Transcranial magnetic stimulation and auditory hallucinations in schizophrenia
Ralph E Hoffman, Nashaat N Boutros, Sylvia Hu,
THE LANCET ? Vol 355 ? March 25, 2000 1073

12 patients with schizophrenia and auditory hallucinations received 1 Hz transcranial magnetic stimulation of left temporoparietlal cortex. In a double-blind crossover trial, active stimulation significantly reduced hallucinations relative to sham stimulation.
Auditory hallucinations are reported by 50?70% of patients with schizophrenia and generally consist of spoken speech or voices. Response to drug treatment is often incomplete or non-existent, and these hallucinations can cause great distress,
functional disability, and lack of behavioural control. Silbersweig and colleagues1 described regional brain activation by use of 15O positron emission tomography when
auditory hallucinations occurred in six patients with schizophrenia. Blood flow activation was detected in left temporoparietal auditory-linguistic association cortex and in
thalamic, hippocampal, and striatal regions. Low frequency (1 Hz), extended duration (15?30 min), repetitive transcranial magnetic stimulation (rTMS) reduces activation in the brain
area directly stimulated as well as in other functionally connected brain areas.2,3 We postulated that low frequency rTMS delivered to the left temporoparietal cortex would
curtail auditory hallucinations by reduction of excitability of distributed neurocircuitry that produce these experiences. 12 right-handed patients with auditory hallucinations who
met Diagnostic and Statistical Manual IV (DSM-IV) diagnostic criteria for schizophrenia (eight paranoid type, four schizoaffective type; ten men; mean age 41?8 years [SD 8?6])
were included. Education level of the participants in grades was mean 14?2 (SD 1?8); a level of 14 grades corresponds to 2 years of college. All patients received antipsychotic drugs and were maintained on these drugs without change in dose throughout the study period. Five patients received concomitant anticonvulsant drugs (four valproate semisodium, one carbamazepine). All patients had daily auditory hallucinations without remission for at least 6 months. Auditory hallucinations were either continuous (three) or
intermittent (nine). Each patient had normal routine laboratory studies, electrocardiogram, and electroencephalogram. Motor threshold was identified as the minimum magnetic
field strength required to produce left thenar muscle activation by single transcranial magnetic stimulation pulses delivered to the motor cortex, confirmed by electromyographic
monitoring, for at least four of eight trials. Site and strength of the motor threshold was redefined each session. 1 Hz stimulation at 80% motor threshold was then given midway
between the left temporal (T3) and left parietal (P3) electroencephalogram electrode sites on the basis of the international 10?20 electrode placement system. Sham following 12 and 16 minutes of active stimulation but not for any duration of sham stimulation (figure 1). In all but one case, hallucination severity was lower after the active stimulation sequence than the sham stimulation sequence (figure 2). Endpoint hallucination ratings were analysed by use of a repeated measure ANOVA with two additional factors; order of stimulation (active or sham first), and concomitant treatment with anticonvulsant drugs. Reductions in
hallucination severity after active compared with sham stimulation were significant (p<0?006), as was the interaction between change in hallucination severity and anticonvulsant drugs (p<0?02) showing reduced treatment effects with these drugs (figure 2). No effect of order of stimulation was seen. Other positive symptoms and negative symptoms did not change much after rTMS. Follow-up assessments of the eight patients classified as responders (ie, hallucination severity improved for active relative to sham stimulation by score >1) indicated that auditory hallucinations returned roughly to
baseline 1 day after the course of active rTMS in two patients, 4 days in two, 5 days in one, 2 weeks in one, 3 weeks, and 2 months in one. Left temporoparietal cortex, the site of rTMS in this study, is a brain area critical in perceiving spoken speech.4 Our findings therefore support the hypothesis that speech perception neurocircuitry plays a part in the generation of hallucinated speech. Not all patients showed robust improvements in
hallucinations after active rTMS. One factor contributing to variable response was concurrent anticonvulsant drug treatment, which seemed to reduce rTMS effects. This
observation suggests that higher levels of signal propagation are required for rTMS to curtail auditory hallucinations, or that symptoms prompting administration of anticonvulsant drugs (eg, mood lability) are negative predictors of rTMS response. Other factors that might contribute to the variability of rTMS effects include individual differences in anatomical location of speech processing areas,5 variable location of
cortical activation producing auditory hallucinations,1 and differences in skull-brain relation, and baseline physiology. stimulation was given at the same location, strength, and
frequency with the coil angled 45° away from the skull to induce scalp stimulation but curtailing brain stimulation. A Cadwell magnetic stimulator system (Cadwell Inc,
Kennewick, WA, USA) with a water-cooled, handheld figureeight coil was used to deliver rTMS. Stimulation was initiated at 4 min for each condition and built up on successive days by 4 min to 16 min on day 4. Psychiatric symptoms were assessed daily with the Positive and Negative Symptom Scale (PANSS). An earlier study showed that factors contributing to severity of auditory hallucinations varied between patients (for instance, frequency, loudness, content, number of voices,emotional distress, and level of distraction). consequently, auditory hallucinations were assessed with an individualised, composite scale. A score of ten corresponded to a narrative description of the patient?s hallucinations at the time of study entry, with zero corresponding to no hallucinations. For reassessments the patient produced a severity rating on the basis of these individualised anchor points. Higher scores were permitted if the patient?s hallucination severity exceeded that
at study initiation. Trials of active versus sham stimulation took place on separate weeks with 2?3 days separating each trial. Baseline hallucination assessments for each trial were
done just before initiation of each stimulation condition and reflected the 24 h before. Reassessments took place the morning after each of the four rTMS sessions and indicated
overall hallucination severity since the last rTMS session. Patients, clinical interviewers, and clinical staff were unaware of stimulation condition. Patients randomised to receive sham stimulation first received active stimulation the second week and viceversa.
Besides complaints of mild headache in two cases after active stimulation, patients tolerated rTMS without difficulty. Mean (SD) baseline hallucination score was 8?5 (2?2) for the active rTMS trial and 7?5 (2?6) for the sham rTMS trial. Symptom improvements relative to baseline were significant
1074 THE LANCET ? Vol 355 ? March 25, 2000




em Psiquiatría