|Home | About | Journals | Submit | Contact Us | Français|
Transcranial magnetic stimulation (TMS) and repetitive TMS (rTMS) have been used increasingly over the past few years to study both the pathophysiology of schizophrenia as well as the utility of focal neuromodulation as a novel treatment for schizophrenia. rTMS treatment studies to date have explored its effect on both positive and negative symptoms by targeting cortical regions thought to underlie these symptom clusters. Studies on auditory hallucinations have been largely positive, while efficacy for negative symptoms is equivocal. A better understanding of the functional abnormalities that accompany symptoms may facilitate the development of rTMS as a treatment modality. Furthermore, schizophrenia patients appear to have abnormal cortical inhibition, consistent with GABA and dopamine abnormalities in schizophrenia. The effect of TMS on GABA and dopamine neurotransmission has not been clearly delineated. Given the variability in cortical response to rTMS in schizophrenia, methods to optimize dosage are essential. Consideration of these factors among others may broaden the scope of utility of TMS for schizophrenia as well as enhance its efficacy.
Transcranial magnetic stimulation (TMS) is a noninvasive method of altering brain activity (Barker et al., 1985) in which time-varying magnetic fields induce current in the cerebral cortex (Barker et al., 1987). Various modes of TMS have been utilized to assess intracortical excitation and inhibition (Pascual-Leone et al., 1998), map cortical representation of motor, sensory, and cognitive function (Wassermann, 1998), explore the functional connectivity of different brain regions (Hoffman et al., 2007), and modulate regional brain activity for therapeutic effect (Wassermann and Lisanby, 2001). Here we review TMS treatment studies that have targeted various symptom domains in schizophrenia patients, theorize on how TMS studies of pathophysiology in these patients may be exploited to maximize efficacy and discuss future directions for TMS research in schizophrenia.
Repetitive TMS (rTMS) produces targeted changes in neurophysiological measures lasting minutes after completion of the rTMS train (Chen et al., 1997; Pascual-Leone et al., 1994). Low-frequency rTMS (≤ Hz) has an inhibitory effect on motor cortical excitability in healthy individuals (Chen et al., 1997; Wassermann and Lisanby, 2001). High-frequency rTMS (>1 Hz) reduces intracortical inhibition in healthy individuals (Ziemann et al., 1996) and promotes long-term plastic effects (Esser et al., 2006).
Initial schizophrenia TMS treatment trials were exploratory in nature and compared schizophrenia patients with depressed patients (Feinsod et al., 1998; Geller et al., 1997). These studies used treatments which were brief, single pulse rather than repetitive (Geller et al., 1997), used a low total number of pulses delivered with non-focal coils, and targeted a limited number of regions [motor or dorsolateral prefrontal cortex (DLPFC)] (Feinsod et al., 1998; Geller et al., 1997). Benefits were found for schizophrenia patients, but were limited to improvements in secondary symptoms (e.g. anxiety, mood). The first randomized controlled study of rTMS in schizophrenia was also the first to assess positive and negative symptoms (Klein et al., 1999). Thirty-five patients received either active or sham 1 Hz rTMS to the right DLPFC at 110% motor threshold (MT) (1200 total pulses). This dose did not produce a therapeutic effect for positive or negative symptoms (as assessed with the Positive and Negative Syndrome Scale, PANSS; Kay et al., 1987). Given the earlier findings that low-frequency rTMS is inhibitory to motor cortical activity in healthy individuals (Hoffman and Cavus, 2002), low-frequency rTMS might further reduce activity in a brain region hypoactive in schizophrenia patients. More recent studies have taken advantage of frequency-specific rTMS effects on cortical excitation and site specificity as described below (see Table 1).
