Augmentative repetitive navigated transcranial magnetic stimulation (rTMS) in drug-resistant bipolar depression


  • The authors of this paper do not have any affiliation or financial interest in any organization that might pose a conflict of interest in connection with this manuscript.

Bernardo Dell’Osso, M.D., Department of Psychiatry, University of Milan, Fondazione IRCCS Ospedale Maggiore Policlinico, Via F. Sforza 35, 20122 Milano, Italy. Fax: 0039 02 50320310; e-mail:


Objectives:  The efficacy of transcranial magnetic stimulation (TMS) has been poorly investigated in bipolar depression. The present study aimed to assess the efficacy of low-frequency repetitive TMS (rTMS) of the right dorsolateral prefrontal cortex (DLPFC) combined with brain navigation in a sample of bipolar depressed subjects.

Methods:  Eleven subjects with bipolar I or bipolar II disorder and major depressive episode who did not respond to previous pharmacological treatment were treated with three weeks of open-label rTMS at 1 Hz, 110% of motor threshold, 300 stimuli/day.

Results:  All subjects completed the trial showing a statistically significant improvement on the 21-item Hamilton Depression Rating Scale (HAM-D), Montgomery-Åsberg Depression Rating Scale, and Clinical Global Impression severity of illness scale (ANOVAs with repeated measures: = 22.36, p < 0.0001; = 12.66, p < 0.0001; and = 10.41, p < 0.0001, respectively). In addition, stimulation response, defined as an endpoint HAM-D score reduction of ≥50% compared to baseline, was achieved by 6 out of 11 subjects, 4 of whom were considered remitters (HAM-D endpoint score ≤ 8). Partial response (endpoint HAM-D score reduction between 25% and 50%) was achieved by 3/11 patients. No manic/hypomanic activation was detected during the treatment according to Young Mania Rating Scale scores (ANOVAs with repeated measures: = 0.62, p = 0.61). Side effects were slight and were limited to the first days of treatment.

Conclusions:  Augmentative low-frequency rTMS of the right DLPFC combined with brain navigation was effective and well tolerated in a small sample of drug-resistant bipolar depressive patients, even though the lack of a sham controlled group limits confidence in the results.

Transcranial magnetic stimulation (TMS) allows a noninvasive electrical stimulation of the cerebral cortex by means of magnetic fields generated by a handheld coil. The powerful magnetic fields act as a vector that passes unimpeded across the skull, and then converts into an electrical field within the brain. Traditionally used in neurophysiology as a research tool, repetitive TMS (rTMS) has since been applied in a variety of psychiatric disorders—mostly major depression—as a potentially therapeutic intervention. In addition, in recent years TMS has been associated with functional imaging methods such as positron emission tomography (1), single photon emission computed tomography (2), and electroencephalography (3), allowing researchers to detect TMS-induced effects on brain perfusion, metabolism, and connectivity. Moreover, when rTMS is combined with brain navigation procedures utilizing patients’ brain magnetic resonance image (MRI), it is possible to target the cortical stimulation area with extreme precision, reliability, and reproducibility over the treatment (4, 5).

Taken as a whole, different meta-analyses assessing the efficacy of rTMS in major depression (6–12) showed superior effects, even though modest, for active stimulation over sham rTMS, and some authors suggested that a two-week treatment course may be too short to achieve optimal results (13). From this perspective, a recent multicenter, randomized, controlled trial (14) reported that rTMS delivered to the left dorsolateral prefrontal cortex (DLPFC) showed a statistically significantly greater efficacy over sham stimulation after four weeks of treatment in a large sample of major depressives.

Nevertheless, major concerns on published clinical trials of rTMS include the wide variability of used setting parameters, the limited samples of patients, the small number of sham-controlled studies, and the complex reproducibility of results. In addition, the majority of published trials with rTMS in major depression have been focused on unipolar depressives, with bipolar depressed subjects usually in the minority of the samples, when present (15–20).

With regard to rTMS trials specifically performed in bipolar depression, three controlled studies have been published to date (21–23), two of which were limited by small samples. The larger study among these, a single-blind, randomized, sham-controlled trial (22), carried over two weeks in 23 bipolar depressed/mixed patients, failed to find a statistically significant superiority for left DLPFC rTMS at 5 Hz over sham stimulation.

The present study aimed to assess the efficacy of augmentative low-frequency rTMS, combined with brain navigation and delivered to the right DLPFC for three weeks, in a sample of bipolar depressed subjects who had not benefited from prior pharmacological treatment.



