Effects of Warm-Needling Acupuncture Plus Tuina on Inflammatory and Cartilage Biomarkers in Knee Osteoarthritis: A Systematic Review and Meta-Analysis

Abstract

Background

Knee osteoarthritis (KOA) involves cartilage degeneration and synovial inflammation. Warm-needling acupuncture (WNA) plus Tuina is used clinically, but its systemic and local anti-inflammatory effects, as well as its potential benefits in modulating cartilage metabolism, lack robust evidence.

Objective

To systematically evaluate WNA plus Tuina’s effects on inflammatory cytokines and cartilage biomarkers in KOA.

Methods

We searched 8 databases (PubMed, Embase, Web of Science, Cochrane, CNKI, Wanfang, VIP, SinoMed) to March 2026 for RCTs comparing WNA plus Tuina against single or conventional therapies. Outcomes included IL-1/IL-1β, TNF-α, IL-6, CRP, IL-17, and MMP-3. We used the Cochrane RoB 2.0 tool and GRADE system.

Results

A total of 19 RCTs were included. Meta-analysis revealed that compared with the control group, WNA combined with Tuina was associated with reduced systemic inflammation levels, including serum CRP (standardized mean difference [SMD] −2.58, 95% confidence interval [CI] [−4.49, −0.67]), IL-1/IL-1β (SMD −1.73, 95% CI [−3.00, −0.47]), IL-6 (SMD −2.19, 95% CI [−3.38, −0.99]), TNF-α (SMD −1.63, 95% CI [−2.92, −0.34]), and IL-17 (SMD −2.06, 95% CI [−3.49, −0.62]). Within the local joint microenvironment, the combined therapy also showed a greater decrease in synovial fluid IL-1/IL-1β (SMD −0.93) and TNF-α (SMD −1.60). Furthermore, the serum level of MMP-3, a cartilage degradation marker, appeared to be downregulated (SMD −2.77, 95% CI [−3.22, −2.31]). Sensitivity analyses confirmed the robustness of these results. GRADE assessment indicated that the certainty of the evidence ranged from very low to low.

Conclusion

WNA combined with Tuina shows potential in reducing local and systemic inflammatory markers in KOA patients and may offer potential benefits in modulating cartilage metabolism. However, limited by the low certainty of current evidence, these findings should be interpreted with caution, and future validation through high-quality, large-scale RCTs is still required.

Introduction

Knee osteoarthritis (KOA) is the most common degenerative joint disease among middle-aged and elderly populations. According to the latest 2021 Global Burden of Disease report, KOA affects approximately 365 million people worldwide (accounting for nearly 4.7%), and its prevalence shows a significant upward trend with increasing age.^1,2 KOA is a whole joint disease, involving not only articular cartilage degeneration but also inflammation and fibrosis of the infrapatellar fat pad, meniscal degeneration, and subchondral bone remodeling.^3,4 Clinically, it manifests as knee joint pain, stiffness, and restricted mobility, ultimately leading to a significant decline in quality of life and functional disability.^5,6

The pathogenesis of KOA involves a complex interaction of multiple biological pathways, among which chronic low-grade inflammation and cartilage matrix degradation are the core pathological links.⁷ Studies have shown that pro-inflammatory cytokines such as interleukin-1/interleukin-1β (IL-1/IL-1β), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) are significantly elevated in the serum and/or synovial fluid of KOA patients.^8–10 By activating the nuclear factor-κB (NF-κB) and mitogen-activated protein kinase (MAPK) signaling pathways, these cytokines promote the expression of matrix metalloproteinases (MMPs), which in turn degrade type II collagen and proteoglycans, thereby accelerating cartilage degeneration.^11,12 Additionally, C-reactive protein (CRP), as a systemic inflammatory marker, is closely associated with the severity of pain and disease progression in KOA.¹³ Interleukin-17 (IL-17), secreted by Th17 cells, can synergize with TNF-α to amplify the inflammatory response, further exacerbating cartilage damage.¹⁴ Therefore, modulating the inflammatory cytokine network and inhibiting cartilage matrix degradation have become crucial targets in the treatment of KOA.

Current first-line treatments for KOA include non-steroidal anti-inflammatory drugs (NSAIDs), acetaminophen, and intra-articular injections of sodium hyaluronate.^15,16 Although these pharmacological agents can provide short-term symptom relief, their long-term use is accompanied by adverse reactions such as gastrointestinal bleeding, cardiovascular risks, and renal impairment, and they are incapable of reversing or halting the progression of cartilage degeneration.^17,18 In this context, non-pharmacological therapies have garnered increasing attention due to their high safety profile, minimal side effects, and excellent patient acceptability.

Warm-needling acupuncture (WNA) is a distinctive traditional Chinese medicine (TCM) therapy that combines acupuncture with moxibustion. The procedure involves placing moxa wool or a moxa stick on the needle handle and igniting it, allowing the thermal energy to conduct along the needle body deep into the acupoint, thereby delivering the dual effects of mechanical needle stimulation and moxibustion’s thermal warmth.¹⁹ Existing systematic reviews and meta-analyses suggest that WNA can ameliorate pain and functional impairment in KOA patients, demonstrating definitive clinical efficacy.^20,21 Tuina, an important branch of TCM external therapies, acts on the muscles and meridians surrounding the knee joint through manual techniques such as kneading, rolling, pressing, and passive flexion/extension. It can release adhesions, improve joint range of motion, and regulate local microcirculation.^22,23 Recent network meta-analyses indicate that Tuina ranks highly in improving joint function (using the Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC] functional score) and mobility in KOA patients.²⁴ According to TCM theory, “if there is no free flow, there is pain; if there is no nourishment, there is pain”. KOA falls under the category of “Bi syndrome”, primarily caused by the invasion of wind, cold, and dampness pathogens, leading to qi and blood stasis and meridian blockages.²⁵ WNA can warm the meridians, dispel cold, unblock collaterals to relieve pain, and tonify the liver and kidneys;²⁶ Tuina focuses on relaxing tendons, activating blood circulation, and releasing soft tissues.²² The combined application of both can exert a synergistic therapeutic effect.