The first studies guided by known frequency-specific effects of rTMS to target a-priori identified symptoms were in the treatment of medication-resistant auditory hallucinations (AH). Evidence suggests that an over-active left temporo-parietal cortex (TPC) is one of several cortical regions associated with the perception of external voices in hallucinating patients (Silbersweig et al., 1995). Since low-frequency rTMS has an inhibitory effect on motor cortical excitability in healthy volunteers (Chen et al., 1997; Wassermann and Lisanby, 2001), it was hypothesized that 1 Hz rTMS to the left TPC may treat overactive receptive language areas that are associated with AH. Hoffman was the first to test this hypothesis in a randomized cross-over trial using 1 Hz rTMS at 80% MT, increasing the number of pulses over 4 d with positive results (Hoffman et al., 2000). In an open trial of nine patients, 10 d of daily low-frequency rTMS (1 Hz at 80 % MT) to the left auditory cortex also decreased AH ratings (d'Alfonso et al., 2002). Reductions in AH frequency, intensity and attentional salience were subsequently found in randomized controlled blinded cross-over studies (Hoffman et al., 2007; Poulet et al., 2005). Parallel design studies (Brunelin et al., 2006; Hoffman et al., 2003, 2005) in which up to 50 subjects were treated for 4–9 d with low-frequency rTMS to the left TPC have replicated this finding of efficacy and have even demonstrated improvements in source monitoring (Brunelin et al., 2006). More recently, greater number of sessions (20 min vs. 10–15 min; Horacek et al., 2007) and longer duration of treatment (15 sessions vs. 5–10; Sommer et al., 2007) (both increasing the total dose of treatment) also reduced refractory AH in schizophrenia patients. These findings provide evidence for a functional role of the TPC in the experience of AH and also point towards a potential clinical application of rTMS to reduce AH in treatment-refractory patients as reviewed in a recent meta-analysis (Aleman et al., 2007).
There have been negative studies of rTMS for AH, with some limitations. One study treated subjects for only 4 d (McIntosh et al., 2004) compared to the more typical 10 d (d'Alfonso et al., 2002; Hoffman et al., 2005; Horacek et al., 2007) or two sessions daily for 5 d (Brunelin et al., 2006; Poulet et al., 2005), suggesting that inadequate dosing may account for the negative finding. Furthermore, in one study five of 16 patients were on anticonvulsants (McIntosh et al., 2004), potentially interfering with the action of rTMS. A more recent study performed in 11 schizophrenia patients with refractory AH had several advantages, including a cross-over design and an active sham condition (Schonfeldt-Lecuona et al., 2004). Subjects received 1 Hz rTMS at 90 % MT to three areas (randomly ordered) for 5 d in each condition, separated by 2 d. The two treatment locations were the superior temporal gyrus and Broca's area, chosen for a subgroup of patients using a fMRI auditory delayed match to sample task to target regions thought to be activated during AH. While there was no significant change in AH overall, the subgroup with fMRI-targeted treatment tended to improve and three of the 11 subjects who received active rTMS were designated complete responders (>50 % reduction in rating). Therefore, individualized targeting of rTMS based on functional imaging may have value, as described by others (Hoffman et al., 2003; Sommer et al., 2007). Consistent with this idea is the finding of only a decrease in loudness of AH between active and sham rTMS in an adequately dosed study (10 d of 15-min 1-Hz sessions; Fitzgerald et al., 2005), i.e. negative results may have been due to a lack of functional targeting. Although coil targeting was based on the International 10–20 EEG electrode placement and not on functional activity, positive results have been found using similar techniques (Chibbaro et al., 2005).
A more recent positive study underscores and innovates on the technique of functional targeting. Hoffman's group (Hoffman et al., 2007) treated 16 patients with 12–24 sessions of 1 Hz rTMS to three or six sites for 16 min with one sham session randomly included. For the 11 subjects who had six target sites, strict attention was paid to response. If there was less than a 10% change in AH after 2 d, the site was changed. Targets were chosen based on hallucinations occurring during fMRI and subjects were grouped into intermittent or continuous hallucinator groups. The most efficacious target was the left temporo-parietal region, with no site order effects. Interestingly, predictors of response related to frequency of hallucinations. In the intermittent group, there was a negative correlation between treatment response (to TPC) and activity in left Broca's area and the right homologous region. In contrast, in the continuous hallucinator group, a negative correlation was found between treatment response (to TPC) and coupling between Wernicke's area and the right homologous Broca's area. Using PET imaging, another group also found that greater activity in the left inferior frontal gyrus was a negative predictor of response to rTMS (Horacek et al., 2007). Using fMRI, rTMS to the left TPC for AH produced greater activation in the inferior frontal gyrus and left temporo-parietal junction during an auditory activation task (Fitzgerald et al., 2007), although reductions in activity were found in other prefrontal regions as well as the left inferior parietal lobe. In contrast, an open study found no difference in response with functional targeting vs. left TPC (Sommer et al., 2007). However, they had positive results in both groups using a high dosage (15 d of 20 min of 1 Hz rTMS).
