Systematic review of repetitive transcranial magnetic stimulation for post-stroke hemiplegic shoulder pain

Abstract

Background

Repetitive transcranial magnetic stimulation (rTMS) has shown potential in alleviating hemiplegic shoulder pain (HSP) and improving upper limb function, yet its efficacy remains debated. This study aims to assess the effectiveness of rTMS for HSP through a systematic review and meta-analysis.

Methods

Four databases were searched with the keywords “rTMS” and “HSP”. Adults aged 18 years and older with post-stroke HSP were included. The primary outcomes were pain scores and upper limb function scores, and the secondary outcomewas the incidence of adverse events. The risk of bias was assessed through the ROB tool in Review Manager 5.4.1, and statistical analysis was primarily conducted through this software.

Results

A total of 52 articles were identified from PubMed, Embase, Cochrane Library, and CNKI. Following literature screening, 11 studies were included in the analysis. The quality of the included studies was moderate.The studies encompassed 584 patients with post-stroke HSP and their average age was 62. The analysis revealed that rTMSwas significantly more effective in relieving pain compared to the control group (SMD = -1.14, p < 0.01), and low-frequency rTMSwas superior to high-frequency rTMS. In terms of improving upper limb function, rTMSwas also significantly more efficacious compared to the control group (SMD = 2.20, p < 0.01), and low-frequency and high-intensity rTMSwere more beneficial.

Conclusions

This study highlights the potential efficacy of rTMS. However, the heterogeneity among included studies, limited sample sizes, and lack of long-term follow-up data restrict the generalizability of the results.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10072-024-07961-3.

Introduction

Stroke is one of the top three causes of increased mortality rates globally [1]. Hemiplegic shoulder pain (HSP) is a common complication among stroke survivors, affecting approximately 30-70% of this population [2]. HSP not only severely impacts patients’ quality of life but also complicates rehabilitation efforts and adds economic burden [3]. Patients with HSP often experience severe shoulder pain and restricted upper limb mobility, leading to a decrease in activities of daily living. The pathogenesis of HSP is multifactorial, involving both central nervous system-related and local lesions. Local lesions include shoulder subluxation, complex regional pain syndrome, biceps tendinitis, rotater cuff tears, adhesive capsulitis, and increased muscle tension. Central nervous system-related causes may share some common and complex mechanisms, involving abnormal neurotransmitter release, neuroinflammatory responses, changes in neuroplasticity, and mood disorders [4]. The complexity and variability of these mechanisms make the treatment of HSP particularly challenging.

Repetitive transcranial magnetic stimulation (rTMS), a non-invasive brain stimulation technique, has recently demonstrated significant potential in the field of neurological rehabilitation. rTMS has been shown to promote neuroplasticity and functional recovery [5–7], and alleviate stroke-related pain and mood disorders [8–10], offering substantial clinical benefits. Despite evidence suggesting that rTMS has some efficacy for HSP, there remains controversy regarding its effectiveness and mechanisms. Some studies indicate that rTMS can significantly alleviate HSP [11], while others have failed to observe notable effects [12]. Some research supports that rTMS improves HSP by enhancing neuroplasticity, promoting the reconstruction of neural networks, and reducing pain while improving motor function [13]. Other studies propose that rTMS alleviates pain by modulating neurotransmitter release (e.g., glutamate, GABA) or suppressing neuroinflammatory responses, though these mechanisms are not uniformly agreed upon [14].

This study, therefore, employed a systematic review and meta-analysis to comprehensively evaluate the effectiveness of rTMS in treating post-stroke HSP and aims to address inconsistencies in existing research. Furthermore, through the analysis of different frequencies, intensities, and treatment parameters of rTMS, the study seeks to identify the optimal regimen for alleviating HSP pain and improving upper limb function, providing evidence-based recommendations for clinical practice.

Materials and methods

Study registration and adherence to PRISMA guidelines

This study was performed in accordance with PRISMA guidelines [15] and pre-registered with PROSPERO (Registration Number: CRD42024525376).

Literature search

A comprehensive search was carried out across PubMed, Embase, Cochrane Library, and China National Knowledge Infrastructure (CNKI) databases. The primary search terms included “rTMS” and “HSP,” with the search covering the period from database inception to June 2, 2024. No language or other restrictions were applied. The search history is provided in Table S1-4.

