Introduction

Brain stimulation techniques such as electroconvulsive therapy, deep brain stimulation or repetitive transcranial magnetic stimulation are often accused by its critics to induce brain damage. In order to facilitate general acceptance for these techniques, the current study tested brain damage markers in repetitive transcranial magnetic stimulation (rTMS). Repetitive TMS has its origin in neurological research and has transformed into a promising treatment option for several neuropsychiatric diseases such as major depressive disorder, tinnitus, auditory hallucinations in schizophrenia, Parkinson disease, Alzheimer dementia, stroke recovery, dystonia, pain and addiction. The US Food and Drug Administration (FDA) approved this noninvasive brain stimulation approach for the treatment of major depressive disorder (MDD) (O’Reardon et al. 2007), yet in other diseases, effectiveness of rTMS has been shown only in a preliminary way. However, further prospective, randomised, well-blinded trials are needed to establish more evidence and to collect more data on optimal parameter choice, safety and tolerance.

Usually, rTMS is considered as well tolerated. The rare incident of seizure is supposed the greatest acute risk. Other adverse events are syncope, minor pains such as headache or local scalp discomfort and minor cognitive changes. Safety guidelines recommend certain ranges of stimulation parameters to minimise side effects (Wassermann 1998).

In order to provide further evidence to the safety of rTMS on a cellular level even when applying exceptional high frequency stimulation (30 Hz), we analysed serum levels of neuron specific enolase (NSE) and protein S-100—sensitive markers for neural or glial brain damage—in patients with depression before and after an ultra high frequency rTMS treatment. Beyond its function as damage marker, it has been shown that peripheral levels of S-100 increased in patients with depression (Schroeter et al. 2008) might predict the response to antidepressant treatment (Arolt et al. 2003) and that successful antidepressant treatment might decrease S-100 levels (Schroeter et al. 2002). Serum levels of NSE were reported to be reduced in drug-naive patients with uni- or bipolar depression compared to controls (Wiener et al. 2013), but there are no studies concerning modulation after treatment. We hypothesise that even the ultra high frequency rTMS treatment approach will not change NSE or S-100 serum concentration.

Methods

The present study has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) and written informed consent was obtained from all participants before inclusion in the study. The clinical data of this randomised, double-blind, sham-controlled study with a detailed description of patient selection and rTMS treatment were already published previously (Ullrich et al. 2012). Shortly, 43 subjects (16 males and 27 females) with a primary diagnosis of at least a moderate depressive episode (MDE), age from 18 to 75 years and stability of antidepressant medication (either venlafaxine or mirtazapine) were recruited. The allowed concomitant medication was lithium, when already established at least 4 weeks before inclusion, lorazepam up to 1.5 mg/day and antipsychotics.

Primary endpoint of the study was clinician-rated measure of depressive symptoms with the Hamilton Depression Rating Scale (HDRS; 21 items version). Evaluation took place before the rTMS treatment and after completion of 3 weeks (15 sessions) of stimulation. Another outcome parameter was if the patient could be considered as “responder”, which is defined as a reduction of at least 50 % in HDRS or as “remitter”, which is defined as a HDRS <8.

Repetitive TMS was administered using a MagPro stimulator (Dantec, Denmark) with a figure-of-eight coil with 97 mm outer diameter over the left dorsolateral prefrontal cortex (DLPFC). We choose two different, alternated randomised stimulus conditions: (1) Left side ultra high frequency (30 Hz) stimulation (20 trains of 3 s duration with 57 s intertrain intervals, total 1,800 pulses per session). This condition was regarded as the active treatment. (2) Left side low frequency (1 Hz) stimulation (11 trains of 90 s duration with 30 s intertrain intervals, total 990 pulses per session). That condition was regarded as not active treatment.

Blood samples for determining NSE and S-100 were taken the day before the rTMS treatment was started, after day 3, 7, 10, 14 and 17 of the treatment, 1 day and 1 week after the last rTMS session. Samples were stored at −20 °C until analysis. Serum NSE and S-100 assays were performed using a radioimmunoassay technique (Santec Medical, Bromma, Sweden) (Pahlman et al. 1984; Agelink et al. 2001). Due to logistic reasons, the samples were only taken from a subgroup of patients (22 subjects) that did not differ from the whole sample in a relevant way.

Statistics were performed using STATA® (StataCorp, Texas 77845, USA, version 11) at a significance level of 0.05. Baseline differences between the two groups were analysed with the two-tailed paired t tests or if appropriate with the Pearson’s Chi square test. Pre–post rTMS comparisons of the outcome parameters (HRSD) were analysed with two-tailed paired t tests. Comparisons between the two stimulation groups were carried out with one-way ANOVA, including possible covariates. Comparison of the time course between baseline and end of treatment was measured by repeated measurement regression analysis. A logistic regression analysis with either response or treatment group was also performed for initial, final and differences (i.e. final − initial) of NSE and S-100.