Other parameters examined have included duration and side of treatment. One study found an effect of time on positive symptoms (assessed with the PANSS) and frequency of AH in groups treated with left vs. right TPC rTMS for 5 d, but found that the only significant differences between active and sham treatments were reductions in Clinical Global Impression (CGI) (Guy, 1976; Lee et al., 2005). Differences between active and sham treatments may have reached significance had the subjects been treated for a longer duration. A similar recent study yielded somewhat different findings (Jandl et al., 2006). Although there was no significant difference between groups (left TPC, right TPC, sham), there was a significant difference in the number of responders by group with most receiving left rTMS (Jandl et al., 2006). Again the short duration of treatment (5 d) may have contributed to these authors' negative findings.
Given fronto-temporal loops have been implicated in the pathophysiology of positive symptoms (Liddle, 1992), it has been suggested that prefrontal cortex rTMS may treat positive symptoms via trans-synaptic action on the temporal lobes. Hypofrontality may disinhibit temporal activity contributing to psychosis. Since rTMS has the ability to alter cortical activity remote from the site of stimulation, high-frequency rTMS to the DLPFC may provide feed-forward inhibition of the temporal lobe to decrease positive symptoms other than AH. Several rTMS studies have examined the role of the DLPFC in positive symptoms. A randomized controlled cross-over study using high-frequency rTMS to the left DLPFC during active psychosis demonstrated a significant decrease in Brief Psychiatric Rating Scale (BPRS; Overall and Gorham, 1962) scores compared to sham at all time-points (Rollnik et al., 2001). The BPRS includes many symptom types making it difficult to determine if the effect was specific to positive symptoms. Studies that assessed changes in the positive subscale of the PANSS (Holi et al., 2004; Sachdev et al., 2005) did not find a reduction in positive symptoms with high-frequency rTMS to the DLPFC when compared to sham. One high-frequency rTMS study targeting the DLPFC found a trend of worsening of positive symptoms (Hajak et al., 2004).
Despite mixed results in clinical trials of TMS for positive symptoms, most studies targeting the TPC for AH do find significant clinical improvement. Seven of nine double-blind, sham-controlled parallel or crossover trials using low-frequency rTMS found a significant and sustained decrease in frequency and/or severity of treatment-resistant AH (Brunelin et al., 2006; Chibbaro et al., 2005; Hoffman et al., 2000, 2003, 2005, 2007; Poulet et al., 2005) as recently reviewed (Aleman et al., 2007). All studies had a similar design, using 1 Hz rTMS, 80–90 % MT, a similar duration of 4–10 d of treatment [(except for Hoffman's group; Hoffman et al., 2007) with up to 24 sessions, and twice daily sessions for 5 d (Brunelin et al., 2006; Poulet et al., 2005)] and a similar number of total trains delivered (usually 900/d). Follow-up assessments found lasting reductions.
The trials reviewed herein were carried out for a brief period of time, most for 1–2 wk. Even though effects were sustained in some trials for 10 wk (Sommer et al., 2007) or 15 wk (Hoffman et al., 2003) and suggest persistent change in cortical activity, it remains unknown whether initial non-responders to rTMS at 1 wk might not have an increased latency to response and would benefit from prolonged application of rTMS, with time-courses more akin to the speed of action of pharmacological agents. Repeated administrations of rTMS for refractory AH have had equivocal results (Chung et al., 2007; Fitzgerald et al., 2006a).
Further studies are warranted to explore this treatment strategy. Most studies which target the frontal or prefrontal areas focus on negative symptoms without taking advantage of trans-synaptic effects of rTMS (see below). Individualized targeting could also further elucidate the differential phenomenology of schizophrenia between patients as well as optimize outcome.
Targeting of negative symptoms with rTMS has been guided by knowledge of frontal pathology in schizophrenia. Hypoactivity in the DLPFC has been shown to correlate with negative symptoms in schizophrenia (Andreasen et al., 1997). It was hypothesized that high-frequency rTMS, which increases cortical excitability in healthy volunteers (Jin et al., 2005), might reverse prefrontal hypoactivity in schizophrenia patients. The first study published on rTMS for negative symptoms was an open study targeting the prefrontal cortex in six patients with 10 d of 20 Hz rTMS at 80 % MT (Cohen et al., 1999). Treatment produced a 12% reduction in the negative symptoms subscale of the PANSS. Two subsequent open trials also led to a reduction in negative symptoms. Jandl and colleagues treated 10 patients for 5 d with 10 Hz rTMS to left DLPFC at 100% MT (Jandl et al., 2004). With a lower frequency and fewer total pulses, they found reductions in negative symptoms, as evidenced by a 9 % reduction in Scale for the Assessment of Negative Symptoms (SANS; Andreasen, 1982) scores and improvements on the CGI scale. In the longest open trial at the highest dose to date (Sachdev et al., 2005), four male patients with prominent negative symptoms received 20 d of 15 Hz rTMS to the left DLPFC at 90 % MT (36 000 total pulses). The mean reduction in PANSS negative symptoms scores at the end of treatment was 33 %, an effect that was still evident 1 month later. They also found a 33 % improvement in social and occupational function. Although the investigators excluded subjects with current diagnoses of depression, they also found a 33 % reduction in the Montgomery–Asberg Depression Rating Scale (MADRS; Montgomery and Asberg, 1979). In this study (Sachdev et al., 2005), there was no change in positive symptoms and the high dose was safe and well tolerated.
