GABRB3 Gene Polymorphisms Influences the Analgesic Response to Acupuncture Treatment in Patients With Chronic Knee Pain: A Randomized Controlled Neuroimaging Transcriptomics Study.

GABRB3 Gene Polymorphisms Influences the Analgesic Response to Acupuncture Treatment in Patients With Chronic Knee Pain: A Randomized Controlled Neuroimaging Transcriptomics Study - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice FASEB J . 2026 Apr 18;40:e71779. doi: 10.1096/fj.202600031R Search in PMC Search in PubMed View in NLM Catalog Add to search GABRB3 Gene Polymorphisms Influences the Analgesic Response to Acupuncture Treatment in Patients With Chronic Knee Pain: A Randomized Controlled Neuroimaging Transcriptomics Study Sai Wu Sai Wu 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Sai Wu 1 , Wanxia Wu Wanxia Wu 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Wanxia Wu 1 , Jun Zhou Jun Zhou 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Jun Zhou 1, 2, 3 , Shirui Cheng Shirui Cheng 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Shirui Cheng 1, 2, 3 , Xiaohui Dong Xiaohui Dong 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Xiaohui Dong 1, 2, 3 , Yang Chen Yang Chen 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Yang Chen 1 , Chenjian Tang Chenjian Tang 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Chenjian Tang 1 , Wenhua He Wenhua He 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Wenhua He 1 , Fang Zeng Fang Zeng 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Fang Zeng 1, 2, 3 , Fanrong Liang Fanrong Liang 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Fanrong Liang 1, ✉ , Zhengjie Li Zhengjie Li 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China Find articles by Zhengjie Li 1, 2, 3, ✉ Author information Article notes Copyright and License information 1 Acupuncture and Tuina School, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China 2 Key Laboratory of Acupuncture for Senile Disease (Chengdu University of TCM), Ministry of Education, Chengdu, Sichuan, China 3 Acupuncture & Brain Research Center, Chengdu University of Traditional Chinese Medicine, Chengdu, Sichuan, China * Correspondence: Fanrong Liang ( [email protected] ), Zhengjie Li ( [email protected] ) ✉ Corresponding author. Revised 2026 Mar 17; Received 2026 Jan 5; Accepted 2026 Apr 1; Issue date 2026 Apr 30. © 2026 The Author(s). The FASEB Journal published by Wiley Periodicals LLC on behalf of Federation of American Societies for Experimental Biology. This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. PMC Copyright notice PMCID: PMC13091185  PMID: 41999225 ABSTRACT The mechanisms underlying individual variability in acupuncture analgesia among patients with chronic pain remain unclear. This randomized controlled trial investigated the core mechanisms of differential responses to acupuncture from genetic, neuroimaging, and transcriptomic perspectives in patients with chronic pain due to knee osteoarthritis (KOA). A total of 180 KOA chronic knee pain patients were randomly assigned to verum acupuncture (VA), sham acupuncture (SA), celecoxib (SC), placebo (PB), or waiting list (WL) groups (36 each). Over 2 weeks, VA/SA received 10 sessions, SC/PB oral medication for 14 days, and WL no intervention. Baseline 3.0T MRI 3D‐T1 scans and genotyping (GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, BDNF rs6265) were performed. Efficacy was assessed via VAS and WOMAC; responders/non‐responders were defined by minimally clinically important difference. Chi‐square test, logistic regression, voxel‐based morphometry (VBM), and Allen Human Brain Atlas‐based partial least squares regression were used. No significant difference in primary outcomes was observed between VA and SA, so they were combined as the acupuncture group (AG) to enhance statistical power. Only AG had a significant association between GABRB3 rs4906902 AG/GG genotype and acupuncture response ( p < 0.05); other loci showed no correlation. AG/GG carriers in AG had lower gray matter volume in caudate head, putamen, and ventral striatum, with higher GABRB3 expression in these regions. Genetic polymorphisms at GABRB3 rs4906902 could influence the analgesic effect of acupuncture treatment in patients with KOA chronic knee pain, with AG/GG genotype carriers exhibiting superior analgesic effects. This finding may be associated with pain‐modulating brain regions' gray matter volume reduction and upregulation of GABRB3 gene expression. Keywords: acupuncture, chronic pain, GABRB3 rs4906902 gene, gene polymorphism, knee osteoarthritis, neuroimaging transcriptomics Study on KOA chronic knee pain patients shows GABRB3 rs4906902 polymorphism influences acupuncture analgesic efficacy, related to gray matter volume and gene expression in pain‐modulating brain regions. 