Background
Methods
Systematic literature search and study selection
Search | k studies | Search terms | Databases (time frame) |
---|---|---|---|
DTMS studies with H1-coil | 24 (with duplicates) | TI (“deep transcranial magnetic stimulation” OR “deep repetitive transcranial magnetic stimulation” OR deep rTMS OR deepTMS OR deep TMS OR H-coil) AND TI (depress* OR dysthymi* OR MDD) | PsycInfo, Medline (EBSCO); any date – 24.06.2016 |
rTMS studies with F8-coil | 236 (with duplicates) | TI (“repetitive transcranial magnetic stimulation” OR rTMS OR HF-rTMS OR TMS OR “transcranial magnetic stimulation”) AND TI (depress* OR dysthy* OR MDD OR antidepress*) AND TX (“high-frequency” OR “20 Hz”) NOT TI (bilateral OR review OR meta-analysis OR meta-analyses OR case OR bipolar OR “Parkinson’s Disease” OR “posttraumatic stress disorder” OR tinnitus OR “deep transcranial magnetic stimulation” OR “deep TMS” OR H-coil) | PsycInfo, Medline (EBSCO); any date – 24.06.2016 |
Data coding
Study | Sham-controlled RCT | Sample size baseline | Age all patients baseline (mean ± SD) | Female patients baseline | Concurrent antidepressants (% patients baseline) | Dropouts daily stimulation phase before session 10 (number of patients and reasons) | Treatment-resistance definition | Mean onset age (years) | Mean illness duration (years) |
---|---|---|---|---|---|---|---|---|---|
DTMS (H1-coil) | |||||||||
Levkovitz et al., 2009 [15]; Israela,b | No | 23 | 46 | 48% | 0% | 3 (all for reasons unrelated to treatment or a good response) | Did not respond to ≥2 antidepressant medications current episode | 29 | 17 |
Rosenberg et al., 2010 [16]; Israel | No | 7 | 47 ± 12 | 14% | 0% | 0 | Failed 2 antidepressant trials current episode | 33 | 14 |
Rosenberg et al., 2010 [17]; Israel | No | 6 | 41 ± 13 | 67% | 50% | 0 | Failed ≥2 antidepressant courses and ≥ 1 course of ECT | 17 | 24 |
Isserles et al., 2011 [18]; Israelc | No | 25 | 45 ± 13 | 45% | 100% | 5 (1 seizure, 1 intolerance, 2 lack of response, 1 high motor threshold) | Failed ≥2 antidepressants | 29 | 17 |
Harel et al., 2014 [19]; Israelb | No | 29 | 41 ± 11 | 48% | 62% | 3 (1 safety reasons <1st treatment, 2 non-compliance with study protocol) | Failed ≥1 pharmacological trial or intolerant to 2 antidepressants | 24 | 17 |
Levkovitz et al., 2015 [9]; multicentreb,d | Yes | 89 | 45 ± 12 | 48% | 0% | 6 (1 seizure, 3 no improvement, 1 withdrawal of consent, 1 missed > 2 treatment days) | 93% failed ≥1 medication | 25 | 20 |
Rapinesi et al., 2015 [20]; Italyb | No | 9 | 54 ± 6 | 44% | 89% | 0 | Failure to respond to ≥3 adequate doses of ≥2 classes antidepressants | 45 | 9 |
Rapinesi et al., 2015 [21]; Italyb,e | No | 12 | 51 ± 8 | 42% | 100% | 0 | Unsatisfactory response to ≥1 adequate course of antidepressant treatment current episode | 34 | 17 |