Results

From 22 patients (6 male, 16 female), data for NSE and S-100 levels are available. 14 subjects were randomised for the ultra high frequency (UHF) group and eight subjects for the low frequency (LF) group. In the UHF group, 85.7 % were female, the average age was 60.4 years (SD 10.0) and baseline HDRS was 30.4 (SD 4.8). In the LF group, 50.0 % were female, the mean age was 57.3 (SD 9.1) and HDRS mean before the treatment was 28.2 (3.9). The stimulus intensity did not differ in the two groups (Stimulus intensity UHF 42.9 % (SD 2.3) and LF 42.9 % (SD 1.5). Both stimulation groups showed an improvement after the treatment (p < 0.001); in the active UHF group, HRDS after treatment was reduced about 7.9 points (SD 4.5) compared to HDSR at baseline, the LF group showed a reduction of the HDRS of 5.5 points (SD 2.3). The difference between both groups was not statistical significant (p = 0.19). 30 Hz UHF was well tolerated by the participants. All patients completed the study and no seizure occurred. No adverse events were reported from the patients.

NSE serum concentration at baseline was 13.3 ng/ml (SD 6.1) and 12.9 ng/ml (SD 1.7) in the UHF and LF group, respectively. After the 3 weeks of stimulation, NSE levels were 11.7 ng/ml (UHF; SD 1.8) and 11.0 ng/ml (LF; SD 1.7). The differences between baseline and end were not significant in both groups (UHF 1.7 ng/ml, SD 1.3, p = 0.23; LF 1.9 ng/ml, SD 1.5, p = 0.24); the difference at baseline (p = 0.86), after the end of stimulation (p = 0.36) and the difference of these two points (0.71) were also not significant between the two groups (Fig. 1). In addition, the HDRS did not correlate with the level of serum NSE at all [F(7,14) = 0.71; p = 0.67], nor did the intake of lithium, antipsychotics or benzodiazepines.

Fig. 1
figure 1

NSE serum concentration: Boxplot of NSE serum concentration in the ultra high frequency (UHF) group and the sham group during the study (t1–t8)

Full size image

S-100 protein level at baseline was 0.077 ng/ml (SD 0.048) in the UHF group and 0.091 ng/ml (SD 0.027) in the LF group. After the treatment, S-100 levels was 0.064 ng/ml (SD 0.018) and 0.079 ng/ml (SD 0.011) in the UHF and LF group, respectively. The differences between baseline and end were not significant in both groups (UHF 0.013 ng/ml, SD 0.046, p = 0.32; LF 0.013 ng/ml, SD 0.028, p = 0.24), neither were the differences between the groups [F(1,20) = 0.00; p = 0.98] (Fig. 2). No other covariate correlated with S-100 concentration in a significant way.

Fig. 2
figure 2

S-100 serum concentration: Boxplot of S-100 serum concentration in the ultra high frequency (UHF) group and the sham group during the study (t1–t8)

Full size image

Neither for NSE nor for S-100, the time courses between baseline and end of treatment reveal any statistical significant difference measured by repeated measurement analyses.

All logistic regression analyses (including response and treatment group) revealed no statistical significance.

Discussion

In this randomised trial, patients with severe major depression got 3 weeks of ultra high frequency rTMS treatment additionally to a concurrent stable antidepressant medication. rTMS was done either with 30 Hz stimulation at the left DLPFC or left low frequency stimulation that was considered the sham condition. Markers for brain damage were analysed before and after the treatment. Both NSE and S-100 serum levels did not change during the stimulation treatment over 3 weeks. In addition, we found neither differences between the levels of the active and the sham group nor intraindividual differences. Treatment response or group did not predict serum levels.

To our knowledge, this is the first study examining markers for neuron and glial damage in rTMS. For another brain stimulation technique, electroconvulsive therapy (ECT) exists more data regarding this topic; in general, ECT does not lead either to elevation of such markers (Zachrisson et al. 2000; Agelink et al. 2001; Palmio et al. 2010), however, a single report found a slight increase of protein S-100 (Arts et al. 2006). The focus in rare rTMS serum studies has been the brain-derived neurotropic factor (BDNF) (Lang et al. 2006; Zanardini et al. 2006; Gedge et al. 2012) and hormones (Evers et al. 2001; Szuba et al. 2001) with heterogeneous results.

Interestingly, the glial protein S-100 can act as both (Arolt et al. 2003) as a marker brain damage (Donato 2001) or as a marker for neuronal plasticity associated with therapeutic response (Whitaker-Azmitia and Azmitia 1994; Busnello et al. 2006). Based on this, the critical discussion of the use of S-100 as a brain damage marker in our study is justified.

Major limitation of the study is the small sample size, thus smaller changes might have been missed. In addition, the half-life of S-100 is only about 30 min (Jonsson et al. 2000), thus acute rTMS effects of S-100 might have been missed too.

Conclusion

In our small study, we found no evidence that rTMS over the left dorsolateral prefrontal cortex with ultra high frequency (30 Hz) stimulation over 3 weeks lead to a rise in the known brain damage markers NSE and S-100. This preliminary negative finding contributes evidence to the general safety of rTMS.