Acupoint Selection Patterns and Potential Mechanisms of Acupuncture in Knee Osteoarthritis: A Combined Data Mining and Network Pharmacology Study Authors: Danning Zhang, Shuangxin Yang, Xingxing Lin, Kaixuan Zhang, Baoqiang Dong Published: 2026 ABSTRACT AbstractObjectiveTo identify the core acupoint prescription and Kellgren-Lawrence (K-L) grade-dependent compatibility patterns of acupuncture for KOA through complex network analysis, and to predict the potential molecular mechanisms underlying the core prescription via network pharmacology.MethodsLiterature was retrieved from PubMed, EMbase, Cochrane Library, Web of Science, CNKI, Wanfang, VIP, and SinoMed (inception to September 3, 2025). Frequency, association rule, complex network, and K-L grade subgroup analyses were applied. Potential targets of the core prescription were identified via network pharmacology and intersected with disease targets from OMIM, Therapeutic Target, GeneCards, and DrugBank. A protein-protein interaction (PPI) network was constructed, and Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to explore the potential molecular mechanisms.ResultsWe included 522 studies, yielding 582 prescriptions involving 123 acupoints. The core prescription comprised 24 acupoints, including Dubi (ST35), Neixiyan (EX-LE4), Liangqiu (ST34), Xuehai (SP10), Zusanli (ST36), Yanglingquan (GB34), Yinlingquan (SP9), among others. K-L subgroup analysis revealed ST35, GB34, SP9, and SP10 as universal core acupoints. The mild-to-moderate subgroup mainly used local acupoints, while the moderate-to-severe subgroup centered on ST35, with increased distal acupoint usage and higher degree values. Network pharmacology analysis identified 77 overlapping targets. Core targets included tumor necrosis factor (TNF), interleukin 6 (IL6), interleukin 1 beta (IL1B), tumor protein p53 (TP53), matrix metallopeptidase 9 (MMP9), signal transducer and activator of transcription 3 (STAT3), transforming growth factor beta 1 (TGFB1), caspase 3 (CASP3), and B-cell lymphoma 2 (BCL2), which were enriched in inflammation and immunity, cartilage metabolism, and tissue repair pathways.ConclusionThe core acupoint prescription for KOA features local acupoints combined with distal ones, exhibiting distinct patterns across K-L grades. Our computational findings suggest that core acupoints may potentially delay knee joint degeneration by synergistically regulating inflammation, cartilage metabolism, apoptosis, and tissue repair, although these predictions require experimental validation. These findings provide preliminary evidence and a theoretical basis for standardized clinical point selection and further mechanistic research.Keywords: knee osteoarthritis, KOA, acupuncture, core acupoints, complex network analysis, network pharmacology Introduction KOA is a complex, highly prevalent chronic degenerative whole-joint disease, involving not only cartilage degeneration and synovitis, but also subchondral bone remodeling, meniscal degeneration, and biomechanical and inflammatory alterations of the infrapatellar fat pad.1–5 Major risk factors include older age, obesity, and prior knee injury.6 Its clinical manifestations include knee pain, deformity, and limited mobility, and may also induce a range of accompanying symptoms such as depression and insomnia.7 KOA is a leading cause of disability worldwide, and its associated burden is particularly significant in developed countries.8,9 It is projected that the number of primary total knee arthroplasty procedures in the United States will rise to 3.5 million annually by 2050,10 and the global prevalence of KOA will reach 642 million.11 Clinical management primarily relies on medication, intra-articular injections, and surgery.12 Among these, long-term use of commonly prescribed non-steroidal anti-inflammatory drugs (NSAIDs) is associated with significant adverse effects, and cannot reverse cartilage degeneration. Repeated intra-articular injections of glucocorticoids may accelerate cartilage damage and increase local complications. Total joint arthroplasty, as an end-stage treatment option, has problems such as large trauma, high costs, and postoperative complications.13–15. Acupuncture, a characteristic non-pharmacological therapy of traditional Chinese medicine (TCM), has the advantages of low cost and few adverse effects. It can effectively alleviate pain and restore joint function, and has become one of the important research directions in the clinical treatment of KOA.16 Currently, most relevant data mining studies on acupuncture for KOA have shortcomings such as simple analytical methods, unclear differences in acupoint selection across disease stages, and failure to explore the mechanism of acupuncture effects.17,18 This study combines multi-dimensional network analysis and network pharmacology. It aims to accurately screen the core acupoint prescription and elucidate the molecular mechanisms of acupuncture treatment for KOA.These efforts are intended to provide objective, evidence-based support for standardized clinical acupoint selection. They also offer new insights and experimental directions for mechanisms exploration. Materials and Methods Literature Search Strategy A systematic search was conducted for published literature on acupuncture for KOA, covering from database inception to September 3, 2025.The Chinese literature sources included CNKI, Wanfang, VIP, and SinoMed. The search terms were “acupuncture,” “needling,” “electroacupuncture,” “body acupuncture,” “filiform needle,” “moxibustion,” “needle,” “catgut embedding,” “knee osteoarthritis,” “degenerative osteoarthritis of the knee,” “randomized controlled trial,” “randomized,” and “RCT.” The English literature sources included PubMed, EMbase, Cochrane Library, and Web of Science.The search terms were “Acupuncture”, “Pharmacopuncture”, “Acupuncture Therapy”, “Acupotomy”, “Electroacupuncture”, “Filiform Needle”, “Moxibustion”, “Needle”, “Acupuncture Point”, “Acupoint”, “Cupping Therapy”, “Knee Osteoarthritis”, “Osteoarthritis of Knee”, “Randomization”, “Random”, “Clinical Trial”, “Intervention Study”, and their synonyms. Search strategies were adjusted to fit the retrieval logic of each database. The full search strategies are provided as File S1. Inclusion and Exclusion Criteria Inclusion criteria: ① Study subjects: Patients with a confirmed diagnosis of KOA. ② Study type: Randomized controlled trials (RCTs). ③ Intervention: The observation group received single acupuncture therapy or a combination of multiple acupuncture modalities (eg, acupuncture combined with electroacupuncture or moxibustion); no restrictions were placed on the interventions in the control group. ④ Studies must adopt recognized