Although previous systematic reviews have explored the efficacy of acupuncture or Tuina alone for KOA,^27,28 a significant knowledge gap remains regarding the anti-inflammatory and cartilage metabolism-modulating effects of the specific combined regimen of WNA and Tuina. Existing meta-analyses predominantly focus on subjective clinical indicators such as pain and function, leaving the objective molecular evidence regarding how this combined therapy modulates the underlying pathophysiological mechanisms largely fragmented. Specifically, there is insufficient attention paid to inflammatory biomarkers (IL-1/IL-1β, TNF-α, IL-6, CRP, IL-17) and cartilage metabolism markers (matrix metalloproteinase-3 [MMP-3]). This study specifically selected these biomarkers to capture complementary components of the inflammatory and tissue-degradation processes involved in KOA.⁷ IL-1/IL-1β and TNF-α represent key pro-inflammatory and catabolic mediators that can initiate and amplify cartilage-degrading responses,²⁹ whereas IL-6 serves as an important inflammatory mediator associated with disease activity and cartilage pathology.³⁰ CRP was included as a systemic marker of low-grade inflammation.¹³ IL-17 was selected to reflect the contribution of adaptive immune responses, particularly Th17-associated inflammation, to OA progression.³¹ MMP-3 was chosen as a key marker of extracellular matrix degradation and cartilage tissue damage.³² Furthermore, variations in control settings (WNA alone, Tuina alone, NSAIDs, conventional care, etc) and outcome measures across primary studies have resulted in fragmented evidence that urgently requires integration through a systematic review and meta-analysis. The novelty of this study lies in its exclusive focus on these objective biochemical outcomes to explore the potential therapeutic effects of WNA and Tuina. To our knowledge, this is the first meta-analysis to systematically synthesize evidence on both systemic (serum) and local (synovial fluid) anti-inflammatory effects, alongside the potential modulation of cartilage matrix metabolism, for this specific non-pharmacological combined therapy.

Therefore, this study aims to comprehensively search for published randomized controlled trials (RCTs) evaluating WNA combined with Tuina for KOA globally, and to systematically assess the impact of this combination therapy on serum/synovial inflammatory cytokines (IL-1/IL-1β, TNF-α, IL-6, CRP, IL-17) and cartilage protection-related biomarkers (MMP-3), in order to provide evidence-based medical support for clinical decision-making and future research.

Methods

This study was conducted and reported strictly in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement,³³ and the completed PRISMA checklist is provided as Supplementary File 1. The study protocol has been registered in the International Prospective Register of Systematic Reviews (PROSPERO) (Registration number: CRD420261350420).

Literature Search Strategy

Two researchers (Y.D. and K.Z.) independently conducted a systematic literature search across mainstream domestic and international electronic databases. English databases included PubMed, Embase, Cochrane Library, and Web of Science; Chinese databases included China National Knowledge Infrastructure (CNKI), Wanfang Data Knowledge Service Platform, VIP Information, and the Chinese Biomedical Literature Database (SinoMed). The search time frame for all databases spanned from their inception to March 2026. The search strategy utilized a combination of Medical Subject Headings (MeSH/Emtree) and free-text terms. Taking PubMed as an example, the search string mainly comprised four core modules: (1) Disease module: (“Osteoarthritis, Knee”[Mesh] OR “Knee Osteoarthritis”[tiab] OR “Gonarthrosis”[tiab] OR “Osteoarthritis of Knee”[tiab] OR “Knee Osteoarthritides”[tiab]); (2) Acupuncture intervention module: (“Acupuncture Therapy”[Mesh] OR “warm needle”[tiab] OR “warm needling”[tiab] OR “warming needle”[tiab] OR “warm acupuncture”[tiab] OR “wen zhen”[tiab]); (3) Manual therapy (Tuina) module: (“Massage”[Mesh] OR “Musculoskeletal Manipulations”[Mesh] OR “Acupressure”[Mesh]) OR (“tuina”[tiab] OR “tui na”[tiab] OR “massage”[tiab] OR “manual therapy”[tiab]); (4) Study design module: (randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR randomly[tiab] OR trial[tiab] OR groups[tiab]).

Furthermore, the reference lists of relevant published meta-analyses and systematic reviews were manually screened to supplement any potentially omitted literature. Detailed search strategies for each database are provided in the supplementary material (Supplementary Table 1).

Inclusion and Exclusion Criteria (PICOS)

The inclusion criteria were formulated based on the Population, Intervention, Comparison, Outcomes, Study Design (PICOS) framework:

• Population: Patients with a definitive clinical and/or radiological diagnosis of primary KOA, without restrictions on age, gender, or ethnicity.

• Intervention: The experimental group received WNA combined with Tuina. Studies where WNA and Tuina were applied as an add-on therapy to conventional basic treatments (eg, NSAIDs, glucosamine, or Chinese herbal medicine) were also eligible, provided that the exact same basic treatments were administered to the control group.