Literature screening

Two independent reviewers conducted the literature screening, and a third expert consolidated the results. The inclusion criteria were as follows: (1) Patients had post-stroke HSP and were 18 years or older; (2) Intervention measures were rTMS, irrespective of frequency, duration, or location; (3) The control group received treatment strategies without rTMS, including but not limited to sham rTMS, acupuncture, intramuscular efficacy patch therapy, cold water immersion, rehabilitation training, extracorporeal shock wave therapy (ESWT), or laser therapy. For studies where non-sham rTMS was the control, the control group must have received the same background treatment as the rTMS group, ensuring comparable treatment conditions. (4) Primary outcomes included pain scores and upper limb function scores, while the secondary outcome was adverse events. (5) Only randomized controlled trials (RCTs) were included. Exclusion criteria were: (1) Full texts cannot be assessed; (2) Conference abstracts, communications, or other non-peer-reviewed sources.

Risk of bias assessment

The risk of bias was assessed independently by two reviewers, with a third reviewer consolidating the results. The Review Manager 5.4.1 ROB tool was utilized for this assessment. The risk of bias was evaluated across seven domains: (1) random sequence generation; (2) allocation concealment; (3) blinding of participants; (4) blinding of outcome assessors; (5) incomplete outcome data; (6) selective reporting; (7) other sources of bias. Each domain was categorized as unclear, low, or high risk. The overall quality of studies was rated as low, moderate, or high based on all these domains.

Data extraction

Data were extracted independently by two reviewers, and a third reviewer consolidated the data. A data extraction form was pre-designed. Key extracted information included: (1) study detasupils such as first author, methodology, year, and country; (2) population characteristics including sample size, gender, age, and disease severity; (3) outcome measures, with pain scores and upper limb function scores extracted as mean and standard deviation.

Statistical analysis

Data analysis was primarily conducted through Review Manager 5.4.1. The heterogeneity of pooled estimates was quantified through the I² statistic; a random-effects model was employed if I² > 50% and P < 0.05; otherwise, a fixed-effects model was used. For continuous outcomes, the standard mean difference (SMD) was used to account for inherent differences in pain and upper limb function scores across studies. 95% confidence intervals (CI) were also calculated. Sensitivity analysis was performed if I² > 50% and P < 0.05. Additionally, publication bias was assessed through funnel plots and Egger’s test if more than ten studies were included. Since Egger’s test could not be performed in Review Manager, R was used for supplementary analysis as needed.

Results

Literature retrieval

A total of 52 articles were selected from PubMed, Embase, Cochrane Library, and CNKI. After duplication removal and initial screening, 18 articles remained. Upon full-text review, seven studies were excluded due to significant methodological flaws, inadequate data analysis, or irrelevance to the current research. Ultimately, 11 studies were included in the analysis (Fig. 1).

Fig. 1.

Flow chart of literature screening

Risk of bias and study characteristics

Overall, the quality of the included studies was moderate. Major quality concerns encompassed: (1) inadequate description of allocation concealment; (2) lack of registration information in some studies, making it difficult to assess study quality and applicability; (3) small sample sizes leading to insufficient statistical power to support the research hypotheses (Fig. 2).

Fig. 2.

Risk of bias

The included studies involved 584 patients with unilateral hemiplegia due to cerebral infarction, with an average age of 62, and a baseline numerical pain rating scale (NPRS) score ≥ 4. Of the included studies, five employed low-frequency rTMS, five used high-frequency rTMS, eight targeted the affected side, three targeted the unaffected side, six used low-intensity (≤ 80% RMT), and four used high-intensity (> 80% RMT) rTMS. Seven studies involved more than ten treatment sessions, while four studies had 10 or fewer sessions (Table 1). In particular, we analyzed the inclusion and exclusion criteria reported in eligible studies. As summarized in Tables 1 and 10 studies excluded focal causes of HSP, while the study by Xue 2019 did not mention it. The criteria for enrolling participants of each study are shown in Table S5.

Table 1.

