Two Weeks of Image-guided Left Dorsolateral Prefrontal Cortex Repetitive Transcranial Magnetic Stimulation Improves Smoking Cessation: A Double-Blind, Sham-Controlled, Randomized Clinical Trial

Abstract

Background

Previous studies have found that repetitive transcranial magnetic stimulation (rTMS) to the left dorsal lateral prefrontal cortex (LDLPFC) transiently reduces smoking craving, decreases cigarette consumption, and increases abstinence rates.

Objective

We investigated whether 10 daily MRI-guided rTMS sessions over two weeks to the LDLPFC paired with craving cues could reduce cigarette consumption and induce smoking cessation.

Methods

We enrolled 42 treatment-seeking nicotine-dependent smokers (>10 cigarettes per day) in a randomized, double-blind, sham-controlled trial. Participants received 10 daily sessions over 2 weeks of either active or sham MRI-guided rTMS (10Hz, 3000 pulses each session) to the LDLPFC concurrently with video smoking cues. The primary outcome was a reduction in biochemically confirmed cigarette consumption with a secondary outcome of abstinence on the target quit date. We also recorded cue-induced craving and withdrawal symptoms.

Results

Compared to sham (n=17), participants receiving active rTMS (n=21) smoked significantly fewer cigarettes per day during the 2-week treatment (mean [SD], 13.73[9.18] vs. 11.06[9.29], P<.005) and at 1-month follow-up (12.78[9.53] vs. 7.93[7.24], P<.001). Active rTMS participants were also more likely to quit by their target quit rate (23.81 %vs. 0%, OR 11.67, 90% CL, 0.96–141.32, x²=4.66, P=.031). Furthermore, rTMS significantly reduced mean craving throughout the treatments and at follow-up (29.93[13.12] vs. 25.01[14.45], P<.001). Interestingly across the active treatment sample, more lateral coil location was associated with more success in quitting (−43.43[0.40] vs. −41.79[2.24], P<.013).

Conclusions

Daily MRI-guided rTMS to the LDLPFC for 10 days reduces cigarette consumption and cued craving for up to one month and also increases the likelihood of smoking cessation.

INTRODUCTION

Worldwide, nearly one billion adults smoke cigarettes despite deleterious effects on health.^{1, 2} Even with the considerable progress in tobacco education and regulation over the past 50 years, in 2016, an estimated 15.5% of the US adults still smoke cigarettes.³ Despite the availability of treatments for nicotine dependence, including nicotine replacement therapy (NRT) medication and cognitive behavioral therapy, only 7.4% of smokers quit during the past year.⁴ Thus, there is an urgent need to develop alternative treatments with higher efficacy for smoking cessation such as neurostimulation techniques.

The prefrontal cortex has been implicated in the control and regulation of drug-seeking behavior.⁵ Impaired inhibitory control is often seen in individuals with addictive behaviors, and has been associated with substance use frequency in addiction. ⁶ High-frequency repetitive transcranial magnetic stimulation (HF-rTMS) over the dorsal lateral prefrontal cortex (DLPFC) has been demonstrated to improve inhibitory control. ^7,8 In earlier work, we found that a single 15-minute session of rTMS over the left DLPFC (LDLPFC) immediately reduced cue-induced craving compared to sham stimulation in smokers.⁹ Subsequently, the results of cue-induced functional MRI showed that one session of LDLPFC rTMS decreased brain activity in both the medial orbitofrontal cortex (mOFC) and nucleus accumbens.¹⁰ Notably, activity in these brain regions has been reported to be increased in cigarette smokers.^{11, 12}

Extending the previous findings of a single session of rTMS^{8, 13, 14}, several multi-session rTMS trials have been carried out for smoking cessation.^15–17 Three trials have shown that 10 or more sessions daily rTMS reduces cue-induced craving, decreases cigarette consumption and increases abstinence rates.^15–18 With these positive findings, however, more research needs to be done to improve rTMS treatment efficacy for smoking cessation as the prior studies have various inter-study differences. For example, treatments for more than 10 days may have a larger effect, so the overall optimal dose needs to be determined.^{15, 16} Additionally, we focused on one brain region.⁹ Other coil locations may offer better clinical responses.¹⁶ Furthermore, the methods of TMS targeting (scalp-based vs. MRI-guided neuronavigation) may affect treatment efficacy.^{19, 20} Finally, it may be that certain patients do better with this treatment than others. Perhaps rTMS combined with smoking cue exposures were more efficacious than rTMS with no smoking cues ¹⁶ since rTMS with simultaneously smoking cue exposure might disrupt brain circuits associated with maladaptive memories that associate a drug addictive habit. ^9,21 Previous studies demonstrated that disrupting the cue-drug memory reconsolidation could reduce the strength of cues in motivating drug-seeking and drug-taking behavio.^21–23 All of these issues require further studies. Hence, using our previously reported rTMS parameters ⁹ with simultaneously cue provocation, we aimed to rigorously test the effects of 10-session daily rTMS on smoking cessation in this MRI-guided neuronavigation, randomized, sham-controlled, and double-blind clinical trial. We hypothesized that compared to sham rTMS, 10 daily sessions of 10Hz individually MRI guided rTMS at 100% resting motor threshold (rMT) over the LDLPFC would significantly reduce cigarette consumption and cue-elicited craving, and increase smoking cessation rates.^{15, 24, 25}

