Description of the studies and bias assessment

The literature search yielded a total of 531 potential studies. After removing 211 duplicates, 320 studies remained. Among these, 303 studies were excluded during the screening phase as they did not meet the inclusion criteria, and one study was excluded later due to its focus on animal models of Alzheimer’s Disease [29]. The remaining 17 studies were retrieved and included in the quality control analysis. Out of these, 2 were animal studies [30, 31], 7 were human longitudinal studies [32,33,34,35,36,37,38], and 8 were human clinical trials [39,40,41,42,43,44,45,46]. While the year of publication was not an exclusion criterion, the papers included in the systematic review covered the time period from 2019 to 2023. Among the 17 selected papers, 2 were published in 2019, 3 in 2020, 5 in 2021, 6 in 2022, and 1 in 2023.

For the included human studies, we used the NHLBI tools for clinical and observational cohort studies to assess and rate their quality. In the case of animal studies, we applied the SYRCLE checklist to evaluate the risk of bias. Two independent blinded raters assessed the included studies after the screening process. Tables S1 to S3 present the inter-rater agreement of quality ratings and the scores assigned to each study for human (observational cohort studies and clinical studies) and animal studies.

Animal studies

We found two animal studies relating to rTMS effects on inflammation in models of psychiatric diseases [30, 31]. The first animal study investigated the anti-inflammatory effects of rTMS in a depression animal model using mice and focusing on glial activation [30]. For the measures of the cytokines, enzyme-linked immunosorbent assay (ELISA) was used to measure the concentrations of the cytokine IL-1β, IL-6 and TNF-α in specific brain regions [30]. In case of microglia activation, immunofluorescence was performed on brain tissue, studying specific cellular markers in different brain regions [30]. Western blot and quantitative real-time polymerase chain reaction was used to study protein and gene expression in brain tissue, respectively [30]. The study revealed that rTMS treatment, following exposure to chronic unpredictable mild stress (CUMS), increased sucrose intake, decreased immobility time in the forced swim test, and enhanced exploratory behavior in the open field test. These findings suggest a reversal of CUMS-induced depressive-like behavior after rTMS treatment. Regarding neuroinflammatory mechanisms, rTMS treatment reduced microglial activation levels. Furthermore, rTMS reversed the polarization of microglia from the M1 to M2 phenotype and increased astrocyte levels. Additionally, rTMS lowered the concentrations of pro-inflammatory cytokines such as IL-6, β, and TNF-α in the hippocampus and prefrontal cortex (PFC) by modulating the TLR4/NF-κB/NLRP3 signaling pathway in the animal model of depression.

The second animal study examined the effects of rTMS on depression and anxiety-like behavior and on Nuclear Factor-E2-related Factor 2-mediated (Nrf2) anti-inflammatory processes in rats [31]. A one-week rTMS treatment successfully prevented CUMS-induced depression- and anxiety-like behaviors. The one-week treatment with real rTMS significantly increased the nuclear expression of the Nrf2 protein compared to sham rTMS. Moreover, after rTMS treatment, the expression of TNF-α, inducible nitric oxide synthase (iNOS), IL-1β, and IL-6 decreased in the hippocampus, which counteracted the initial elevation induced by CUMS. All the above-mentioned effects of rTMS were nullified after Nrf2 knockdown by transfection of Nrf2 small interfering RNA. These results imply that Nrf2 may play a crucial role in the antidepressant and anti-inflammatory effects of rTMS.

Human studies - longitudinal observational studies

Of the 15 human studies included, we identified seven longitudinal studies, as described in Table 1 [32,33,34,35,36,37,38]. The quality control of these clinical studies is reported in Table S2 and the overview of the rTMS-induced changes in Table 2. Generally, these studies assessed peripheral cytokines and other inflammation-related markers, including BDNF, metabolites of the kynurenine pathway, oxidized phosphatidylcholines and oxylipins, as well as soluble Triggering receptor expressed on myeloid cells-2 (sTREM2) during rTMS treatment in patients with depression. Only one study reported findings with microRNA involved in inflammatory processes (miRNA-16-5p, miRNA-93-5p, and miRNA-146a-5p). Initially, all studies reported a significant reduction in depression severity following rTMS therapy. However, the reported inflammatory parameter changes after rTMS treatment exhibited broad variability.