Eligible subjects were right-handed outpatients, aged 18–70, with a diagnosis of bipolar disorder, either type I or type II, and current major depressive episode, according to the criteria of the DSM-IV-TR (24), with a duration of six weeks or more. To be included in the study, patients had to have a severity of the current depressive episode score of ≥4 on the Clinical Global Impression severity of illness scale (CGI-S) (25) and a total score ≥18 on the 21-item Hamilton Depression Rating Scale (HAM-D) (26). In addition, patients were required to have not responded to at least one but no more than three adequate antidepressant trials during the current episode. Adequate antidepressant trial was defined as a treatment with antidepressants at full or best tolerated doses for at least six weeks. Nonresponder patients were defined as subjects who had not achieved a significant reduction of depressive symptoms, as assessed by an improvement of the HAM-D total score ≤ 25% compared to pretreatment scores. Patients had to be on stable treatment with valproate (at least six months, with plasma levels ranging from 50 to 100 μg/ml) as a mood-stabilizing agent.

Exclusionary criteria for study participation were: comorbid organic or psychotic disorders, mental retardation, substance abuse or dependence within the past three months, personality disorders severe enough to interfere with study participation, risk of suicide, obsessive-compulsive disorder, posttraumatic stress disorder, eating disorders, lack of response to an adequate trial of electroconvulsive therapy (ECT), prior treatment with rTMS or vagus nerve stimulation, pregnancy, and a personal or confirmed family history of seizures in first-degree relatives. All subjects gave written informed consent to participate into the study after receiving a full explanation of the study protocol, which had been approved by the local ethics committee and institutional review board.

Assessment and outcome measures

Diagnoses were performed by trained psychiatrists using a semi-structured clinical interview based on DSM-IV-TR criteria (SCID) (27), during which patients’ main demographic and clinical characteristics were collected. These included: age, age at onset, marital status, occupational status, diagnosis subtype, duration of illness, number of failed antidepressant treatments during the current episode, and family history for mental disorders in first-degree relatives.

After baseline assessment, patients’ depressive symptoms were assessed weekly throughout with the HAM-D, the Montgomery-Åsberg Depression Rating Scale (MADRS) (28), and the CGI-S. Manic/hypomanic symptoms were also assessed weekly with the Young Mania Rating Scale (YMRS) (29). Safety and tolerability were assessed at each weekly visit after baseline using spontaneously reported events and rates of discontinuation for adverse events. MADRS and HAM-D scores were the primary outcome measures and CGI-S score was the secondary outcome measure.

Brain navigation

Before starting the stimulation, a brain MRI for each patient was obtained and installed on brain-navigation computer software (30). The software was connected to an infrared system that recognizes the exact position of the stimulating coil and patient’s brain through a system of trackers positioned on the coil and on a pair of glasses that the patient wears during the session of rTMS. Once the patient’s brain MRI is installed on the software, the registration of the scalp position is performed in order to match the MRI to the real position of the patient’s brain. The brain navigation procedure allows targeting of the DLPFC with extreme precision, reliability, and replicability during the entire cycle of rTMS.

TMS parameters

Stimulation setting was fixed at low frequency (1 Hz) and at 110% magnetic field intensity relative to the patient’s observed resting motor threshold (MT). MT was defined as the minimum magnetic intensity required to elicit five motor evoked potentials (100 μV) out of 10 consecutive stimuli in a target hand muscle (abductor pollicis brevis). Right DLPFC was chosen as stimulation site on the basis of positive results from previous studies stimulating this area (31, 32). According to the cytoarchitectonic Brodmann’s classification, DLPFC corresponds to areas 9 and 46. Area 46 in humans occupies approximately the middle third of the middle frontal gyrus and the most rostral portion of the inferior frontal gyrus. By means of the MRI-guided brain navigation system, we precisely targeted the middle third of the right middle frontal gyrus (Fig. 1) in each subject and stored its coordinates in the software. This procedure allowed the reproduction of the target with extreme accuracy during each stimulation session.

Figure 1.

 Target area of stimulation with anatomical landmark positioned on the right prefrontal dorsolateral cortex (middle third of right middle frontal gyrus, corresponding to Brodmann Area 46), as displayed by the brain navigation software.