MATERIALS and METHODS

Study Design

We conducted this randomized, double-blind, sham-controlled trial at the Medical University of South Carolina (MUSC) in Charleston, South Carolina, USA. The study consisted of 10 daily rTMS sessions over the left DLPFC over 2 weeks, with follow-up visits for another 3 months after the treatments stopped. Outcome measures included the following: urine cotinine, CO level, self-reported cigarette consumption, smoking cessation on the target quit date, cue-induced craving measures, and withdrawal symptoms. The Institutional Review Board at MUSC approved all study procedures, and the study was registered on ClinicalTrials.gov (NCT02401672).

Participants

We recruited participants via public media advertisements, flyers, email broadcast messages, and word of mouth with the inclusion and exclusion criteria (Table 1).

Table 1:

Study Inclusion and Exclusion Criteria

Inclusion Criteria	Exclusion Criteria
Between the ages of 18 and 60 years old	Current substance use of any psychoactive substances other than nicotine or caffeine
Smoking 10 or more cigarettes per day and a CO level > 10 ppm	Contraindications to MRI and TMS
Motivated to quit smoking	Use of other forms of nicotine delivery, such as nicotine patch, electronic cigarettes
If female, a negative pregnancy test and use of adequate birth control	Currently taking smoking cessation medications, including varenicline and bupropion

Screening, Randomization and General Procedures

Participants who met the inclusion criteria during an initial telephone screening came for an additional in-person screening where they signed written informed consent. Initial assessments included: medical history, physical examination, Mini International Neuropsychiatric Interview (MINI), Mini-Mental State Exam (MMSE), Transcranial Magnetic Stimulation Adult Safety Screening Questionnaire (TASS), drug screening (One Step Drug of Abuse Urine Test) and other contraindications to TMS and MRI. A baseline high-resolution Tl-weighted anatomical images were acquired for each participant on a 3T Siemens TIM Trio MRI scanner (Siemens, Erlangen, Germany) (TR = 2250 ms, TE = 4.18 ms, voxel dimensions 1.0 × 1.0 × 1.0 mm, 177 slices, full-brain, and cerebellar coverage, no gap). Each subject’s rMT was obtained. The Data Coordination Unit at MUSC randomly assigned each subject’s treatment condition.

TMS Procedures

We scheduled participants at the same time for daily treatment. They were instructed to abstain from smoking at least two hours before each treatment, which was hoped to increase the degree of craving during the treatment.^{9, 11}

rTMS Therapy System

We used a modified version of the NeuroStar XPLOR Clinical Research System (Neuronetics, Inc), which was loaned to MUSC for this research. (e Figure 2)

Determining Resting Motor Threshold (rMT)

We determined the resting motor threshold (rMT) for all participants 3 times over the treatment course: (1) at the screening visit, (2) prior to 1st rTMS treatment 1, and (3) prior to 6^th rTMS treatment. The motor threshold was ascertained using the active NeuroStar coil placed over the area of the skull corresponding to the motor cortex and adjusted until each pulse resulted in isolated movements of the right thumb (Abductor Pollicis Brevis -APB). rMT was determined with the Neurostar algorithm, which provides an iterated estimated of the rMT.^{26, 27}

Targeting of the left DLPFC

Targeting of the stimulation site (LDLPFC, used a modified method by Mylius) ²⁸ (Supplement #1, e Figure 1)) was performed from structural MRI scans using Brainsight™ TMS neuronavigation (Rogue Research, Inc) on the first session. In order to save TMS procedure time, the individual participant used a tightly fitted cloth cap initially positioned by using the BrainSight system. The close-fitting cloth cap was personalized by marking the midline and the outline of both ears. The front edge was fitted 0.5 cm above the eyebrow. This assured that repositioning would not have errors in pitch, roll or yaw. At the first treatment, we used Brainsight with the individual MRI to locate the left DLPFC and then marked the TMS coil location on the cap. This location was used at all following visits. We have used this cap system for many studies and find that it has high reliability, by repeatedly testing the indicated motor knob marked on the cap, or reconfirming with Brainsight. To make sure patients received the right dose to the right location every time, we used contact sensors for each treatment with a rotation point about the tip of the subject’s nose (e Figure 1).

