Parkinson’s disease (PD) is a neurodegenerative disorder prevalent among middle-aged and elderly populations (1). The primary pathological characteristic of PD is the degeneration and loss of dopaminergic neurons in the substantia nigra-striatal pathway (2), resulting in motor symptoms (MS) and non-motor symptoms (NMS). Resting tremor, bradykinesia, and rigidity are predominantly features of MS, while NMS typically present as depression, hyposmia, cognitive impairment, sleep disorders, and constipation (3). Furthermore, the appearance of non-motor symptoms often precedes motor symptoms and dominates as PD progresses, exacerbating the burden on PD patients and complicating their management strategies (4). According to the Global Burden of Disease (GBD) report, approximately 6.1 million individuals were diagnosed with PD worldwide, and an additional 1.02 million new cases were reported in 2017, with this number continuing to rise alongside global aging trends (5–7). Moreover, it is foreseeable that by 2030, the number of PD patients is estimated to reach 4.94 million in China, with the burden caused by PD expected to rise substantially over the coming decades (8, 9).

It is widely recognized that depression represents the most prevalent mood disorder in PD patients, with at least one-third of individuals experiencing symptoms of depression (10, 11). This condition significantly impairs patients’ quality of life and may lead to varying degrees of disability (12–14). Consequently, early detection and timely intervention of depression are crucial to ensure positive treatment effects. Depression may be associated with the decreased dopamine levels in the anterior cingulate cortex (15). Pharmacological treatments, including dopamine agonists, selective serotonin reuptake inhibitors (SSRIs), and serotonin/norepinephrine reuptake inhibitors (SNRIs), have been employed to alleviate PD with depression (PD-D) patients (16, 17). However, the tolerance and adverse effects of drug therapy undermine the efficacy of these treatments (18).

Clinically, transcranial magnetic stimulation (TMS) modalities are typically classified according to stimulation paradigms and neurophysiological effects, including single-pulse TMS, paired-pulse TMS, deep TMS, and repetitive TMS (rTMS). Among these, rTMS induces small intracranial electric currents by stimulating neurons through alternating magnetic fields, which has been widely recognized for its safety and efficacy. It influences neurotransmitter activity and synaptic plasticity, for instance, it enhances the secretion and release of endogenous dopamine in the ipsilateral striatum, resulting in sustained alterations of neural circuits and subsequent improvement in depressive symptoms. Thus, it serves as a valuable adjunctive therapy within clinical treatment regimens for depression (19–21). The pattern of rTMS varies according to specific frequencies and patterns. Low-frequency stimulation (below 1 Hz) suppresses neuronal activity and reduces cortical excitability. In contrast, high-frequency stimulation (above 5 Hz) promotes neuronal depolarization, thereby increasing cortical excitability (22–24). Intermittent theta burst stimulation delivers high-frequency pulses within a short time frame, mimicking theta rhythms to achieve excitatory effects. Bilateral rTMS combines high-frequency stimulation targeting the left dorsolateral prefrontal cortex (LDLPFC), with low-frequency stimulation targeting the right dorsolateral prefrontal cortex (RDLPFC), modulating both excitatory and inhibitory effects across hemispheres. Aftanas et al. (25) reported that high-frequency rTMS (HF-rTMS) to be significantly more effective than placebo in improving motor and mood symptoms in PD patients. Zhuang et al. (26) demonstrated that low-frequency rTMS (LF-rTMS) significantly improved cognitive function and alleviated depression in PD patients, resulting in stable and long-term therapeutic benefits. Chen (27) comparing the effects of HF-rTMS and LF-rTMS on PD-D patients, concluded that although both modalities effectively alleviate depression, LF-rTMS exhibits superior therapeutic efficacy.

However, rTMS setting parameters which include stimulation frequency, location, and duration vary widely in clinical practice, so optimizing rTMS treatment protocols is a pressing issue. This study incorporates randomized controlled trials (RCTs) of rTMS with different frequencies and stimulation sites for treating PD patients with depression, aiming to assess the therapeutic efficacy and acceptability of various rTMS modalities.

