NEUROIMAGING CENTER
Brain Imaging Center
LFMS: Low Field Magnetic Stimulation
Low Field TMS Treatment for Bipolar Depression Using a MRI Based Stimulator
Michael Rohan*†, Aimee Parow*, Andrew L. Stoll*‡†, Christina Demopulos˚‡,
Seth Friedman1, Stephen Dager1, John Hennen*‡, Bruce M. Cohen*‡†,
Perry F. Renshaw*‡† *McLean Hospital Brain Imaging Center, ‡Harvard Medical School
Department of Psychiatry, ˚Massachusetts General Hospital,
†Stanley Foundation Bipolar Studies Research Center at McLean Hospital,
1University of Washington
Background
Existing treatment approaches for bipolar disorder primarily utilize pharmacologic agents, such as lithium and valproic acid or antipsychotic and antidepressant drugs, that too often are of limited efficacy and frequently have objectionable side effects. ECT is usually effective for bipolar disorder but involves general anesthesia, some degree of memory loss, and its effects can be transient. Transcranial Magnetic Stimulation (TMS), initially developed to test gross central nervous system function, has more recently been applied with some success in the treatment of bipolar disorder (1). TMS uses a small magnetic coil and large power supply to apply a short, strong magnetic field pulse to the motor cortex for assessment of nerve function; for depression treatments the fields are applied to the left prefrontal cortex. The coil is a loop of several centimeters, with a correspondingly local field extent. A short pulse induces a local electric field pulse that prompts neuron firing. Therapeutic psychiatric uses for TMS have involved repetitive TMS (rTMS) which has a train of field pulses typically at 1 - 20 Hz. rTMS treatment can be unpleasant, with some patients declining participation due to scalp pain induced by the apparatus (2). It also carries a small risk of seizure (3).
At the McLean Hospital Brain Imaging Center (BIC) there are several ongoing proton echo-planar magnetic resonance spectroscopic imaging (EP-MRSI) studies of subjects with bipolar disorder. These studies employ the oscillating magnetic fields associated with functional MRI (fMRI), but which differ from the usual fMRI scan in field direction, waveform frequency and duration. This EP-MRSI scan has longer acquisition times and lower readout gradients than the standard fMRI echo-planar imaging sequences, and these readout gradients are employed in the anterior-posterior (A/P) direction. A/P gradients induce electric fields in the prefrontal cortex and throughout the brain. Following the serendipitous observation of mood improvement during EP-MRSI studies of depressed bipolar subjects, we obtained clinical data of such mood changes systematically and prospectively.
In this study we test the hypothesis that the EP-MRSI scan has mood enhancing effects on subjects with bipolar disorder, compared to bipolar subjects given a sham EP-MRSI scan, and compared to healthy subjects given the EP-MRSI scan.
Methods: Study Population
The study population was comprised of participants in three studies of subjects with bipolar disorder that shared the same EP-MRSI scan prescription and clinical interview scheme. These studies were investigating the effects of conventional and non-conventional (omega3 fatty acids supplements) on mood over a period of time, and involved EP-MRSI scans and clinical interviews on a monthly basis. These studies used the results of the EP-MRSI scan and clinical rating scales to evaluate the effects of specific medication interventions; subjects were blinded to the hypothesis that the EP-MRSI exam itself was being investigated for correlation with mood change. Subjects in the three source studies came in for multiple visits; in this study we examine results from only the first visits by these subjects, because many of the subjects had medication changes after the first visit as part of their participation.
Subjects with bipolar disorder had a diagnosis of bipolar I or II disorder, currently fulfilled the DSM-IV criteria for either depression or mania, and were between the ages of 18 and 65. They were either currently on a course of medication limited to the conventional mood stabilizers lithium and depakote, or were medication naïve at the start of the study. Subjects that were given anxiolytic medication during the scan sessions or that were taking medication in addition to that listed above were not considered in this study. Thirty (n=30; 16 female, 14 male) subjects with bipolar disorder received EP-MRSI scans; 11 of these subjects were medication na¨ve. Ten (n=10; 5 female, 5 male) subjects with bipolar disorder received sham EP-MRSI scans; 2 of these subjects were medication naïve. Fifteen (n=15; 9 female, 6 male) healthy subjects received EP-MRSI scans. Subject demographic data is summarized in Table 1.
Written informed consent obtained from all subjects and study protocols were approved by the McLean Hospital Institutional Review Board.
