Reduced TMS-evoked fast oscillations in the motor cortex predict the severity of positive symptoms in first-episode psychosis.

Abstract

Accumulating evidence points to neurophysiological abnormalities of the motor cortex in Schizophrenia (SCZ). However, whether these abnormalities represent a core biological feature of psychosis rather than a superimposed neurodegenerative process is yet to be defined, as it is their putative relationship with clinical symptoms, in this study, we used Transcranial Magnetic Stimulation coupled with electroencephalography (TMS-EEG) to probe the intrinsic oscillatory properties of motor (Brodmann Area 4, BA4) and non-motor (posterior parietal, BA7) cortical areas in twenty-three first-episode psychosis (FEP) patients and thirteen age and gender-matched healthy comparison (HC) subjects. Patients underwent clinical evaluation at baseline and six-months after the TMS-EEG session. We found that FEP patients had reduced EEG activity evoked by TMS of the motor cortex in the beta-2 (25-34Hz) frequency band in a cluster of electrodes overlying BA4, relative to HC participants. Beta-2 deficits in the TMS-evoked EEG response correlated with worse positive psychotic symptoms at baseline and also predicted positive symptoms severity at six-month follow-up assessments. Altogether, these findings indicate that reduced TMS-evoked fast oscillatory activity in the motor cortex is an early neural abnormality that: 1) is present at illness onset; 2) may represent a state marker of psychosis; and 3) could play a role in the development of new tools of outcome prediction in psychotic patients.

1. Introduction

Abnormal motor behavior, including catatonia, hyperkinetic movements, and neurological soft signs, is commonly observed in patients with Schizophrenia (SCZ) (1–4). Furthermore, recent evidence points at an involvement of the motor system early in the course of the disorder (5,6) and even in prodromal states and proneness to psychosis (7,8). This suggests that abnormalities in the neurodevelopmental trajectory of the motor system represent a core feature implicated in the pathogenesis of SCZ and related psychotic disorders (9,10). Recent evidence indicates that motor impairment might in fact be closely associated with defining clinical features of SCZ spectrum disorders, including psychotic symptoms (11). Moreover, functional connectivity within the sensory-motor system was recently shown to predict clinical outcome at the beginning of psychosis (12). Thus, investigating neural dynamics within the motor system may help identify key pathophysiological mechanisms underlying the clinical manifestations of psychotic disorders.

The primary motor area (Brodmann area 4, BA4) was identified as a first-choice target for investigating the neurophysiological properties of the cerebral cortex with non-invasive brain stimulation (13), and especially with Transcranial Magnetic Stimulation (TMS) (14). Because it is easily accessible through the scalp and is able to provide a functional readout of an effective transcranial stimulation (i.e., a muscle contraction), BA4 has been targeted with TMS to explore intrinsic brain properties of several neuropsychiatric populations, including individuals with SCZ and related psychotic disorders (15). Earlier studies in SCZ have investigated motor cortical excitability by quantifying the amplitude of the Motor Evoked Potential (MEP) and of the resting Motor Threshold (rMT, the intensity required to induce a ≥50 μV MEP in 5 out of 10 single-pulse TMS trials) (16). Several TMS-MEPs paradigms have been developed to assess more specifically excitatory and inhibitory neurons within the motor cortex (17,18). However, these paradigms relied on modulating MEPs amplitude, thus providing only an indirect assessment of cortical neuronal activity, and could not be used outside of the motor cortex due to the lack of measurable outputs.

More recently, coupling single pulse TMS with high-density electroencephalography (TMS-EEG) has provided a reliable tool for the objective assessment of spatiotemporal dynamics of neural circuits (19). TMS-EEG allows probing the neurophysiological properties of cortical areas beyond the motor cortex (20,21). Several studies have employed TMS-EEG to investigate abnormalities in cortical oscillations and connectivity in SCZ (22,23). For instance, SCZ patients showed a decrease in the amplitude and propagation of the early neural response to TMS of frontal areas (24,25). Conversely, late TMS-evoked oscillations (400-700ms) following stimulation of BA4 were increased in the same patients (26). This study also reported a direct correlation between the aberrantly increased propagation of brain activity evoked by TMS of the motor cortex and the severity of positive psychotic symptoms. Furthermore, in another TMS-EEG study investigating four cortical areas, we found that the main oscillatory frequency evoked by TMS in a given cortical area (21) was slowed in frontal cortical areas, including BA4, in chronic patients affected by SCZ (27).