diagnostic criteria, efficacy evaluation criteria (eg, WOMAC, VAS, Lequesne index, or overall response rate), with a definite treatment course that allows assessment of the effectiveness of acupuncture. Studies with significant improvement (P < 0.05) or overall response rate were included. ⑤ Studies must have an explicit acupoint prescription, and all needling points used must belong to the fourteen meridians. Exclusion criteria: ① Studies enrolling non-KOA patients, KOA patients complicated with other specific diseases (eg, diabetes, bone marrow edema, ischemic stroke with hemiplegia, etc), or patients after total knee arthroplasty. ② RCTs with improper randomization methods (eg, allocation based on patient preference, date of birth, or alternating assignment) or those that did not provide any description of the randomization procedure (eg, merely stating “randomized” without specifying the method) were excluded. ③ Studies of other types, including basic experimental research, reviews, theoretical discussions, meta-analyses, retrospective surveys, case reports, statements, guidelines, and other non-RCT studies. ④ Studies in which the intervention in the observation group was non-acupuncture therapy (eg, acupotomy, transcutaneous electrical acupoint stimulation, tuina) or acupuncture combined with non-acupuncture therapies. ⑤ Studies in which the observation group lacked a complete acupoint prescription or the treatment points did not belong to the fourteen meridians (eg, meridian sinew therapy, holographic therapy, etc). ⑥ Studies with unavailable full text or duplicate publications (only the most recent version was retained). ⑦ Studies with a treatment duration of less than 10 days. The methodological quality of the included RCTs was assessed using the Cochrane Risk of Bias 2.0 (RoB 2) tool, and the results were used for informational purposes only, not as an exclusion criterion. Data Extraction and Standardization Using NoteExpress software, the literature retrieved in Literature Search Strategy was screened and categorized according to the inclusion and exclusion criteria.A “KOA Acupuncture Treatment Database” was then constructed using Microsoft Excel 2021. Data extracted from each eligible study included title, authors, publication year, sample size, interventions, acupuncture prescription, K-L radiographic grade, outcome measures, and treatment duration. Standardization of acupuncture prescriptions: ① For studies containing multiple acupuncture treatment protocols, only the prescription with the greatest improvement in the study-defined primary outcome was retained. If no primary outcome was specified, the WOMAC total score or VAS pain score was prioritized. Conflicting outcomes were resolved through discussion between two reviewers, with arbitration by a third reviewer if needed. ② When different acupoint combinations were used based on traditional Chinese medicine (TCM) syndrome differentiation (eg, liver–kidney deficiency, cold-dampness obstruction, blood stasis) in the same study, they were recorded as multiple prescriptions by syndrome type. ③ Acupoint names were standardized according to Nomenclature and location of meridian points (GB/T 12346–2021)19 and Nomenclature and location of extra points in common use (GB/T 40997–2021),20 and acupoint codes were assigned following the WHO Standard International Acupuncture Nomenclature.21 All the above steps were performed jointly by two researchers. Any disagreement was resolved through consultation with a third researcher. Data Mining Analysis All data mining analyses were performed at the prescription level rather than the study level, as one study could yield multiple prescriptions. Frequency Analysis After standardizing the acupuncture prescription data included in the “KOA Acupuncture Treatment Database”, the data were converted into basket format. The usage of acupoints was statistically analyzed using R software (Version 4.5.1). High-frequency acupoints were identified after sorting by frequency in descending order, and a frequency distribution chart of the top 20 high-frequency acupoints was generated. Association Rule Analysis The Apriori algorithm was used to analyze association rules of acupoint compatibility and explore strongly correlated acupoint combinations, providing preliminary evidence for the compatibility logic of core acupoint prescription. The minimum support was set to 0.1, the minimum confidence to 0.8, and the lift was required to be >1 as the screening criteria.22 The top 20 core rules were extracted based on “count,” “support,” “confidence,” and “lift”.23 Parallel coordinate plots and network diagrams were constructed to visualize the association strength and distribution characteristics of acupoint combinations from multiple dimensions. K-Core Hierarchical Analysis and Community Structure Analysis of Complex Networks K-core hierarchy analysis was used to identify the core acupoint prescriptions, and community structure analysis was applied to divide functional compatibility modules.24 A weighted undirected acupoint-acupoint co-occurrence network for acupuncture treatment of KOA was constructed, and core topological parameters including degree, weighted degree, betweenness centrality, and closeness centrality were calculated25 K-core hierarchical analysis was used to compute the k-core value of each node, extract the core acupoint subnetwork, and identify core acupoints. Core acupoints were defined as the nodes remaining in the network at the highest k-core level. The Louvain algorithm was then employed to partition the core subnetwork into compatibility communities, clarifying the compatibility logic and hierarchical structure of the core acupoint prescriptions. Cluster Analysis and Phi Coefficient Calculation Hierarchical cluster analysis based on Jaccard similarity was performed on the core acupoints to validate their functional clustering characteristics. The Ward.D2 method was used. The elbow method, gap statistic method, and silhouette coefficient method were respectively applied to determine the optimal number of clusters.26,27 A hierarchical clustering dendrogram and a similarity heatmap of core acupoints were plotted, revealing the clustering and compatibility patterns of acupoints for KOA. The phi coefficient was calculated to quantify the strength of pairwise associations between core acupoints, and to verify the stability of compatibility relationships. Analysis of Core Acupoint Utilization Rate and Degree Centrality in K-L Grade Subgroups The included studies were divided into mild-to-moderate (K-L ≤ II) and moderate-to-severe (K-L > II) subgroups according to the K-L grade. The utilization rate and degree centrality of the 