All patients were stimulated with an eight-figured coil for a total of 15 days (Monday to Friday, every week, for three weeks). A single stimulation session consisted of five trains of 60 stimuli, with each train separated from the subsequent by one minute of pause, and a total of 300 stimuli per session (4,500 stimuli over the three weeks of treatment).

Concomitant treatment

During the stimulation period, all patients were maintained on their current pharmacological treatment, consisting of selective serotonin reuptake inhibitors, duloxetine, or bupropion. These treatments had been administered at adequate and stable doses for at least six weeks without providing significant improvement of symptoms (HAM-D total score improvement ≤ 25% compared to pretreatment scores). Valproate treatment was kept stable for the whole duration of the study.

Data analysis and statistics

Baseline demographic and clinical characteristics of the samples were tabulated with descriptive statistics. Baseline to endpoint changes in primary and secondary outcome measures—HAM-D, MADRS, and CGI-S—were analyzed using ANOVAs with repeated measures. Bonferroni’s pairwise comparisons across the four different assessment times were done as well. Finally, the total numbers of responders (HAM-D total score reduction ≥50% with respect to baseline), partial responders (HAM-D total score reduction >25% and <50% with respect to baseline), and remitters (HAM-D total score at endpoint ≤ 8) were computed. For all the statistical analyses, the alpha level of significance was set at 0.05, and was not adjusted. All the statistical analyses were performed using SPSS software for Windows (version 14.0; SPSS, Chicago, IL, USA).


Demographic and clinical characteristics of study subjects at baseline are summarized in Table 1. All subjects completed the three weeks of treatment. ANOVAs with repeated measures performed on HAM-D, MADRS, and CGI-S total scores showed a statistically significant time effect for all the outcome measures: = 22.36, p < 0.0001; = 12.66, p < 0.0001; and = 10.41, p < 0.0001, respectively (Fig. 2). None of the patients developed manic/hypomanic episodes during the three weeks of treatment. In addition, repeated-measures ANOVA done on the YMRS scores showed no effect of the treatment (repeated-measures ANOVA: = 0.62, p = 0.61). Stimulation response, defined as an endpoint HAM-D score reduction of ≥ 50% compared to baseline, was achieved by 6/11 subjects, 4 of whom were considered remitters (HAM-D endpoint score ≤ 8). Partial response (endpoint HAM-D score reduction between 25% and 50%) was achieved by 3/11 patients. Taken as a whole, 9/11 patients (82% of the sample) showed a benefit from the stimulation. There were no serious adverse events during the trial. Side effects were reported by three subjects and consisted of early insomnia and headache. However, these side effects were mild and limited to the first week of treatment, and thus did not require any specific intervention.

Table 1.   Baseline demographic and clinical characteristics of study subjects
CharacteristicsValue (n = 11)
  1. aOral doses varied according to valproate plasma levels that had to be between 50 and 100 μg/ml.

  2. TMS = transcranial magnetic stimulation.

Age (years), mean ± SD 54.36 ± 10.82
Gender, n (%)
 Women 8 (72.7)
 Men 3 (27.3)
Marital status, n (%)
 Single 1 (9.1)
 Married 9 (81.8)
 Divorced 1 (9.1)
Occupational status, n (%)
 Employed, full time 5 (45.5)
 Employed, part time 1 (9.0)
 Unemployed 5 (45.5)
Age at onset (years), mean ± SD 39.64 ± 13.47
Age at first treatment (years), mean ± SD 40.36 ± 13.46
Diagnosis, n (%)
 Bipolar disorder type I 5 (45.5)
 Bipolar disorder type II 6 (54.5)
Pharmacological treatment during TMS trial, n (%)
 Duloxetine (60 mg/day) plus valproatea 5 (54.5)
 Fluoxetine (40 mg/day) plus valproatea 1 (9.1)
 Sertraline (150 mg/day) plus valproatea 2 (18.2)
 Paroxetine (40 mg/day) plus valproatea 1 (9.1)
 Citalopram (40 mg/day) plus valproatea 1 (9.1)
 Bupropion (150 mg/day) plus valproatea 1 (9.1)
Number of pharmacological trials failed before entering the study, mean ± SD 2.3 ± 1.2
Duration of current depressive episode (weeks), mean ± SD 53.3 ± 18.4
Time since last manic/hypomanic episode (weeks), mean ± SD 62.2 ± 25.7
Figure 2.