Importantly, most existing TMS-EEG studies in SCZ were performed in chronic, medicated subjects, often with long illness duration. Thus, several questions concerning the neurophysiological abnormalities of motor areas in SCZ remain unanswered. Specifically, it is still unclear whether oscillatory deficits as those seen in the motor cortex of SCZ patients: 1) are present at illness onset 2) are specific to the frontal lobe or rather involve other cortical areas at the beginning of a psychotic illness; and 3) are associated with clinical variables such as psychotic symptoms and can predict symptoms severity longitudinally, i.e. at a follow-up clinical assessment.

The current poverty of available evidence in early psychosis reflects the notable challenge of engaging these individuals, given the acuity of illness and the generally poor compliance with mental health services (15). At the same time, first-episode psychosis (FEP) patients represent a unique window into the neurobiology of SCZ and an opportunity to gaining precious insight into disturbances in neural dynamics that are implicated in the neurobiology of psychotic disorders, rather than superimposed alterations attributable to confounding factors such as chronic exposure to medications, length of psychiatric history, or comorbidities. Thus, to show the feasibility of TMS-EEG recordings in this population, in a recent pilot study we recorded the EEG responses evoked by TMS of a BA4 and found that FEP patients had reduced TMS-evoked high beta/low gamma frontal oscillations relative to healthy control subjects (28).

Building on this preliminary evidence, in the present study we used single pulse TMS-EEG aiming to: 1) corroborate that TMS-evoked oscillatory deficits were observed in the motor cortex of FEP patients; 2) examine whether these oscillatory deficits were also present in a non-frontal area, such as the posterior parietal cortex; and 3) investigate whether altered TMS-evoked EEG responses were associated with illness severity at baseline and could predict changes at a 6 months clinical follow-up assessment in these patients.

2. Methods

2.1. Participants

Twenty-three FEP patients and thirteen healthy control subjects (HC) were recruited. Exclusion criteria for both groups included major medical or neurological illness affecting the central nervous system, diagnosed intellectual developmental disorder, and inability to complete magnetic resonance imaging (MRI) scans or TMS. Exclusion criteria for the control population included a history of treatment with antipsychotic medications; personal or family history of SCZ spectrum disorder or mood disorder with psychotic features; current medication affecting brain structure or function. FEP subjects were untreated or minimally treated (i.e. <1 month of exposure to treatment). A summary of the participant population is provided in Table 1. All subjects were evaluated by an expert rater from the Psychosis Recruitment and Assessment Core (PRAC) team with the Structured Clinical Interview for DSM Disorders (SCID) (29). The PRAC team is a team of expert clinical raters from our institution that performs rigorous evaluations of psychopathology while remaining blinded to study designs and expected outcomes. FEP patients’ evaluation included the Scale for the Assessment of Positive Symptoms (SAPS) and the Scale for the Assessment of Negative Symptoms (SANS) (Table 1). Six months after the TMS-EEG procedure, patients were revaluated with the SAPS/SANS. The baseline clinical evaluation was performed before the TMS-EEG procedure. The clinical rater was blinded to the results of the TMS-EEG recordings for the entire duration of study. This study was approved by the University of Pittsburgh Institutional Review Board, and all participants provided written informed consent prior to completing any study procedures.

Table 1. Participants’ demographics.

FEP: First-Episode Psychosis. SAPS: Scale for the Assessment of Positive Symptoms; SANS: Scale for the Assessment of Negative Symptoms. Antipsychotic dose at the time of Transcranial Magnetic Stimulation - electroencephalography (TMS-EEG) assessment is expressed in chlorpromazine equivalents.

