging in clinical TMS at present.

Nevertheless and in the area of addiction in particular, neuroimaging could help identifying those areas that become hyper-reactive to the presentation of certain stimuli (internal or external) which could theoretically guide the choice of the preferred modulatory method (putatively excitatory or inhibitory) in an individualized way (see section 5.3 below). It is possible that the information obtained from activity-designed fMRI techniques could confer different treatment predictability compared to the information from rsMRI techniques, but this is an area that requires further research.

5.2. Other potential TMS targets in addiction:

5.2.1. The Salience Network (SN)

Most of the studies commented above have used the DLPFC as the main target area based among different factors on the possible beneficial effects of “activating” a hypofunctional executive control network in the preoccupation/anticipation stage. From a theoretical perspective, however, the SN, formed by different cortical nodes such as the anterior insula and the dorsal anterior cingulate cortex (Seeley et al., 2007) is central in the physiopathology of addiction. The SN plays an essential function in the detection of important interoceptive and external stimuli thanks to its connection to other areas relevant in the addiction pathology such as the VTA, the ventral striatum and/or the extended amygdala (Peters et al., 2016). These deeper structures are impossible to be targeted with the current non-invasive neuromodulation techniques, although some promising research with low intensity focused ultrasound might eventually lead to a change in paradigm (Philip and Arulpragasam, 2023). The SN (anterior insula and/or dorsal ACC) however is reachable with deep TMS and its neuromodulation could have an effect in these deeper structures and hence in the three different stages of the neurocircuitry of addiction (Harmelech et al., 2023).

The importance of the possible beneficial effects of modulating the SN is highlighted in a recent multicenter double blind clinical trial of TMS done in more than 250 people with tobacco use disorder (TUD) and previous unsuccessful quitting attempts (Zangen et al., 2021). The H4 coil used in this study (deep TMS) was hypothesized to activate the mPFC and the lateral PFC and insula bilaterally. The results of this study led to the clearance of the H4 coil as an aid in short-term smoking cessation by the FDA in 2020, first time a TMS modality is cleared for addiction medicine.

Connectome studies such as the recent “lesion network mapping” developed from patients with TUD that had achieved remission after having suffered a stroke in different cortical areas (Joutsa et al., 2022) have shed more light into the role of the SN. This study identified cortical areas positively connected to the dorsal cingulate, insula and lateral PFC as potential areas of interest in neuromodulation. Interestingly, the neuroimaging data obtained in the already commented multicenter study (Zangen et al., 2021) showed a decrease in insula activity and changes in the resting-state connectivity between the insula and the DMN after active TMS was applied with the H4 coil. In other words, it is not clear if a stimulatory or an inhibitory effect is needed within the SN, but these studies highlight the complexity and the indirect multi-effects when modulating different areas in the brain.

5.2.2. The frontopolar cortex

The frontopolar cortex, understood as the Brodmann area 10, is a vast region of the brain that includes areas in the frontal pole and the ventromedial cortex (Ongur et al., 2003) and it is involved among other functions in decision making (Koechlin and Hyafil, 2007) and cognitive flexibility. Evidence from lesion-based mapping, fMRI maps and different TMS studies points towards the frontopolar cortex as an important area worth to consider in the treatment of MUD and addiction in general (Soleimani et al., 2023). For example, the already mentioned lesion network mapping study (Joutsa et al., 2022) identified that tobacco remission was more likely after strokes in areas that had negative functional connectivity to the medial frontopolar and temporal cortices.

From a neuroimaging point of view, cue-induced exposure, thorough the view of pictures or images related to specific drugs or materials (Maas et al., 1998), seems to activate frontostriatal and frontolimbic networks that include regions such as the vmPFC (Goldstein and Volkow, 2011), ventral and dorsal striatum, anterior cingulate and the insula (Kearney-Ramos et al., 2018). The mPFC (Hanlon et al., 2018) as such has been found to be an important node across multiple substance use disorders. The ventral subregion, as described above in the IRISA model, is an important area involved in salience attribution and craving phenomenology. Kearney-Ramos (Kearney-Ramos et al., 2018) has indeed showed a possible transdiagnostic effect of this area in 50 subjects with alcohol or cocaine use disorder. After the use of a cue-induced craving paradigm, cTBS (1 session, 3600 pulses), a putatively inhibitory intervention, produced a decrease in functional connectivity between the vmPFC and regions of the striatum, the ACC, and the insula. A subsequent preliminary study in 19 patients with cocaine use disorder (Kearney-Ramos et al., 2019), showed that those subjects with higher striatal response to cocaine cues had a more positive response to cTBS (1 session, 3600 pulses) when applied to the mPFC as manifested in the form of attenuation in the striatal activity.