24 core acupoints were calculated separately for each subgroup. Independent-sample non-parametric test was used for comparison. Visual comparison charts were plotted to analyze acupoint selection patterns for KOA across different K-L grades. Network Pharmacology Analysis Target Mining of the Core Acupoint Prescription To further explore the mechanisms underlying the core acupoint prescription, The literature retrieved in Literature Search Strategy was re-screened to identify studies reporting molecular targets (proteins or genes). A “Core Acupoint Target Database” was constructed using Microsoft Excel 2021. To maximize coverage of potential targets, both human RCTs and animal experiments reporting molecular targets were included in the database. Inclusion criteria: ① Study subjects: patients meeting the clinical diagnostic criteria for KOA, or animal models meeting corresponding modeling standards; ② Study type: randomized controlled trials (RCTs) or animal experiments; ③ Acupoints studied: all acupuncture points in the observation group were from the core acupoint prescription identified in K-Core Hierarchical Analysis and Community Structure Analysis of Complex Networks; ④ Intervention: the observation group received acupuncture alone; animal experiments were required to include both a control group and a model group; ⑤ Outcome measures: the study reported at least one molecular target (protein or gene) with a statistically significant difference between the acupuncture group and the control group (P < 0.05). Exclusion criteria: ① involved subjects without KOA, subjects with KOA plus comorbidities, or patients after total knee arthroplasty; ② used improper randomization methods (eg, allocation based on patient preference, date of birth, or alternation) or did not specify the randomization method; ③ were neither RCTs nor animal experiments; ④ used non-acupuncture interventions or combined acupuncture with non-acupuncture therapies in the observation group; ⑤ did not clearly specify the acupoints used or used acupoints not belonging to the core prescription; ⑥ had unavailable full text or were duplicate publications (only the most recent version was retained); or ⑦ did not report any molecular target with statistically significant intergroup differences (P < 0.05). Screening and categorization were performed using NoteExpress software according to the above criteria. Data were extracted using Microsoft Excel 2021, including title, authors, publication year, sample size, interventions, acupuncture prescription, and molecular targets (proteins or genes). Acupoint name standardization followed the procedures described in Data Extraction and Standardization. All identified targets were standardized using the UniProt database (http://www.uniprot.org/) to ensure consistent gene nomenclature. All the above steps were performed jointly by two researchers. Any disagreement was resolved through consultation with a third researcher. Identification of Potential Targets of Core Acupoints for KOA Intervention Potential therapeutic targets related to KOA were retrieved from the OMIM (https://omim.org/), GeneCards (https://www.genecards.org/, version 5.22), Therapeutic Target (http://db.idrblab.net/ttd/, version 2025),28 and DrugBank (https://go.drugbank.com/, version 5.1.13) databases on September 3, 2025. The search term “knee osteoarthritis” was used for all databases. In OMIM, the “Gene Map” was searched and all entries were extracted, deduplicated using Microsoft Excel 2021, and sorted in descending order. In GeneCards, all retrieved targets were downloaded, and only those with a Relevance score above the median were retained. In DrugBank, the “Indications” section was searched, and targets were extracted from the “Drugs and Targets” section; only targets associated with “approved” or “investigational” drugs were retained. UniProt IDs were manually copied and compiled into a list. In TTD, all targets associated with osteoarthritis were directly retrieved and included. The results from the four databases were then consolidated, and duplicates were removed using Microsoft Excel 2021 based on official gene symbols and UniProt IDs. The resulting list was established as the “KOA Disease Target Database.” Overlapping targets between this database and the “Core Acupoint Target Database” were identified using the EVenn platform (https://www.bic.ac.cn/test/venn/#/), and a Venn diagram was plotted to obtain the potential targets of acupuncture intervention in KOA. A “Core Acupoint-KOA-Target” interaction network was then constructed using Cytoscape software (version 3.10.0). Construction of PPI Network The potential targets of acupuncture intervention in KOA were imported into the STRING database (https://string-db.org/). The species was set to Homo sapiens, and the interaction confidence score was set to medium confidence (score ≥ 0.400). This threshold is the default recommended by STRING for constructing general protein-protein interaction networks, providing a balanced trade-off between sensitivity and specificity. Free nodes were hidden before analysis. The resulting PPI network was imported into Cytoscape 3.10.0 for visualization. Topological analysis of the PPI network was performed using the Network Analyzer function, and the top 10 targets ranked by degree centrality were selected as core targets. Degree centrality reflects the number of direct interactions each protein has in the network and is a widely accepted measure of node importance in protein-protein interaction studies. GO Functional Enrichment and KEGG Pathway Enrichment Analysis GO and KEGG enrichment analyses were performed on potential targets of acupuncture intervention for KOA using the Metascape database (https://metascape.org/gp/) with its default parameters (minimum count for enrichment = 3; enrichment factor = observed/expected genes). The Benjamini-Hochberg (BH) method was applied to control the false discovery rate (FDR), and only terms with an adjusted P-value (q-value) < 0.05 were considered statistically significant. Four analyses were performed: GO Biological Processes (BP), GO Molecular Functions (MF), GO Cellular Components (CC), and KEGG pathway. The top 10 GO terms and top 20 KEGG pathways were selected based on p-value. Visualization was performed using the Bioinformatics platform (https://www.bioinformatics.com.cn/). A “core acupoint–target–pathway” network diagram for acupuncture treatment of KOA was constructed using Cytoscape 3.10.0 software. Statistical Analysis All data were analyzed using R software (Version 4.5.1) and its functions. A P value < 0.05 was considered statistically significant. Results Literature Search and Screening Results A preliminary