• Comparison: To comprehensively evaluate the clinical efficacy and synergistic effects of the combined therapy, the control group could receive: (1) conventional active treatments (eg, NSAIDs, glucosamine, or Chinese herbal medicine); (2) a single component of the experimental intervention (ie, WNA alone or Tuina alone); or (3) other conventional acupuncture-related therapies (eg, ordinary acupuncture or specific electromagnetic wave therapy [TDP]). These control settings were deliberately chosen to evaluate the pure efficacy, comparative effectiveness against first-line pharmacological treatments, and the specific synergistic add-on effects of the combined therapy.

• Outcomes: At least one of the following objective biomarkers from blood or synovial fluid was reported (pre- and post-treatment comparison or post-treatment endpoint data): pro-inflammatory cytokines (IL-1/IL-1β, TNF-α, IL-6, IL-17), acute-phase reactants (CRP/hs-CRP), or cartilage degradation markers (MMP-3).

• Study Design: RCTs.

The exclusion criteria were as follows: (1) Non-RCTs (eg, case series, retrospective studies, cross-over designs without first-phase data); (2) Animal experiments, reviews, conference abstracts, commentaries, and letters; (3) Studies where WNA and Tuina were combined with other distinct TCM interventions (eg, electroacupuncture, acupotomy) that were not equally applied to both the experimental and control groups, making it impossible to isolate the specific effect of the WNA plus Tuina component; however, studies were included if the same basic therapy (eg, conventional treatment, herbal medicine, or NSAIDs) was applied to both groups; (4) Failure to report any of the inflammatory or cartilage-related biomarkers of interest in this study; (5) Duplicate publications (only the one with the most complete data was included); (6) Inability to access the full text or unextractable data.

Data Extraction and Quality Control

Two independent reviewers (K.Z. and Y.L.) conducted the literature screening. EndNote X9 software was first used to remove duplicates, followed by a preliminary screening based on titles and abstracts to exclude overtly irrelevant articles. The remaining articles were then subjected to full-text review for final inclusion. Any discrepancies were resolved through discussion and consensus with a third researcher (Y.D).

Data extraction was performed using a pre-designed standardized Excel spreadsheet, independently extracted and cross-checked by two researchers (K.Z. and Y.L). Extracted information included: first author, year of publication, sample size (experimental/control group), patient age, gender ratio, disease duration, specific intervention protocols (parameters for both groups), treatment duration, and the targeted biomarkers. Regarding the timing of sample collection, we extracted the post-treatment endpoint data exactly as reported in the primary trials. It should be noted that while these data generally reflect the biomarker status following the designated treatment course, the original studies rarely specified the precise timing of collection. We excluded any data explicitly labeled as long-term follow-up. Furthermore, we verified that baseline biomarker levels were statistically comparable between the experimental and control groups in all included trials prior to data extraction. Disagreements during data extraction were resolved via discussion, with arbitration by a senior third researcher (G.X.) if necessary.

Risk of Bias Assessment

The methodological quality of the included studies was assessed using the revised Cochrane Risk of Bias 2.0 (RoB 2) tool.³⁴ The assessment dimensions covered five domains: (1) bias arising from the randomization process; (2) bias due to deviations from intended interventions; (3) bias due to missing outcome data; (4) bias in measurement of the outcome; and (5) bias in selection of the reported result. Each domain was rated as “low risk”, “high risk”, or “some concerns”. The overall bias judgment for each study was determined holistically: if all domains were at low risk, the overall risk was low; if at least one domain had some concerns but no high risk, the overall risk was some concerns; if at least one domain was at high risk, the overall risk was high. Two reviewers (K.Z. and Y.L.) independently completed the evaluation, with discrepancies resolved by discussion or consultation with a third reviewer (Y.D).

Statistical Analysis

All data analyses were conducted using the meta package in R software (version 4.3.1). Given that all the inflammatory and cartilage metabolism indices included were continuous variables, and different studies might have used different assay kits or units (eg, pg/mL, ng/L, mg/L) to detect the same biomarker (detailed in Supplementary Table 2), the standardised mean difference (SMD) and its 95% confidence interval (CI) were uniformly adopted as the pooled effect size. We explicitly utilized absolute post-treatment endpoint concentrations for meta-analysis rather than change-from-baseline scores, given that standard deviations for change scores were rarely reported in the primary literature. The synthesis of endpoint data is methodologically sound, as we rigorously confirmed that baseline biomarker levels were statistically comparable between the experimental and control groups across all included trials. This was adjusted for small-sample bias using Hedges’g.

Cochran’s Q test and the I² statistic were used to evaluate statistical heterogeneity across the included studies. An I² > 50% or P < 0.1 indicated significant statistical heterogeneity. Considering the clinical heterogeneity in intervention details (eg, Tuina techniques, needle retention time, acupoint selection) and patient baselines, a random-effects model was employed for all meta-analyses. Furthermore, subgroup analyses based on the type of control intervention and the exact treatment duration were pre-specified in the PROSPERO protocol to explore potential sources of heterogeneity.

To verify the robustness of the pooled results, a leave-one-out sensitivity analysis was performed by iteratively removing one study at a time and recalculating the pooled effect size to observe whether the results fundamentally changed. Since the number of included studies for each single outcome measure was less than 10, funnel plots and Egger’s tests were not conducted to quantitatively assess publication bias, as recommended by the Cochrane Handbook; however, potential small-study effects were considered during the GRADE quality assessment.

GRADE Evidence Quality Assessment

Two independent evaluators (K.Z. and Y.L.) used the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system to grade the certainty of evidence for each key outcome.³⁵ The quality of evidence was downgraded based on five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. The final quality of evidence was rated as “High”, “Moderate”, “Low”, or “Very Low”.