Active rTMS

rTMS was administered at 100% rMT, at 10 Hz for 5-second trains, with an intertrain interval of 10 seconds. Treatment sessions lasted for 15 minutes (60 trains) with 3000 pulses/session.

Sham rTMS

The sham system used in this trial is considered an “active” sham in which the sensation of active rTMS is mimicked using time-locked electrical stimulation at the target treatment site without a magnetic intervention. ^{29, 30}(Supplement #2 and e Figure 2)

Cue Provocation

We used structured 1.5 min exposure and interactions with real-life smoking paraphernalia (cigarettes, ashtray, lighter)³¹ immediately before each rTMS session. While rTMS was administered, subjects watched 15-minute smoking cued video⁹ (scenes of individuals smoking in various environments) displayed on an iPad placed on a tripod at the foot of the treatment chair. (Supplement #3 & 4, e Figure 3, and Video still)

Evaluation of Cigarette Consumption, Nicotine Dependence, Cued Craving and Biomarkers

The number of cigarettes smoked per day (printed pre-prepared cigarette diary brought to the TMS lab for each session)³² was evaluated by subjective self-report. The subject recorded their daily cigarette use each morning before each rTMS session and every morning during the 1- month follow-up.

Cued craving was measured at three times (Pre-cue provocation, post-cue provocation, and post-TMS session) with a subjective visual analog scale (VAS).³³ The scaled Fagerstrom Test for Nicotine Dependence (FTND)³⁴, the Questionnaire of Smoking Urges-Brief (QSU-B)³⁵, and the Minnesota Nicotine Withdrawal Scale (MNWS)³⁶ were measured (at baseline, prior to the 6^th TMS session, prior to the 10^th TMS session, 1-week follow-up, and at 1-month follow-up). Carbon monoxide (CO) levels were measured before each TMS treatment using the Micro Smokerlyzer Breath Carbon Monoxide Monitor.

At the MUSC Clinical Neurobiology Laboratory, urinary sample cotinine levels were measured with homogeneous enzyme immunoassay (EIA, DRI Cotinine Assay) at baseline, before the 6^th TMS session, before the 10^th TMS session, at 1-week and 1-month follow-up.

On the 3-month phone follow-up assessment, participants were asked if they continued smoking cigarettes or not, and the number of cigarettes smoked per day. FTND evaluated nicotine dependence, the QSU-B measured cued craving, and the MNWS evaluated nicotine withdrawal symptoms.

Smoking Cessation on Target Quit Date

Setting a quit date is often a central element of tobacco cessation treatments.³⁷ Recently, Saladin and colleagues asked participants to select a target quit date (TQD) after 1-week of medication titration before 4 weeks of active treatment.³⁸ Using this model, we asked participants to select a TQD within days 7–10 of the rTMS sessions.

Subjects were instructed to quit smoking on their selected TQD. We reminded them before the TQD. A successful quit was defined as at least 2 days of abstinence³⁹ and CO < 5 parts per million (ppm)⁴⁰. Continuous abstinence was assessed at 7-day and 4-week visits after the last TMS session.^{41, 42} The smoking cessation status on TQD was evaluated at the end of rTMS treatment.

Statistical Analysis

Data analysis was performed using IBM SPSS Statistics 22 (IBM, Endicott, New York). T-test or the Chi-squared (x²) test was performed to test for baseline differences in demographic and clinical variables between the two treatment arms. Continuous variables (outcomes) including cigarettes per day, computerized QSU-B and VAS, and CO level were analyzed with a two-tailed, parametric 2 (treatment: sham vs. active) x 2 times (pre- and post- sessions) x 11 visits (baseline and TMS session number). Mixed-Models for Repeated Measures (MMRM) weekly measures including FTND, QSU-B, MNWS, and urine cotinine were analyzed using a two-tailed, parametric 2 (treatment: sham vs. active) x 5 (visit number: baseline, prior to 6^th TMS session, prior to 10^th TMS session, 1-week follow-up, and 1-month follow-up) model. To adjust the potential difference of the baseline characteristics, we included the years of smoking and the number of previous quit attempts as covariates in the analyses. Categorical variables (outcomes), including the quit rate on TQD and the response rate, were analyzed with binary logistic regression analysis. To adjust the potential difference of the baseline characteristics between groups, we performed binary logistic regression with quit rate on TQD or response rate as the outcome and the years of smoking and the number of previous quit attempts as covariates.

RESULTS

Enrollment

Figure 1 shows the study flow for both cohorts, including the intention-to-treat (ITT) and completer samples.

Figure 1. Consort Flow Diagram.

The Flow of Participants from a Longitudinal Cohort in Trial of Active vs. Sham TMS. The timeline of the trial with the number of participants and dropouts in each phase.