This systematic review and network meta-analysis was conducted following the Preferred Reporting Items of the Guidelines for Systematic Reviews and Meta-Analysis (PRISMA) (28). The protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO), number: CRD42024564867.

This systematic review and network meta-analysis of 35 RCTs (2,353 participants) demonstrates that both HF-rTMS and LF-rTMS are effective in alleviating depressive symptoms in patients with PD. LF-rTMS appears to offer greater benefits, as evidenced by a higher SUCRA probability (96.7% vs. 70.0% for HF-rTMS). These findings suggest that LF-rTMS, which inhibits cortical excitability, may be particularly effective in downregulating hyperactive neural circuits associated with PD-D. Clinically, LF-rTMS could serve as a valuable adjunctive therapy, especially for patients with poor response to pharmacological treatments.

In this study, we observed variations in the scales used to assess depression during rTMS treatment across the included studies. These scales included the HAMD-17, HAMD-24, and the Hamilton Rating Scale for Depression (HRSD). Notably, various versions of the HAMD scale may have a certain impact on the final result. To address this, we conducted further analysis, which revealed that scores from the HAMD-17 scale aligned more closely with the overall findings. This may be due to several factors: As the most commonly used scale to evaluate the efficacy of antidepressants (30, 31), HAMD-17 is widely recognized for its reliability and comprehensive assessment, so the number of studies utilizing this scale has increased. Additionally, some studies may exclude patients with severe depression, which would make a more concentrated range of depressive severity among participants (32), and the therapeutic effect after receiving the intervention may be more similar. A meta-analysis (33) has suggested that the HAMD-6 may exhibit superior clinical properties compared to the HAMD-17, with its efficacy well-supported in clinical settings. This makes it a promising instrument for assessing core depressive symptoms in hospitalized patients and potentially assisting in treatment decisions (34–36). Employing multiple versions of depression assessment scales could improve the objectivity and robustness of efficacy evaluations. Among these tools, the BDI, well known for its high sensitivity and specificity, serves as a valuable tool for early detection of depression in PD patients (37, 38). In our analysis, we also included studies that used the BDI in selecting outcome measures to reduce the potential for measurement error. Nevertheless, the observed differences in efficacy may be constrained by the limited number of included studies, highlighting the need for further high-quality research to enhance the precision and generalizability of rTMS outcome assessments. It is also important to note that the use of different scales can introduce measurement bias across populations from diverse regions and cultural backgrounds. Since the majority of the included studies were conducted in China, our subgroup analysis indicated that while both studies from China and those from other countries demonstrated significant intervention effects, the effect size was larger in Chinese studies. This regional difference, potentially attributable to cultural or methodological factors, may limit the generalizability of our findings. Therefore, such variability should not be overlooked (39).

Recent studies have demonstrated that rTMS enhances global brain functional network connectivity, particularly by strengthening the connections between the prefrontal cortex and limbic structures such as the amygdala and cingulate cortex, thereby facilitating top-down modulation of emotional responses (40–43). Furthermore, rTMS promotes the release of key neurotransmitters, including dopamine and serotonin, which play a critical role in the amelioration of depressive symptoms (44–46). In addition, rTMS increase synaptic plasticity by inducing synaptogenesis and improving neurocircuitry involved in emotional regulation (47, 48). Based on the above, we analyzed various rTMS parameter settings, including the stimulation target, frequency, intensity, and total number of pulses. The dorsolateral prefrontal cortex (DLPFC) plays a central role in regulating emotional, social, and cognitive functions (49). Studies have shown that emotional regulation in individuals with depression often shifts from the lateral frontal pole cortex to the DLPFC (50, 51). Moreover, individuals with major depressive disorder exhibit reduced frontoparietal network connectivity, with the DLPFC implicated in this dysfunction. Depression is generally characterized as an exaggerated response to external stress and exhibits specific symptoms. It has been associated with hyperactivity of the RDLPFC (52–55). Several studies (26, 56, 57) have indicated that low-frequency stimulation targeting the RDLPFC, the most commonly used region in antidepressant rTMS protocols, has shown favorable efficacy, the findings are similar to our analysis results. Considering the limited understanding of the specific neural circuitry involved in PD with depression, as well as the mechanisms through which rTMS exerts its neuromodulatory effects, it is hypothesized that cortical activation may extend beyond the stimulated regions and propagate through neural networks to distant areas (58). LF-rTMS targeting the hyperactive RDLPFC can induce effects resembling long-term depression (LTD), leading to the suppression of cortical excitability. This inhibitory effect may contribute to the downregulation of hyperactive neural circuits associated with depressive symptoms (59). LF-rTMS may promote striatal dopamine release via the fronto-parietal-striatal-cortical pathway, thereby improving the neurochemical basis of PD-D (26).

