Effects of 10 add‐on HF‐rTMS treatment sessions on alcohol use and craving among detoxified inpatients with alcohol use disorder: a randomized sham‐controlled clinical trial

Abstract Background and aims Alcohol use disorder (AUD) is a chronic disorder with high relapse rates. There are currently few clinical trials of high frequency repetitive transcranial magnetic stimulation (HF‐rTMS) to reduce alcohol use among AUD patients, and results are mixed. The current study tested the effect of 10 add‐on sessions of HF‐rTMS over the right dorsolateral pre‐frontal cortex (DLPFC) on alcohol use and craving. Design Single‐center, single blind sham‐controlled parallel‐group RCT (n = 80), with 3 and 6 months follow‐up. Setting Clinical treatment center in Amsterdam, the Netherlands. Participants Eighty detoxified and abstinent AUD inpatients in clinical treatment (20 females, average age = 44.35 years). Intervention Ten sessions of active or sham HF‐rTMS (60 10 Hz trains of 5 sec at 110% motor threshold) over the right DLPFC on 10 consecutive work‐days. Measurements The primary outcome measure is the number of abstinent days over 6‐month follow‐up (FU). Secondary outcome measures are craving over 6‐month FU (alcohol urge questionnaire and obsessive‐compulsive drinking scale), time to first relapse over 6‐month FU and grams of alcohol consumed over 6‐month FU. Additional outcome measures: full abstinence over 6‐month FU and treatment success over 12‐month FU. Findings HF‐rTMS did not affect the number of abstinent days over 6 months FU [sham = 124 ± 65.9 days, active = 115 ± 69.8 days, difference: 9 days, 95% confidence interval (CI) = Poisson model: 0.578–3.547]. Moreover, HF‐rTMS did not affect craving (AUQ/OCDS) (sham = 15.38/5.28, active = 17.48/4.75, differences = 2.1/−0.53, 95% CI mixed‐effects model = −9.14 to 2.07/−1.44 to 2.40). Conclusions There was no clear evidence that high‐frequency repetitive transcranial magnetic stimulation over the right dorsolateral pre‐frontal cortex treatment has a long‐term positive effect on alcohol use or craving as add‐on treatment for alcohol use disorder. High treatment response at 6‐month follow‐up could have limited the possibility to find an effect.


