One of the unusual aspects of the LFMS program is that this clinical effect was discovered by chance, without being developed at a laboratory level beforehand. It was not thought that these lower level electromagnetic fields could have an effect on the brain, but the dramatic and immediate effects on mood that have been observed suggest that at particular frequencies this occurs. It is not known exactly how LFMS causes its mood improving effects, why they occur so rapidly, or whether the current protocol is the most effective one possible. The main goals of the studies of mechanism described below are the optimization and improvement of the LFMS protocol, and finding an explanation of the cellular effects of LFMS as a new direction in the general study of depression.
LFMS is unique because of its rapid effect on mood. We can observe this rapid effect in action using Functional Magnetic Resonance Imaging (fMRI) and Electroencephalography (EEG). In the imaging studies described below we are observing this effect as it happens, an observation that is not possible with more slowly acting antidepressants. The imaging project also benefits because the effects are observed in healthy control subjects, and the use of real time observations will allow investigations to be completed quickly. The imaging work will support protocol optimization relying on maximizing the imaging effect, and will be extended to populations suffering from depression in a confirmation phase.
The results of imaging based optimization can be applied directly to clinical studies and protocols. A more basic effort will be directed towards explaining how the electromagnetic fields of LFMS are interacting with neuronal function. Theoretical work relying on field calculations and their interaction with neuronal simulations will help to identify the structures that are supporting the imaging results. Projects in global calculations as well as in local network simulation are planned, which will lead towards future pre-clinical studies in single-cell recording and optogenetic based studies of synaptic function.
Goals Using fMRI and EEG we will perform observations of brain activity before and after LFMS treatment to observe changes in brain activity. We will quantify them, and hypothesize that by maximizing the imaging effect we will increase the clinical effect. Phase 1 will confirm and quantify the imaging effects in control subjects. Phase 2 will confirm the imaging effects in subjects with depression. Phase 3 will optimize the LFMS protocol, using the imaging effect as a measure, by varying its parameters.
Background fMRI and EEG can provide a first look at how LFMS directly affects brain activity. There are characteristic changes in particular brain networks that are seen with fMRI and that are associated with depression and response to treatment. There are also characteristic changes in EEG power in particular brain regions and in particular frequency bands that are associated with depression. In our preliminary data we have observed changes in fMRI network activity, and in EEG power, in these brain regions after LFMS. These observations were made in healthy control subjects.
Facilities The LFMS group is located within the McLean Hospital Neuroimaging Center and has access to a Siemens 3T Prisma MRI Systems and a Siemens 3T TIM Trio MRI system. The Prisma system has been designed for simultaneous fMRI/EEG acquisitions. We currently use a 128 channel EEG system (EGI Inc.) with a 64 channel net for easier placement during experiments.
The link between imaging effects in control subjects and effects on mood in subjects with depression will be confirmed as part of this investigation. The possibility that optimization efforts could be pursued in control subjects, allowing faster study turnaround, is exciting.
Preliminary Imaging Results
Data has been acquired in 3 healthy subjects using fMRI and EEG in separate acquisitions. The fMRI and EEG protocols included included pre and post LFMS resting state scans in a double blinded, sham controlled experiment. Following this protocol, each subject participated in 8 fMRI scans and 4 EEG acquisitions, so that each group result is based on 24 fMRI acquisitions and 12 EEG acquisitions.
fMRI Methods fMRI data were analyzed using seed based connectivity. In this method the activity in the Medial PreFrontal Cortex (mPFC) was compared to timecourses in the rest of the brain, and the similarity quantified and compared between active and sham, pre and post acquisitions.
fMRI Results The fMRI data show a decrease in connectivity between the mPFC and the Posterior Cingulate Cortex (PCC), with an increase in connectivity between the mPFC and the bilateral insula. These changes are consistent with network models of depression in which a decrease in activity within the default mode network (mPFC and PCC) is associated with a reduction in depression, accompanied by an increase in activity within the salience network (mPFC and insulae, among others).
Images of activation with fMRI and EEG: left is seed-based connectivity to the dorsal anterior cingulate, right is gamma band EEG current density. Both are post-minus pre-LFMS, active vs. sham, paired analyses.