Methods: MRI
Studies were conducted at the McLean Hospital Brain Imaging Center. Scanning was performed on a General Electric 1.5T Signa MRI scanner (5.8 EchoSpeed version). At each clinic visit, 4 EP-MRSI scans were acquired. The EP-MRSI pulse sequence used (PEPSI) has been extensively described (4). The complete MR exam consisted of a conventional double echo - spin echo T2 scan, 4 EP-MRSI scans totaling 20.5 minutes, a T1 anatomic scan and an echo-planar T2 imaging acquisition; the entire study took about 1 hour. The sham EP-MRSI exam was identical to the exam described above, except that the EP-MRSI scans were omitted.
Methods: Clinical Ratings
Based on anecdotal reports of subjects experiencing enhanced mood following the EP-MRSI scanning session an additional rating scale was added to the study. The "Brief Affect Scale" (BAS) (Stoll, personal communication) was administered to all subjects immediately before and after the MR scanning session. The BAS is based on the Clinical Global Impression scale (5), and measures immediate mood state.
Please circle how you feel since the last time you filled out this questionnaire:
3. Very much improved
2. Much improved
1. Minimally improved
0. No change
-1. Minimally worse
-2. Much worse
-3. Very much worse
These numerically ranked responses are grouped into the ordinal groups of "improved" (3 to 1), "same" (0) and "worse" (-1 to -3) for statistical treatment. This grouping is referred to as "ordinal BAS ratings." At each visit the 17 item Hamilton Depression Rating Scale (HAMD) and the Young Mania Rating Scale (YMRS) were administered. All HAMD, YMRS and BAS ratings were administered by the same research assistant.
Methods: Statistics
Ordered logistic regression modeling methods were used to examine the differences in BAS scores among the study groups. Robust estimators of standard errors were obtained. Data were summarized as means (±SD) or by means with 95% confidence intervals (95% CI). Two sided significance tests, requiring p<0.05 for statistical significance, were employed. Statistical software systems used included Statview®, SAS Institute Inc., Cary NC and Stata®, Stata Corporation, College Station, TX.
Figure 1
Electric field magnitude. Each color band is 0.07 V/m light green is
0.63 to 0.70 V/m; the color bands in the two graphs show the same
values. The rTMS plot increases to 120 V/m too quickly for display.
Figure 2
Timing for the electric field waveforms in the two systems, at a point in the center of
the brain. 10 msec is shown; rTMS system repetition times vary from 25msec to 100 msec.
Results: EP-MRSI Electric Fields
The EP-MRSI scan is a spectroscopic imaging sequence; it combines an EPI-like readout train with CSI style spatial encoding to obtain a proton CSI image. The alternating readout gradient encodes the first spatial direction in the usual EPI manner. Spectroscopic information is acquired across the readout acquisition lines. The second spatial direction is encoded with a stepped phase encode pulse across excitations in CSI fashion. After water suppression, slice selective excitation, and a spatial phase encode, it applies a train of 512 alternating trapezoidal magnetic field pulses for a readout gradient. For this study the readout gradient was placed in the anterior-posterior (A-P) direction, using the ‘Y’ gradient coil. This is not the usual readout direction for an echo-planar style fMRI sequence, because an A-P readout is more likely to cause peripheral nerve stimulation in the body (6), (7). This is not a concern for EP-MRSI scans because the readout gradient, at 1/3 G/cm, is only 1/10 the strength of an echo-planar fMRI scan.
The EP-MRSI readout gradient waveform consists of 512 trapezoids each about 1 millisecond long. Each trapezoid has an amplitude of ± 0.33 G/cm, ramp up and ramp down times of 128 msec and a constant portion duration of 768 msec The trapezoids alternate in sign. The maximum magnetic field in the cortex delivered by this gradient is about 5G.
The vector potential produced by the readout gradient in this sequence coil was calculated from the wire pattern for the ‘Y’ gradient coil in the Signa MRI system using a Biot-Savart style integration (8), and the resulting electric field was determined from this vector potential and from the readout gradient waveform. The resultant electric field in the head is modified by surface charge at the edge of the brain (9), (10); this modification does not change the order of magnitude of the result, and for purposes of estimation we present the fields in free space The electric field in this study is spatially uniform to 5% throughout the head, aligned in the R-L direction, and has a magnitude of about 0.7 V/m. It is delivered in bursts of 512 pulses of alternating polarity; each pulse is a square shaped pulse with a duration of 256 msec delivered once every millisecond. This &fract12; second pulse train is repeated every 2 seconds for the duration of the scan. The electric field distribution for the EP-MRSI system is shown in Figure 1. The electric field waveform for the EP-MRSI system is shown in Figure 2.