		Healthy controls	FEP patients Baseline	6-months	P value
Age (years)		22.3	22.7		0.831
Gender (#/%Male)		9/68%	14/61%		0.25
Handedness (#/%Left)		1/7.7	4/17.4		0.41
Antipsychotic dose		-	137.87	-	-
SAPS	Item	-	21.1 ± 13.6	12.1 ± 10.1	<0.001
	Global	-	7.1 ± 3.4	4.4 ± 3.4	<0.001
SANS	Item	-	20.9 ± 12.0	20.5 ± 11.5	0.595
	Global	-	9.1 ± 2.6	7.7 ± 3.9	0.817

2.2. Procedure

Study participants sat comfortably on a chair with their arms in resting position. A TMS neuronavigation system (Localite, LTD) was used to identify and reliably stimulate the left primary motor cortex (BA4) and the left posterior parietal cortex (Brodmann area 7, BA7) on T1-weighted individual MRIs. BA4 was identified both anatomically and functionally. Firstly, BA4 was identified anatomically using a TMS neuronavigational system that allowed 2D and 3D visualization of individuals MRIs, along with a freely available brain atlas based on Montreal Neurological Institute (MNI) coordinates (30). Then, BA4 location was confirmed functionally by assessing the individual rMT. Once rMT was established, BA4 and BA7 single-pulse stimulation sessions were acquired. EEG responses were recorded using a 64-electrode cap connected to a TMS-compatible amplifier (BrainAmp, BrainVision) at a sampling rate of 5000 Hz. TMS was delivered at 110% of rMT at both stimulation sites, keeping an angle of stimulation orthogonal to the targeted brain gyrus. The rMT, expressed as % of Maximum Stimulator Output (%MSO), did not differ significantly between the two groups (HC: 52.7±4; FEP: 49.9±5.6; p=0.277). Importantly, to avoid re-afferent somatosensory activity (31,32), a motor region medial to the hand area was targeted when stimulating BA4. This resulted in targeting of an area corresponding to the motor representation of the lower limb, which notably shows a higher rMT than the hand motor area (33,34). No muscle twitch or movement was observed in - or reported by - any participant. In order to mitigate auditory artifacts (35), participants wore earbuds playing a noise-masking track specifically designed to include components of the TMS “click” sound embedded with white noise. To ensure the quality of the elicited EEG response (35), we used an online, real-time graphical interface displaying the signal from averaged trials referenced to the average of all channels, as in Casarotto et al. (36). Before recording each session, the average TMS-evoked potential (TEP) resulting from 20 TMS pulses was visually inspected. The recording was started if the average response to TMS showed a minimum of 5 μv peak-to-peak amplitude of the first component. Otherwise, the coil orientation was slightly adjusted to elicit an early TEP response at or above the 5 μv threshold. Finally, we confirmed the accurate location of the final targeting spot on BA4 by entering the coordinates in MNI space. For each session, 250–300 stimuli were delivered at 0.4 to 0.6 Hz, following international safety guidelines (37).

2.3. Data analysis

Data analysis was performed with Matlab R2015a (The Mathworks, Natick, MA). Channels making poor contact with the scalp and/or contaminated by environmental noise were removed and signals were segmented in epochs of 800ms before and after stimulus onset. Ocular artifacts were automatically reduced by subtracting the first Principal Component of the EEG when the correlation between the first component and the electrooculogram (EOG) was ≥0.9. Epochs contaminated by environmental noise and/or muscle activity were rejected by visual inspection, as in previous studies (27,38). After this step, a comparable number of epochs between the two groups was analyzed for both stimulation of BA4 (HC: 207±32; FEP: 196±29; p=0.36) and BA7 (HC: 205±32; FEP: 186±27; p=0.12, Wilcoxon rank-sum tests). TMS pulse artifacts were removed from each trial by replacing the interval around the stimuli (from - 2ms to 6ms, time 0 corresponding to the TMS pulse) with the data immediately before (from −10ms to −2ms, reflected around −2ms) and a fifth-order moving-average filter was applied between 4 and 8ms to reduce high-frequency edges. Bad channels were then interpolated using the spherical function of the public license toolbox EEGLAB and EEG data were downsampled to 1000 Hz, bandpass filtered (1-80Hz), re-referenced to the average, and baseline corrected. Independent Component Analysis (ICA) was applied to remove residual artifactual components resulting from eye blinks, eye movements, muscle activations, and the TMS pulse.