search of eight databases yielded 20171 records. After removing duplicates, 11135 records remained. After screening the titles, abstracts, and full texts of the above records strictly according to the inclusion and exclusion criteria, 522 studies were finally included in the “KOA Acupuncture Treatment Database”, including 63 English studies and 459 Chinese studies. A total of 582 acupuncture prescriptions were extracted, which served as the basis for all subsequent data mining analyses. The detailed literature screening process is shown in Figure 1. The full study list and risk of bias results are provided in Supplementary Table S1 and Figure S1. Flowchart of the literature screening. Frequency Analysis Results A total of 123 acupoints were involved, with a total usage frequency of 3895. The top three most frequently used acupoints were ST35 (476 times, 12.22%), EX-LE4 (469 times, 12.04%), and GB34 (384 times, 9.86%). Each of the top nine acupoints was used more than 100 times, with a cumulative frequency of 3009, accounting for 77.25% of the total. The frequencies and cumulative percentages of high-frequency acupoints (frequency ≥ 20) are shown in Figure 2. High-frequency acupoints (frequency ≥ 20) in acupuncture prescriptions for KOA. Association Rule Analysis Results Association rule analysis was performed on all acupoints with the minimum support set to 0.1, the minimum confidence to 0.8, and lift >1. The results showed that the top three acupoint combinations by support were: EX-LE4-ST35 (support 72.68%, confidence 90.19%), GB34-EX-LE4 (support 55.84%, confidence 84.64%), and GB34-ST35 (support 54.81%, confidence 83.07%). Visualization revealed strong correlations among ST36, SP9, GB34, Ashi point (EX-LE0), ST35, SP10, ST34, Heding (EX-LE2), and EX-LE4, which represented the principal acupoints for acupuncture treatment of KOA. The top 20 association rules are presented in Table 1, and the visualizations are shown in Figure 3. Table 1Association Rule Analysis of Acupoints in Acupuncture for KOAN0.LHSRHSSupport (%)Confidence (%)LiftCount1EX-LE4ST3572.6890.191.104232GB34EX-LE455.8484.641.053253GB34ST3554.8183.071.023194SP10ST3552.5886.441.063065SP10EX-LE450.1782.491.022926EX-LE4, GB34ST3549.4888.621.082887GB34, ST35EX-LE449.4890.281.122888ST36ST3547.7782.491.012789ST34SP1047.4287.621.4427610ST36EX-LE447.0881.311.0127411EX-LE4, SP10ST3546.9193.491.1427312SP10, ST35EX-LE446.9189.221.1127313ST34ST3544.8582.861.0126114ST34EX-LE444.5082.221.0225915EX-LE4, ST36ST3540.8986.861.0623816ST35, ST36EX-LE440.8985.611.0623817GB34, SP10ST3540.7285.871.0523718EX-LE4, ST34ST3540.2190.351.1023419ST34, ST35EX-LE440.2189.661.1123420GB34, SP10EX-LE440.2184.781.05234 Association Rule Analysis of Acupoints in Acupuncture for KOA Association rule analysis of acupoints: (A) Parallel coordinate plot. (B) Bubble chart. Complex Network Analysis Results The acupoint co-occurrence network contained 125 nodes and 1468 edges, with a network diameter of 6, indicating good connectivity of the acupoint selection system. The average degree of the network was 23.488, reflecting extensive acupoint compatibility. The network density was 0.0947, and the average clustering coefficient was 0.6829, indicating significant “small-world” characteristics and a tendency to form tight compatibility modules. These topological parameters confirmed the rationality of the network structure. The k-core value of each node was calculated. When k ≥ 21, the complex network collapsed; therefore, the maximum k-core value of the network was 21, involving 24 core acupoints that constituted the core acupuncture prescription for KOA. These 24 core acupoints were divided into two communities. The blue community included ST35, EX-LE4, ST34, SP10, ST36, GB34, SP9, EX-LE2, and EX-LE0. The orange community included Sanyinjiao (SP6), Fengshi (GB31), Weizhong (BL40), Ququan (LR8), Xiyangguan (GB33), Waiqiu (GB36), Xuanzhong (GB39), Kunlun (BL60), Xiguan (LR7), Taichong (LR3), Taixi (KI3), Fenglong (ST40), Zulinqi (GB41), Jiexi (ST41), and Hegu (LI4). The core acupoint subnetwork and community structure are shown in Figure 4. The topological parameters of core acupoints were sorted in descending order of weighted degree. The top five key clinical acupoints were ST35, EX-LE4, GB34, SP10, and ST36. Specific parameters are presented in Table 2. Core acupoint subnetwork and community structure analysis diagram of acupuncture for KOA. Table 2Topological Parameters of Core Acupoints in Acupuncture for KOA (Sorted in Descending Order of Wdeg)N0.AcupointDegWdegBetClo1ST359828520.042438250.4962EX-LE48928050.039480690.497991973GB3410125390.067758140.532188844SP1010024110.064333970.534482765ST3610522590.072827540.516666676ST349921420.076166080.556053817SP98218180.029637640.478764488EX-LE29015950.051176500.516666679EX-LE05413490.012703660.4366197210GB33475660.006715400.3850931711LR8585250.024280450.4460431712SP6484990.008228220.4133333313KI3434400.023096250.4175084214BL40614330.031719550.4509090915GB39373720.010124680.4025974016LR3542430.040308270.4696969717GB31452140.005563420.3899371118LR7351930.016935260.4217687119ST40421700.018960230.4335664320BL60331610.002634350.3936507921LI4371570.010165300.3974359022GB4128870.003097900.3815384623ST4128710.002112500.3974359024GB3622660.000230720.35327635 Topological Parameters of Core Acupoints in Acupuncture for KOA (Sorted in Descending Order of Wdeg) Cluster Analysis and Phi Coefficient Calculation Results Cluster analysis was performed on the 24 core acupoints, and the optimal number of clusters was determined to be 2. The clustering results were consistent with those of the community analysis. The phi coefficient was used to assess the degree of association between acupoint variables.20 The analysis results showed that the combinations with strong correlations included: GB41-GB36 (phi coefficient 0.673), ST41-BL60 (phi coefficient 0.607), and SP10-ST34 (phi coefficient 0.596), as shown in Figures 5 and 6. Hierarchical clustering of core acupoints for KOA: (A) Dendrogram. (B) Heatmap. Phi coefficient of core acupoints for KOA. Results of Usage Rate and Degree Analysis in K-L Grading Subgroups A total of 78 acupuncture prescriptions with definite K-L grades were identified, including the mild-to-moderate subgroup (K-L ≤ II, 20 prescriptions) and the moderate-to-severe subgroup (K-L > II, 58 prescriptions). Results showed that the utilization rates in the mild-to-moderate subgroup were generally higher than those in the moderate-to-severe subgroup. Four core acupoints had a utilization rate ≥60% in the mild-to-moderate subgroup, compared with only one in the moderate-to-severe subgroup. Only the mild-to-moderate subgroup contained one core acupoint with a