Demographic characteristics of included trials

Study	Group	Sample Size	Age (years, mean, SD)	Sex (M, F)	HSP Regional Etiology Excluded	Baseline Pain Score	Maximum Field Strength (T)	Stimulation Intensity	Frequency (Hz)	Stimulation Site	Duration	Course
Choi 2017	rTMS	12	60.30, 7.10	7, 5	Yes	NRS: 6.3, 1.3	1	90%RMT	10	Ipsilesional M1	20 min	qd, 2w, 10 times
	Sham	12	57.80, 8.90	6, 6		NRS: 5.8, 1.5
Liu 2021	SGB	30	56.13,8.17	15,15	Yes	SF-MPQ:18.27,5.66	6	80%RMT	3	Ipsilesional M1	20 min	qd, 4w, 20 times
	rTMS&SGB	30	54.27,7.38	14,16		SF-MPQ:18.27,5.79
LiuQ2021	CWBT	50	62.70,10.70	27,23	Yes	VAS:7.3,1.2	NR	NR	> 5	Contralateral M1	20 min	qd
	rTMS	50	60.20,10.80	29,21		VAS:7.1,1.3
Tao 2023	Sham	20	57.85,3.17	8,12	Yes	NRS:7.25,1.55	2.5	80%RMT	10	Ipsilesional M1	20 min	qd, 4w, 20 times
	rTMS	20	58.85,2.90	11,9		NRS:6.85,1.50
Tian 2021	Sham	29	65.90,9.50	14,15	Yes	VAS:6.18,1.59	NR	80%RMT	5	Ipsilesional DLPFC	20 min	qd, 2w, 12 times
	rTMS	28	65.97,10.51	15,13		VAS:6.00,1.39
Wang 2017	AT	20	65.45,11.12	10,10	Yes	VAS:6.80,0.95	2	100%RMT	1	Contralateral M1	20 min	qd, 4w, 20 times
	rTMS&AT	20	62.13,12.54	11,9		VAS:6.85,0.99
Xiang 2015	rTMS&Rb	28	52.20,	15,13	Yes	VAS:7.64,0.42	2.2	80%RMT	1	Ipsilesional M1	35 min	qd, 10 times
	rTMS	26	53.80,	16,10		VAS:8.07,1.76
Xue 2019	rTMS&IEPT	35	NR	NR	No	NR	2	80%RMT	20	Ipsilesional M1	12.5 min	qd, 5 times
	IEPT	35	NR	NR		NR
Zhang 2019	AT	30	64.82,5.87	14,16	Yes	VAS:6.32,0.24	2	100%RMT	1	Contralateral M1	20 min	qd, 4w, 20 times
	rTMS&AT	30	63.18,4.17	16,14		VAS:6.26,0.41
Zhen 2022	Rb	30	64.03,3.12	20,10	Yes	VAS:5.49,1.10	2	80%RMT	1	Ipsilesional M1	35 min	qd, 4w, 20 times
	rTMS&Rb	30	64.39,3.08	18,12		VAS:5.32,1.03
Ay 2024	rTMS	7	63	4,3	Yes	NRS: 6–8	1	90%RMT	5	Ipsilesional, M1	20 min	qd, 3w, 15 times
	Sham	11	66	7,4		NRS: 6–10

Abbreviation: rTMS, repetitive transcranial magnetic stimulation; SGB, stellate ganglion block; CWBT, cold-water bathing therapy; AT, acupuncture therapy; IEPT, intramuscular efficacy patch therapy; Rb, rehabilitation; SD, Standard Deviation; M, male; F, female; NR, Not Reported; NRS, Numeric Rating Scale; RMT, Resting Motor Threshold; qd, quaque die; SF-MPQ, Short-Form McGill Pain Questionnaire; VAS, Visual Analogue Scale

Statistical analysis results

In the analysis of pain scores, 10 studies involving 514 patients were included. Results indicated a significant difference between the rTMS group and the control group (I² = 87%, P < 0.01; random-effects model; SMD = -1.14 [-1.69, -0.58], P < 0.01; Fig. 3). Subgroup analysis revealed that low-frequency rTMS (-1.67 [-2.69, -0.65]) was significantly more effective than high-frequency rTMS (-0.60 [-0.91, -0.28]) (P < 0.01). No significant differences were observed between unaffected and affected sides, different stimulation intensities, or different sessions of treatments. Sensitivity analysis confirmed the stability of the results. Egger’s test yielded a P-value of 0.12, suggesting no significant publication bias.

Fig. 3.