INTRODUCTION
Alcohol use disorder (AUD) is frequently a chronic relapsing disorder in those in addiction treatment, characterized by loss of control over intake, despite awareness of the harmful consequences [1]. Furthermore, a negative emotional state can arise when alcohol is not available [2,3], and the mental health of the patient can be severely affected with comorbidities such as depression, anxiety and insomnia [4]. Often AUD patients experience intense urges or an irresistible desire to consume alcohol, also known as craving [5][6][7]. It is a world-wide health problem, with 2.6% of the world population meeting criteria for AUD [8]. Each year in the Netherlands alone, already more than 30 000 individuals are hospitalized for the treatment of AUD [9].
Current AUD treatment mainly consists of psychological and pharmacological interventions [10]. Pharmacological treatment consists of anti-craving medication, such as naltrexone and acamprosate [11,12]. Some of the available psychological treatments are cognitive behavioral therapy (CBT) (focusing upon improving problem-solving skills, enhancing drinking refusal skills and developing effective coping strategies by cognitive-behavioral treatment methods), motivational enhancement therapy (MET) (with a focus on enhancement of motivation to change behavior through motivational interviewing) and acceptance and commitment therapy (ACT) (which focuses upon acceptance of emotions and commitment to values that are set at the beginning of treatment) or contingency management (consisting of providing rewards when patients remain abstinent) [13]. The effectiveness of MET, CBT and ACT in alcohol and substance use disorder is small to moderate when compared to no treatment [14][15][16]. The effectiveness of contingency management is moderate when provided during treatment; however, this effect is abandoned after longer follow-up periods [17,18]. Despite these various treatment options, relapse rates are still approximately 60% 1 year after treatment [19,20]. These high relapse rates indicate a need of novel treatment options for AUD.
From a neurobiological perspective, AUD is characterized by neuroplastic changes in several brain circuits. The executive function, reward and stress circuits are usurped by drugs of abuse, including alcohol, via multiple neurotransmitter-specific neuroplasticity circuits [21]. For example, cue-exposure can elicit recruitment of the reward circuit, which has been associated with subjectively experienced craving (reflected by increased neuronal activity in the reward circuit) [7]. Moreover, reductions in executive functioning in combination with decreased pre-frontal cortex activity is often found in patients with AUD [21,22]. This can lead to loss of control over the reward circuit, and thereby over impulses and behavior. Together, loss of control over the reward circuit and diminished executive control often results in issues managing craving and consequently relapse [1,23].
Transcranial magnetic stimulation (TMS) has gained attention as a new treatment option for substance use disorders, including AUD [24], because it has the potential to interfere with the disturbed neurobiology of these conditions. With TMS, magnetic pulses are derived from an electromagnetic coil wherein an alternating current is running.
When placed over the skull the magnetic pulse induces a current in the underlying neuronal tissue [25]. When pulses are applied in multiple trains, it is referred to as repetitive TMS (rTMS). Depending upon whether the frequency is low (LF) or high (HF), stimulation is inhibitory or excitatory, respectively [26][27][28].
In substance use disorder studies, HF-rTMS is often applied over the dorsolateral pre-frontal cortex (DLPFC) [29], as it is part of the executive function circuit [30]. The hypothetical effect of stimulation of these superficial brain regions is, first, to increase pre-frontal cortex activity and strengthen the hypofunctional pre-frontal cortex, thereby increasing executive control [31,32]. Secondly, by indirect divergence into deeper structures of the brain, such as the nucleus accumbens (NAcc) and ventral tegmental area (VTA) due to direct connectivity between the DLPFC and the mid-brain, rTMS influences subcortical neuronal activity [33]. The changes in activity in these areas may potentially decrease craving and decrease the chances for relapse [24,32,34]. rTMS has been shown to increase emotion regulation in AUD [35] and to influence functional connectivity in AUD [36]. Thus, neuromodulation of the DLPFC could primarily strengthen the functioning of the PFC by increasing executive control. Secondarily, it could modulate activity and dopamine release in subcortical areas such as the NAcc and VTA which could decrease craving. Together, this may ultimately lead to decreased relapse rates, when applied in clinical trials in AUD [1,31].
Thus far, studies addressing the effect of HF-rTMS treatment on clinical substance use outcome measures, including relapse and alcohol use, in AUD are non-existent. This is surprising, as relapse is suggested as the primary outcome measure for clinical trials [29,37].
However, previous studies on the effect of HF-rTMS on craving in various substances are abundant (for reviews see [38][39][40]). Few studies focus upon alcohol as a substance of abuse. Some of these studies applied one single session of HF-rTMS over the right DLPFC, which did not result in decreased craving [41][42][43]. Several studies applied multiple sessions of HF-rTMS over the DLPFC. In one study, 10 sessions of HF-rTMS over the right DLPFC resulted in decreased craving immediately after stimulation, but not at 4-week follow-up [44]. In addition, two other studies with 10 HF-rTMS sessions (left DLPFC) [45] or 15 accelerated (multiple sessions on 1 day) HF-rTMS sessions (right DLPFC) [43] did not observe reduced craving. Concluding, the effect of multiple sessions of HF-rTMS over the DLPFC on craving is unclear, especially after a longer follow-up period.
Considering the lack of studies addressing the effect on subsequent alcohol use, and the mixed results of HF-rTMS on craving, the current study applied 10 add-on sessions of HF-rTMS over the right DLPFC. We included hospitalized abstinent AUD inpatients in a randomized sham-controlled clinical trial design to test whether HF-rTMS would lead to decreased alcohol use and less craving. We hypothesized that the HF-rTMS add-on treatment increases number of abstinent days and decreases experienced craving. Thus, we expected a higher number of abstinent days in the active group compared to the sham group. Furthermore, we expected lower craving levels in the active group compared to the sham group.