utilization rate >90%. The top five core acupoints by utilization rate in the mild-to-moderate subgroup were GB34 (95%), ST35 (85%), SP9 (70%), SP10 (60%), and ST36 (55%). In the moderate-to-severe subgroup, they were ST35 (89.7%), GB34 (56.9%), SP10 (51.7%), ST34 (50%), and SP9 (44.8%). The acupoints with high specific utilization rates in the mild-to-moderate subgroup were GB33 (35.0%) and BL40 (15.0%). Specific compatible acupoints in the moderate-to-severe subgroup were LR8 (32.8%), GB39 (19.0%), KI3 (19.0%), and BL60 (12.1%). Degree analysis showed that ST35, GB34, SP9, and SP10 ranked among the top five in both subgroups. In the mild-to-moderate subgroup, GB34 (degree = 80) served as the central acupoint, with ST36 being a subgroup‑specific acupoint. In the moderate-to-severe subgroup, ST35 (degree = 234) became the absolute core, and the degree values of all core acupoints increased markedly with disease progression (ST35 by 239.1%, GB34 by 140.0%, SP10 by 191.8%, and SP9 by 157.6%). Additionally, ST34 was added to strengthen local compatibility (Figure 7). Comparison of core acupoint parameters between K-L grade subgroups: (A) Usage rate of core acupoints in different K-L grades. (B) Top 5 degree values of core acupoints in different K-L grades. Establishment of the Core Acupoint Target Database A total of 221 studies were included in the “Core Acupoint Target Database”, comprising 71 human RCTs and 150 animal experiments. A total of 223 acupuncture prescriptions were extracted. The flowchart of the literature search is shown in Figure 8, and the list of included studies is provided in Table S2. These involved 23 core acupoints: ST35 (172 studies), EX-LE4 (166), GB34 (103), SP10 (79), ST36 (125), ST34 (68), SP9 (43), EX-LE2 (51), EX-LE0 (31), GB33 (7), LR8 (6), SP6 (5), KI3 (12), BL40 (4), GB39 (11), LR3 (3), GB31 (2), LR7 (1), ST40 (3), BL60 (3), GB41 (1), ST41 (2), and GB36 (1). No relevant effect studies were found for LI4; therefore, this acupoint was excluded. After data integration and target screening, a total of 151 potential targets corresponding to the core acupoints were identified (Table S3). Flowchart of literature screening for the core acupoint target database. Identification of Potential Targets of Core Acupoints in KOA Intervention A total of 1,002 potential therapeutic targets were screened and included in the “KOA Disease Target Database” (Table S4). The intersection with Establishment of the Core Acupoint Target Database yielded 77 overlapping targets, which constituted the set of potential molecular targets for acupuncture intervention in KOA (Table S5 and Figure 9). Venn diagram of overlapping targets between core acupoints and KOA. The “Core Acupoint-KOA-Target” interaction network contained two types of nodes: 20 core acupoint nodes and 48 overlapping target nodes, with 123 interaction edges. The target with the highest degree value was TNF (degree = 8), suggesting that it may be a key molecular target of acupuncture for KOA (Figure 10). “Core Acupoint-KOA-Target” interaction network. Construction of the PPI Network The 77 potential targets of acupuncture intervention in KOA were imported into the STRING database and visualized using Cytoscape 3.10.0 software, forming a PPI network containing 77 nodes and 1,482 edges (Figure 11). The top ten targets by degree value were TNF, IL6, IL1B, TP53, MMP9, STAT3, HIF1A, TGFB1, CASP3, and BCL2, which were regarded as the key molecular targets for the treatment of KOA with core acupoints. PPI network of core targets for KOA treatment by core acupoints. Notes: Node color, size, and edge thickness are all ranked in descending order of degree value. GO Functional Enrichment and KEGG Pathway Enrichment Analysis GO functional and KEGG pathway enrichment analyses of the 77 potential targets were performed using the Metascape database. The GO enrichment analysis identified 1484 BP terms, 47 CC terms, and 95 MF terms. The top 10 terms in each category, ranked by adjusted p-value, were selected and visualized as bar charts (Figure 12). BP and MF results were highly consistent and were significantly enriched in functions such as cytokine receptor binding, signal receptor-ligand regulation, and growth factor repair. CC terms were primarily localized to structures such as membrane receptor complexes, extracellular matrix, and membrane microdomains. Bar chart of GO functional enrichment analysis of core acupoints for KOA treatment. KEGG pathway enrichment analysis identified a total of 160 signaling pathways. The top 20 pathways, ranked by adjusted p-value, were selected and visualized as a bubble plot (Figure 13). The main enriched categories included: (1) Inflammation and immune-related pathways, such as the IL-17 signaling pathway and Th17 cell differentiation; (2) Cartilage metabolism and tissue repair pathways, such as the PI3K-Akt signaling pathway, MAPK signaling pathway, and AGE-RAGE signaling pathway. Bubble plot of KEGG pathway enrichment analysis of core acupoints for KOA treatment. A “core acupoint–target–pathway” network for acupuncture treatment of KOA was constructed using Cytoscape 3.10.0 software. This network contained 101 nodes and 714 edges, with core targets highly enriched in pathways related to inflammation and immunity, cartilage protection, oxidative stress, and metabolic regulation (Figure 14). “Core acupoint–target–pathway” network diagram. Note: Green squares represent core acupoints, yellow arrows represent pathways, and blue circles represent targets. Discussion KOA is increasingly viewed as a multifaceted joint disorder. Cartilage, bone, meniscus, synovium, and fat pad jointly shape the degenerative process through complex tissue interactions.1,3,29 Modern medicine regards KOA as a multifactorial disease closely associated with biomechanical loading, chronic low-grade inflammation, metabolic disorders, oxidative stress, and mitochondrial homeostasis imbalance.30–33 In traditional Chinese medicine, KOA falls under the category of “knee bi syndrome”, which is caused by invasion of external pathogenic factors obstructing the meridians, or deficiency of qi and blood.34 Acupuncture has been recommended as a Grade I strong recommendation for non-pharmacological treatment of KOA by the China Association of Chinese Medicine in its Guideline for Integrated Traditional Chinese and Western Medicine Diagnosis and Treatment of Knee Osteoarthritis (2023 edition), and is widely used in clinical intervention, with a continuously growing number of relevant studies.35 Modern studies have confirmed that acupuncture can exert analgesic, anti-inflammatory, and cartilage-protective effects by regulating pain transduction pathways, inhibiting inflammatory cascades, and