Results

Literature Search Results

Figure 1 illustrates the detailed literature screening process. The initial search yielded 748 records. After removing duplicates, 473 articles remained. Title and abstract screening excluded 427 clearly irrelevant articles (including reviews, animal experiments, case reports, non-RCTs, and unmatched interventions). A full-text review was conducted for the remaining 46 articles, resulting in the exclusion of 27 articles due to: non-randomized design (5 articles), non-compliant interventions (WNA and Tuina not combined or mixed with other therapies, 9 articles), absence of targeted biomarkers (11 articles), or incomplete/unextractable data (2 articles). Ultimately, 19 RCTs met all inclusion criteria and were incorporated into the systematic review and meta-analysis.^36–54

Figure 1.

PRISMA flow diagram of the study selection process. The flowchart details the identification, screening, eligibility, and inclusion of randomized controlled trials evaluating warm-needling acupuncture combined with Tuina for the treatment of knee osteoarthritis. Bold text indicates the main phases and headings of the literature search and selection process.

Baseline Characteristics of Included Studies

A total of 1733 KOA patients were included. The sample sizes for the experimental and control groups across studies ranged from 21/21 to 65/66. All studies were conducted in China, with publication years spanning from 2012 to 2026. The average age of the patients ranged from 39.29 to 64.31 years. While a few studies included younger patients, the vast majority focused on the middle-aged and elderly population (50–65 years). Both males and females were included across all studies, though females comprised a slightly higher proportion in some. The experimental interventions all involved WNA combined with Tuina; some studies superimposed this on conventional basic therapies (eg, herbal medicine, NSAIDs). The control interventions included WNA alone, Tuina alone, or conventional NSAIDs. The treatment duration varied from 2 to 8 weeks, with 4 weeks being the most common. Detailed baseline characteristics of the included studies are shown in Table 1.

Table 1.

Baseline Characteristics of the Included Randomized Controlled Trials

Study (Year)	Sample Size (E/C)	Age, Years (E/C)	Gender, M/F (E/C)	Disease Duration, Months (E/C)	Intervention (Experimental)	Intervention (Control)	Treatment Course	Outcomes
Li 202636	40/40	57.63±5.41/58.25±5.13	27/13/25/15	32.04±6.6/31.8±6.36	WNA + Tuina	WNA	4 weeks	①②③④
Ye 202537	40/40	56.19±4.56/57.23±4.09	23/17/21/19	8.35±1.62/8.23±1.49	WNA + Tuina + NSAIDs	Tuina + NSAIDs	4 weeks	⑥⑪
Wang 202538	41/41	56.43±3.75/57.12±3.95	18/23/17/24	26.04±13.89/25.33±14.26	WNA + Tuina	NSAIDs	8 weeks	⑦⑧
Li, Z 202539	50/50	64.31±5.04/64.25±4.87	19/31/17/33	55.44±12.96/54.84±12.24	WNA + Tuina + NSAIDs	NSAIDs	8 weeks	③④⑪
Li, T 202540	45/45	58.82±4.63/58.14±4.98	26/19/24/21	35.4±5.76/34.08±6.24	WNA + Tuina + CHM	CHM	4 weeks	①②⑤
Yin 202441	21/21	59.50±11.42/58.40±10.31	8/13/9/12	11.14±2.95/10.03±3.13	WNA + Tuina	Ordinary acupuncture	2 weeks	①②
Wang 202442	65/66	57.81±10.21/59.42±10.05	38/27/36/30	38.64±10.69/41.10±11.54	WNA + Tuina	NSAIDs	4 weeks	⑦⑧⑨⑪
Zhao 202343	50/50	56.61±2.54/57.21±2.63	20/30/22/28	64.92±12.36/69.84±13.68	WNA + Tuina	WNA	4 weeks	④⑪
Peng 202344	50/50	39.29±9.58/39.33±9.60	27/23/28/22	57.6±13.8/57.48±13.92	WNA + Tuina	Tuina	4 weeks	③④⑪
He 202345	30/30	58.7±6.8/59.3±6.5	16/14/15/15	NR/NR	WNA + Tuina + NSAIDs	NSAIDs	4 weeks	④⑪
Tang 202246	46/46	58.04±4.35/57.32±5.03	19/27/21/25	68.88±27.96/72.24±24.48	WNA + Tuina	WNA	4 weeks	①②⑥
Cao 202247	38/38	55.82±9.16/56.29±9.57	19/19/17/21	52.92±12.96/57.84±13.68	WNA + Tuina	Tuina	4 weeks	①②
Yang 202148	50/50	54.12±7.97/54.25±8.26	24/26/23/27	74.52±18.36/73.92±19.32	WNA + Tuina	WNA	2 weeks	③④
Li 202049	48/48	56.5±7.8/56.6±8.5	26/22/25/23	54.0±15.6/50.4±14.4	WNA + Tuina + NSAIDs	Tuina + NSAIDs	4 weeks	③④
Yue 201850	43/43	57.48±6.42/57.26±6.33	16/27/17/26	25.56±10.32/24.48±9.48	WNA + Tuina + Glucosamine	Glucosamine	2 weeks	③④
Xu 201851	60/50	55.6±2.9/54.9±3.1	25/35/20/30	57.6±19.2/52.8±18.0	WNA + Tuina	Tuina	4 weeks	⑤⑪
Wang 201852	50/50	56.72±8.13/56.67±8.20	21/29/23/27	78.12±45.0/78.84±44.4	WNA + Tuina	WNA + TDP	4 weeks	①②⑥⑪
Zhou 201753	54/54	52.5±5.2/50.4±4.8	30/24/26/28	NR/NR	WNA + Tuina	NSAIDs	4 weeks	⑦⑧⑪
Lin 201254	50/50	59.8±11.6/60.7±12.1	19/31/22/28	82.8±61.2/84.0±63.6	WNA + Tuina	Ordinary acupuncture + TDP	2 weeks	①②