(A) Subgroup analysis of different treating frequencies in pain score; (B)Subgroup analysis of different treating courses in pain score; (C) Subgroup analysis of different treating sides in pain score; (D) Subgroup analysis of different treating intensity in pain score

In the analysis of upper limb function, nine studies involving 542 patients were included. Results demonstrated a significant difference between the rTMS group and the control group (I² = 93%, P < 0.01; random-effects model; SMD = 2.20 [1.37, 3.04], P < 0.01; Fig. 4). Subgroup analysis indicated that low-frequency rTMS (3.05 [1.97, 4.12]) was significantly more effective than high-frequency rTMS (SMD = 1.17 [0.40, 1.94]) (P < 0.01). The high-intensity group (> 80% RMT, SMD = 4.23 [2.83, 5.63]) also showed significantly better outcomes compared to the low-intensity group (≤ 80% RMT, SMD = 1.88 [1.14, 2.62]) (P < 0.01). No significant differences were found among different stimulation sites or treatment sessions. Sensitivity analysis was performed to validate the stability of the results. Due to fewer than 10 studies, Egger’s test was not performed.

Fig. 4.

(A) Subgroup analysis of different treating frequencies in Upper Limb Function; (B) Subgroup analysis of different treating courses in Upper Limb Function; (C) Subgroup analysis of different treating sides in Upper Limb Function; (D) Subgroup analysis of different treating intensity in Upper Limb Function

Discussion

This study demonstrates that rTMS significantly relieves pain and improves upper limb motor function in patients with post-stroke HSP. Additionally, treatment frequency and intensity may influence the efficacy of rTMS.

The primary etiological factors contributing to HSP include abnormal neurotransmitter release, neuroinflammatory responses, changes in neuroplasticity, and mood disorders. Research indicates that neurotransmitters such as glutamate and GABA are abnormally released in the central nervous system (CNS) of stroke patients, a phenomenon associated with the inhibition of pain perception [16, 17]. Previous studies have shown that rTMS can modulate the function of GABAergic neurons through cortical inhibition, thereby ameliorating pain symptoms [18–20]. Therefore, the relief of HSP pain by rTMS may be attributed to its ability to suppress the abnormal release of neurotransmitters. Local and systemic inflammatory responses, characterized by elevated levels of tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6), have been well-established in post-stroke patients. Microglia, astrocytes, and neurons, affected by these inflammatory responses, may heighten pain perception [16, 21]. Studies have proved that rTMS reduces neuroinflammation by inhibiting microglial polarization and decreasing inflammatory cytokines (e.g., TNF-α, IL-1β, and IL-6), thus alleviating pain symptoms in HSP patients [22]. Neuroplasticity is closely linked to early neural repair following stroke and subsequent symptoms and functional outcomes [23–25]. rTMS has been demonstrated to activate connections between different parts of the brain’s signal network, increase serum levels of brain-derived neurotrophic factor (BDNF), and modulate long-term potentiation (LTP) and long-term depression (LTD) mechanisms [26–30]. Therefore, the alleviation of HSP pain by rTMS may result from its effects on enhancing neuroplasticity. HSP is frequently accompanied by depression and anxiety, which exacerbate pain and hinder the rehabilitation process. Studies have indicated that rTMS not only alleviates central post-stroke pain (CPSP) but also improves accompanying depressive and anxious symptoms [31]. Thus, rTMS may enhance pain perception in patients by improving emotions.

The primary causes of upper limb dysfunction in HSP patients include shoulder joint subluxation, adhesive capsulitis (frozen shoulder), and neuropathic pain. Research has shown a correlation between shoulder joint subluxation and upper limb dysfunction in HSP [32]. Reduced movement in hemiplegic patients causes stiffness in the muscles, tendons, and ligaments surrounding the shoulder joint, which ultimately restricts the movement. Weakness of shoulder muscles, particularly levator scapulae, trapezius, and deltoid, compromises shoulder joint stability, thereby leading to subluxation, pain, and limited range of motion [33]. Previous studies suggest that rTMS may indirectly reduce the incidence and severity of shoulder joint subluxation by improving neural function, muscle strength, and coordination [34, 35]. Thus, the improvement in upper limb function in HSP patients due to rTMS may be ascribed to increased shoulder muscle strength and stability, which reduces shoulder joint subluxation. The incidence of adhesive capsulitis and its impact on upper limb function in stroke patients have been well-documented [36, 37]. Studies have found that high-frequencyr TMS can effectively relieve chronic pain [20]. For patients with adhesive capsulitis, rTMS may improve upper limb functional activities by alleviating pain and inflammation [38]. Research evaluating the role of neuropathic pain in post-stroke shoulder pain and its impact on shoulder joint function has highlighted the restrictive effects of chronic pain on joint mobility [39]. Neuropathic pain not only directly causes pain but also indirectly leads to restricted joint movement by affecting patients’ willingness to move and their movement patterns [40, 41]. rTMS has been reported to modulate cortical excitability and neural network activity to alleviate chronic pain, including neuropathic pain [42]. This mechanism is particularly relevant for HSP treatment, as HSP is commonly associated with chronic pain. By altering the processing of pain signals in the brain, rTMS helps to reduce pain and improve the comfort and function of upper limb joint movements [43, 44].