Six larger, longer, parallel design, randomized clinical trials (RCTs) had contrasting results. They had several methodological improvements over prior studies given their parallel design and higher number of pulses (Hajak et al., 2004; Holi et al., 2004; Mogg et al., 2007; Novak et al., 2006; Prikryl et al., 2007). Despite this, high-frequency rTMS (20 trains, 10 Hz, 10 sessions, and 40 trains, 20 Hz, 10 sessions, respectively) to the left DLPFC did not lead to an improvement in negative symptoms in samples of 22 (Holi et al., 2004) and 16 (Novak et al., 2006) chronic schizophrenia patients. A study with lower numbers of subjects (20 trains, 10 Hz, 10 sessions) perhaps not surprisingly failed to show an effect of time or group (real vs. sham) (Mogg et al., 2007).
In contrast, Hajak's group reported improvement in negative symptoms using the higher dosage (20 trains, 10 Hz, 10 sessions; Hajak et al., 2004). They found significant improvement in total PANSS negative symptoms over time compared to sham treatment. This study was the first to assess depressive symptoms, which can cause secondary negative symptoms and might themselves respond to DLPFC rTMS (Gershon et al., 2003). They used depressive scales validated in schizophrenia patients, which are designed to prevent confounding by extrapyramidal symptoms (EPS) and negative symptoms (Calgary Depression Scale for Schizophrenia (CDSS; Addington et al., 1990). Although rTMS was associated with an improvement in depression, this effect did not account for improvement in negative symptoms. A more recent study that treated patients at the same dose, but for 3 wk instead of 2 wk, did not find a change in affective symptoms with treatment, although they found a benefit of active treatment over sham (15 trains, 10 Hz, 15 sessions; Prikryl et al., 2007).
The mechanism by which DLPFC rTMS may improve negative symptoms is not yet known. Abnormalities in α-EEG (8–13 Hz) may be particularly relevant to the negative symptoms of schizophrenia (Jin et al., 2005). A randomized (cross-over) trial was used rTMS to investigate this relationship by assessing the therapeutic effects of stimulation set individually at each subject's peak α-frequency (α-TMS) (Jin et al., 2005). In a cross-over design, 27 subjects were randomized to groups receiving 2 wk of daily treatment of two of four possible treatments: α-TMS, low-frequency rTMS (3 Hz), high-frequency rTMS (20 Hz), or sham stimulation bilaterally over the DLPFC. Individualized α-TMS demonstrated a significantly larger therapeutic effect (29.6% reduction in negative symptoms) than the other three conditions. This study suggests the importance of understanding the resonant features of rTMS and the potential need for individualized frequencies and locations to target rTMS. It is also in line with another report showing that rTMS set relative to individual α-frequency improved performance on a working-memory task in healthy volunteers (Klimesch et al., 2003).
Important differences in these six RCTs (Hajak et al., 2004; Holi et al., 2004; Jin et al., 2005; Mogg et al., 2007; Novak et al., 2006; Prikryl et al., 2007) may explain the conflicting results. One positive study included bilateral (i.e. bifrontal) stimulation (Hajak et al., 2004), which may allow for symmetry differences between patients. The intensity of rTMS was higher in this study (110 % vs. 90 % MT) and included patients with schizoaffective disorder, who tend to do better than schizophrenia patients and may also respond better to rTMS. This superior effect was not explained by the effect of rTMS on depressed symptoms, which can be conflated with negative symptoms. Gender differences in the samples may also explain differential efficacy as males predominated in the negative results studies (Holi et al., 2004; Mogg et al., 2007). In a study of high-frequency rTMS on cognitive symptoms in schizophrenia, only females improved (Huber et al., 2003). A gender effect could be due to sex differences in prevalence of the deficit syndrome, which is more prevalent among males (Malaspina et al., 2000) and these clinical trials did not assess for the presence of the deficit syndrome.