Notes: Data are presented as Mean ± SD or n. Outcomes evaluated in the meta-analysis and narrative synthesis: ① Serum IL-1/IL-1β; ② Serum TNF-α; ③ Serum IL-6; ④ Serum CRP/hs-CRP; ⑤ Serum IL-17; ⑥ Serum MMP-3; ⑦ Synovial IL-1/IL-1β; ⑧ Synovial TNF-α; ⑨ Synovial IL-6; ⑩ Synovial hs-CRP; ⑪ Other exploratory biomarkers (eg, PCT, IL-10, OPG, BGP, CT, TGF-β, IGF-1, FGF-2, TIMP-1, COX-2, or synovial IL-17/MMP-3).

Abbreviations: CHM, Chinese herbal medicine; E, Experimental group; C, Control group; M/F, Male/Female; NR, Not reported; NSAIDs, Non-steroidal anti-inflammatory drugs; TDP, Teding Diancibo Pu (Specific electromagnetic wave therapy); WNA, Warm-needling acupuncture.

Risk of Bias Assessment of Included Studies

The results of the risk of bias assessment based on the Cochrane RoB 2.0 tool are detailed in Figure 2 and Supplementary Figure 1. For the randomization process domain, 14 studies (73.7%) were rated as low risk, while 5 studies (26.3%) had some concerns due to insufficient descriptions of allocation concealment. For deviations from intended interventions (Domain 2), all studies were rated as having some concerns. In accordance with RoB 2.0 criteria, this rating was assigned not merely due to the lack of participant and personnel blinding, but because this unblinded context introduces a potential for unbalanced co-interventions or protocol deviations, and it was unclear whether the primary studies employed appropriate intention-to-treat analyses to fully mitigate these risks. For missing outcome data and selection of the reported result, all studies were rated as low risk. For measurement of the outcome (Domain 4), all studies were rated as low risk. Although outcome assessors were unblinded, the target outcomes were objective biochemical markers measured using standardized, automated laboratory assays (eg, ELISA) uniformly applied across experimental and control groups, which are fundamentally resistant to subjective detection bias. Overall, all 19 studies (100%) were judged as having some concerns, with no study rated as low or high risk overall.

Figure 2.

Risk of bias graph. Review authors’ judgments about each risk of bias domain presented as percentages across all included studies according to the RoB 2.0 tool. Green represents a low risk of bias, yellow indicates an unclear risk (some concerns), and red indicates a high risk of bias. Bold text designates the specific bias domains evaluated and the risk assessment categories.

Systemic Anti-Inflammatory Effects of WNA Combined with Tuina

To comprehensively evaluate the regulatory effect of the treatment on the systemic immune microenvironment, we extracted data on key pro-inflammatory cytokines in patient serum and performed subgroup analyses based on the type of control intervention (Figure 3). As the core initiating factors of the osteoarthritis inflammatory cascade, serum IL-1/IL-1β and TNF-α were each reported in 7 studies (N = 580). Meta-analysis showed that the combined intervention was associated with lower expression levels of serum IL-1/IL-1β (SMD = −1.73, 95% CI [−3.00, −0.47]; I² = 95.0%, P < 0.0001; Figure 3A) and TNF-α (SMD = −1.63, 95% CI [−2.92, −0.34]; I² = 93.9%, P < 0.0001; Figure 3B) in KOA patients. Subgroup analyses further revealed that while WNA combined with Tuina showed significant downregulatory trends when compared against Tuina alone, conventional acupuncture, or herbal medicine, these specific subgroups frequently comprised only a single study, rendering these comparisons exploratory. Furthermore, the specific comparisons against WNA alone yielded wider confidence intervals that crossed the null effect line (P > 0.05). This suggests that while the combination appears broadly effective, its absolute superiority over WNA alone and other single interventions requires larger sample sizes for definitive confirmation.

Figure 3.

Forest plots of systemic inflammatory markers in patient serum. Meta-analysis of warm-needling acupuncture combined with Tuina versus control groups for (A) Serum Interleukin-1/Interleukin-1β (IL-1/IL-1β), (B) Serum Tumor necrosis factor-α (TNF-α), (C) Serum Interleukin-6 (IL-6), (D) Serum C-reactive protein (CRP), and (E) Serum Interleukin-17 (IL-17). Subgroup analyses were performed based on the type of control interventions. Bold text designates specific control interventions, column headers, statistical models, and distinct panels within the figure (eg, A, B). The solid diamond at the bottom of each section represents the pooled effect size.

Abbreviations: CI, confidence interval; SMD, standardized mean difference; NSAIDs, non-steroidal anti-inflammatory drugs.

Analyses targeting the major downstream effector IL-6 and the acute-phase non-specific inflammatory marker CRP similarly supported these anti-inflammatory benefits. The pooled effect size of 6 studies on serum IL-6 (N = 562) indicated a significant decrease in IL-6 concentrations in the intervention group (SMD = −2.19, 95% CI [−3.38, −0.99]; I² = 94.6%, P < 0.0001; Figure 3C). In the analysis of serum CRP comprising 8 studies (N = 722), the combined therapy also demonstrated a notable inhibitory potential (SMD = −2.58, 95% CI [−4.49, −0.67]; I² = 97.1%, P < 0.0001; Figure 3D). Notably, in subgroup analyses for these two markers, WNA combined with Tuina (or as an add-on therapy) showed a trend of maintaining a significantly favorable systemic anti-inflammatory profile even when compared directly with conventional Western medicines like oral NSAIDs (all P < 0.05). However, it is crucial to note that these specific NSAID-controlled subgroups were highly underpowered, containing only 1 or 2 trials. Furthermore, similar to IL-1/IL-1β and TNF-α, the subgroup comparing the combined therapy against WNA alone for CRP did not reach statistical significance.