Cortical evoked activity in response to TMS was quantified in the time domain by the Mean Field Power (MFP) (39). MFP was calculated locally (LMFP), by averaging the square voltages across the channels surrounding the stimulator (Parietal stimulation: channels CP5, CP3, CP1, P5, P3, P1; Motor stimulation: channels FC5, FC3, FC1, C5, C3, C1), as well as globally, across all channels (GMFP). TMS-evoked oscillatory activity between 8 and 45Hz was assessed by a Morlet time-frequency decomposition, with a constant 3.5 cycles (27), as implemented in the EEGLAB toolbox (40). Event-related spectral perturbation (ERSP) matrices were computed as the ratio of the spectral power (μV²) of individual response epochs and their respective mean baseline spectra. After averaging across epochs, a two-tailed bootstrap significance probability level with respect to the baseline was constructed from 500 permutations and applied to the full-epoch window resulting in time-frequency matrices of significant ERSP values (α<0.05, corrected for multiple comparison using False Discovery Rate). To extract the spectral profile of the TMS response, ERSP values were averaged in time (20-300ms) and the resulting spectra were expressed as the percentage of power in a given frequency (relative spectral power, RSP). Inter-trial coherence (ITC) matrices were also calculated from the Morlet decomposition and averaged between 20 and 300ms. ERSP, RSP, and ITC were averaged across electrodes surrounding the stimulator to assess TMS-evoked oscillations over the site of stimulation. Finally, ERSP, RSP and ITC were averaged over frequency in alpha (8-12 Hz), beta-1 (13-24Hz), beta-2 (25-34Hz) and gamma (35-45Hz) bands.

2.4. Statistical analysis

A two-tailed T-test was used to compare age between groups. χ-squared tests were used to assess for differences between groups for dichotomous variables (i.e. gender, handedness). Two-tailed unpaired Wilcoxon-ranksum test were used to establish statistical differences in TMS-EEG measures between FEP and HC participants. Spearman’s correlation coefficients (p<0.05) were calculated between TMS/EEG measures and the positive symptoms, assessed with the SAP scores, in FEP patients.

Receiver operating characteristic (ROC) analysis was employed to define an optimal RSP cutoff for predicting the severity of positive symptoms at six months assessments. For both SAPG and SAPI scales, optimal cutoff values of TMS-evoked beta-2 RSP were calculated by maximizing the total accuracy rate in discriminating patients with higher SAP scores (above the median value) at follow-up evaluations from patients with lower scores (below the median value). Prediction accuracy was then estimated by quantifying the area under the ROC curve (AUC) and the rates of true positives (TP), true negatives (TN), and classification errors (ER) in a leave-one-out cross-validation procedure.

3. Results

3.1. FEP patients showed unaltered TMS-evoked activity in parietal areas.

When TMS was administered over the parietal cortex, FEP patients showed TEPs comparable to those observed in HC subjects, both in the time and frequency domains (Figure 1–A). No significant differences were found in GMFP (calculated across all channels), LMFP (measured at the frontal channels surrounding the TMS coil), nor in any frequency band averaged spectral power between the FEP and HC groups (see supplementary table).

Fig 1.

First-episode psychosis (FEP) patients showed a reduction in Transcranial Magnetic Stimulation (TMS)-evoked beta-2 electroencephalographic (EEG) oscillations in the motor cortex (B), but not parietal areas (A) relative to healthy controls (HC). Data are shown for a representative HC subject (A-B, left panels) and FEP patient (A-B, right panels). Top panels show the location of the stimulator (green shape) relative to the participant’s brain image. Middle panels display the evoked responses at all electrodes (butterfly plots, grey traces) and at the electrodes closest to the TMS coil (black traces). Lower panels show the TMS event-related spectral perturbations (ERSP) and relative spectral powers (RSP, grey bars), averaged in the alpha (8-12Hz), beta-1 (13-24Hz), beta-2 (25-34Hz), and gamma (34-45Hz) bands.

3.2. FEP Patients showed reduced TMS-evoked activity in the beta-2 band in BA4.

Single pulse TMS of the motor cortex evoked brain responses that were comparable between FEP patients and HC participants in the time domain but not in the frequency domain (Figure 1–B). Specifically, FEP patients showed a significant reduction in TMS-evoked EEG activity in the beta-2 band, which was localized in the frontal channels enclosing the TMS coil. Mean beta-2 ERSP value for HC subjects was 1.607 (SD: 0.562), while FEP patients’ mean beta-2 value was 1.241 (SD: 0.217, p=0.003, Wilcoxon-ranksum test, Supplementary Table). We also established that the relative power in the beta-2 frequency band, assessed with the RSP, was markedly decreased in FEP patients (Mean±SD=0.0270±0.002 relative to HC participants (Mean±SD=0.030±0.002, p<0.001, Figure 2). Compared to HC subjects, a reduction in evoked beta-2 RSP in the frontal channels around the TMS coil was confirmed in FEP patients by cluster analysis (not shown). Figure 2 shows individual RSP values in the beta-2 frequency range for HC and FEP groups (left) as well as boxplots and p-values for significant comparisons (right).