Study design
In this single-center, parallel, single-blind RCT, the effects of 10 sessions of add-on HF-rTMS treatment on alcohol use and craving was assessed. We included 80 hospitalized abstinent recently detoxified AUD inpatients, which were randomly assigned (1:1) to either treatment as usual (TAU) plus 10 sessions of active HF-rTMS treatment or TAU plus 10 sessions of sham HF-rTMS treatment on 10 consecutive work-days as add-on treatment. TAU consisted of behavioral therapy, being either cognitive-behavioral therapy (CGT) or acceptance and commitment therapy (ACT). Some patients also received additional psychopharmacological treatment (see Table 1 for an overview). Three months (90 days), 6 months (180 days) and 12 months (360 days) after the final HF-rTMS treatment sessions patients were contacted by telephone to assess outcome measures.
As described in our protocol [46], the primary outcome measure was number of abstinent days over 6 months, secondary outcome measures included craving over 6 months, time until first relapse over 6 months and grams of alcohol consumed over 6 months. Additional outcome measures were complete abstinence over 6 months and treatment success over 12 months. This protocol was approved by the Medical Ethical Committee of the Academic Medical Centre Amsterdam (2015_064) and registered in the Netherlands Trial Register (NTR) (number 5291). The design of this study, and all the aimed primary and secondary outcome measures and analyses, have been described elsewhere in more detail [46]. For a schematic overview, see Supporting information, Figure S1. comorbidities and other patients' problems that occurred during treatment. Additional weekly CBT or ACT mentor sessions were given, focusing upon maintaining abstinence. Finally, pharmacotherapy was prescribed to a subset of patients (see Table 1

Sample size
Sample size was pre-registered [46]. Because at the time this study was the first, to our knowledge, to investigate the effect of multiple HF-rTMS sessions on the number of abstinent days across 6 months, we could not perform a proper power analysis and our estimate was based upon previous rTMS studies which used craving as primary outcome that showed a moderate effect size. Hence, in this study, with the current sample size, a power of 0.8 and a moderate effect size, we should be able to detect a difference of 35 abstinent days between groups.

Inclusion and exclusion
Inclusion criteria were as follows: (1)
Single-pulse TMS over the motor cortex was used to determine the resting MT. When five of 10 stimuli resulted in a muscular (left abductor pollicis brevis) response of the thumb muscular abduction, this stimulation intensity was taken as MT. The actual treatment was given at an intensity of 110% of the MT. During the sham intervention identical procedures and stimulation parameters were applied; however, the coil was tilted 90 relative to the scalp.

Pre-intervention participant characteristics
The following sample characteristics were assessed: age, gender,

Secondary outcomes
As described in Schluter et al. [46], the secondary outcome measure 'craving' was measured by means of the alcohol urge questionnaire (AUQ) [52], assessed at baseline, post HF-rTMS treatment and at 3and 6-month follow-up. Higher AUQ scores signal higher craving levels. The additional measure regarding craving was measured with the short version of the obsessive-compulsive drinking scale (OCDS) [53], which contained five questions. Higher OCDS scores were indicative of higher craving levels.
The secondary outcome measures regarding alcohol use were as follows: (1) grams of alcohol consumed across the 6-month follow-up period (from days 1 to 180 after the final HF-rTMS treatment).
Patients indicated the day and quantity of the type of beverage they consumed on the TLFB during a telephone interview at 3-and 6-month follow-up; (2) number of days until first relapse, which was counted from the TLFB assessed at 3-and 6-month follow-up.
Relapse was defined as a heavy drinking day (> 60 g alcohol per day for men/ > 40 g of alcohol per day for women) [54]. For details on censoring, see the Analyses subsection.

Additional outcomes
The additional measures of alcohol use were as follows: (1)  participant did not answer the telephone during the 3-month followup, they were called at 6-month follow-up. Therefore, in some cases, data were not collected at 3-month follow-up, but at a later moment (6 months or later); for some patients we only have data at 3-month follow-up but not at 6-month follow-up. In those patients who were reached later, the craving questionnaires were not assessed as these were outdated, but measures derived from the TLFB were obtained.
When patients were lost to follow-up, but were later re-admitted to Jellinek Addiction Treatment Center, they were then asked to fill in the TLFB. Again, in these cases no craving questionnaires were assessed.
For an overview of missing data for the primary outcome measure, see Figure 1. measures, see Supporting information, Tables S1 and S2. As being lost to follow-up could have various reasons (e.g. either changed telephone number to part from contacts related to their addictive disorder in the past, or relapsed to daily drinking), we assumed a missing at random (MAR) data process. For our primary intent-to-treat analyses we assumed a worst-case scenario in which participants with missing data had returned to daily drinking, but have also performed additional complete case analyses in which we take into account the data as they are. In all analyses similar results were found, suggesting that our findings do not depend upon differences in missingness between groups.