Participant Characteristics

Participant demographics and smoking-related variables are listed in Table 2. There were no statistically significant baseline differences between the groups in demographic or clinical variables.

Table 2:

Demographic Information (Baseline)

	Sham rTMS (n = 17)	Active rTMS (n = 21)	P-Value
Age, mean (SD)	44.12 (9.1)	41.19 (11.8)	0.39
Gender, M/F	8/9	9/12	0.79
Years Education, mean (SD)	16.6 (10.9)	14.3 (1.8)	0.35
Years smoked, mean (SD)	28.3 (9.5)	21.9 (10.8)	0.06
Smoking Start Age, Mean (SD)	15.3 (3.7)	16.7 (5.0)	0.69
Cigarettes per day, mean (SD)	19.9 (10.7)	18.6 (5.4)	0.59
FTND score, mean (SD)	5.3 (1.4)	4.6 (1.8)	0.18
Urine Cotinine, mean (SD)	1190.4 (124.6)	1109.4 (129.4)	0.68
Desire to quit smoking, mean (SD)a	8.4 (1.9)	8.9 (1.0)	0.27
Previous Trials (SD)b	4.1 (1.1)	3.2 (1.5)	0.06
Previous Success, (Y/N)c	11/6	11/10	0.52

^aRange: 0 –10

^bPrevious trials to quit

^cPrevious abstinent for more than 1 month

For analysis, patients were classified as intent to treat (ITT) or Completers. The ITT population was defined as all randomized patients who started at least 1 treatment session and had a postbaseline assessment. The completer sample was defined as randomized patients who were treated according to the protocol for all 10 sessions. This classification was done before breaking the blind and performing the final statistics. Here we present the completer analysis. The ITT data can be found in supplement #5.

Subjective Self-reported Cigarette Consumption

For the completed sample, the number of cigarettes smoked per day, as self-reported by the participants on each treatment day, is presented in Figure 2. The MMRM (2 [treatment conditions] x 11 [measured points]) revealed that active TMS treatment reduced cigarette consumption compared to the sham stimulated group (sham: mean [SD] 13.73 [9.18]; active: 11.60[6.93], F_{1, 394}=9.43, P<.005). The results also showed that active rTMS treatment continuously reduced cigarette consumption with 10 TMS sessions (baseline: 19.40[8.04]; 10^th session: 8.63[6.75]; F_{10, 394}=5.14, P<.001). Pairwise comparisons revealed a significantly greater effect of active rTMS than that of sham stimulation for session 10 (sham: 11.12[6.99]; active: 6.62[5.97]; P=.043). The years of smoking significantly affected the model of analysis (F_{1, 394}=4.93, P<.05). The number of previous quit attempts did not affect the model of analysis (F_{1, 394}=0.76, P>.05). However, with adjustment for the years of smoking, we found that rTMS sessions showed the treatment efficacy in cigarettes per day.

Figure 2. rTMS Treatment Effects on Daily Mean Cigarette Consumption.

The mean (± SEM) number of cigarettes smoked per day is presented. The Mixed-Model for Repeated Measures reveals active rTMS treatment (n=21) significantly reduces cigarette consumption compared to sham (n=17) (F-i, ₃₉₆ = 7.95, P =.005). Active rTMS longitudinally reduced daily cigarette consumption over the rTMS treatment course (F_{9, 451} = 5.10, P <.001). Pairwise comparisons revealed significantly greater effects of active rTMS than that of sham stimulation for session 10 (sham: 11.12[6.99]; active: 6.62[5.97]; P=.034).

During the one-month follow-up period, for the completer analysis, the active treatment group (n=20) showed significantly lower daily cigarette consumption compared to the sham group (n=15) (sham 12.78[9.53]; active: 7.93[7.24]; F_{1, 137}=10.66, P<.001).

The response was defined as a reduction of at least 50% in the self-reported number of cigarettes smoked in the last TMS session relative to the baseline. At the end of 10 rTMS treatments, the response rate for the active group (76.19%) was superior to the sham group (35.29%) (odds ratio [OR] 5.87, 95% CI 1.43 to 24.11, x²=6.45, P=.011). (Figure 3) When the years of smoking and the number of previous quit attempts were used as covariates in the binary logistic regression, the active group still showed a significantly better response rate than the sham group over the two-week treatment (OR 9.73, 95% CI 1.82 to 52.01, x²=7.08, P=.008). The logistic regression analysis also showed that neither the years of smoking (OR 0.97, 95% CI 0.89 to 1.05, x² = .69, P =.41) nor the number of previous quit attempts (OR 1.48, 95% CI 0.79 to 2.76, x² = 1.48, P = .22) significantly affected the response rate of the reduction of cigarettes.