Peripheral cytokines

Three longitudinal studies evaluated changes in peripheral cytokines after rTMS therapy in patients with TRD [35,36,37]. In a pilot longitudinal study, following rTMS therapy, IL-1β, TNF-α, and IL-6 did not change significantly in the 11 participants with TRD [35]. Further, no significant differences were observed between responders and non-responders after rTMS therapy [35]. However, the decrease in IL-1β after rTMS therapy was associated with an improvement in part 3 of the Color Stroop Test. Changes in other cytokines and cognitive tests did not show significant correlations [35]. In the second longitudinal study [36], the authors reported no significant effects of rTMS on TNF-α concentration in patients with TRD. However, TNF-α concentrations showed a positive significant correlation with the final HAM-D scores in the rTMS group [36]. Finally, the third longitudinal study [37] reported significant differences in IL-18 concentrations between both treatment groups (rTMS and medication groups) and the control group (baseline only assessed). Upon excluding the control group from the analysis, the authors found significant differences for IL-8 and IFN-γ when comparing the rTMS to the medication group [37]. Furthermore, rTMS treatment increased the TGF-β1 level. In contrast, the medication groups exhibited increased IL-18, IL-6, and IL-1β after two weeks of treatment [37]. However, in the last study, is unclear whether the changes in inflammatory parameters were associated with improvement in psychopathology, finding only significant results between baseline inflammation and symptom improvement, as described in Table 1 [37].

Metabolites of the kynurenine pathway

One pilot longitudinal study in patients with TRD, which compared three different treatment options (pharmacotherapy, ECT, and rTMS), showed no significant reductions of metabolites of the kynurenine pathway (kynurenine, kynurenic acid, 3-hydroxykynurenine or quinolinic acid), which is related to inflammatory processes [47], before and after rTMS treatment [32]. Despite the lack of significant changes in kynurenine pathway metabolites, a reduction in TNF-α was observed during rTMS treatment. Similarly, the rTMS group showed a significant increase of 3-hydroxykynurenine-to-kynurenine (HK/KYN) ratio after four weeks of treatment. However, no analyses were performed to establish an association between the decrease in TNF-α or HK/KYN and the improvement of depressive symptoms.

Lipid mediators and steroids

In a longitudinal study with an exploratory design, oxidized phosphatidylcholines (OxPc) and oxylipins were assessed before and after rTMS treatment in patients with major depressive disorders [33]. As reported by the author, oxylipins is produced by enzymes such as cyclooxygenase or lipoxygenase and plays a crucial role in inflammation [33, 48]. At baseline, non-fragmented OxPc and long-chain products were significantly elevated in patients compared to healthy controls [33]. When comparing rTMS remitters and non-remitters at baseline, the remitters exhibited significantly higher total OxPc than the non-remitters, with significant differences observed in fragmented OxPCs, aldehydes, carboxylic acids, PONPC, and 1-palmitoyl-2-azelaoyl-sn-glycero-3-phosphocholine [33]. However, there were no significant alterations in OxPCs following rTMS treatment in patients with MDD. Further, no significant changes in OxPC concentration were observed after rTMS treatment when categorizing the patient group into remitters and non-remitters. Additionally, in identifying 32 different oxylipin species, no differences were discerned between remitters and non-remitters concerning pre-rTMS concentrations [33].

Testosterone, a steroid hormone, presents an inhibitory effect on the expression of TNF-α, IL-6, and IL-1 [49]. One included study evaluated if testosterone changed after rTMS along with symptom improvement. However, testosterone levels did not show significant group, time, or group-by-time interaction effects. Moreover, the decrease in testosterone was not associated with the rTMS treatment [38].

Growth factors

One longitudinal study assessed the concentrations of BDNF and VEGF before and after rTMS treatment in participants with major depression [36]. As previously noted, BDNF and VEGF have been reported to be associated with inflammatory processes [17, 50]. In this study [36], rTMS revealed significant increase between the third and the first measurement time point for BDNF concentrations. Conversely, changes in VEGF during rTMS treatment were non-significant in pairwise differences tests [36]. Nevertheless, initial BDNF serum levels correlated negatively with BDNF increase after five rTMS procedures. Furthermore, the overall percentage of BDNF increase during treatment correlated negatively with the initial MADRS and HAM-D scores [36]. BDNF correlations with clinical scores after treatment were not significant. Similarly, no significant correlations were observed between VEGF and clinical scores [36].