Comparison: rTMS Electric Fields
For comparison, a typical rTMS coil consists of a small figure-8 coil (each loop about 4cm diameter) placed next to the scalp. It produces an electric field in the head as a consequence of the large magnetic field it generates. The magnetic field is highly non-uniform, ranging from 10,000G near the scalp to 11G on the contra-lateral side of the head.
The electric field induced by this magnetic field is calculated in a similar fashion from the vector potential of the rTMS coil and the rate of change of its magnetic field. The rTMS vector potential was calculated from a figure-8 model, tilted at 45° and placed at 6.7 cm vertically and horizontally from a position equivalent to the center of the head. The coil was modeled using an amount of current that produced a 1Tesla (10,000G) magnetic field at the position (6cm, 6cm), the lower range of a typical operating point for depression treatment with rTMS (2). The electric field is very non-uniform, ranging from 120 V/m at the scalp to less than 0.1 V/m on the contra-lateral side of the head, with about 1.8 V/m in the center of the brain. The electric field is delivered in a series of biphasic (single cycle cosines) pulses, each 560 msec in duration, at a rate of 1 to 20 per second. Treatment typically lasts about 20 minutes. The electric field distribution for the rTMS system is shown in figure 1. The electric field waveform is shown in figure 2.
Results: Clinical
Twenty-three of thirty subjects with bipolar disorder reported improvement in mood of at least 1 point on the BAS scale after EP-MRSI exams. "No change" was reported by 7 subjects, and a worsening of mood was reported by 1 subject. The mean BAS score for subjects with bipolar disorder who received EP-MRSI scans was 0.87 ± 0.68. In the subgroup of medication naïve subjects with bipolar disorder who received EP-MRSI, 11 of 11 subjects reported improvement in mood (mean BAS score = 1.18 ± 0.41), compared to reports of improvement by 12 of 19 subjects with bipolar disorder in the subgroup that was taking mood stabilizing medication (mean BAS score = 0.68 ± 0.75).
Three of ten subjects with bipolar disorder who received sham EP-MRSI scans reported improvement in mood after the exam, with 2 reports of worsening in mood. The mean BAS score for subjects with bipolar disorder who received sham EP-MRSI scans was 0.30 ± 1.06. Four of fifteen healthy subjects reported improvement in mood after an EP-MRSI exam, with no reports of worsening. The mean BAS score for healthy subjects who received EP-MRSI scans was 0.29 ± 0.47. Table 2 summarizes the BAS improvement scores.
Ordinal BAS ratings were compared between subjects with bipolar disorder who received EP-MRSI scans and subjects with bipolar disorder who received sham EP-MRSI using ordered logistic regression methods; this difference was statistically significant (z=2.63, p=0.009). Ordinal BAS ratings were also compared between subjects with bipolar disorder who received EP-MRSI and healthy subjects who received EP-MRSI; this difference was also statistically significant (z=2.61, p=0.009). The contrast between subjects with bipolar disorder who received sham EP-MRSI and healthy subjects who received EP-MRSI was not significant (z=0.030, p=0.77). A summary of these effect sizes is listed in Table 3.
Table 1. Summary of subjects in the MRSI study of mood effects.
| #subjects | Age | Female/Male | |
|
Bipolar, EP-MRSI
|
30 |
37±12
|
16/14 |
|
Bipolar, sham EP-MRSI
|
10 |
45±8
|
5/5 |
|
Comparison, EP-MRSI
|
14 |
30±6
|
8/6 |
| Total | 54 |
37±11
|
29/25 |
Table 2. Mood Improvement by BAS Score for Visit 1
| N | # improved | # worse | mean BAS | |
|
Bipolar, EP-MRSI
|
30 |
23
|
1 |
0.87±0.68
|
| subgroup: medication naïve | 11 | 11 | 0 |
|
|
|
19 | 12 | 1 |
0.68±0.75 |
|
Bipolar, sham EP-MRSI
|
10 |
|
2 |
0.30±1.06
|
|
Comparison, EP-MRSI
|
14 |
|
0 |
0.29±0.47
|
Table 3. Effect Size
| Z | p | |
|
Bipolar, EP-MRSI vs Bipolar, sham EP-MRSI
|
2.63 |
0.009
|
|
Bipolar, EP-MRSI vs Healthy, EP-MRSI
|
2.61 |
0.009
|
|
Healthy, EP-MRSI vs Bipolar, sham EP-MRSI
|
0.03 |
0.77
|
Discussion
We have found significant improvement of mood in subjects with bipolar disorder who received EP-MRSI scans. This change is absent in subjects with bipolar disorder who received sham EP-MRSI scans, and is also absent in healthy subjects given the EP-MRSI scans. A greater effect was evident in medication naïve subjects with bipolar disorder. Additionally, decreased symptom severity over time was observed in HAMD depression scores. However, these data are somewhat difficult to interpret because of medication confounds and the lack of a time trend data from a comparison group.