Additionally, 2 included studies (N = 200) indicated that serum IL-17, a marker of the Th17 cell-mediated immune response, was also highly significantly suppressed (SMD = −2.06, 95% CI [−3.49, −0.62]; I² = 93.7%, P < 0.0001; Figure 3E). While this downregulatory trend was observed in individual trials when compared with Tuina alone or herbal medicine, each of these control types was represented by only a single study, rendering these specific subgroup findings exploratory.

In addition to intervention-based subgroups, we further explored the impact of treatment duration (eg, 2, 4, or 8 weeks) on inflammatory factors (Supplementary Table 3). The results indicated that the downregulation of pro-inflammatory cytokines was generally consistent across different treatment courses. However, because the 2-week and 8-week subgroups were significantly constrained by small sample sizes, definitive conclusions regarding the optimal treatment duration cannot yet be established. Overall, despite high statistical heterogeneity in the pooled analyses of various serum inflammatory factors, the direction of the effect sizes generally favored the combined therapy across most subgroups. This observation suggests a potential association between WNA combined with Tuina and a general reduction in circulating pro-inflammatory markers, although robust mechanistic conclusions regarding its precise impact on the systemic immune microenvironment cannot be drawn from the current evidence.

Local Synovial Anti-Inflammatory Effects of WNA Combined with Tuina

In addition to assessing systemic hematological indicators, this study further extracted inflammatory data from synovial fluid to directly reflect the local pathological microenvironment of the affected knee joint (Figure 4). Meta-analysis results revealed that in interventions targeting the synovial microenvironment, WNA combined with Tuina also displayed promising local targeted anti-inflammatory capabilities. Specifically, the inclusion of 3 studies (N = 321) indicated that the combined therapy significantly reduced the secretion level of IL-1/IL-1β in synovial fluid (SMD = −0.93, 95% CI [−1.50, −0.36]; I² = 84.1%, P = 0.0018; Figure 4A). Simultaneously, the pooled effect size based on 3 studies (N = 321) showed that local synovial TNF-α levels were significantly inhibited (SMD = −1.60, 95% CI [−2.27, −0.92]; I² = 84.1%, P = 0.0006; Figure 4B). It is worth emphasizing that the control interventions for these synovial biomarkers exclusively consisted of oral NSAIDs. Therefore, the data specifically demonstrates that WNA combined with Tuina provides a significant advantage over conventional NSAIDs in modulating the local inflammatory microenvironment within the knee joint.

Figure 4.

Forest plots of local synovial inflammatory markers. Meta-analysis of warm-needling acupuncture combined with Tuina versus control groups for (A) Synovial Interleukin-1/Interleukin-1β (IL-1/IL-1β) and (B) Synovial Tumor necrosis factor-α (TNF-α). Bold text designates specific control interventions, column headers, statistical models, and distinct panels within the figure (eg, A, B). The combined therapy demonstrated a significant reduction in local pro-inflammatory cytokines compared to conventional NSAIDs.

Abbreviations: CI, confidence interval; SMD, standardized mean difference.

Modulation of the Cartilage Degradation Biomarker by WNA Combined with Tuina

To evaluate the direct impact of the interventions on cartilage matrix metabolism, we analyzed the expression changes of MMP-3 (Figure 5). The pooled results of 3 studies (N = 272) regarding serum MMP-3 indicated that WNA combined with Tuina could substantially downregulate MMP-3 expression levels in serum (SMD = −2.77, 95% CI [−3.22, −2.31]). Notably, the meta-analysis of this core index exhibited low statistical heterogeneity (I² = 45.4%, P = 0.1602).

Figure 5.

Forest plot of the cartilage degradation marker. Meta-analysis of warm-needling acupuncture combined with Tuina versus control groups for serum Matrix metalloproteinase-3 (MMP-3). Bold text designates specific control interventions, column headers, and statistical models. The pooled result indicates a substantial downregulation of this cartilage degradation enzyme.

Abbreviations: CI, confidence interval; SMD, standardized mean difference.

Sensitivity Analysis and Publication Bias

To verify the robustness of the results, we performed a leave-one-out sensitivity analysis (detailed results in) Supplementary Figures 2–4. Whether for systemic inflammatory indicators, local synovial fluid markers, or cartilage biomarkers, the re-fitted pooled effect sizes (SMD) and their 95% CIs remained stable after the sequential exclusion of any single study. Because the number of included articles for each single outcome was less than 10, the statistical power for testing funnel plot asymmetry was too low, so funnel plots are not presented. However, considering the publication characteristics of TCM clinical trials, the potential for small-study positive bias cannot be ruled out.

Furthermore, to rigorously validate the comparative effectiveness of the combined therapy, we conducted a strict sensitivity analysis (Supplementary Table 4). This analysis exclusively retained trials where the control interventions were conventional pharmacological or distinct active therapies (eg, NSAIDs, Chinese herbal medicine, or ordinary acupuncture), by systematically excluding any trial where the control group received a partial component of the experimental intervention (ie, WNA alone, Tuina alone, or their combinations with baseline drugs). Even under these stringent criteria, the combined therapy consistently demonstrated a statistically significant reduction in core biomarkers (eg, TNF-α, IL-6, and CRP), confirming the robustness of the primary findings against conventional non-physical baselines.