modulating joint biomechanics.36 According to traditional Chinese medicine theory, acupuncture and moxibustion can unblock meridians and harmonize qi and blood, thereby achieving holistic regulation of KOA. However, existing research still faces key issues, including unclear core acupoint combinations and compatibility patterns, insufficient elucidation of acupoint selection across disease stages and the underlying molecular mechanisms. This study integrated complex network analysis with network pharmacology methodology to systematically identify the core acupoint prescription for KOA, analyze compatibility characteristics across disease stages, and explore potential molecular mechanisms, thereby providing a scientific basis for precise clinical acupoint selection and mechanistic research. Analysis of Core Acupoints Our network analysis identified the high-frequency acupoints for acupuncture treatment of KOA, including ST35, EX‑LE4, GB34, SP10, and ST36, which are highly consistent with the main acupoints for KOA listed in Acupuncture and Moxibustion Therapy,37 and are primarily located in the local knee area and adjacent meridians. It should be noted that high usage frequency does not necessarily indicate superior efficacy. Nevertheless, the consistent selection of these acupoints across a large number of independent studies suggests their clinical relevance and practical value in KOA management. ST35 and EX-LE4 are located in the medial and lateral depressions of the patellar ligament, with deep layers corresponding to the infrapatellar fat pad, joint capsule, and synovial tissue. Acupuncture at these points can simultaneously stimulate the anterior cutaneous branch of the femoral nerve and the branch of the saphenous nerve, activating mechanoreceptors to inhibit pain signal transmission, improve local microcirculation, and promote the absorption of inflammatory substances.38–40 Studies have shown that the ST35 acupoint is significantly sensitized in KOA model rats, with a decreased threshold of C-fibers in the acupoint area.41 GB34 has been described in TCM as a convergence point for tendons, meridians, qi, and blood, and possesses the effects of relaxing tendons, unblocking meridians, and facilitating joint movement. It is anatomically adjacent to the common peroneal nerve and the origin of the vastus lateralis muscle. Acupuncture at GB34 can regulate the lateral mechanical balance of the knee joint and relieve tendon spasm, thereby promoting the recovery of limb motor function.42 SP10 is closely related to the regulation of blood. Acupuncture at SP10 can accelerate the clearance of blood stasis-related metabolites, reduce plasma viscosity and coagulation status, and simultaneously regulate the neuro-immune network, decreasing substance P and its related receptors, thereby alleviating inflammatory pain and soft tissue edema in knee joint.43,44 ST36 is an important strengthening and regulating acupoint. Its acupuncture signals can be transmitted via the spinal cord to the central nervous system, modulating autonomic nerve function and exerting systemic regulatory effects.45 Our association rule analysis identified the “EX-LE4-ST35-GB34” combination as the basic core compatibility framework for acupuncture treatment of KOA, and this combination showed high stability across the included prescriptions. This is consistent with previous clinical observations that ST35 and EX-LE4 can directly target the local knee joint; when combined with SP10, ST34, and GB34, they further synergistically unblock meridians, promote the circulation of qi and blood in the local knee joint and the lower limb, effectively alleviating knee pain, swelling, and mobility limitation.46 Our community analysis revealed that the 24 core acupoints can be divided into two functional modules, a finding corroborated by both community structure detection and Jaccard similarity-based clustering. The blue community centers on local acupoint selection, embodying the classic principle of “the location of the acupoint determines the primary treatment indication.” Most included acupoints are located around the knee joint. ST35, ST34, and ST36 belong to the Foot Yangming Stomach Meridian; needling these points can replenish qi and blood and nourish the tendons and bones. SP10 and SP9 are located on the Foot Taiyin Spleen Meridian; needling these points can strengthen the spleen, remove dampness, reduce swelling, and relieve pain. The Foot Shaoyang Gallbladder Meridian is closely related to tendon and meridian function, and GB34 on this meridian is a key point for treating tendon contractures and joint dysfunction. The anatomical locations of acupoints in this community are mostly in areas dense with knee joint mechanoreceptors, and mainly exert local analgesic and joint biomechanical regulatory effects.47 fMRI studies have shown that acupuncture at local acupoints can activate the primary and secondary sensory cortices and enhance functional connectivity of the descending pain inhibitory pathway.48 Acupuncture at tendon spindle sites can excite tendon organs, transmit signals via type Ib fibers to the spinal cord, reduce the excitability of spinal anterior horn α-motor neurons, and improve abnormal muscle tension. This is consistent with the TCM pathogenesis of KOA, which posits that “obstruction leads to pain”.49 The orange community is characterized by distal acupoint selection and pattern-based compatibility, reflecting the TCM theory that “the pathway of the meridian determines the disorders it treats.” The acupoints in this community are distributed among the distal segments of the Foot-Three-Yin, Foot-Tai-Yang, Foot-Shao-Yang, and Hand-Yang-Ming meridians, with greater emphasis on systemic regulation. Consistent with our community findings, a previous study using infrared thermography and principal component analysis reported a “regional thermal sensitization” pattern at specific acupoint locations in lower limbs of KOA patients. Acupoints including ST34, SP10, EX-LE2, BL40, and GB39 showed significantly elevated skin temperatures and were clustered into the same factor group.50 This finding is highly consistent with the orange community, suggesting that distal acupoints exhibit functional activation under the pathological state of KOA, providing supportive evidence for their use as therapeutic targets. SP6 is the meeting point of the three yin meridians of the foot, coordinately regulating the liver, spleen, and kidney. It addresses the internal cause of “qi and blood deficiency” in KOA. Acupuncture at SP6 can transmit signals via the tibial and saphenous nerves, modulate autonomic nerve and immune functions, thereby improving the pathological state of