Primary outcome
The primary outcome measure of total number of abstinent days across the whole 6-month follow-up period was calculated as a single

Secondary outcomes
First, to determine whether craving at baseline of the active group differed from the sham group, craving scores of the baseline session were compared between groups using a Mann-Whitney U-test.

Safety and tolerability
To assess whether the reported side effects differed between the treatment groups, χ 2 tests (or Fisher's exact tests in case the expected counts were less than 5) were performed for each reported side effect separately. A P-value of < 0.05 was considered significant.

Blinding
In order to determine whether blinding succeeded, percentage of individuals who guessed their treatment allocation correct was calculated.
A binomial test was used to determine whether this percentage significantly differed from chance level (50%). A P-value < 0.05 of was considered significant.

Sample characteristics
In total, 100 individuals were screened for eligibility, from whom 82 were included. Before the start of the add-on treatment two individuals withdrew their informed consent. Therefore, 80 participants started the study (see Figure 1, Table 1).

Dropout and data loss during treatment and follow-up
As some participants were lost to follow-up, we aimed to first compare the completeness of data regarding the primary outcome between the treatment groups, where we split up participants depending on whether they had complete or incomplete data on the primary outcome. A χ 2 test showed that there were no differences in completeness of data between the treatment groups (nine incomplete cases in the active group, 13 incomplete cases in the sham group; χ 2 = 0.5643, P = 0.452). Thus, importantly, there are no differences in dropout or data loss between the groups, which could potentially explain our further findings. Moreover, when testing for differences in baseline characteristics between the complete data and incomplete data group, no significant differences were found that survived multiple testing corrections.

Number of abstinent days
The primary intent-to-treat analysis using a generalized mixed-effects model showed that treatment group did not predict the number of abstinent days (  Figure 2b).

Grams of alcohol
The mixed-effects model showed that there was no effect of group, a significant effect of time (showing more grams of alcohol over time), but no interaction effect between group and time on grams of alcohol consumed (Table 4a). Our second analysis, using a Mann-Whitney U-test, also showed no significant differences in complete cases in terms of grams of alcohol consumed between the active treatment group and the sham treatment group, with an effect size of 0.0587, indicating that 5.8% of the variance in grams of alcohol consumed was explained by type of add-on treatment (Table 4b).
Days until first relapse The Cox regression showed that days until first relapse did not significantly differ between the active treatment group and the sham group, showing that there was no increased hazard for relapse in the sham T A B L E 2 Results of the generalized linear mixed-effects models regarding the effects of treatment group on the primary outcome measure of number of abstinent days.  Figure 3).

Full abstinence
The χ 2 test in the intent-to-treat analysis did not show a significant difference in full abstinence between the active treatment group (nine reached full abstinence, whereas 31 did not) and the sham treatment group (seven reached full abstinence, whereas 33 did not) (Table 4e).
Similar results were found with complete case analysis (Table 4f). The effect size indicated that approximately 3% of the variance in full abstinence rate was explained by an effect of add-on HF-rTMS (Table 4f).

Treatment success
Furthermore, Fisher's exact test showed in the intent-to-treat analysis that treatment success (measured at 12-month follow-up) did not  differ significantly between the active treatment group (and the sham treatment group (Table 4g). Similar results were found with complete case analysis (Table 4h). Effect size showed that approximately 2% of the variance in treatment success was explained by an effect of the add-on HF-rTMS treatment.

Safety and tolerability
Patients in the active group reported headache after stimulation, pain or beep in the ear, tiredness after stimulation and uncomfortable sen-  Table S3).

Blinding
In total, 68 patients reported their treatment beliefs, of which 39 patients (57,4%) believed to have received active treatment.
Approximately two-thirds of participants (63.2%) guessed their treatment allocation accurately. The binomial test revealed that the proportion of patients who guessed their treatment allocation correctly (0.63) was significantly (P = 0.038) higher than chance level (0.50).