Given the small number of RCTs to date on the efficacy of rTMS for negative symptoms in schizophrenia, no definitive conclusions regarding the treatment potential of this modality can be drawn, and further investigation is needed. Three out of six trials (Hajak et al., 2004; Jin et al., 2005; Prikryl et al., 2007) showed significant improvement in negative symptoms, which is promising. However, unlike studies in AH, these trials varied in design, duration and follow-up period; the optimal parameters for clinical treatment remain to be determined. It is also necessary to demonstrate the sustainability of symptom reduction, especially given that schizophrenia is a chronic disorder that will probably require continued treatment. The efficacy of maintenance rTMS in schizophrenia has not been explored. Although statistically significant effects have been shown, the clinical significance (i.e. degree of functional improvement) of these effects remains largely unknown. Yet, as negative symptoms are generally more refractory to conventional pharmacological treatment, rTMS paradigms deserve further exploration as an alternative treatment tool for this specific symptom complex (for a summary of findings, see Table 1).
While there is evidence that rTMS may be effective in schizophrenia with dose (Hajak et al., 2004; Holi et al., 2004) and frequency specificity (Hajak et al., 2004; Hoffman et al., 2007), further work is needed to determine whether rTMS will have practical clinical value for the treatment of the symptoms of schizophrenia. The significant heterogeneity in schizophrenia may impact treatment response and be an important design consideration for future studies.
Most schizophrenia studies have not taken into account clinical heterogeneity. Baseline patient characteristics of treatment responders, including demographic and cognitive factors may be useful determinants of treatment response. In a recent negative study of rTMS for negative symptoms, the active treatment group was significantly older with a longer duration of illness (Mogg et al., 2007). Degree of resistance of AH to antipsychotic medications has not been assessed in TMS treatment studies and may contribute to variability in treatment response. One schizophrenia study found that response (to the same target) was differentially predicted depending on frequency of hallucinations (Hoffman et al., 2007).
Diagnostic subgroup and symptom expression are particularly important for negative symptoms. Primary enduring negative symptoms, defined as ‘deficit symptoms’ (DS), may not respond the same as negative symptoms secondary to other symptoms (e.g. depression, psychosis), medications (e.g. antipsychotics), environmental deprivation or other illness factors. Data show that DS vs. non-DS are distinct subgroups (Kirkpatrick et al., 2001) with distinct neuroanatomical features (Kirkpatrick et al., 1999; Tamminga et al., 1992; Vaiva et al., 2002), suggesting that subgroup, brain metabolism, cognitive deficits and symptom profiles may be necessary in guiding individual treatment. Given the neuroanatomical differences in DS vs. non-DS schizophrenia patients, it is plausible that different patient subgroups may respond differentially to a given rTMS treatment paradigm. Other distinctions that may be of value include the effect of aetiological subgroup (e.g. family history of schizophrenia, advanced paternal age) on treatment response and whether there is an interaction between aetiological group and treatment location since these patient subpopulations differ clinically and cognitively (Malaspina et al., 2000).
By understanding how patient subgroup and characteristics influence treatment response, one may select patients who will have the best outcome with treatment.
Research literature from diverse disciplines suggests the presence of an abnormality in cortical inhibition (CI) in schizophrenia patients (Frith et al., 2000). The strongest evidence comes from studies of event-related potentials, which show impaired inhibition of the P50 auditory evoked potential with paired auditory stimuli (Freedman et al., 1996; McCarley et al., 1991) and from post-mortem studies which demonstrate morphological and other abnormalities in inhibitory cortical GABA interneurons (Akbarian et al., 1995; Benes et al., 1998; Hoffman and Boutros, 2001; Selemon and Goldman-Rakic, 1999; Simpson et al., 1989). These GABAergic changes are linked to reduced grey-matter volume in schizophrenia patients (Benes, 2000). GABA abnormalities in schizophrenia patients (Lewis et al., 1999; Volk et al., 2000, 2002; Woo et al., 1998) may relate to cognitive deficits (for review see Lewis and Gonzalez-Burgos, 2006) and negative symptoms (Carpenter et al., 1999; Wolkowitz and Pickar, 1991). A better understanding of the relationship of CI to GABA transmission may be useful in guiding rTMS treatment.