Fig 2.

In first-episode psychosis (FEP) patients, the relative spectral power (RSP) in the beta-2 frequency band (25-34 Hz) evoked by Transcranial Magnetic Stimulation (TMS) of the motor cortex was significantly reduced compared to HC participants. Boxplots with median, quartiles, and extreme values are displayed for both groups, together with the p-value after Wilcoxon’s ranksum test.

3.3. Frontal beta-2 band deficit predicted positive symptoms severity at 6 months.

In FEP patients, TMS-evoked frontal beta-2 RSP values were inversely correlated with the severity of the positive symptoms, as quantified by the global (SAPG) and total (SAPI) scores (Figure 3, SAPG r=−0.57 p=0.006, SAPI r=−0.51 p=0.017). Reduced frontal RSP in the beta-2 band was also inversely correlated with the levels of positive symptoms at their six-month follow-up clinical assessments (Figure 4, SAPI r=−0.66 p=0.006 and Fig S1, SAPG r=−0.59 p=0.017). No correlation was found between beta2 RSP values and the difference between baseline and follow-up SAP scores (Fig. S2–3). We thus employed ROC analysis to estimate whether RSP at baseline could be used to predict the severity of symptoms at follow-up in FEP participants. For both SAPI (Figure 4A) and SAPG (Figure S1A), an optimal RSP cutoff was computed to discriminate patients presenting worse positive psychotic symptoms at follow-up clinical evaluations (above the median values of SAPI=4.5 and SAPG=3) from those with less severe symptoms. With an area under the ROC curve of 0.77, 12 patients were correctly classified using the optimal RSP cutoff of 0.028 (accuracy of 75%), resulting in a sensitivity of 87.5% and a specificity of 62.5% (Figures 4B and S1B). We then employed a leave-one-out cross-validation procedure to estimate the robustness and reliability of this prediction. For SAPI scores, the AUC in the training dataset was found between 0.73 and 0.94 (median = 0.87), with a median true positive rate of 83.3%, a median true negative rate of 88.2%, and a median error rate of 16.7% (Fig 4C). Finally, cross-validation results for the test dataset confirmed a sensitivity of 87.5% and a specificity of 62.5% in predicting the severity of symptoms at follow-up evaluations from the TMS-evoked high beta frontal activity at baseline. Similar results were also obtained for SAPG scores (Fig S1C).

Fig 3.

In first-episode psychosis (FEP) patients, beta 2 relative spectral power (RSP) evoked by Transcranial Magnetic Stimulation (TMS) of the motor cortex was inversely related to positive symptom severity at baseline assessment. Panels show Spearman’s correlation coefficients (R) and p-values (p) for SAP-General (left) and SAP-Item (right) scores.

Fig 4. The relative spectral power (RSP) evoked by Transcranial Magnetic Stimulation (TMS) of the motor cortex in the beta-2 range predicted the severity of positive symptoms at follow-up evaluations.

A) Receiver operating characteristic (ROC) analysis applied to beta-2 RSP for discriminating patients with higher SAP-Item scores at six months follow-up from those with lower scores. B) Scatter plot displaying the relationship between beta-2 RSP values at baseline and SAP-Item scores at six months assessments. ROC’s optimal cutoff value (RSP = 0.028, yellow star, horizontal dashed line) detected 7 from 8 patients (red squares) with SAP-Item scores above the median value (vertical dashed line) at six-months evaluations and 5 from 8 patients (red hexagons) bellow the median value. 4 from 16 patients (red triangles) were misclassified. C) Results of the leave-one-out cross-validation procedure. The panel shows boxplots (median, quartiles, maximum and minimum values) for the area under the ROC curve (AUC), true positive rate (TP), true negative rate (TN), and classification error rate (ER) in the training data, as well as sensitivity (TP) and specificity (TN) for the left-out patients (test data).