DISCUSSION
The aim of the current study was to test whether 10 add-on sessions of HF-rTMS over the right DLPFC of hospitalized abstinent AUD patients would reduce alcohol use and diminish experienced craving.
Results of the current study did not show significant differences in alcohol use between the active and the sham group at 6-month follow-up, nor a significant difference in treatment success at 12-month follow-up. Furthermore, results of the current study did not show an effect of session or treatment group, nor an interaction of the two, on experienced craving levels. This suggests no beneficial add-on effect of active HF-rTMS treatment compared to sham treatment on relapse and craving in AUD patients receiving treatment as usual. This was confirmed by low effect sizes and Bayesian analyses indicating lack of effects and moderate evidence for the null hypotheses (i.e. no effect of HF-rTMS treatment).
The current study was one of the first, to our knowledge, to address the effect of HF-rTMS treatment on alcohol use outcome measures such as number of abstinent days and relapse, which is surprising, as relapse is suggested as primary outcome measure of clinical trials [29,37]. Contrary to our hypotheses we did not find evidence for a beneficial effect of HF-rTMS add-on treatment on alcohol use and relapse rates. repeated-measures analysis approach should be used to enable valid conclusions to be drawn [61]. Another small pilot study (n = 11) using deep rTMS on the bilateral DLPFC found that drinking behavior improved in the active treatment group, but not in the sham group [62]. However, again no direct statistical tests were performed to test for a group × time interaction effect. Both these studies were pilot studies, and thus results are preliminary. A larger study targeting the insula using deep rTMS in 56 AUD patients showed no evidence of a group × time interaction on alcohol use, and thus showed no effect of the active over the sham condition on alcohol use over a period of 12 weeks' follow-up [63]. A recent proof-of-concept study (n = 51) showed that dTMS targeting mid-line frontal areas reduced heavy drinking days in the active versus sham condition over a period of 12 weeks without an interaction effect with time [64]. Overall, most studies reported a lack of a positive effect of HF-rTMS add-on treatment on drinking behavior and alcohol use in the long term, which is in line with the findings of this larger RCT. Studies addressing the effect of HF-rTMS over the DLPFC as add-on treatment in other substances of abuse also report mixed results. For example, two studies indicated an acute reduction of cigarette consumption [65,66] and one study reported higher abstinence from smoking at 3-month follow-up [67]. Furthermore, two pilot studies in cocaine use disordered patients showed mixed results, one indicating acute beneficial effects of HF-rTMS treatment compared to TAU [68], but the other not reporting long-term (6 months) effects of HF-rTMS treatment compared to sham [69]. Together, our findings are in line with the negative findings reported in the relatively larger studies in AUD [63] and studies conducted in nicotine use disorder [65] and cocaine use disorder [69]. However, our findings are contrary to other smaller studies that were either pilot studies [60,62] or did not report on follow-up measures of alcohol use [59]. Furthermore, results of the current study are hard to compare to other studies as they did not include follow-up periods [66,68]. Hence, currently, there is limited evidence to suggest that HF-rTMS treatment has a positive clinical treatment impact in AUD; however, in nicotine use disorder, the first results are more promising [67].
F I G U R E 3 Kaplan-Meier plot of relapse [relapse was defined as a heavy drinking day (> 60 g alcohol per day for men/> 40 g of alcohol per day for women)] in both treatment groups. The x-axis represents follow-up time in days after the treatment and the y-axis shows the proportion of participants who have experienced a relapse.
The upward trend indicated that over time more participants (a larger proportion) experienced a relapse. The red line represents the active treatment group, whereas the blue line represents the sham treatment group. No differences were found between the treatment groups With regard to craving, the current study also did not find an effect of 10 HF-rTMS sessions on acute and long-term (up to 6 months) experienced craving. These results are in line with earlier studies showing no evidence for an effect of HF-rTMS on craving in AUD patients. A relatively small RCT (n = 19) showed no acute craving differences between active and sham rTMS on the left DLPFC [70].