GRADE Evidence Quality Rating

The evidence summary profile generated based on the GRADE system (Supplementary Table 5) indicates that the overall certainty of evidence for the outcomes ranges from “Very Low” to “Low”. Notably, because the objective laboratory biomarkers evaluated are highly resistant to subjective measurement bias (as verified by the uniform “low risk” in RoB 2.0 Domain 4), we did not downgrade the certainty of evidence for the lack of blinding. Instead, the certainty ratings were primarily constrained by suspected publication bias across all outcomes, combined with varying degrees of inconsistency and imprecision. Specifically, most inflammatory markers were downgraded due to significant statistical heterogeneity. Serum MMP-3 was a notable exception, exhibiting low heterogeneity; however, its final rating was still limited to “Low” due to imprecision (optimal information size < 400). Endpoints penalized for both high heterogeneity and imprecision dropped to “Very Low”. Ultimately, these ratings advise readers to exercise caution when interpreting the magnitude of the effect sizes, as future research may alter the current estimates.

Discussion

This study systematically evaluated the dual anti-inflammatory and the potential modulation of cartilage metabolism by WNA combined with Tuina in the treatment of KOA. Meta-analysis results demonstrate that compared to single physical interventions or conventional active treatments, this combined non-pharmacological therapy is associated not only with a reduction of key pro-inflammatory cytokines (IL-1/IL-1β, TNF-α, IL-6, and IL-17) and acute-phase proteins (CRP) in the systemic circulation, but also with decreased levels of local pro-inflammatory markers in the synovial fluid. Furthermore, the combined intervention showed a robust correlation with reduced serum concentrations of MMP-3, a biomarker associated with cartilage matrix degradation.⁵⁵ Although the overall certainty of the evidence from the included studies was rated low primarily due to high statistical heterogeneity, imprecision, and suspected publication bias, the highly consistent direction of effects across different subgroups and sensitivity analyses suggests that WNA combined with Tuina may help modulate local and systemic inflammatory markers in KOA and potentially influence cartilage metabolism.

The findings of this study provide crucial biomolecular support for previous systematic reviews that primarily focused on clinical symptom assessment (eg, pain and functional disability). In a recent network meta-analysis, Wang et al (2023) confirmed that Tuina possesses a significant advantage in improving joint function, yet its standalone anti-inflammatory efficacy is relatively moderate.²⁴ The data from our study further indicates that the addition of WNA greatly enhances the systemic anti-inflammatory potential of Tuina. For example, in the assessment of IL-6 and CRP, the combined therapy was associated with more pronounced reductions in these inflammatory markers compared to single physical interventions. Furthermore, compared to Liu et al (2025)’s report on the downregulation of inflammatory factors by superficial acupuncture (Fu’s subcutaneous needling),⁵⁶ the combined intervention presented in our study falls within a similar or even superior effect range in the magnitude of inhibition on multi-target pro-inflammatory networks (especially TNF-α and IL-1β). This not only corroborates the inference by Wang et al (2025) that multi-modal combined external therapies are clinically superior to Tuina alone,⁵⁷ but also provides preliminary molecular biomarker evidence supporting the combined application of the thermal impact of WNA and the mechanical stimulation of Tuina in anti-inflammatory mechanisms.

The anti-inflammatory mechanisms and the potential to mitigate cartilage degradation exhibited by WNA combined with Tuina may stem from a bidirectional interaction between biochemical signal blockade and local biomechanical remodeling. Cartilage degradation in KOA is highly dependent on the overexpression of MMPs (especially MMP-3) mediated by IL-1β and TNF-α, a pathological cascade typically driven by the abnormal activation of the NF-κB and MAPK signaling pathways.⁵⁸ Basic research suggests that the thermal stimulation of moxibustion can effectively inhibit the nuclear translocation of NF-κB p65, thereby cutting off the synthesis pathway of pro-inflammatory factors at the transcriptional level;⁵⁹ meanwhile, the mechanical stimulation of acupuncture can exert endogenous anti-inflammatory effects by activating the adenosine A1 receptor.⁶⁰ Concurrently, Tuina maneuvers (such as kneading, grasping, and stretching), by releasing tense peri-articular muscle groups and fasciae, can directly reduce the abnormal mechanical shear stress transmitted to the cartilage surface—this abnormal mechanical loading is precisely the key physical stimulus inducing chondrocytes to secrete MMPs.^61,62 Additionally, Tuina can promote the circulation and secretion of synovial fluid, improving the nutrient supply to chondrocytes.²² Existing clinical and basic research has shown that such physical-mechanical interventions not only downregulate MMPs but also significantly upregulate the expression of tissue inhibitors of metalloproteinases (such as tissue inhibitor of metalloproteinase-1),³⁷ thereby remodeling the dynamic metabolic balance of MMPs/tissue inhibitors of metalloproteinases. Therefore, the targeted biochemical immunosuppression by WNA and the improvement of the mechanical environment brought by Tuina complement each other, jointly contributing to the significant downregulation of MMP-3 observed in this study, providing a plausible theoretical basis for the clinical observations of this combined therapy. Recent magnetic resonance imaging (MRI) -based studies have highlighted the relationship between structural pathologies, such as meniscus damage and cartilage loss, and clinical symptoms in KOA.⁶³ Our findings on MMP-3 reduction align with the hypothesis that this therapy may influence cartilage metabolism; however, as our analysis was limited to serum biomarkers, further research integrating MRI-based morphological assessments is essential to establish a direct link between biomarker changes and structural joint preservation.