4. Discussion

In this study, we investigated the TMS-evoked EEG responses of motor (BA4) and non-motor (BA7) cortical areas in a group of untreated or minimally treated FEP patients and a sample of age and gender-matched HC participants. No differences in EEG responses to TMS of BA7 were found between FEP and HC groups. In contrast, in FEP patients, brain responses to TMS of BA4 showed reduced EEG activity in the beta-2 frequency band in a cluster of electrodes overlying the motor cortex, relative to HC participants. Furthermore, in FEP patients, beta-2 deficits were associated with worse positive psychotic symptoms at baseline and predicted psychotic symptoms severity at a six-month clinical follow-up.

The first important finding of this study was a reduction in the oscillatory activity of the motor cortex in FEP patients. In previous TMS-EEG work, our group reported a reduction of motor oscillatory activity in SCZ patients (27). Conversely, other studies employing similar approaches did not report a reduction in fast oscillatory activity evoked by stimulation of BA4 in SCZ (26,41,42). Possible explanations for this discrepancy include the TMS protocol employed (i.e., two of three studies utilized a paired-pulse, rather than single-pulse approach) and the heterogeneity of the populations examined. In this regard, all previous TMS-EEG studies were conducted in chronically medicated SCZ patients. Here we showed that a reduction in EEG beta-2 range oscillations evoked by single pulse TMS of the motor cortex is observed in first-episode, antipsychotic naive or minimally treated psychotic patients. This finding, therefore, indicates that motor oscillatory deficits are unlikely to be the consequence of a long-standing illness or prolonged antipsychotic treatment and may, conversely, reflect a primary biological mechanism associated with psychosis.

Altered motor oscillatory activity has been previously reported in SCZ. For example, impaired synchronization of motor-sensory oscillations has been previously reported in chronic SCZ patients both before and after movement onset (43), and it is thought to reflect defective corollary discharges (44) (CDs). CDs represent the putative neural mechanism responsible for distinguishing between environmental and self-generated stimuli (so-called “self-monitoring”) (45). In psychosis, disturbances in CDs are hypothesized to underlie the false attribution to the outer world of internally generated percepts and thoughts, which could explain symptoms, including hallucinations, delusions (46–48), and self-disorders (49). Intriguingly, in our patients’ group, we reported an association between the reduced intrinsic ability of the motor area to generate beta-2 frontal oscillations, assessed by TMS-EEG, and the severity of positive symptoms (Fig 3). This finding is also consistent with the theoretical framework that links deficits in synchronized neural activity, as a sign of defective CDs, to impaired self-monitoring and psychotic symptoms (47,50).

Our results appear in line with accumulating evidence of altered structural and functional properties of the motor system in SCZ and its prodromal states (51–55). Hyperconnectivity between BA4 and the thalamus has been associated with abnormal motor behavior such as catatonia or diskinesia in SCZ (53) and was found to be higher in subjects at clinical risk who eventually convert to full-blown psychosis (54). Furthermore, recent evidence suggested a link between motor system pathology and clinical features of psychoses, including positive symptoms. For example, one study found that abnormal cerebello-thalamo-motor connectivity was able to track over time (1 year) the worsening of positive symptoms in subjects at clinical risk for developing psychosis (52). Considering this body of evidence, our finding that specific intrinsic oscillatory properties of the primary motor cortex are altered in FEP subjects and appear to predict the trajectory of positive psychotic symptoms may help identifying new targets for characterizing motor dysfunction in psychosis. Specifically, coupling TMS-evoked oscillations with measures of structural and functional connectivity may shed light on the relationship between intrinsic (i.e. local) and network (i.e. distant) aberrant activity and their association with clinical features of psychosis across stages of SCZ.