It remains to be tested in MUD whether vmPFC modulation could offer therapeutic benefit. There is only one published study done in 74 patients with MUD that shows that the use of cTBS in the vmPFC might be at least as effective as DLPFC modulation (Chen et al., 2020b). These options in any case have been more widely tested in other stimulant use disorders such as cocaine use disorder (Hanlon et al., 2017), but the neurotoxic effects of cocaine vs methamphetamine may be quite different (Jayanthi et al., 2021) including the possibility of seizures (Sanchez-Ramos, 2015). It could be possible for example that the seizure threshold and the motor cortex excitability differ in patients with MUD which would impact certain parameters including the intensity of the TMS delivery.

5.3. Activity dependence and Hebbian principles

Elucidating the net overall TMS effects in neuronal populations, those being excitatory or inhibitory in nature, is a challenging task as already commented previously. Both anatomical and technical factors (frequency and type of the applied pulses, intensity of the applied field) and other less obvious factors such as baseline “state” activity of the intended area of modulation are increasingly being recognized as fundamental elements in the overall efficacy of this treatment. Initial studies on motor regulation (Siebner et al., 2004) followed by studies on visual perception and behavior (Silvanto et al., 2008) demonstrated how the baseline activity of a certain neuronal network can influence the intended outcome of the TMS stimulation. In other words, an expected inhibitory effect could paradoxically become a facilitatory one if the area was previously inhibited (Silvanto et al., 2008). These effects are explained by homeostatic mechanisms and the presence of a balance between inhibitory and excitatory circuits in the general brain organization.

Different brain networks might be more susceptible for neuromodulation when previously activated. Studies in TUD have found that TMS (deep TMS with an H-coil) is effective in tobacco reduction when delivered after cue-induction of cravings (Dinur-Klein et al., 2014). Indeed, the FDA protocols for both OCD and tobacco cessation include symptom provocation. However, not every psychiatric condition might benefit from this potential activation as suggested in a PTSD study with deep TMS in n=125 outpatients (Isserles et al., 2021). There could also be a window-period after provocation or even after the use of TMS where the possibility of neuronal modifications is optimal, following Hebbian principles (Bear et al., 1987; Harmelech et al., 2023). Psychotherapeutic or environmental interventions and/or certain medications that might play a role in mechanisms of neuroplasticity could in theory activate or facilitate (but also hinder) the neuromodulation potential of TMS in certain neurocircuitries, but more research is clearly needed in this area.

5.4. Biomarkers and treatment prediction

There is an increasing effort led by the National Institute of Mental Health to move towards a more precise and individualized form of medicine which might involve the use of genetic information, biomarkers and lifestyle data to gather a deeper understanding of several complex multifactorial disorders, including of course substance use disorders (NIDA, 2018)

5.4.1. Using neuroimaging as a biomarker

It is still unclear whether individual functional connectivity relationships may predict TMS treatment responses in MUD. We hypothesize that differences in the degree of intra- or inter-connectivity between the ECN, the SN and other areas of the brain could be an important factor in predicting treatment response (see general review by Zhang and Volkow (Zhang and Volkow, 2019)). Like in the resting state studies, it is also unknown if a higher reactivity to drug-cue stimuli could represent a useful biomarker for TMS response in people with MUD.