Moreover, multiple trials with sample sizes ranging from 24 to 31 participants showed no effects of rTMS on the right DLPFC on acute craving [41][42][43]. Del Felice and colleagues [58], in a small sample (n = 17) showed no improvements of acute craving or craving after 1-month follow-up using rTMS on the left DLPFC [58]. A larger study (n = 39 patients) using rTMS on the right DLPFC also failed to find an acute effect on craving [35]. Contrary to these null findings, two studies using rTMS on the right DLPFC reported positive results for craving: Mishra and colleagues reported reduced acute craving, whereas Belgers and colleagues showed a group × time interaction, showing decreased craving in the active group up to 3 months' follow-up [44,59]. Differences in study samples could have had an influence on the outcomes, as Belgers et al. [59] collected data from outpatients whereas the current study only included inpatients. Other neuromodulation methods also show mixed findings in AUD; no effects of theta burst stimulation (TBS) targeting the medial PFC were found on acute craving (n = 24) [71] and, similarly, no effects on acute craving were found using TBS targeting the ventromedial PFC (n = 24) [72].
Also, a lack of effect on acute craving using deep TMS was found by Addolorato and colleagues [62]. Deep TMS over the insula also failed to improve craving over a 12-week follow-up period in a fairly large sample (n = 56) [63]. However, Ceccanti and colleagues showed a decrease in craving in the active condition up to 1-month follow-up which they did not find in the sham condition, but again did not directly test the group × time interaction effect [60,61]. Finally, Harel and colleagues [64] found that mid-line dTMS reduced acute craving without any long-term follow-up effects.
Several studies in other substances (cocaine and methamphetamine) have shown acute effects of HF-rTMS on craving [73,74]; however, these effects were not found at 6-month follow-up [68].
Therefore, future studies with longer-term craving measures are required in order to be able to conclude whether or not HF-rTMS add-on treatment may have beneficial effects in reducing experienced craving.
Despite the absence of effect in our study, the potential effectiveness of rTMS in AUD could be studied with other parameters.
Other methods of applying rTMS might be more promising for substance abuse, such as deep rTMS [60,62,63]. First, deep rTMS also resulted in decreased acute craving in patients with cocaine use disorder; however, no longer follow-up periods were conducted [75].
Secondly, changing the target location may result in increased effectiveness of the HF-rTMS treatment. The one study that reports longterm beneficial effects on abstinence rates in nicotine dependence stimulated the left instead of the right DLPFC [67]. This is in line with a meta-analysis that found favorable acute effects only for left-sided DLPFC stimulation [76]. However, right-sided stimulation has previously also been found to tend to have a bigger effect size [77], and another meta-analysis found that active stimulation was only favorable over sham stimulation in trials focusing on the right DLPFC [78].
Thirdly, the method used to assess craving might also determine whether an effect is found. For example, studies that applied rTMS after provoking craving by showing pictures of substance-related cues [74], touching instruments related to use or smelling and using fake substances [73] found The results of the current study should be seen in the light of several limitations. For instance, in the current study, relapse as well as craving were measured by means of self-report. This required patients to remember whether they have been drinking, what day and how much for the past 3 months, which might appear to be unreliable.
However, in practice, patients were very well able to recall when, and how much, they were drinking [80,81]. In addition, in the current sham condition, wherein the coil is tilted away from the scalp, stimulation of neuronal tissue does not take place, but also the somatosensory effect of stimulation is removed [82]. Indeed, unpleasant sensations of the skin after stimulation occurred less in the sham group. As a result, patients were able to identify above chance level to which of the treatment groups they were allocated. Although results of the current study were not affected by this, as no placebo response was detected, future studies should consider using a sham coil [83] which mimics somatosensory effects of stimulation. Moreover, the single -blind design could also be considered a limitation; however, minimal to no influence is expected on the outcome measure as the outcome measures were self-reports on which the researcher had no influence.
Taken together, we found no evidence of an effect of 10 sessions of HF-rTMS treatment over the right DLPFC on relapse rates and experienced craving in hospitalized AUD patients. These findings suggest that HF-rTMS over the right DLPFC might not have added value in the treatment of these patients, despite some indications of efficacy in nicotine use disorder [67]. Future studies must determine whether other means of measuring craving, other stimulation techniques and other target areas do improve relapse rates and experienced craving.

CLINICAL TRIAL REGISTRATION DETAILS
This clinical trial is registered in the Netherlands Trial Register, trial number 5291: https://www.trialregister.nl/trial/5151.