Figure 3. rTMS Treatment Effects on Response and Quit Rate.

The treatment response rate (at the end of the intervention phase) was significantly higher for the active group (76.19%) than for the sham group (35.29%) (OR 5.87, 95% CI 1.43 to 24.11, x²=6.45, P=.011). The smoking cessation rate was also higher for the active group (23.81%) than for the sham group (0%) (OR 11.67, 90% CL, 0.96–141.32, x²=4.66, P=.031).

Smoking Quit Rate on Target Date and Continuous Abstinence Rate

Five of 21 active TMS participants (23.81%) and 0 out of 17 sham TMS participants (0%) stopped smoking on the target quit date. Because there was no smoker who quit on TQD in the sham group, we used Haldane-Anscombe correction that added 0.5 to each of the cells and then calculated the odds ratio over these adjusted cell counts.⁴³ Active TMS produced significantly higher quit rate than sham TMS (OR 11.67, 90% CL, 0.96–141.32, x²=4.66, P=.031) (Figure 3). When the years of smoking and the number of previous quit attempts were used as covariates in the binary logistic regression, only the treatment condition significantly affected the quit rate on TQD over the two week treatment (Score test, Score = 5.16, P = .023). The logistic regression analysis also showed that neither the years of smoking (OR 0.95, 95% CI 0.85 to 1.06, x² = .87, P = .35) nor the number of previous quit attempts (OR 0.43, 95% CI 0.38 to 1.63, x² = 1.48, P = .51) significantly affected the smoking quit rate on TQD. Four successful quitters in the active TMS group and no one in the sham group demonstrated 7 days of continuous abstinence at a 1-month follow-up after the last TMS session (x²=3.62, P=.057, nearing significance). Regarding logistic regression results, we did not find either the years of smoking (OR 0.94, 95% CI 0.84 to 1.07, x² = .89, P = .34) or the number of previous quit attempts (OR 1.02, 95% CI 0.47 to 2.19, x² = .002, P = .96) significantly affect the smoking quit rate on TQD. At a 3-month follow-up, three in the active group and no one in the sham group reported continuous abstinence (x²=2.64, P=.10).

Biochemical Measures (completer sample)

Urine Cotinine

A MMRM analysis (2 [treatment conditions] x 5 measured points [baseline, 6^th TMS, 10^th TMS, 1-week follow-up, and 1-month follow-up]) revealed that active TMS (1008.33[557.46] ng/ml) significantly reduced urine cotinine levels compared to sham treatment (1205.64 [631.01]) (95% CI 11.93–382.70; F_{1, 164}=5.22, P=.024). No main effect of time points was found between the 5 measures (F_{4, 164}=.46, P=.764). No significant interaction between treatment and time was found (F_{4, 164}=0.94, P=.44). Pairwise comparisons showed a significantly greater effect of active rTMS than that of sham stimulation for a 1-week follow-up (sham: 1250.13[602.39]; active: 825.67[620.23]; P=.046). Neither the years of smoking nor the number of previous quit attempts significantly changed the analysis results of urine cotinine (F_{1, 164}=1.94, P=16; F1, 164=0.1, P=.75).

CO Level

Compared to the sham group, the MMRM analysis (2 [treatment conditions] x 10 [measure points]) showed that active rTMS significantly reduced daily mean CO levels (sham: 10.75[5.52] ppm; active: 9.25[5.21]; F_{1, 358}=2.25, P=.019). Compared to the baseline, CO levels continued to decrease during 10 TMS treatment course (baseline: 10.64[4.84]; 10^th TMS: 7.35[5.39]; F_{9, 35}=12.81, P<.001). Pairwise comparisons revealed a significantly greater effect of active rTMS than that of sham stimulation for rTMS session 7 (sham: 11.88[6.62]; active: 8.57[4.49]; P=.024) and session 9 (sham: 10.41[6.17]; active: 6.95[4.79]; P=.019). (Figure 4) Neither the years of smoking nor the number of previous quit attempts significantly changed the analysis results of CO level (F₃_ 358=3.45, P=.06; F1, 358=1.68, P=.19).

Figure 4. Treatment Effects on CO Levels.

Mean (± SEM) daily CO levels are presented. Compared to the sham group, the active rTMS significantly reduced CO levels (F^ ₃₆₀=2.22, P=.021). Furthermore, CO levels were longitudinally reduced (daily) over the rTMS treatment course (F_{9, 360}=7.49, P=.007). Pairwise comparisons reveal significantly greater effects of active rTMS than that of sham stimulation for rTMS session 7 (sham: 11.88[6.62]; active: 8.57[4.49]; P=.037) and session 9 (sham: 10.41 [6.17]; active: 6.95[4.79]; P=.045). TX= Treatment; *<.05).