KOA patients.44 K-L grading51 is an internationally recognized radiographic standard for evaluating KOA severity. Based on features such as the degree of joint space narrowing, osteophyte formation, subchondral sclerosis, and joint deformity on X-ray, KOA is classified into grades 0 to IV. The K-L grade subgroup analysis in this study further confirmed that the common core acupoints for KOA are ST35, GB34, SP9, and SP10, and that acupoint selection exhibits distinct adaptation patterns across K-L grades. Mild-to-moderate KOA (K-L ≤ II) is mainly characterized by synovial inflammation, superficial cartilage degeneration, and muscle tension.29,52 Acupoint selection in this stage centers on GB34, with ST36 as an important compatible acupoint. The utilization rate of these acupoints is high, and the compatibility is concentrated. By stimulating local acupoints to regulate muscle tone and local blood circulation, acupuncture promotes the release of endogenous opioids (eg, β-endorphin), aiming to rapidly relieve muscle spasm and pain. This aligns with the TCM treatment principle of “treating the symptoms in acute conditions”.53 Moderate-to-severe KOA (K-L > II) is accompanied by cartilage defects, osteophyte proliferation, chronic persistent inflammation, and functional impairment.51,54,55 Acupoint selection in this stage centers on ST35 as the absolute hub, with the addition of distal acupoints such as ST34, LR8, GB39, and KI3. The degree values of core acupoints increase significantly, reflecting a more complex acupoint compatibility network as the disease worsens. While strengthening the local intervention, the acupuncture strategy also expands the scope of systemic regulation, which highly aligns with the functional characteristics of the “distal acupoint selection and pattern-based compatibility” in the orange community mentioned above, embodying the holistic regulatory feature of acupuncture. From the perspective of TCM, LR8 and KI3 belong to the Liver and Kidney meridians respectively, and can tonify the liver and kidney, strengthen tendons and bones, corresponding to the pathogenesis of “malnourishment of tendons and bones” at this stage. From a modern medical perspective, this may be related to factors such as aggravated joint structural damage, lowered pain threshold,56 and neuroplastic changes57 in patients with moderate-to-severe KOA, necessitating a more complex acupoint combination to address these multiple pathological changes. Analysis of Core Targets and Pathway Mechanisms In this network pharmacology analysis, the core acupoint prescription was derived exclusively from human clinical data to ensure the integrity of clinical conclusions. Animal experiments were included solely for mechanistic support to maximize coverage of potential targets, as invasive molecular detection (eg, cartilage or synovial tissue sampling) is difficult to perform in human subjects, and animal models are widely recognized as standard carriers for mechanistic research in the field. Acupoint localization in animal studies follows standardized translational protocols, with well-established anatomical and functional correspondences to human acupoints. Nevertheless, species differences should be considered when extrapolating animal findings to human KOA. Our PPI network analysis identified TNF, IL6, IL1B, TP53, MMP9, STAT3, TGFB1, CASP3, and BCL2 as core targets. These targets are not acupuncture-specific, but rather well-known inflammatory and stress-related molecules. This is not unexpected, given that KOA involves multi-pathological processes. This suggests that acupuncture may not act through “acupuncture-specific” targets. Instead, it may work by stimulating core acupoints to synergistically regulate multiple targets, including inflammatory immunity, apoptosis, and matrix degradation, to intervene in KOA progression. Previous studies have shown that TNF, IL6, and IL1B are upstream core pro-inflammatory cytokines mediating chronic inflammation, and STAT3 serves as a key transcriptional hub in inflammatory signal transduction, playing a widespread role in various pathological stages of KOA.58 Mechanistic studies have demonstrated that TNF-α is produced by damaged articular chondrocytes, directly affects cellular gene expression, inhibits the synthesis of proteoglycans and matrix collagen, promotes their catabolism, and directly destroys cartilage structure.59,60 Meanwhile, it induces the activation of synovial fibroblasts, leading to the release of large amounts of pro-inflammatory and chemotactic factors, thereby exacerbating inflammation and joint effusion.61 TNF-α can also induce the production of IL-6, continuously amplifying the local inflammatory response.62 Studies have confirmed63 that serum concentrations of TNF-α and IL-6 are positively correlated with knee pain severity in KOA patients over a five-year period.64,65 In the inflammatory microenvironment, STAT3 is persistently activated by p-JAK2, accelerating the transcription of the above-mentioned inflammatory genes; it also promotes endothelial cell proliferation, induces osteogenic differentiation, and contributes to chondrocyte pathology.66,67 Previous studies have demonstrated that acupuncture at core acupoints such as ST35, GB34, ST36, and SP10 may reduce the levels of these pro-inflammatory cytokines in the serum and synovial fluid of KOA patients, inhibit STAT3 phosphorylation and activation, block the cascade amplification of inflammatory signals, and consequently alleviate synovial inflammation and relieve joint pain and swelling.68,69 Further evidence has shown that MMP9 is a key effector enzyme that mediates cartilage matrix degradation. It is released by chondrocytes, participates in the degradation of type II collagen and aggrecan, and directly mediates cartilage degeneration.70,71 TGFB1 mainly regulates the synthesis and repair of the cartilage matrix; it has also been reported to be activated by acupuncture to reduce IL-6 expression and delay cartilage degeneration.72,73 TP53 is a tumor suppressor gene that regulates chondrocyte cell cycle, senescence, and apoptosis. Its expression level in cartilage tissue has been shown to be positively correlated with K-L grade.74,75 CASP3 is often located at the terminal end of the mammalian cell apoptosis pathway, and its high expression has been reported to disrupt cartilage structure and biochemical homeostasis.76 Previous studies have reported that the decreased BCL2/Bax ratio in KOA cartilage activates CASP3 and accelerates apoptosis,77 and that acupuncture at ST35 may inhibit chondrocyte apoptosis by regulating the expression of CASP3 mRNA in cartilage tissue, while blocking the NF-κB pathway to reduce inflammation, thereby