The findings in structural MRI have been inconclusive based on multiple factors that can affect variance such as sex differences, age, length of abstinence, use of other drugs, use of medications, or psychiatric comorbidities. For example, no specific differences in gray matter volumes were found in a univariate analysis of more than 200 patients with MUD when compared to controls (Mackey et al., 2019). A similar finding (lack of significant differences) was found by the ENIGMA addiction group (Chye et al., 2020). On the other hand, a cross-sectional T1-MRI study showed positive relationships between duration of abstinence for up to 2 years in males with MUD with volumes of the OFC, parietal cortex, and the hippocampus (Nie et al., 2021). In relation to white matter changes, the ENIGMA group has described changes manifested by lower fractional anisotropy in tracts related to cognitive control, impulsivity, and emotion regulation such as the cingulum, genu and splenium of the corpus callosum and superior longitudinal fasciculus (Ottino-Gonzalez et al., 2022). We hypothesize that, in line with changes in some neuropsychological domains such as executive functions (see below), successful treatment outcomes with TMS could correlate with increases in gray matter volumes in some prefrontal cortical regions or with normalization of parameters in certain white matter tracts. If this is the case, then pre-treatment differences in these biomarkers could predict the efficacy of this treatment.

5.4.2. Neuropsychological Deficits

People with MUD suffer from deficits in different cognitive domains, especially in social cognition and impulsivity reward/processing, but also in attention, working memory, verbal and visual memory, speed processing and visuo-spatial abilities (Potvin et al., 2018). As in the general addiction field, it is unknown whether these deficits are predisposing factors or neuropsychological effects of methamphetamine use and, hence, partially or totally reversible with long term abstinence time (Basterfield et al., 2019).

A study by Chen (Chen et al., 2020a) in n=90 subjects found, for example, that those patients with MUD who had better spatial working memory (measured by the Groton Maze Learning Test) had also higher craving reductions in response to methamphetamine cues when 20 daily iTBS sessions was used over the L-DLPFC. A recent study by Wang et al in n=66 patients with MUD (Wang et al., 2021) (already described above) has found that the Working Memory and the Global Executive Composite subscales of the BRIEF-A executive function battery had both predictive power to iTBS therapeutic efficacy, defined as more than 2 points change in the VAS.

5.4.3. Peripheral Biomarkers

Methamphetamine effects extend far beyond its primary actions in the synaptic cleft and different groups in neuroscience are describing actions in different molecular cascades including oxidative stress, neuro- and excitotoxicity, and neuroinflammation (Yang et al., 2018). A study focused on possible plasma metabolomics biomarkers, more specifically in peripheral blood mononuclear cells, in patients with MUD found that an active vs sham treatment with TMS was associated with lower levels of alpha-tocopherol, glyceric acid and fumaric acid (Su et al., 2022). These intermediates have antioxidant properties via direct or indirect mechanism of actions. The area of peripheral biomarkers is vast and very preliminary but worth to explore based on its convenience (easy access) and potential applications.

6. Conclusions.

In the same way that the introduction of electroconvulsive therapy was revolutionary in the field of Psychiatry, after a gap of several decades, we might be witnessing another revolution in non-invasive neuromodulation with the creation of more precise techniques such as TMS. This coincides with a change in the conceptualization of addiction as a brain disease with well-defined neuronal circuitries and hence open to the possibility of intervention. The neurocircuitry model of addiction (Koob and Volkow, 2010) has defined three different stages in the pathophysiology of addictive disorders, each one supported by a different psychological theory, corresponding to three different areas in the brain. MUD, and other stimulant use disorders, is one of the most difficult to treat addiction pathologies as evidenced by the paucity of effective psychotherapeutic and pharmacological interventions (Ballester et al., 2017; Paulus and Stewart, 2020). The most evidence-based interventions in MUD consist at present on psychosocial interventions, especially the use of CM, cognitive behavioral therapy or a combination of CM with social interventions (community reinforcement approach) (De Crescenzo et al., 2018). There is some evidence that the use of pharmacological interventions, such as a combination of the long acting naltrexone preparation and daily bupropion (Trivedi et al., 2021), or the use of mirtazapine (Coffin et al., 2020) might aid in decreasing the use of methamphetamine. Recently the American Society of Addiction Medicine (ASAM) and the American Academy of Addiction Psychiatry (AAAP) have released a clinical practice guideline on the management of stimulant use disorder where different clinical factors such as the presence of comorbidities are taken into account (ASAM/AAAP, 2024).