Micro-RNA involved in inflammatory processes

Regarding miRNAs involved in inflammatory processes, all three tested miRNAs (miRNA-16-5p, miRNA-93-5p, and miRNA-146a-5p) demonstrated higher ΔCt values in the rTMS group compared to the medication group, indicating lower expression of these miRNAs in the rTMS group [37]. Additionally, lower expression of miRNA-146a-5p at baseline correlated with less successful rTMS treatment as assessed with HAM-D-17, MADRS, and PHQ-9 scores. Interestingly, in the rTMS group, ΔCt values of miRNA-93-5p positively correlated with IL-18 concentrations [37].

Other inflammatory and immune response markers Soluble triggering receptor expressed on myeloid cells – 2 (sTREM2)

No significant reduction of sTREM2 level could be observed before and after rTMS treatment. The changes in sTREM2 were not correlated with age, sex, BMI, smoking status and symptom improvement [34].

Serum amyloid A (SAA)

Serum amyloid A (SAA), an acute-phase protein [51], did not showed changes in a longitudinal study during rTMS treatment in patients with MDD [38]. In addition, the authors reported no significant group, time, or group-by-time interaction effects. The decrease of SAA was not associated with the rTMS treatment. However, the authors reported a significant relationship between the reduction of SAA levels and the decrease in HAM-D in the rTMS group during the second week of treatment [38].

Study limitations

Overall, the included longitudinal studies had several limitations that must be considered when evaluating the quality of the results. Firstly, none of the studies reported a power analysis. Additionally, there were non-uniform rTMS protocols across the studies, comparisons between different diseases using various stimulation methods, exploratory designs with small sample sizes (i.e., n < 30), no correction for medication effects, a loss of information exceeding 20%, and a lack of adjustment for sample characteristics (i.e., sex, age, or BMI).

Human studies - clinical studies

Out of the 15 included human studies, we identified eight clinical studies [39,40,41,42,43,44,45,46]. The quality control of these clinical studies is reported in Table S3. Similar to the included longitudinal studies, these clinical trials evaluated peripheral cytokines as well as other inflammation-related markers, such as white blood cells (WBC), acute-phase proteins (e.g., CRP, antithrombin III), metabolites of the kynurenine pathway, and growth factors (e.g. BDNF), and their relations to clinical outcomes. Six out of eight trials included patients with depression, one with schizophrenia, and one post-stroke patients. The latter was included since the authors also evaluated the effects of rTMS on depressive symptoms assessed with the geriatric depression scale (GeDS) after the ischemic event. Short descriptions of the clinical studies are summarized in Table 3 and the overview of the rTMS-induced effects is summarized in Table 4. In the clinical trials, the reported changes of inflammatory parameters after rTMS treatment exhibited similar variability as in the longitudinal studies.

Peripheral cytokines

One prospective clinical study in patients with post-stroke cognitive impairment demonstrated by performing a reverse transcription polymerase chain reaction just after the rTMS treatment a downregulated expression of mRNA IL-1β, IL-6, TNF-α, and TGF-β, in comparison to the mRNA expression before the rTMS-treatment of the above-mentioned cytokines [39]. Twelve weeks after competition of the rTMS treatment, a sustained reduction in gene expression of IL-1β was found [39]. Furthermore, changes in mRNA expression of IL-6 correlated negatively and significant with changes in the Auditory Verbal Learning Test and with negatively with changes in Complex Figure Test scores. However, rTMS treatment did not reduce depressive symptoms in the GeDS in post-stroke participants, and rTMS showed only marginal effects on participants with moderate-to-severe depression. In addition, there were no correlations between changes in inflammatory parameters and depression symptomatology based on the GeDS scores in the study participants with post-stroke cognitive impairment.

In another prospective clinical study [40] examining 58 elderly patients with treatment-resistant depression (29 in rTMS group and 23 in control group with medication), the effects of rTMS on serum levels of IL-1β and TNF-α and their association to changes in depressive symptoms were investigated. HAMD-24 scores significantly reduced in the rTMS group during the treatment. Additionally, both IL-1β and TNF-α concentrations were significantly reduced in the rTMS group. Finally, HAMD-24 scores were positively correlated with both IL-1β and TNF-α concentrations. Nevertheless, it was not specified which HAMD-24 scores (e.g., at the beginning or the end of the trial) correlated with IL-1β or TNF-α.