Nicotine Dependence (FTND)

The MMRM analysis (2 [treatment conditions] x 5 measure points [baseline, 6^th TMS, 10^th TMS, 1-week follow-up, and 1-month follow-up]) revealed a significant treatment effect on the mean FTND scores over 5 longitudinal measurements (sham: 4.63[2.08]; active: 3.42[2.34]; F_{1, 163} = 10.60, P = .001). The results also revealed the main effect of treatment time points on the FTND score which continuously decreased during TMS treatment course and remained at the new score at two follow-up visits (baseline 4.95[1.66], 6^th TMS: 4.16[2.39], 10^th TMS: 3.21[2.21], 1-month follow-up:3.64[2.56]; F₄₁₆₃ =3.48, P = .009). Pairwise comparisons revealed a significantly greater effect of active rTMS than that of sham stimulation at the 1-month follow-up (sham: 4.90[2.28]; active: 2.84[2.39]; P<.001). Neither the years of smoking nor the number of previous quit attempts significantly changed the analysis results on FTND (F_{1, 163}=3.45, P=.50; F_1,163=.18, P=.67).

Subjective Cued Craving

Questionnaire of Smoking Urges-Brief (QSU-B)

The MMRM analysis (2 [treatment conditions] x 5 measure points [baseline, 6^th TMS, 10^th TMS, 1-week follow-up, and 1-month follow-up]) revealed that the active TMS group had significantly lower mean craving ratings over 5 longitudinal measurements than did the sham group (sham: 29.93[13.12]; active: 25.01[14.45]; F₁₁₆₈=10.37, P=.002), and that the craving rating continuously decreased during the TMS treatment course and remained at this level at two follow-up visits (baseline: 37.85[13.84], 6^th TMS: 28.83[12.81], 10^th TMS: 23.28[11.80], 1-week follow-up: 22.00[11.73], 1-month follow-up: 25.50[13.74]; F ₄₁₆₈=9.10, P<.001). Pairwise comparisons revealed a significantly greater effect of active rTMS than that of sham stimulation at the 1-month follow-up (sham: 32.00[15.00]; active: 19.00[10.81]; P=.008). Neither the years of smoking nor the number of previous quit attempts significantly changed the analysis results in subjective cravings (F_1,166=.04, P=.84; F_{1, 166}=18, P=.73).

Immediate Effect of TMS on Craving and Provocation Effect

VAS (0–7): An MMRM analysis (2 [treatment conditions] x 3 [pre-provocation, post-provocation, after TMS] x 10 [TMS session]) was conducted for VAS data. The analysis revealed a significant treatment effect (sham: 4.03[1.94]; active: 3.60[2.21]; F₁₁₀₇₆=23.73, P<.001), significant effect of the number of TMS sessions (1^st TMS: 4.14 [2.37]; 10^th: 2.25[2.43]; F_{9, 1076}= 5.79, P <.001), and significant pre-post effect within each session (pre: 3.81[2.05], post-provocation: 4.26[2.07] and post-TMS: 3.39[2.09]; F_{2, 1076}= 17.71, P<.001). Pairwise analysis showed greater immediate reductions in VAS for the active group than the sham group (2.36[0.26] vs. 0.69[0.28]; F _{9, 1076}= 3.13, P=.001). Furthermore, the greater reductions of active treatment in VAS showed in TMS sessions 6–10 (P<.05). Pairwise analysis also showed a significant effect for provocation (P<.001), which did not differ between treatment conditions. The years of smoking significantly affected the model of analysis (F_{1, 1076}=5.47, P=.02). The number of previous quit attempts did not affect the model of analysis (F_{1, 1076}=0.005, P=.95). However, with adjustment for the years of smoking, we found that rTMS sessions showed the treatment efficacy in cravings.

Withdrawal Symptoms (MNWS)

No main treatment effect was found between treatment groups in the mean MNWS scores over 5 longitudinal measurements (baseline, 6^th TMS, 10^th TMS, 1-week follow-up, 1-month follow-up) (sham: 9.16[6.83] vs. active: 7.43[6.81]; F₁₁₆₀ =2.62, P=. 11). No difference between visiting times was found in the MNWS analysis (F_{4, 160}=.57, P=.69). Factor analyses showed that no significant effect of TMS was found in the Negative Effect Factor (depressed mood, irritability, frustration or anger, and anxiety), while a significant difference between active TMS and sham treatment was found on the urge to smoke at the end of ten sessions of TMS (P =.04). Neither the years of smoking nor the number of previous quit attempts significantly changed the analysis results of withdrawal symptoms (F_{1, 160}=1.27, P=.26; F_{1, 166}=0.25, P=.62).