Among the studies included in our review, most utilized devices provided by RAPID-II. The primary motor cortex (M1) is the most commonly targeted region used for addressing both pain and motor impairments. Our findings suggest that lower frequencies and higher treatment intensities may be associated with better therapeutic outcomes. Different frequencies and intensities of rTMS are believed to have distinct effects. Low-intensity rTMS and low-frequency rTMS are commonly employed to inhibit hyperexcited neuronal activity. This inhibitory effect can help alleviate pain by reducing the transmission of pain signals [10, 45]. Concurrently, low-intensity rTMS can achieve therapeutic effects by modulating neural network activity without excessively stimulating neurons, potentially making it more suitable for sensitive or easily provoked patients [46]. Conversely, high-intensity rTMS and high-frequency rTMS are typically used to enhance neuronal activity. This excitatory effect can promote neuroplasticity and functional recovery, such as improving upper limb function by activating the motor cortex [47]. The stronger neural stimulation provided by high-intensity rTMS might lead to more pronounced therapeutic effects but also comes with a higher risk of side effects [48]. A possible explanation for our findings is that rTMS may have multifaceted effects, rather than a singular inhibitory or excitatory effect. Low-frequency and high-intensity stimulation may improve function through different mechanisms. Various frequencies or intensities of rTMS may exert beneficial regulatory effects on neural networks at different levels. Additionally, in the included studies, Chinese researchers often employed low-intensity rTMS, while researchers outside of China more frequently used high-intensity rTMS. However, neither group provided specific justifications for the choice of different intensities. Given the potential impact of intensity on outcomes, further attention to this aspect is warranted in future research.

This systematic review pooled the data and findings from multiple studies. By analyzing rTMS parameters and therapeutic effects across different studies, this study elucidates the advantages and limitations of various parameter combinations, potentially aiding clinicians in making more informed decisions in practice. Nonetheless, there are several limitations in this study. First, the included studies exhibit heterogeneity in research design, treatment parameters, and assessment methods. Therefore, it is necessary to carefully consider this heterogeneity when interpreting and applying these findings. Future research may need to explore more refined assessment and monitoring methods for individualized treatments. Second, although this review incorporates data from multiple studies, the overall sample size remains insufficient to fully represent the HSP population. The limited sample size may affect the statistical significance and generalizability of the results. Future research should increase the sample size to enhance the reliability and representativeness of the findings. Lastly, most studies primarily focus on short-term treatment effects, with insufficient follow-up data on the long-term effects and safety of rTMS. Long-term side effects and tolerance issues need to be further investigated to ensure the safe application of rTMS in HSP treatment. In particular, the long-term impact of high-intensity rTMS needs to be further validated. Finally, although our study has demonstrated the benefits of low-frequency and high-intensity rTMS, further research is needed to verify our findings due to a limited number of available studies.

Conclusion

This study indicates that rTMS is effective in relieving pain and improving upper limb motor function in patients with HSP. Additionally, lower frequencies and higher intensities of rTMS may provide greater therapeutic benefits. However, due to limitations such as the small sample size and significant differences in treatment protocols among the included studies, the findings should be interpreted with caution. Future clinical research on individualized treatment protocols with robust designs, larger sample sizes, and long-term follow-up is necessary.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Writing - original draft preparation: Zhenchao Ma; Writing - review and editing: Zhenchao Ma, Huijuan Pan, Ranran Bi, Zhenhua Li, Weichen Lu, Ping Wan; Conceptualization: Ping Wan; Methodology: Ping Wan; Formal analysis and investigation: Zhenchao Ma, Huijuan Pan, Ranran Bi, Zhenhua Li, Weichen Lu, Ping Wan; Resources: Ping Wan; Supervision: Ping Wan, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

The authors did not receive support from any organization for the submitted work.

Data availability

All data generated or analysed during this study are included in this published article and its supplementary information files.

Ethics approval

As this was a systematic review, ethical approval was not necessary.

Competing interests

The authors have no competing interests to declare that are relevant to the content of this article.

Consent to participate

Not applicable.