CI can be studied with TMS and has been in schizophrenia patients, supporting the presence of reduced CI in patients. Following TMS-elicited motor-evoked potentials (MEPs), there is a subsequent period of tonic suppression of EMG activity that endures beyond the neuronal refractory period (Cantello et al., 1992), the cortical silent period (CSP). CSP has been attributed to inhibitory interneurons which synapse onto α-motor neurons, which appear to be GABAB dependent (Sanger et al., 2001; Siebner et al., 1998; Werhahn et al., 1999). A shorter CSP in patients compared to controls has been demonstrated in several studies (Daskalakis et al., 2003; Fitzgerald et al., 2002, 2003, 2004). The role of GABAB in long-term depression (Kamikubo et al., 2007) may explain its role in CI. If regional changes in CI predict treatment response, they may be a useful tool in treatment targeting or identifying patient subgroups that respond best to TMS. A case report of improved AH with low-frequency rTMS found lengthening of the CSP following treatment (Langguth et al., 2006), suggesting it may be a useful marker of response.
The motor cortex, easily accessed with TMS producing an observable outcome, has been used as a probe of pathophysiology as the neurological soft signs of schizophrenia (e.g. dyscoordination and decreased fine motor skills) may be linked to deficits in central inhibition (Davey et al., 1997). However, these deficits do not contribute significantly to the morbidity of the illness. More striking are the abnormal findings in the prefrontal cortex and limbic structures underlying psychosis and negative symptoms, which may still be achieved (and studied) with TMS via its transsynaptic effects that alter the function of distributed neural circuits (Speer et al., 2000). Strafella and colleagues have found that stimulation of the prefrontal and premotor cortical areas leads to the release of dopamine in the striatum of healthy individuals (Strafella et al., 2001, 2003). Such pre-treatment imaging studies, although costly, may play a role in guiding treatment or predicting outcome. Other methods of combining imaging with rTMS should be considered for targeting and probing the underlying physiology of treatment effects of TMS (Hoffman et al., 2007; Horacek et al., 2007; Langguth et al., 2006; Sommer et al., 2007).
Primary motor cortex may not be the most salient cortical region to study in schizophrenia. Cerebellar pathology is implicated in schizophrenia patients in phenomenological (Andreasen et al., 1999) and histological (Ichimiya et al., 2001) studies. Further, individuals with cerebellar lesions have demonstrated cognitive deficits similar to schizophrenia patients (Schmahmann and Sherman, 1998). Cerebellar inhibition can be studied using TMS by administering a conditioning pulse to the cerebellar cortex 5–15 ms prior to a test pulse over motor cortex. Cerebellar neurons projecting to the contralateral primary motor cortex have demonstrated reduced inhibitory control of motor cortex in schizophrenia patients when compared to healthy individuals (i.e. patients had almost 30 % less inhibition of the MEP) (Daskalakis et al., 2005). This study demonstrates how the cerebellum may cause cognitive deficits in schizophrenia patients via abnormal cortical control. Therefore, studies of cerebellar CI in schizophrenia may be relevant for the development of TMS for cognitive deficits. Other TMS measures may also be useful in studying pathophysiological findings in schizophrenia that involve cortical regions outside of the motor system. For example, our group has demonstrated the use of TMS over the occipital cortex in evaluating cortical involvement in visual masking deficits of schizophrenia (Luber et al., 2007).
Frequency-specific relationships have not been established for schizophrenia patients. To date, two studies have examined the plastic brain responses to rTMS in patients compared to controls and found deficient inhibitory response to low-frequency rTMS in schizophrenia (Fitzgerald et al., 2004; Oxley et al., 2004). The first study did not find changes in cortical excitability from baseline in patients. Healthy subjects demonstrated a significant increase in the resting MT after 1 Hz rTMS to the primary motor cortex. Active MT also increased in the control group, as well as in medication-treated patients, but did not increase in the medication-free patient group. Control subjects also had a significantly reduced CSP compared to baseline, unlike the patient groups, who demonstrated a nonsignificant increase in CSP. Thus, medicated and unmedicated patients demonstrated reduced plastic responses in motor cortex to low-frequency trains of rTMS that are commonly thought of as inhibitory in healthy volunteers.
In another study by the same group, targeting the premotor cortex (Oxley et al., 2004), healthy control subjects demonstrated an increase in resting MT following low-frequency rTMS. Patients did not have this increase in resting MT, but instead a significant decrease (in the previous study they had no change). These changes remained significant after controlling for baseline variability in MT. If the goal of an intervention is to dampen cortical excitability to target positive symptoms, 1 Hz rTMS may not be the ideal approach to achieving that end in schizophrenia patients with deficits in mounting an inhibitory response to low-frequency rTMS. Sustained changes in cortical excitability are of clinical significance for the development of treatments in schizophrenia, given the abnormalities in underlying cortical excitability. Interestingly, a recent review of the effect of studies of the effect of single-train low-frequency rTMS on CI found the studies were equivocal (Fitzgerald et al., 2006b) tempering the above results in schizophrenia patients.