A prospective, double-blind, controlled trial involving 57 patients with depression (29 in the rTMS group and 28 in the sham-stimulation group) and 30 healthy controls focused on differences in baseline inflammatory cytokines between patients and healthy controls, changes in inflammatory cytokines during rTMS treatment, and correlations between changes in HAMD-24 scores and inflammatory cytokines after rTMS treatment [41]. Regarding baseline inflammatory cytokines, TNF-α concentrations were significantly lower in the healthy controls than in the patients with depression; however, IFN-γ, IL-4, and IL-2 were significantly higher in the healthy controls. Concerning changes in inflammatory cytokines during rTMS treatment, the rTMS group exhibited significant increase in IL-2 and CRP-hs levels after two weeks of treatment. Between weeks 2 and 12, the levels of TNF-α and IFN-γ underwent significant decrease. Finally, after 12 weeks of treatment, significant increases were observed in the levels of CRP-hs. Furthermore, the change in depressive symptoms (measured with the HAMD-24) was negatively correlated with the changes of TNF-α after two weeks and negatively correlated with the changes of IL-2 after 12 weeks; however, changes in depressive symptoms positively correlated with changes of IL-4 in both the rTMS and the sham group.

In another randomized sham-controlled trial involving 53 patients with depression (27 in the rTMS treatment group and 26 in the sham-stimulation group), the study aimed to investigate the effects of theta-burst stimulation (TBS), a variant of rTMS treatment, on inflammatory cytokine levels and depressive symptoms. No significant changes in serum inflammatory cytokines, including IL-1β, IL-2, IL-4, IL-5, IL-6, IL-12p40, and IP-10, were observed after the treatment in either group. However, an increase in CRP was observed in the sham-stimulation group [42]. There was a significant difference in changes in HAMD-21 scores between the groups (sham > TBS) after four weeks of treatment.

Cellular immunity

One randomized double-blinded controlled trial [46] investigated the clinical effects of high-frequency rTMS on 47 long-term hospitalized veterans with schizophrenia (27 in the rTMS group, 20 in the sham group). The study aimed to identify possible correlations between clinical improvement and inflammatory parameters. No significant correlation was observed between WBC count and overall changes in the Positive and Negative Syndrome Scale (PANSS). However, an in-depth analysis of specific items within the PANSS revealed a negative association between WBC count at baseline and the improvement of a subgroup of positive symptoms, which was also confirmed by a regression analysis.

Acute phase proteins

One randomized double-blinded controlled trial reported no significant correlations between CRP and overall changes in the PANSS in patients with schizophrenia [46]. Finally, one clinical trial [43] examined the effects of occipital rTMS treatment on four different inflammation-related proteins (ATIII, CRP, ITIH4, and VDB) in 74 patients with Major Depressive Disorder (MDD), who were drug-naive or had been drug-free for at least two weeks. In this study, participants were allocated into three groups: a standard rTMS group, an individualized rTMS group, and a sham-stimulation group (see Table 1). After individualized rTMS therapy, the concentrations of all proteins were significantly reduced, with the exception of CRP. Participants in the standard rTMS group showed significant reductions in all four proteins. Sham-group did not show significant differences for the four parameters mentioned above. Regarding psychometric assessments, in the individualized rTMS group, only changes in ATIII showed significant positive correlations with changes in HAMD-24, SDS, BHS, and SAS. Moreover, changes in ATIII showed a significant positive effect in a multiple linear regression model with changes in HAMD-24 in the individualized rTMS group. Finally, baseline ATIII in both the standard and the individualized rTMS groups exhibited good predictive performance for the response after five days of rTMS treatment and predicted the values of HAMD-24 at follow-up. HAMD-24 values at follow-up were also predicted by changes in ATIII following rTMS treatment.

Metabolites of the kynurenine pathway

Two randomized controlled trials [44, 45] studied the effects of changes of metabolites of the kynurenine pathway along changes in the psychopathology. In both trials, no significant changes on the metabolites of the kynurenine pathway correlated with changes in psychopathology. As a secondary finding for