Personalized prediction tools to estimate transition to psychosis for high-risk subjects are typically based on a broad set of social (56), clinical (57), and biological variables (58). In FEP patients, structural MRI has been extensively studied due to its relative independence from the individuals’ cognitive condition and level of cooperation. Abnormalities in medial temporal and prefrontal cortices as well as in the networks connecting these cortical areas with subcortical regions are considered promising markers of poor symptomatic and functional outcome (59). However, at the individual level MRI measures alone have failed to inform accurate prediction of illness course in early psychosis cohorts (60). Therefore, the development of novel approaches that gauge functional brain responses without directly engaging patients on experimental tasks seems worthwhile. Here, we found that lower baseline RSP beta-2 frontal activity predicted worse clinical symptoms at six-month clinical assessments with high sensitivity and good specificity. If replicated in larger samples, TMS-evoked frontal deficits could be employed as a state biomarker to design individualized treatment plans and allocate additional resources towards patients who are expected to endure a poorer clinical outcome.GABA-ergic cortical interneurons play a critical role in the generation of spontaneous fast oscillatory activity in the human brain (61,62). Consistently, recent evidence shows that GABA-ergic transmission is also implicated in cortical responses to TMS (63). Interestingly, alterations in parvalbumin-positive, GABA-ergic inhibitory interneurons have been reported in SCZ by human post-mortem and animal studies (64,65) and were linked to the cognitive dysfunctions associated with the disorder (66,67). Recently, in a cortical network involved in visual working memory functions, gene expression of GABA-ergic transmission’s markers was found to be abnormally reduced in the frontal lobe of subjects with SCZ, as opposed to more caudal areas, compared to healthy controls (68). These findings are consistent with previous TMS-EEG findings from our research group (27) and may contribute to account for the intrinsic deficit of frontal areas in psychotic patients to generate the synchronous neural activity responsible for EEG fast oscillations.

Together with our present findings in FEP patients, this body of evidence points at deficits of GABA-ergic activity of cortical interneurons as a putative pathophysiological mechanism sustaining the clinical picture of psychosis since its earliest stages.Future work will help to address some of the limitations and questions left unanswered by this study. First, we recruited a relatively small sample of FEP patients. Thus, the findings reported here will need to be confirmed in larger cohorts of psychotic individuals at illness onset. Second, in the present study, we provided initial evidence that this measure was a reliable readout of psychotic symptoms at baseline and could also predict symptom severity at 6-month follow-assessments. Future work collecting longitudinal clinical evaluations in FEP patients will help to establish whether the TMS-evoked beta-2 RSP in frontal areas can reliably predict clinical course in these individuals, thus representing a predictive biomarker of psychosis. This approach will also clarify whether the TMS-evoked beta2 RSP reflects and predicts the persistence of psychotic symptoms after treatment (i.e., partial remission) rather than changes in clinical symptoms after a given period of time (i.e. treatment response), as our data seem to suggest (Fig. 3–4 as opposed to Fig. S2–3). Furthermore, by performing repeated TMS-EEG assessments it will be possible to assess whether these frontal oscillatory deficits can consistently track the course of illness in FEP patients, thus representing a potential monitoring biomarker of psychosis. Third, treatment trials with established antipsychotic compounds will determine whether TMS-evoked frontal beta-2 RSP may contribute to predict treatment response in FEP patients. For instance, a relatively preserved beta-2 RSP might correlate with a better response to medications, whereas lower beta-2 RSP values may be associated with treatment resistance. Also, the ability of an antipsychotic compound to ameliorate these oscillatory deficits may serve as an objective assessment of the efficacy of pharmacological trials in FEP individuals. One further limitation is that, similarly to other TMS-EEG studies in SCZ and psychosis (26,28,42), we based the intensity of stimulation on a functional readout of activation of the motor cortex, i.e. the rMT. In principle, this approach may be partially biased by the cortical thinning that has been reported in FEP subjects across multiple brain regions (69), including the motor cortex (70). However, consistent with previous findings (71), we report no significant differences in rMT between FEP patients and healthy comparison subjects. Furthermore, we ensured that all subjects showed a minimum peak-to-peak amplitude of 5 μv of the TEPs’ first component by using an online, real-time monitoring of the EEG response to TMS, as in previous studies (36). Nonetheless, future TMS studies in FEP should combine these innovative strategies with more objective measures for defining the intensity of stimulation, including modeling of the electric field induced by TMS (72). Finally, future studies employing neuromodulatory interventions, including transcranial direct or alternate current stimulation (tDCS/tACS) and theta-burst stimulation (TBS) protocols, could provide more focal, targeted approaches to improve frontal oscillatory impairments and related clinical symptoms in psychotic individuals in the early stages of illness.

Highlights.

• Evidence suggests that the motor system plays a key role in Schizophrenia

• We applied TMS-EEG to the motor and parietal cortices in FEP and controls

• We found reduced oscillatory activity in the beta2 range in the motor cortex in FEP

• This deficit correlated with SAP scores and predicted SAP scores at 6 months