Treatment Adherence, Visit Attendance, and Adverse Events

Across all subjects, 22 of 38 (57.9%) participants reported a side effect during at least one visit. There was no significant difference between the active (66%) and sham groups (47%) (x²=.79, df=1, P=.375). Headache (n=16), discomfort or unpleasant (n=4) feelings, and scalp pain (n=4) were the most common side effects. However, the discomfort or pain was never severe enough for any subject to receive treatment for the side effect.

TMS Coil Placement

The mean targeting coordinates (the Montreal Neurological Institute [MNI]) did not differ between active and sham arms (Table 3). The average coordinates of coil placement in the successfully quitting group (n = 5) were more lateral in coil location than in the unsuccessful group (MNI coordinates, X-axis: mean successful group = −43.42, mean unsuccessful group = - 41.79). The average coordinates of coil placement in the response group (n = 16) were also more lateral than in the non-response group (MNI coordinates, X-axis: mean response group = - 42.90, mean non-response group = −39.89).The results suggest that a more lateral coil placement carried potentially greater clinical effects, even though the exact spot for each individual was determined based on their structural MRI scan. (Table 3)

Table 3:

Comparison of the targeted left DLPFC MNI coordinates in different treatments and different effects.

DLPFC Coordinates	Group 1(N)	Group 2 (N)	P-Value
Treatment Groups	Sham (17)	Active (21)
X (mean ± SD)	−41.95 ± 2.20	−41.35 ± 2.68	0.456
Y (mean ± SD)	37.67 ± 2.68	37.53 ± 2.80	0.875
Z (mean ± SD)	30.86 ± 2.26	30.80 ± 2.47	0.937
Active TMS Quit or Not	Quit (5)	Non-quit (16)
X (mean ± SD)	−43.43 ± 0.40	−41.79 ± 2.25	0.013
Y (mean ± SD)	37.67 ± 3.86	37.17 ± 2.87	0.799
Z (mean ± SD)	33.38 ± 1.97	31.06 ± 2.44	0.062
Active TMS Quit or Not	Response (5)	Non-response (16)
X (mean ± SD)	−42.90 ± 0.93	−39.89 ± 3.12	0.002
Y (mean ± SD)	37.71 ± 2.71	35.93 ± 3.94	0.264
Z (mean ± SD)	31.86 ± 2.49	30.81 ± 2.67	0.043

Blind Integrity

As a group, participants were successfully blinded to the rTMS treatment condition. The blinding of the study was effectively maintained during the two-week treatment. The patient guesses were significantly affected by clinical responses. (S Table 1).

DISCUSSION

The findings from this randomized, sham-controlled trial demonstrate that ten daily sessions of active MRI-guided- rTMS over the LDLPFC combined with concurrent cue-elicited craving reduced cigarette consumption and increased the likelihood of smoking cessation when compared to sham (0% sham vs. 23.81% active). Importantly, these self-reported smoking findings were secondarily corroborated by assessments of urine cotinine and CO levels. Using structural MRI guidance targeting LDLPFC, those with better smoking cessation had a more lateral coil placement. At the one-month follow-up, the active rTMS group continued to have lower cigarette consumption and cued-induced craving than did the sham group. These results are consistent with, and extend, our previous findings using one session of rTMS in smokers,⁹ as well as other rTMS trials on craving and cigarette consumption.^{13, 15, 16} Corroborating the findings of previous studies, the present study demonstrates that ten sessions of rTMS over the LDLPFC can significantly reduce cigarette consumption and increase smoking abstinence rates.

Compared to sham treatments, the main effect of active TMS in attenuating cigarette craving started at the first TMS session and accumulated throughout the treatment course. Furthermore, the results of pre- and post- TMS session VAS measures demonstrate that rTMS induced a significant immediate therapeutic effect on cued craving. The current study supports the findings of our previous rTMS study which found that a single session treatment reduced cue-induced craving.⁹ In addition, previous studies suggest that medications known to increase smoking cessation rates could decrease cued craving during a smoking cessation attempt.^{44, 45} Taken together, the current results are consistent with a theory that prefrontal TMS decreased craving, which may then prevent relapse to smoking and enhance the likelihood that a smoker will successfully quit smoking.⁴⁶

The total duration of the sessions, the frequency of stimulation employed, the intensity, and the pattern of stimulation are potentially critical factors in determining long-lasting TMS effects.⁴⁷ The literature from rTMS in the treatment of depression overall suggests that the total number of rTMS pulses delivered to the patient is positively associated with treatment efficacy, up to a certain cap.⁴⁸ In the current study, we found that rTMS produced a gradual reduction in both cigarette consumption and cued craving over two weeks of treatment, which was statistically significant by the tenth session. These findings are consistent with the deep rTMS study completed by Abraham Zangen’s group,¹⁶ in which they reported that 10 Hz deep rTMS decreased cigarette consumption gradually. Furthermore, the current study revealed that the effect of rTMS persisted over the 3-month follow-up.