rTMS (F8-coil) | |||||||||
George et al., 1997 [22]; USAb,f | Yes | 7 | 42 ± 16 | 86% | some | 0 | Received 1–21 previous medications | – | – |
Berman et al., 2000 [23]; USAb,d | Yes | 10 | 45 ± 10 | 20% | 0% | 0 | Failed ≥1 adequate pharmacologic trial current or previous episodes | 20 | 25 |
Catafau et al., 2001 [24]; Spainb | No | 5 | 50 ± 8 | 60% | 100% | 0 | HDRS17 score > 18 after ≥6 weeks of treatment | – | – |
Garcia-Toro et al., 2001 [25]; Spainb,d | Yes | 17 | 52 ± 16 | 41% | 100% | 0 | Failed 2 antidepressant trials for ≥6 weeks current episode | 36 | 16 |
Garcia-Toro et al., 2001 [26]; Spaind | Yes | 11 | 43 ± 13 | 55% | 100% | 3 (no treatment-related adverse reactions) | Failed 1 antidepressant trial | – | – |
Boutros et al., 2002 [27]; USAd | Yes | 12 | 49 ± 8 | 33% | 100% | 1 (worsening of depression) | Failed ≥2 prior medication trials of adequate length and dosages | – | – |
Bajbouj et al., 2005 [28]; Germany | No | 30 | 46 ± 12 | 37% | 40% | 0 | Some non-responders to current treatment with antidepressants | – | – |
Yukimasa et al., 2006 [29]; Japan | No | 26 | 53 ± 18 | 58% | some | 0 | Failed ≥2 antidepressant trials of adequate duration and dosages | – | – |
Luborzewski et al., 2007 [30]; Germanyb | No | 17 | 46 ± 11 | 12% | 53% | 0 | Completed 3–19 antidepressant treatment trials | – | – |
Bakim et al., 2012 [31]; Turkeyb,g | Yes; 80% | 12 | 39 ± 10 | 83% | 100% | 0 | No response to adequate courses (≥6 weeks) of ≥2 different classes of antidepressants at optimal doses | 36 | 3 |
Bakim et al., 2012 [31]; Turkeyb,g | Yes; 110% | 11 | 43 ± 8 | 91% | 100% | 0 | No response to adequate courses (≥6 weeks) of ≥2 different classes of antidepressants at optimal doses | 40 | 3 |
Chen et al., 2013 [32]; Taiwand | Yes | 10 | 44 ± 4 | 70% | 100% | 0 | No response to 2 different antidepressants administered for 6 weeks each | – | – |
Study | PFC location | Location definition | Frequency (Hz) | Intensity (% MT) | Coil type | Total stimuli | Stimuli/session | Trains/session | Inter-train interval (s) | No. of sessions |
---|---|---|---|---|---|---|---|---|---|---|
DTMS (H1-coil) | ||||||||||
Levkovitz et al., 2009 [15] | L | 5.5 cm | 20 | 120 | H1 | 16,800 | 1680 | 42 | 20 | 10 |
Rosenberg et al., 2010 [16] | L | 5.5 cm | 20 | 120 | H1 | 16,800 | 1680 | 42 | 20 | 10 |
Rosenberg et al., 2010 [17] | L | 5.5 cm | 20 | 120 | H1 | 16,800 | 1680 | 42 | 20 | 10 |
Isserles et al., 2011 [18] | L | 5.5 cm | 20 | 120 | H1 | 16,800 | 1680 | 42 | 20 | 10 |
Harel et al., 2014 [19] | L | 6 cm | 20 | 120 | H1 | 16,800 | 1680 | 42 | 20 | 10 |
Levkovitz et al., 2015 [9] | L | 6 cm | 18 | 120 | H1 | 19,800 | 1980 | 55 | 20 | 10 |