The absence of an inhibitory response to 1 Hz rTMS trains over motor cortices in schizophrenia patients (Fitzgerald et al., 2004; Oxley et al., 2004) may explain some of the variance in treatment results, especially in studies of AH which used 1 Hz. Although the generalizability of findings from acute studies to chronic effects of rTMS is unclear, it cannot be assumed that schizophrenia patients will have the predicted physiological response to rTMS. Differences in the treatment effects of 1 Hz rTMS trains might also be accounted for by differences in number of trains (or treatment intensity, etc.), site of stimulation (motor cortex vs. TPC) or medication exposure.
Associations between TMS-generated neurophysiological measures and various symptom domains in schizophrenia have not been determined. Given the heterogeneity of schizophrenia and the large variance in physiological measures, it may be most revealing to evaluate such putative associations.
There are also known EEG abnormalities in schizophrenia that may be useful in guiding treatment. Spencer and colleagues reported that visual Gestalt stimuli elicited abnormally slow oscillations in schizophrenia patients and oscillations in the parietal cortex absent in controls (Spencer et al., 2004). α-EEG has already been used as a means of individualizing the appropriate frequency of stimulation (Jin et al., 2005) with positive effects.
Techniques such as high-density EEG and functional brain imaging in addition to TMS-generated neurophysiological measures could be helpful to document further evidence of abnormal plasticity in cortical regions outside of the motor cortex in schizophrenia patients (Esser et al., 2006; Pascual-Leone et al., 1998).
Overlap has been identified between symptoms of depression and negative symptoms. PANSS scores (Brebion et al., 2000; Nakaya et al., 1997), but not SANS scores (Andreasen, 1982; Brebion et al., 2000; McKenna et al., 1989; Sax et al., 1996; Selemon and Goldman-Rakic, 1999), correlate with depressive symptoms measured by the Hamilton Depression Rating Scale (Hamilton, 1960). Since there is a high incidence of depression in schizophrenia patients (Siris, 1991), it is important to differentiate symptoms to understand the nature of response. Furthermore, the DLPFC plays a crucial role in modulation of deeper brain structures involved in regulating mood and affective display, which can be altered in both depression and negative symptoms (Alexander and Goldman, 1978; Kalivas and Barnes, 1993). To minimize this confound, one might exclude subjects who meet criteria for a current depressive episode and assess for depressive and negative symptoms throughout the study. The CDSS is a measure designed to assess depressive symptoms unrelated to negative symptoms and EPS in schizophrenia patients (Addington et al., 1994) and has differentiated outcomes (Sachdev et al., 2005).
Case reports have demonstrated improvements in positive symptoms when targeting the prefrontal cortex with rTMS (Grisaru et al., 1998; Saba et al., 2002). Future studies may benefit by examining which positive symptoms improve under these conditions.
Important TMS schizophrenia study design issues include: (1) appropriate dosage; (2) accurate targeting/coil localization; and (3) coil-to-cortex distance.
A higher number of pulses (Hoffman et al., 2000) and longer duration of treatment (Aleman et al., 2007) may result in better effect on AH. Such dosage effects have yet to be assessed for negative symptoms and other aspects of dosage (frequency, intensity, number of sessions, etc.) have not been systematically examined. Approaches such as simultaneously combining different frequencies of stimulation at different cortical sites for synergistic effects may be beneficial in schizophrenia. Other dosing paradigms such as priming (administering θ-frequency rTMS prior to a 1 Hz train) (Iyer et al., 2003) and θ-burst stimulation (administering γ-frequency rTMS pulses at θ rate) (Huang et al., 2005) have shown more powerful and lasting effects on cortical excitability than mono-frequency paradigms in healthy volunteers and might be useful in overcoming the plasticity deficits seen in schizophrenia.
Variance in site of stimulation can be reduced by using structural MRI combined with frameless stereotaxy to target coil placement. One rTMS study suggested that patients who had MRI-targeted treatment for AH might have better response (Schonfeldt-Lecuona et al., 2004). Functional targeting may be useful to identify regions active during AH (Hoffman et al., 2007; Langguth et al., 2006; Schonfeldt-Lecuona et al., 2004). A recent case report used FDG PET to identify hyperactive cortex for rTMS targeting of AH (Langguth et al., 2006). Not only did they find a decrease in symptoms, but also a decrease in the local metabolism following rTMS treatment, which has since been replicated in an open study (Horacek et al., 2007). In open (Horacek et al., 2007) and randomized controlled studies (Hoffman et al., 2007), functional activity during hallucinations predicted response. Functional targeting of negative symptoms has yet to be developed.