exerting a cartilage-protective effect.78 Our GO analysis suggested that the therapeutic effect of acupuncture on KOA may begin with the regulation of signal recognition and transduction both inside and outside the cell membrane. Through the cascade transmission of multiple key pathways including inflammation and immunity, cartilage metabolism and tissue repair, and oxidative stress, it ultimately influences gene expression and cellular biological behavior. Our KEGG enrichment analysis showed that the IL-17 signaling pathway was the most highly enriched among the identified pathways. Previous studies have reported that IL-17 is a pro-inflammatory cytokine that plays an important role in cytokine imbalance and connective tissue remodeling in joint inflammation.79 Under inflammatory conditions, synovial T lymphocytes continuously differentiate, promoting the generation of Th17 cells and the secretion of IL-17; simultaneously, they stimulate the production of cytokines such as IL-6 and MMPs, exacerbating synovial inflammation and cartilage matrix degradation.69,80 Abnormal expression of IL-17 has also been closely associated with the severity of pain in KOA, and its pathway inhibitors have been successfully applied in the clinical treatment of various inflammatory joint diseases.81,82 The PI3K-Akt and MAPK signaling pathways play critical roles in chondrocyte survival, apoptosis, and matrix metabolism. Previous studies have shown that upon activation of the PI3K-Akt pathway, Akt is rapidly phosphorylated, downregulating the BCL2/Bax ratio and inhibiting apoptosis.83 It also acts on its downstream target, the mechanistic target of rapamycin (mTOR), to regulate autophagy and maintain chondrocyte physiological function and matrix homeostasis. Activation of the MAPK signaling pathway promotes downstream NF-κB activation and the expression of inflammatory factors, promoting chondrocyte apoptosis and exacerbating cartilage degeneration.84–86 Stimulation of core acupoints including ST35, GB34, and ST36 in KOA patients has been suggested to exert regulatory effects on the PI3K-Akt and MAPK signaling pathways.87–89 The AGE-RAGE signaling pathway is a key driver of age-related KOA. With advancing age, advanced glycation end products (AGEs) accumulate continuously in articular cartilage. Upon binding to RAGE, they activate the NF-κB and MAPK signaling pathways, inducing the expression of pro-inflammatory cytokines and the generation of reactive oxygen species, exacerbating oxidative stress injury and apoptosis of chondrocytes, and promoting cartilage matrix degradation.90 Acupuncture may indirectly affect the activity of this pathway by resisting oxidative stress and regulating inflammatory responses, although this predicted effect requires experimental verification. Limitations and Future Perspectives This study has certain limitations: First, obtaining K-L grade information was relatively difficult, and the sample size for subgroup analysis was limited, which may affect the statistical power for some low-frequency acupoints. Second, some core acupoints (eg, LI4) lack mechanistic research and thus could not be included in the analysis. Third, the current mechanism analysis is mainly based on literature mining and bioinformatics prediction. While this approach provides a systematic overview of potential targets and pathways, the predicted mechanisms require experimental validation. The conclusions drawn from this computational analysis should therefore be interpreted as hypothesis-generating rather than definitive. Future research could conduct multicenter, prospective, large-sample clinical studies with uniform collection of K-L grade imaging data, and employ Bayesian hierarchical models to enhance the stability of results and reduce bias risk. Multi-omics technologies and gene-knockout animal models can be integrated to validate the therapeutic advantages of the core prescription and the regulatory mechanisms of its key targets. Concurrently, advanced techniques such as neural tracing and optogenetics can be used to further elucidate the differences in neural pathways and the specific regulatory effects of different acupoints. Conclusion The core prescription for acupuncture treatment of KOA is characterized by a predominance of local acupoints combined with distal acupoints, and the compatibility ratio of local to distal acupoints exhibits significant differences across K-L grades. Core acupoints may target and regulate inflammation, cartilage matrix metabolism, and apoptotic factors, affecting the activation of multiple core signaling pathways related to inflammatory immunity, cartilage metabolism and tissue repair, and oxidative stress, further delaying knee joint degeneration and improving pain and dysfunction. The findings of this study systematically elucidate the structural characteristics of the core acupoint prescription, the acupoint selection patterns across disease stages, and the potential molecular mechanisms underlying acupuncture intervention for KOA. This provides evidence-based support for the clinical standardization and precision of acupuncture treatment for KOA, and also establishes a replicable methodological paradigm for mechanistic research. Funding Statement General Program of the National Natural Science Foundation of China (81574050); Team Development Project for Major and Key Research Programs of the Liaoning Provincial Department of Education (2024-JYTCB-053); Basic Scientific Research Project of Universities under the Department of Education of Liaoning Province (2024-JYTCB-026); Liaoning Provincial Science and Technology Joint Program Project (2024-MSLH-283). Abbreviations KOA, knee osteoarthritis; K-L, Kellgren-Lawrence; PPI, protein-protein interaction; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; TNF, tumor necrosis factor; IL6, interleukin 6; IL1B, interleukin 1 beta; TP53, tumor protein p53; MMP9, matrix metallopeptidase 9; STAT3, signal transducer and activator of transcription 3; TGFB1, transforming growth factor beta 1; CASP3, caspase 3; BCL2, B-cell lymphoma 2; NSAIDs, non-steroidal anti-inflammatory drugs; TCM, traditional Chinese medicine; RCTs, Randomized controlled trials; BP, Biological Processes; MF, Molecular Functions; CC, Cellular Components; mTOR, mechanistic target of rapamycin; AGEs, advanced glycation end products. Data Sharing Statement The data supporting the findings of this study are available from the corresponding author upon reasonable request. Disclosure The authors declare no conflict of interest. References Associated Data Data Availability Statement The data supporting the findings of this study are available from the corresponding author upon reasonable request.