The impact of coil placement on the effect of rTMS for smoking cessation has not been well studied. Our putative target of stimulation for smoking cessation was the LDLPFC, which was related to the inhibitory and control function.⁵ In addition, the imbalanced neuronal circuits, including driving-reward circuit and executive control circuit may cause smoking relapse and difficulties in quitting.⁴⁹ In the current study, we hypothesized that increased prefrontal cortex function would interact with the driving-reward circuit involved in nicotine dependence.^{9, 49, 50} Most trials of rTMS for smoking cessation have targeted the LDLPFC.^{9, 14, 15, 51} Unfortunately previous evidence shows the conventional placement rule (5 or 6 cm anterior to M1) does not always reach the DLPFC. ^ In the current study, we used modified Mylius’ methods with personalized MRI (Supplement #1, e Figure 1) to target the LDLPFC. We found that actively treated rTMS smokers who quit had a more lateral coil location than did those who could not quit. This small sample result bears testing in a larger cohort. Hence, MRI-targeted rTMS may be an important consideration for future brain stimulation in the addiction population. Our results, if replicated, indicate that using modified Mylius’ targeting method²⁸, the targeted LDLPFC should be a more lateral site by 1.6 mm (mean difference) to be even more effective for smoking cessation. The more lateral stimulation site might be more close to an individualized DLPFC-NAc circuit. Therefore, future research might combine E field modeling⁵³, structural MRI guidance, and an individualized circuit-based method to identify DLPFC for improvement in smoking cessation⁵⁴.

rTMS appears safe and tolerable in a nicotine-dependent smoker population. The most common side effects of rTMS were headaches and scalp pain. However, there was no significant difference in side effects between the active and sham groups. The side effects were typically reported in the first two treatment sessions and then disappeared. The tolerability of daily rTMS in this population is consistent with the findings in depression treatment.⁵⁵

The major strengths of this study include the double-blind, randomized design, active sham stimulation, individualized stimulation site via neuronavigation, daily CO measures, and convergent validity of the findings across multiple outcomes and measures. However, several limitations of the study should also be noted. First, this study had a relatively small sample size (in the completer sample, 21 active rTMS vs. 17 sham rTMS) but reveals a large effect of the reduction of cigarette consumption. Furthermore, we found that the number of cigarettes per day was significantly correlated with urine cotinine (Supplement #6). The second limitation is that the 3-month follow-up by phone is insufficient to evaluate smoking cessation fully (e.g., assessing smoking cessation rate at 6 months post-treatment may be more informative of true cessation). It is unclear if subjects might need additional TMS sessions over an extended period to quit smoking entirely. Future studies should increase the sample size, extend the follow-up time to 6 months, and verify follow-up cigarette consumption by biological measures. The third limitation is a relatively 3-month low quit rate of rTMS for smoking cessation (14%). The abstinence rate of rTMS is lower than available tobacco addiction treatments(e.g., varenicline, 24% and nicotine replacement therapy, 23%).^{56, 57} Therefore, combining rTMS with other therapies may be the best clinical treatment for smoking cessation if it were to gain FDA approval.²⁵ Fourth, the sham group showed nearly significantly longer the years of smoking and more the number of previous quit attempts. However, having adjusted the two covariates, we found that the active rTMS treatment produced the therapeutic efficacy for smoking cessation in the reduced cigarette consumption and the increased abstinence rate. Nevertheless, stratified randomization may be helpful in reducing the likelihood of such imbalances between treatment groups in future research. Finally, researchers have used heterogeneous treatment parameters that make the results difficult to pool and to overall compare the efficacy of rTMS on nicotine dependence.^{58, 59} Previous studies in smokers with comorbid schizophrenia showed that multiple sessions of rTMS did not affect the cue craving and cigarette consumption.^{60, 61} Nevertheless, most of the previous studies showed the evidence of treatment efficacy of rTMS for smoking cessation.

Conclusions

In conclusion, this study demonstrates that ten daily sessions of MRI-guided rTMS over the left DLPFC can reduce cigarette consumption, decrease cue craving, and increase abstinence rates. Individually MRI-targeted DLPFC methods may improve treatment efficacy for future brain stimulation studies in addictions. If these results are replicated in more sessions, longer-follow up, and larger sample trials, daily prefrontal rTMS might become a new treatment for smoking cessation. In addition, future research should identify optimal parameters (targets, intensity, frequency, duration of stimulation, neuronavigation, concomitant treatments).

Highlights.

• A single session of rTMS over the left LDLPFC reduces cue-induced craving.

• Two weeks of daily MRI-guided rTMS to the LDLPFC aids smoking cessation.

• The more lateral stimulation, the better the rTMS efficiency for smoking cessation.