Finally, studies suggest that frontal deficits in schizophrenia are bilateral, but to date only one TMS study in schizophrenia involved bilateral application (with a decrease in negative symptoms) (Cohen et al., 1999), in lieu of the left side (Hajak et al., 2004; Holi et al., 2004; Jandl et al., 2004, 2006; Jin et al., 2005; Novak et al., 2006; Sachdev et al., 2005) and no studies have examined treating the right DLPFC alone.
The variable of coil-to-cortex distance has a major impact on the delivered dose of rTMS. In a study of depressed patients aged >55 yr, adjusting for distance to scalp produced a higher rate of responders than other studies (Nahas et al., 2001). Since studies have demonstrated smaller frontal volumes in schizophrenia patients (Andreasen et al., 1986; Wible et al., 1995), it might be particularly important in the study of schizophrenia to adjust for cortex-to-coil distance even in younger patients. Such intensity adjustment has not previously been used in schizophrenia rTMS clinical trials.
TMS studies in schizophrenia have begun to explore the possible efficacy of both low- and high-frequency rTMS for refractory symptoms. Low-frequency rTMS to the left temporal lobe has been studied as a treatment for refractory hallucinations. The inhibitory effects of 1 Hz rTMS have demonstrated efficacy in reducing intensity and attentional salience of refractory AH in some studies (Hoffman et al., 2000, 2003, 2007; Hoffman and Cavus, 2002; Horacek et al., 2007; Poulet et al., 2005; Sommer et al., 2007). Left-sided stimulation and fMRI-guided targeting may improve response (Hoffman and Cavus, 2002; Hoffman et al., 2007; Schonfeldt-Lecuona et al., 2004), while under-dosing (McIntosh et al., 2004) may attenuate treatment response.
In studies of negative symptoms, the DLPFC has largely been targeted with high-frequency rTMS. Reductions in negative symptoms have been found in some open (Cohen et al., 1999; Sachdev et al., 2005) and randomized studies (Hajak et al., 2004; Jin et al., 2005; Prikryl et al., 2007). Interestingly, these findings could not be accounted for by depressive symptoms (Hajak et al., 2004) nor were they accompanied by changes in functional imaging (Cohen et al., 1999; Hajak et al., 2004) or cognitive deficits (Sachdev et al., 2005). Although response appears to correlate with dose of treatment (Hoffman et al., 2000), the optimal dosages for low- and high-frequency rTMS for schizophrenia have yet to be determined. Considerations such as gender, patient subtype, pathophysiology (e.g. CI, EEG, etc.), accurate anatomical and functional coil localization, and duration of treatment may all improve response. There have been no adverse events reported in these trials, including seizure, despite a greater theoretical risk given the ability of antipsychotics to lower seizure threshold.
Schizophrenia is an illness that is only partially responsive to currently available pharmacological treatments (Lieberman, 2006). Given medication non-compliance, treatment resistance and low function, there is a pressing clinical need for the development of novel therapeutics for this devastating illness. TMS offers the opportunity to probe underlying pathophysiology and the potential to develop targeted treatments. Further, a better understanding of the heterogeneity in this disease may lead to more effective treatment. To maximize the power of rTMS to treat schizophrenia, studies of cortical excitability combined with phenomenological assessment of patients need to be carried out and may lead to improved treatment outcome and novel treatment sites. Also the analogue of dose-finding studies for medications – i.e. optimize the multiple TMS parameters (e.g. frequency, intensity of stimulation) – need to be determined. Emphasis on strategies that optimize TMS technique including controlling for coil-to-cortex distance and better anatomical and functional target identification may also impact therapeutic response.
This research was supported by a NARSAD Brian Bass Young Investigator Award and Janssen Translational Neuroscience Fellowship (Dr Stanford), NIH grants MH60884 and MH069895, the Stanley Foundation, the American Foundation for Aging Research, and the Dana Foundation.
Statement of Interest
Dr Stanford received a consulting honorarium from Lehman Brothers and a lecture honorarium from Magstim Inc. Dr Sharif is a speaker for BMS and Janssen. Dr Corcoran received lecture fees from iCME and the France Foundation. Dr Lisanby has received research support from Neuronetics, Magstim, and Cyberonics, and an honorarium from Magstim. Columbia University has filed an invention disclosure on a novel TMS device developed in Dr Lisanby's laboratory.