This prospective pilot study has a number of limitations. Changing medication states in the component studies prevented the use of longer time course observations, and a change in MRI system midway through the study with a corresponding suspension in EP-MRSI exams limited the size of the group receiving the sham EP-MRSI. Also, the serendipitous use of existing study samples may be considered to be a limitation. The elevated rates of reported mood improvement, however, suggest that there is a significant effect present.
The immediate improvement shown in the BAS scores indicates a surprising response to this treatment, particularly in the medication naïve subjects (100%) vs the subjects taking medication (63%) vs the subjects receiving sham EP-MRSI (30%). There were no adverse effects to this treatment. Out of 135 total visits made by subjects in the course of the component studies, only 8 were associated with a worsening of mood.
The mood improvement effect we have reported seems specific to this EP-MRSI scan. McLean Hospital is a psychiatric hospital, and has performed over 10,000 MRI and EPI MRI exams of psychiatric subjects over the last 10 years prior to the use of the EP-MRSI exam without observing a similar effect. The EP-MRSI gradient fields are the outstanding difference from the previous MRI exams.
These induced electric fields differ from rTMS fields in 4 ways. The first is timing; the EP-MRSI fields are applied on this scanner at 1 kHz, and are monophasic pulses while the rTMS fields are applied at 1Hz to 20Hz, and each pulse is a fast biphasic pulse. The second difference is the strength of the electric fields, reduced by a factor of 200 in the cortex relative to fields in the rTMS treatment. The third difference is in the uniformity of the electric field. The EP-MRSI fields are relatively uniform, up to a factor of about 2, throughout the brain while the rTMS fields vary by a factor of more than 200 from one side of the brain to the other. It is interesting to note that the electric field values for the two techniques are of similar strength in the midbrain regions. The last difference is the electric field direction; in the EP-MRSI treatment it is in the RL direction, while the rTMS fields resemble the fields from a point source.
We think that the monophasic pulses delivered at 1kHz in this system may interact with neuronal processes more efficiently than the biphasic pulses delivered by rTMS at 1 to 20 Hz. Further testing of this effect is planned, with attention to optimization of the field distribution and timing. We are also testing these proposals outside of an MRI system using a small whole head magnetic stimulation apparatus designed to duplicate the EP-MRSI fields. This system is sited in a standard clinical exam room, and allow a focused, double blinded clinical trial.
This work was supported in part by MH-58681, the Poitras Foundation, the Stanley Foundation Bipolar Disorders Research Center at McLean Hospital, and gifts from John and Virginia Taplin.
- Wassermann EM, Lisanby SH:. Biol Psychiatry 2000; 48(10):962-70.
- George MS, Nahas Z, Molloy M, Speer AM, Oliver NC, Li XB, Arana GW, Risch SC, Ballenger JC:. Biol Psychiatry 2000; 48(10):962-70.
- Wassermann EM: Electroencephalogr Clin Neurophysiol 1998; 108(1):1-16.
- Posse S, Dager SR, Richards TL, Yuan C, Ogg R, Artru AA, Muller-Gartner HW, Hayes C: Magn Reson Med 1997; 37(6):858-65.
- National Institute of Mental Health: CGI (Clinical Global Impression) scale. Psychopharmacol Bull 1985; 21:839-43
- Ehrhardt JC, Lin CS, Magnotta VA, Fisher DJ, Yuh WT: J Magn Reson Imaging 1997; 7(2):405-9.
- Reilly JP: Med Biol Eng Comput 1991; 29(6):571-9.
- Jackson JD: Classical Electrodynamics. New York, Wiley, 1975
- Roth BJ, Saypol JM, Hallett M, Cohen LG: Electroencephalogr Clin Neurophysiol 1991; 81(1):47-56.
- Post A, Keck ME: J Psychiatr Res 2001; 35(4):193-215.