Rapinesi et al., 2015 [20] | L | 5.5 cm | 18 | 120 | H1 | 19,800 | 1980 | 55 | 20 | 10 |
Rapinesi et al., 2015 [21] | L | 5.5 cm | 18 | 120 | H1 | 19,800 | 1980 | 55 | 20 | 10 |
rTMS (F8-coil) | ||||||||||
George et al., 1997 [22] | L | 5 cm | 20 | 80 | F8 | 8000 | 800 | – | 58 | 10 |
Berman et al., 2000 [23] | L | 5 cm | 20 | 80 | F8 | – | – | 20 | 58 | 10 |
Catafau et al., 2001 [24] | L | 5 cm | 20 | 90 | F8 | 12,000 | 1200 | 30 | 30 | 10 |
Garcia-Toro et al., 2001 [25] | L | 5 cm | 20 | 90 | F8 | – | – | 30 | 30 | 10 |
Garcia-Toro et al., 2001 [26] | L | 5 cm | 20 | 90 | F8 | 12,000 | 1200 | 30 | 30 | 10 |
Boutros et al., 2002 [27] | L | 5 cm | 20 | 80 | F8 | 8000 | 800 | 20 | 58 | 10 |
Bajbouj et al., 2005 [28] | L | 5 cm | 20 | 100 | F8 | 20,000 | 2000 | 50 | – | 10 |
Yukimasa et al., 2006 [29] | L | 5 cm | 20 | 80 | F8 | 8000 | 800 | – | – | 10 |
Luborzewski et al., 2007 [30] | L | 5 cm | 20 | 100 | F8 | 20,000 | 2000 | 50 | – | 10 |
Bakim et al., 2012 [31] 80% | L | 5 cm | 20 | 80 | F8 | 8000 | 800 | 20 | 60 | 10 |
Bakim et al., 2012 [31] 110% | L | 5 cm | 20 | 110 | F8 | 8000 | 800 | 20 | 60 | 10 |
Chen et al., 2013 [32] | L | 5 cm | 20 | 90 | F8 | – | – | 20 | 10 | 10 |
Study | Response rate (session 10) | Remission definition | Remission rate (session 10) | Scale | Baseline daily phase; Mean ± SD (n) | Session 10 daily phase; Mean ± SD (n) |
---|---|---|---|---|---|---|
DTMS (H1-coil) | ||||||
Levkovitz et al., 2009 [15]a,b | 45% (9/20) | HDRS≤10 | 20% (4/20) | HDRS24 | 31 ± 5 (20) | 19 ± 8 (20) |
Rosenberg et al., 2010 [16] | 29% (2/7) | HDRS≤10 | 14% (1/7) | HDRS24 | 27 ± 4 (7) | 18 ± 6 (7) |
Rosenberg et al., 2010 [17] | 50% (3/6) | HDRS≤10 | 17% (1/6) | HDRS24 | 31 ± 4 (6) | 17 ± 7 (6) |
Isserles et al., 2011 [18]c | – | – | – | HDRS24 | 29 ± 6 (20) | 16 ± 4 (20) completers |
Harel et al., 2014 [19]b | – | – | – | HDRS21 | 23 ± 4 (29) | 17 ± 3 (26) completers |
Levkovitz et al., 2015 [9]b,d | 15% (13/89) | HDRS≤10 | 7% (6/89) | HDRS21 | 24 ± 4 (89) | 18 ± 6 (83) |
Rapinesi et al., 2015 [20]b | 0% | HDRS≤10 | 11% (1/9) | HDRS21 | 24 ± 3 (9) | 15 ± 3 (9) |
Rapinesi et al., 2015 [21]b,e | 0% | HDRS≤7 | 0% | HDRS17 | 27 ± 6 (12) | 22 ± 5 (12) |
rTMS (F8-coil) | ||||||
George et al., 1997 [22]b,f | 14% (1/7) | HDRS≤10 | 14% (1/7) | HDRS21 | 30 ± 4 (7) | 23 ± 9 (7) |
Berman et al., 2000 [23]b,d | 10% (1/10) | HDRS≤10 | 10% (1/10) | HDRS25 | 37 ± 10 (10) | 25 ± 9 (10) |
Catafau et al., 2001 [24]b | 40% (2/5) | HDRS≤7 | 0% | HDRS17 | 22 ± 4 (5) | 19 ± 9 (5) |
Garcia-Toro et al., 2001 [25]b,d | – | – | – | HDRS21 | 27 ± 7 (17) | 20 ± 6 (17) |
Garcia-Toro et al., 2001 [26]d | 36% (4/11) | – | – | HDRS21 | 26 ± 6 (11) | 16 ± 8 (11) completers |
Boutros et al., 2002 [27]d | 25% (3/12) | HDRS≤10 | 8% (1/12) | HDRS25 | 41 ± 10 (12) | 29 ± 14 (12) LOCF |