When interpreting the results of this study, its inherent limitations must be fully considered. Firstly, the included studies exhibited high statistical heterogeneity across multiple inflammatory outcomes. In-depth analysis suggests that this variation primarily stems from the broad scope of control group settings (encompassing NSAIDs, acupuncture alone, or Tuina alone), variations in treatment duration (ranging from 2 to 8 weeks), and differences in baseline disease severity (Kellgren-Lawrence grading) of the patients. In particular, different treatment durations may lead to significant disparities in the cumulative dose of the intervention, subsequently affecting the depth of modulation on the inflammatory network. Despite conducting our pre-specified subgroup analyses and rigorous sensitivity analyses, the high statistical heterogeneity could not be fully resolved. This indicates that inherent clinical variations intrinsic to physical therapies—such as distinct Tuina maneuvers applied by different practitioners, variations in WNA thermal intensity, and specific acupoint selections—play a substantial role in driving the observed heterogeneity. Moreover, many of our subgroups ultimately contained only a single study. This limited statistical power precludes robust meta-regression and renders the interpretations of these specific subgroups exploratory. Secondly, constrained by the physical operational characteristics of TCM external therapies such as acupuncture and Tuina, none of the included trials were able to implement double-blinding for operators and subjects. While this inherent lack of blinding resulted in an overall RoB 2.0 assessment of “some concerns” (primarily driven by potential deviations from intended interventions in Domain 2), its actual impact on the strictly objective laboratory biochemical indices focused on in this study is minimal. Because automated laboratory assays are fundamentally resistant to subjective detection biases, as evidenced by the uniform “low risk” ratings in the outcome measurement domain (Domain 4), we did not penalize the GRADE certainty ratings for risk of bias. Instead, our confidence in the pooled estimates is primarily limited by the unexplained statistical heterogeneity, suboptimal sample sizes (imprecision), and suspected publication bias. Additionally, our meta-analysis utilized absolute post-treatment endpoint concentrations rather than change-from-baseline scores. Although all included trials reported comparable baseline levels between groups, utilizing change scores might theoretically provide a more precise estimation of the intervention’s true effect size. Lastly, existing studies generally lack mid-to-long-term follow-up data post-treatment and have not synchronously employed MRI or other modalities for longitudinal morphological assessment of joint space narrowing or cartilage volume. This limits the ability to draw definitive conclusions regarding the long-term clinical efficacy and structural joint preservation of this therapy. Additionally, the regional limitation of the included literature implies that its universal applicability requires further validation through global, multi-center, large-sample studies.

In summary, this meta-analysis suggests that WNA combined with Tuina is associated with significant reductions in systemic and local inflammatory markers in KOA, and may downregulate MMP-3, a key cartilage degradation biomarker, at an evidence-based medicine level. In clinical practice, for elderly KOA patients, those with multiple comorbidities, or those intolerant to the gastrointestinal side effects of long-term NSAID use, this non-invasive/minimally invasive combined protocol serves as a highly promising alternative or adjunctive therapeutic strategy, possessing significant health-economic value, especially in establishing a step-wise conservative treatment pathway for KOA. There is an urgent need for future multi-center clinical trials characterized by rigorous sham-acupuncture/sham-Tuina controls, sufficient sample sizes, and long-term follow-up designs. It is also recommended to conduct multi-dimensional correlation analyses combining subjective scales, blood/synovial fluid biomarkers, and MRI structural parameters to further integrate imaging endpoints and establish the core evidence-based status and standardized clinical application guidelines for this intervention regimen in the step-wise treatment of KOA.

Conclusion

In conclusion, the existing evidence from this systematic review and meta-analysis indicates that WNA combined with Tuina may offer anti-inflammatory benefits and help modulate cartilage matrix metabolism in KOA. This therapy shows potential in reducing key pro-inflammatory cytokines in the synovial fluid and systemic circulation, and might lower levels of MMP-3, a biomarker associated with cartilage destruction. However, considering the low overall certainty of evidence, these conclusions should be interpreted with caution. Future high-quality, sham-controlled, multicenter RCTs with long-term follow-up are urgently needed to definitively validate the clinical value of this non-pharmacological combined intervention.

Funding Statement

This work was supported by the Scientific Research Project of Guangdong Provincial Bureau of Traditional Chinese Medicine (Grant no. 20272055), and the Scientific Research Start-up Program for Talent Introduction of the Third Affiliated Hospital of Southern Medical University (Grant no. RC2025005).

Abbreviations

CI, Confidence interval; CNKI, China National Knowledge Infrastructure; CONSORT, Consolidated Standards of Reporting Trials; CRP, C-reactive protein; GRADE, Grading of Recommendations Assessment, Development and Evaluation; IL-1β, Interleukin-1β; IL-6, Interleukin-6; IL-17, Interleukin-17; KOA, Knee osteoarthritis; MAPK, Mitogen-activated protein kinase; MeSH, Medical Subject Headings; MMP-3, Matrix metalloproteinase-3; MMPs, Matrix metalloproteinases; MRI, Magnetic resonance imaging; NF-κB, Nuclear factor-κB; NSAIDs, Non-steroidal anti-inflammatory drugs; PICOS, Population, Intervention, Comparison, Outcomes, Study Design; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PROSPERO, International Prospective Register of Systematic Reviews; RCTs, Randomized controlled trials; RoB, Risk of Bias tool; SMD, Standardised Mean difference; TCM, Traditional Chinese medicine; TNF-α, Tumor necrosis factor-α; WNA, Warm-needling acupuncture; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no competing interests.