Still and despite decades of research, methamphetamine addiction is an elusive and extremely difficult to treat condition where, as stated above, potential treatments have only shown modest efficacy. Consequently, there is a marked interest in the use of novel intervention and neuromodulation techniques, particularly TMS, is showing some promise due to the low profile of side effects and the possibility of an individualizing approach.

Most of the different published studies have found TMS to be effective in decreasing craving for methamphetamine use when applied to the left DLPFC (see Potential Efficacy of TMS Treatment in MUD section above), in line with a possible change in PFC functioning as proposed in the preoccupation/anticipation stage. Other studies are also showing that the frontopolar cortex could as well be susceptible for intervention (Soleimani et al., 2023). The FDA-clearance of deep TMS in tobacco use disorder (Zangen et al., 2021) has opened the door to the possibility of modulating the SN, an intervention theoretically supported by different connectomic, lesion-based (Joutsa et al., 2022) and neuroimaging studies. All of these interventions would theoretically lead to an improvement in the preoccupation/anticipation stage and so those patients who are afflicted by craving intensity/frequency could benefit the most.

Within the withdrawal/negative affect stage, current TMS techniques do not permit neuromodulation of the extended amygdala, due to the depth of its location and the possibility to modulate other unintended regions (trade-off between depth and specificity). Other non-invasive neuromodulation techniques such as low intensity focused ultrasound is a promising intervention and could in theory circumvent this problem but the evidence in this area is very preliminary (Philip and Arulpragasam, 2023). Still, the effects of modulating certain nodes extend beyond their initial location and different studies are showing the potential benefits of stimulating the DLPFC in MUD in sleep (Chen et al., 2020b; Liang et al., 2018b; Su et al., 2020a), withdrawal (Liang et al., 2018b) and depressive symptoms (Chen et al., 2020b; Liang et al., 2018b). In other words, an intervention intended initially to target the anticipation/preoccupation stage could as well lead to an improvement in other stages, which speaks about the interconnectivity of the brain and the need for further research.

It is possible that certain patients with MUD might need different interventions (frontopolar, salience network, DLPFC or a combination of them) depending on the severity of their addiction stage or the pathological predominance of certain stages. We believe that the use of biomarkers (neuroimaging, peripheral, neuropsychological) would aid in individualizing and optimizing their treatment. For example, those patients with higher predominance of the withdrawal-negative affect stage might benefit from different forms or different areas of stimulation, or even the same patient might require different interventions depending on the longitudinal stage of the addiction process.

We are pending to see the applicability of these techniques to real world clinical scenarios, such as in outpatients with different comorbidities and it is unknown if this will become a treatment by its own or an augmenting modality to other treatments. For example, certain medications or psychotherapeutic techniques might activate or not different neurocircuitries, opening the window to a more effective modulation following Hebbian neuroplasticity mechanisms. We believe that due the complex nature of addiction, a combination of different treatments (psychotherapeutic, psychopharmacological and neuromodulatory) will be required to individualize and optimize outcomes. In the case of TMS, due to the benign profile of side effects, and the more recent improvements in its administration and time commitment (accelerated TMS), its clearance by the FDA could in any case advance and improve the treatment of this population in desperate need.

Highlights.

• Despite decades of research, there is no clear evidence for specific pharmacological interventions for MUD and psychosocial interventions only have modest long-term efficacy.

• Neuromodulatory techniques and TMS in particular are promising interventions in the area of addiction medicine capable of individualize treatments within a neurocircuitry approach.

• The neurocircuitry model of addiction has defined three different stages in the pathophysiology of addictive disorders, corresponding to three different areas in the brain.

• The published clinical trials using TMS for MUD have so far focused on the neurocircuitry effects induced by the PFC stimulation, either the dorsal or the ventral networks and its clinical correlate in craving reduction.

• In the area of addiction, some studies are now showing that the frontopolar cortex could as well be susceptible for intervention, and other studies, in line with the recent FDA-clearance of deep TMS in tobacco use disorder, are highlighting the role of the salience network.

• Different biomarkers (neuroimaging, peripheral, neuropsychological) could aid in individualizing this treatment. It is possible that certain patients with MUD might need different interventions (frontopolar, salience network, DLPFC or a combination of them) depending on the severity of their addiction stage or the pathological predominance of certain stages.