Bajbouj et al., 2005 [28] | 33% (10/30) | – | – | HDRS24 | 26 ± 7 (30) | 18 ± 9 (30) |
Yukimasa et al., 2006 [29] | 19% (5/26) | HDRS≤7 | 27% (7/26) | HDRS17 | 21 ± 5 (26) | 16 ± 7 (26) |
Luborzewski et al., 2007 [30]b | 35% (6/17) | HDRS≤10 | 29% (5/17) | HDRS28 | 25 ± 7 (17) | 19 ± 11 (17) |
Bakim et al., 2012 [31] 80%b,g | – | – | – | HDRS17 | 23 ± 4 (12) | 16 ± 6 (12) |
Bakim et al., 2012 [31] 110%b,g | – | – | – | HDRS17 | 24 ± 3 (11) | 17 ± 5 (11) |
Chen et al., 2013 [32]d | 70% (7/10) | – | – | HDRS17 | 24 ± 2 (10) | 10 ± 2 (10) |
Outcome measures
Data analysis
Results
Study designs and patient characteristics
Stimulation parameters
Acceptability
Antidepressant outcomes: all studies (rTMS with F8-coil and DTMS with H1-coil)
Random-effects analyses | Primary outcome (depression severity); Hedges’ g (95% CI) | Secondary outcome (response rates); responders/total n (95% CI) | Secondary outcome (remission rates); remitters/total n (95% CI) |
---|---|---|---|
Pooled weighted effect | |||
Mean (95% CI); k; n | 1.20 (.96–1.44); k = 20; n = 351 | 29% (20–39%); k = 15; n = 271 (66/271) | 15% (10–22%); k = 12; n = 220 (28/220) |
Heterogeneity statistics | Q (df 19) = 49.12; p < .001*; I2 = 61% | Q (df 14) = 27.26; p = .018*; I2 = 49% | Q (df 11) = 12.19; p = .350; I2 = 10% |
Publication bias analysis | |||
Fail-Safe N Rosenthal/Orwin | NRosenthal = 1372; NOrwin = 182 | NRosenthal = 124; NOrwin = 25 | NRosenthal = 187; NOrwin = 8 |
Funnel plot symmetric? | No: k = 4 missing with small effect sizes | No: k = 2 missing with large response rates | No: k = 6 missing with large remission rates |
Mean effect (95% CI) adjusted for missing studies | 1.05 (.78–1.32) | 31% (22–42%) | 23% (14–34%) |
Subgroup analysis | |||
Study design | |||
RCT | 1.17 (.86–1.48); k = 9; n = 173 | 26% (13–47%); k = 6; n = 139 (29/139) | 8% (4–14%); k = 4; n = 118 (9/118) |
Open-label | 1.24 (.87–1.61); k = 11; n = 178 | 32% (23–42%); k = 9; n = 132 (37/132) | 21% (14–31%); k = 8; n = 102 (19/102) |
RCT vs. open-label | Q (df 1) = .14, p = .712 | Q (df 1) = 3.03, p = .082 | Q (df 1) = 7.31, p = .007* |
Therapy | |||
Add-on | 1.20 (.89–1.50); k = 16; n = 231 | 31% (22–43%); k = 11; n = 145 (41/145) | 20% (13–31%); k = 8; n = 94 (16/94) |
Monotherapy | 1.21 (.98–1.44); k = 4; n = 120 | 23% (10–45%); k = 4; n = 126 (25/126) | 11% (6–19%); k = 4; n = 126 (12/126) |
Add-on vs. monotherapy | Q (df 1) = 1.15, p = .283 | Q (df 1) = 3.09, p = .079 | Q (df 1) = 3.63, p = .057 |
Meta-regression predictors | |||
Mean age | b = −.05; p = .019*; R2 = 45%; k = 18 | b = −.11; p = .123; R2 = 2%; k = 15 | b = .06; p = .337; R2 = 23%; k = 12 |
% female | b < .01; p = .642; R2 = 0%; k = 19 | b < .01; p = .491; R2 = 0%; k = 15 | b < −.01; p = .696; R2 = 0%; k = 12 |