EEG Methods EEG resting state data was acquired before and after either active or sham LFMS, using the same protocol as the fMRI. Standard subject-wise analysis was performed and the group voxel wise significance estimated using permutation methods in sLORETA. Voxels are displayed with a threshold of |z|>2.3 (p<0.02). An increase in activity is red and a decrease is blue.
EEG Results The EEG data show an increase in gamma power in the mPFC, accompanied by a decrease in gamma power in the PCC. These regions correspond to the connectivity change observed with fMRI. While gamma power has not been studied in connection with depression, it is considered an indicator of cortical activity and could represent the immediate effects of LFMS on the network activity that is seen with fMRI.
Goals We will calculate the electric field that LFMS induces in the brain, and the effects on neuronal function that it produces. This will be done at several levels. First, closed form and Finite Element Method calculations have been performed that describe the distribution of field in the head during treatment. Second, calculations are being performed to determine what effect these fields within the brain have on the neuronal membrane voltage and function. Third, simulations of neuronal networks are being planned to assess the effects of changed neuronal function on standard models of neuronal network activity.
Background LFMS is in an unusual situation in that its clinical effects have been observed and are being tested in advance of the investigation of its mechanism of action. Most new therapies in the last few decades have been based on the translation of existing therapies to new molecular mechanisms in the brain. LFMS was discovered serendipitously, and investigations of its mechanism and clinical investigations will occur at the same time.
Goal The finite element method (FEM) was used to evaluate the electromagnetic fields within the head of a subject. The source of the fields was a continuous current distribution based on the LFMS coil, and a standard MRI based model of the human head was used to represent the patient. The purpose of this calculation was to determine the likely penetration and distribution of the electric fields induced by LFMS in a patient’s brain, given the effects of conductive tissue and shielding charge density. It was expected that the head, a weakly conducting object, would screen a significant portion of the free-air electric field of the device.
Methods An anatomic MR image (T1 contrast, 1mm resolution) of a 26-year old woman was used as a model for the head in this calculation. This model, which was truncated at the neck, was segmented into 7 tissue types that were used to construct an electrical conductivity map using published conductivity values. The volume within the head was represented as a mesh of nodes, obtained by using the standard geometric meshing algorithms of the FEM program; a conductivity value was assigned to each node to form a continuous electrical conductivity distribution.
Result The magnitude of electric fields in seven transverse slices of the brain during LFMS is shown with a wireframe model of the outside surface of the head superimposed to provide context. While “free space” electric field values reach 0.75 V/m at the surface of the head, and 0.51 V/m in the center, shielding by the head reduces the cortical fields to 0.25 V/m, and the fields in the center of the head are reduced to less than 0.05 V/m. These calculations suggest that electric field penetration during LFMS treatment can be expected to be strongest in cortical regions such as the prefrontal and orbitofrontal brain regions, as shown.
Return to top of page
Goal To assess the effects of the electric fields induced by LFMS on the membrane voltages in neurons.
Background Several calculations of the effects of constant fields and of low frequency fields (5Hz) have been performed in models of neurons that show a constant or slowly varying polarization of the membrane voltage along the axis of the dendrite. The effects of fields at higher frequencies such as 1 kHz, the frequency of LFMS, have not been evaluated. The high frequency results are expected to be different because the fields will change more quickly than the recovery time for action potentials, but within the times required for ion channels to operate. This suggests that specific neuronal components of structures are involved in mediating the effects of LFMS.
Methods Boundary value calculations for electric fields will be performed using a simple geometric model of a dendrite. The electric field of LFMS will be applied both parallel to and transverse to the axis of the dendrite. A first model will incorporate only the passive cable elements of the cell membrane and geometry. A second model will include active conductance elements to model ion channel thresholds.
Status: In planning.
Goal To investigate the effects of the electric fields of LFMS on cortical network function, using mathematical simulation. We will translate the changes caused on individual neuron function to the effects on network function in cortical regions using mathematical simulation.
Methods We will use existing network models that rely on simplified neurons. We will incorporate the modifications introduced into single neurons by LFMS into small-scale network models.
Status: In planning.