The parameters of rTMS, particularly stimulation intensity and total pulse number, were also preliminarily analyzed in this study. Rossini et al. (60) compared stimulation intensity at 80 and 100% of the RMT, finding that 100% RMT produced better antidepressant effects. Since stimulation intensity is adjusted relied on the minimum threshold needed to induce a minimal muscle contraction response, it varies among individuals. Caulfield et al. (61) developed a new approach, termed “A Personalized E-field X Motor Threshold” (APEX MT), to make stimulation intensity more personalized and accurate. In addition, the findings indicated that rTMS yielded optimal antidepressant effects with 1,200–1,500 high-frequency pulses or 360–450 low-frequency pulses, with no additional benefits observed from increasing the total pulse count (62, 63). This phenomenon may stem from a plateau effect in pulse quantity, whereby exceeding a certain threshold does not further enhance cortical excitability and may even induce inhibitory effects or neural fatigue (64, 65). Generally, the optimal number of pulses may be influenced by the specific rTMS frequency utilized, as well as individual patient response and tolerance. Achieving the appropriate balance is crucial for maximizing therapeutic efficacy while minimizing patient burden. The distribution and intensity of the electric field generated within the target brain region during TMS are affected by coil positioning, as the coil induces intracortical currents that modulate neuronal firing patterns (66). Currently, anatomical localization methods rely on scalp landmarks, but these are susceptible to individual anatomical variability. Neuronavigated localization, utilizing neuronavigation systems, enables more precise coil placement; however, existing studies have not conclusively demonstrated the significant superiority of neuronavigation (67). Notably, Moser et al. used a novel multi-task deep neural network to rapidly determine the optimal coil position for inducing maximal electric field at the target site (68).

Compared to existing meta-analyses, our study not only provides a more detailed analysis of rTMS parameters (frequency, target location, stimulation intensity, and total pulses count), but retrospectively assesses the impact of rating scales on outcomes, thereby offering clinical guidance for rTMS use in PD-D and insights into assessment scale selection in practice.

This study has several limitations: Influenced by the included studies, the range of TMS modalities investigated is limited, and the predominance of studies from Chinese databases, combined with cultural differences and varied evaluation scales may introduce bias, potentially affecting the robustness of the results. Additionally, both the intervention and control groups received antidepressant medications in most included studies, which may affect the accurate evaluation of the efficacy of TMS. Furthermore, the heterogeneity in rTMS parameter combinations and the lack of information on coil positioning and localization methods restrict the standardization of treatment recommendations. The absence of reported Hoehn–Yahr stages in baseline characteristics in most studies further hinders the evaluation of disease severity’s impact on treatment outcomes. These limitations underscore the need for more high-quality, multi-national randomized controlled trials in future research, in order to enhance the generalizability, reliability, and clinical applicability of the findings.

This study indicates that both HF-rTMS and LF-rTMS are effective for improving depressive symptoms in PD patients, with LF-rTMS potentially exhibiting greater benefits compared to other interventions. Moreover, the application of specific intervention parameters, such as targeting applying no more than 1,200 pulses per session, may yield enhanced outcomes. Nevertheless, further high-quality RCTs are required to validate and refine this conclusion.