Mean illness duration | b < −.01; p = .681; R2 = 0%; k = 12 | b = .05; p = .629; R2 = 0%; k = 7 | b = −.04; p = .620; R2 = 0%; k = 7 |
Mean onset age | b < .01; p = .913; R2 = 0%; k = 12 | b = −.06; p = .327; R2 = 0%; k = 7 | b = .01; p = .811; R2 = 0%; k = 7 |
Stimuli/session | b < .01; p = .415; R2 = 0%; k = 17 | b < .01; p = .964; R2 = 0%; k = 13 | b < −.01; p = .313; R2 = 25%; k = 11 |
Trains/session | b < .01; p = .804; R2 = 0%; k = 17 | b = −.03; p = .132; R2 = 32%; k = 13 | b < −.01; p = .858; R2 = 0%; k = 10 |
Intensity (%MT) | b = .02; p = .007*; R2 = 37%; k = 19 | b < −.01; p = .897; R2 = 0%; k = 15 | b = −.02; p = .087; R2 = 100%; k = 12 |
Inter-train interval (s) | b = −.01; p = .102; R2 = 0%; k = 17 | b = −.02; p = .234; R2 = 0%; k = 12 | b < .01; p = .996; R2 = 0%; k = 10 |
Primary outcome (depression severity)
Secondary outcome (response rates)
Secondary outcome (remission rates)
Antidepressant outcomes: rTMS with F8-coil vs. DTMS with H1-coil
Mixed-effects analyses | Primary outcome (depression severity); Hedges’ g (95% CI) | Secondary outcome (response rates); responders/total n (95% CI) | Secondary outcome (remission rates); remitters/total n (95% CI) |
---|---|---|---|
Pooled weighted effects | |||
Mean (95% CI); k; n | |||
DTMS (H1-coil) | 1.55 (1.17–1.94); k = 8; n = 183 | 24% (11–44%); k = 6; n = 143 (27/143) | 10% (6–17%); k = 6; n = 143 (13/143) |
rTMS (F8-coil) | .97 (.70–1.25); k = 12; n = 168 | 31% (22–43%); k = 9; n = 128 (39/128) | 22% (14–33%); k = 6; n = 77 (15/77) |
H1-coil vs. F8-coil | Q (df 1) = 9.32, p = .002* | Q (df 1) = 2.46, p = .116 | Q (df 1) = 4.46, p = .035* |
Heterogeneity statistics | |||
DTMS (H1-coil) | Q (df 7) = 16.89, p = .018*; I2 = 58% | Q (df 5) = 14.13, p = .015*; I2 = 65% | Q (df 5) = 4.05, p = .543; I2 = 0% |
rTMS (F8-coil) | Q (df 11) = 22.91, p = .018*; I2 = 52% | Q (df 8) = 10.67, p = .221; I2 = 25% | Q (df 5) = 3.68, p = .596; I2 = 0% |
Sensitivity analyses | |||
Open-label studies only | |||
DTMS (H1-coil) | 1.67 (1.24–2.11); k = 7; n = 100 | 29% (12–53%); k = 5; n = 54 (14/54) | 15% (8–28%); k = 5; n = 54 (7/54) |
rTMS (F8-coil) | .74 (.50–.98); k = 4; n = 78 | 30% (21–41%); k = 4; n = 78 (23/78) | 26% (16–41%); k = 3; n = 48 (12/48) |
H1-coil vs. F8-coil | Q (df 1) = 19.14, p < .001* | Q (df 1) = .34, p = .559 | Q (df 1) = 1.76, p = .184 |
Add-on studies only | |||
DTMS (H1-coil) | 1.77 (1.12–2.42); k = 5; n = 73 | 14% (2–62%); k = 3; n = 27 (3/27) | 10% (3–30%); k = 3; n = 27 (2/27) |
rTMS (F8-coil) | .96 (.67–1.26); k = 11; n = 158 | 33% (24–44%); k = 8; n = 118 (38/118) | 23% (14–35%); k = 5; n = 67 (14/67) |
H1-coil vs. F8-coil | Q (df 1) = 11.46, p = .001* | Q (df 1) = .52, p = .473 | Q (df 1) = 1.66, p = .197 |