 Baseline to endpoint changes of 21-item Hamilton Depression Rating Scale (HAM-D)a, Montgomery-Åsberg Depression Rating Scale (MADRS)b, and Clinical Global Impression severity of illness scale (CGI-S)c total scores. Statistical measurements are presented as ANOVAs with repeated measures. aHAM-D: = 22.36; p < 0.0001; pairwise comparisons T2 versus T1: mean difference = −6.9; p = 0.01. bMADRS: = 12.66; p < 0.0001; pairwise comparisons T3 versus T1: mean difference = −10.1; p = 0.02. cCGI-S: = 10.41; p < 0.0001; pairwise comparisons T3 versus T1: mean difference = −1.36; p = 0.05.


Results from this preliminary open-label study showed that augmentative low-frequency rTMS of the right DLPFC combined with brain navigation was effective and well tolerated in a small sample of drug-resistant bipolar depressive subjects.

With regard to stimulation parameters used, the choice of low-frequency rTMS was motivated by theoretical reasons such as a lower risk of accidental seizure (33) and a better tolerability (13). In terms of stimulation site, besides positive results reported with right DLPFC stimulation in major depressives (13, 23), some neurophysiological and imaging studies have pointed out a contrasting role in mood regulation between right and left hemispheres (34, 35). Thus, it has been hypothesized that right DLPFC stimulation at low frequency may produce the same antidepressant effect as left DLPFC stimulation at high frequency (13). With regard to stimulation intensity, studies utilizing higher intensities, as in the present study, seemed to be more likely to give positive results (36), even though a systematic investigation of this aspect is still lacking. Of clinical interest, a longer duration of treatment (three weeks) was used in the present trial in comparison to the standard duration (two weeks) of most acute clinical trials with rTMS. Converging evidence, in fact, indicates that a stimulation course longer than two weeks may be necessary to achieve a better response (13, 14, 23). In fact, studies reporting an rTMS course longer than two weeks have generally demonstrated continued improvement over the third and fourth week. However, most of these studies, but not all (14, 23, 37), were performed with no sham control (38–40). Moreover, it is well established that any antidepressant treatment, either pharmacological or nonpharmacological, with the exception of ECT, has a latency of efficacy not shorter than three weeks (41). Thus, it is to be expected that a treatment period longer than two weeks would be necessary for optimal outcomes with rTMS.

In the present study, a lower number of stimuli per session (300) were delivered, compared to the number of stimuli per session administered in the majority of clinical trials with rTMS. Considering the low frequency of stimulation, this resulted in a good tolerability with no dropouts and few, mild side effects. From this perspective, a recent small sham-controlled trial (23) of low-frequency rTMS delivered for four weeks to the right DLPFC in bipolar depressive subjects used an even lower number of stimuli per session (100) than the present study (300), with treatments scheduled twice a week for four consecutive weeks (total number of 800 stimuli versus 4,500 in the present study). Taken as a whole, the present trial and the study by Tamas et al. (23), though limited by the small samples, may suggest that it is not only the number of stimuli per session that is relevant to the outcome, but the duration of the trial as well, and that a lower number of stimuli per session, with sessions distributed over a stimulation course longer than two weeks, may translate to good efficacy and safety.

Another point that deserves to be mentioned is that study participants were nonresponders to pharmacotherapy and that, during the three weeks of stimulation, they were maintained on their previous treatment consisting of antidepressants and valproate (in therapeutic plasmatic concentration) at the same doses.

Finally, a neuronavigational system using patients’ brain MRI was used to position the magnetic coil above the selected cortical region. Actually, in most published clinical trials with rTMS stimulating the DLPFC, this region is usually determined by identifying the optimal site on the primary motor cortex for the stimulation of the controlateral peripheral hand muscle, and from there the coil is placed 5 cm forward. However, this empirical method for locating the DLPFC seems to be anatomically imprecise and may be improved by navigating procedures taking individual anatomy into account (42). From this perspective, the use of brain navigation in the present study and the more precise target area localization might have contributed to the efficacy of treatment.

Nevertheless the open-label design of the present study with the lack of a sham-controlled group needs to be considered as a methodological limitation, and further sham-controlled studies are needed to confirm these preliminary results. In particular, it would be useful to assess the duration of the improvement obtained with rTMS.


The authors thank Federico Resta, Radiology Unit, Hospital ‘L. Sacco,’ and Monica F. Bosi, Department of Clinical Sciences ‘L. Sacco,’ Milan, Italy. The present study was partially supported by European Grant STREP LSHM-CT-2205-51818.