1. Introduction Knee Osteoarthritis (KOA) is a chronic degenerative osteoarthropathy primarily characterized by persistent knee pain and functional disability [ 1 , 2 ]. According to the 2019 Global Burden of Disease Study [ 3 ], approximately 364 million individuals are affected by KOA globally, making it the leading cause of disability in individuals over 60 years of age [ 4 , 5 ], imposing a substantial socioeconomic burden [ 6 ]. Multiple international clinical guidelines have designated pain control and functional improvement as the primary objectives for KOA treatment [ 7 , 8 , 9 ]. Acupuncture, which is recommended as a nonpharmacological therapy for KOA by the current clinical guidelines [ 7 , 10 , 11 ], has demonstrated significant efficacy and safety in pain management in patients with KOA [ 12 , 13 ]. However, large‐scale clinical studies have shown significant individual variability in the analgesic effects of acupuncture. Specifically, clinical studies have reported that 52%–58% of KOA patients experience a pain reduction of 50% or more [ 14 , 15 ]. These differences in the analgesic efficacy of acupuncture may be jointly driven by multiple factors, including genetic background, central nervous plasticity, and epigenetic regulation, among others [ 16 , 17 , 18 ]. However, the specific underlying mechanisms remain poorly understood. Further elucidation of the biological mechanisms underlying the individual differences in acupuncture analgesia is therefore crucial to facilitate individualized and precise treatment. Genetic polymorphisms are key determinants of individual variation in the efficacy of acupuncture. Multiple recent studies have shown that functional single nucleotide polymorphisms (SNPs) in genes highly expressed within the central nervous system could significantly influence the analgesic effects of non‐pharmacological therapies such as acupuncture. For example, current research has suggested that GABRB3 rs4906902 may predict clinical response to acupuncture in patients with knee osteoarthritis (KOA) [ 19 , 20 , 21 , 22 , 23 ]; OPRM1 rs1799971 (A118G) has further been linked to suboptimal pain relief and restricted improvement in cortical inhibitory function following acupuncture analgesia [ 24 , 25 , 26 ]; COMT rs4680 (Val158Met) modulates the efficacy of individualized interventions for KOA via dopamine‐related reward processing and anti‐inflammatory/analgesic mechanisms [ 18 , 27 ]; whereas the BDNF rs6265 (Val66Met) polymorphism mediates the elevation of peripheral blood BDNF levels following electroacupuncture intervention, dynamically influencing KOA treatment outcomes through its involvement in the “inflammation‐neurotrophic” bidirectional regulatory relationship [ 28 , 29 ]. Among these, GABRB3 rs4906902 (located at 15q11.2‐q12) is one of the most extensively studied candidate gene loci [ 30 , 31 , 32 ]. This locus has further been shown to significantly downregulate GABRB3 transcription and protein expression by interfering with the binding of transcription factors (such as SP1 and NF‐KB) [ 33 ]. GABRB3, a core subunit of the GABAA receptor [ 30 , 31 ], is enriched in the postsynaptic membranes of neurons within key pain modulation brain regions, including the anterior cingulate cortex, caudate nucleus, and amygdala [ 34 , 35 ]. Here, it inhibits neuronal excitability by mediating Cl‐ influx, thereby maintaining the balance of pain modulation [ 32 , 36 ]. It has further been demonstrated that the GABRB3 rs4906902 polymorphism, particularly the G allele, can downregulate GABRB3 gene expression in a dose‐dependent manner, with the following genotype effect intensity: AA > AG > GG (where A denotes the major allele and G the minor allele) [ 37 ]. This alteration in expression subsequently influences the inhibitory tone of the γ‐aminobutyric acid (GABA)‐ergic nervous system. This compromises the functional integrity of pain modulation circuits, ultimately fostering the development of central pain sensitization and hypersensitivity [ 38 , 39 ]. While these studies were predominantly based on animal and in vitro experiments, the mechanism by which this gene locus polymorphism influences acupuncture analgesia in patients with chronic pain is yet to be fully elucidated. Neuroimaging transcriptomics, an emerging interdisciplinary technology that has developed in recent years, offers a novel approach for non‐invasively deciphering the cross‐scale mechanisms of acupuncture, particularly by bridging “macro‐level neuroimaging and molecular mechanisms” [ 40 , 41 , 42 ]. By integrating macro‐level neuroimaging phenotypes with spatial transcriptomics data, this method can establish a spatial association between brain structural/functional abnormalities and gene expression patterns [ 40 , 43 , 44 ], thus revealing potential molecular mechanisms. This technology has previously been extensively applied in research on chronic pain and neuropsychiatric disorders [ 45 , 46 , 47 ]. Thanks to the ongoing expansion of neuroimaging transcriptomics methodologies, this approach has shown considerable promise in the investigation of the central mechanisms underlying various interventions, such as acupuncture [ 48 , 49 , 50 ], pharmaceuticals [ 51 , 52 ], and exercise [ 53 , 54 , 55 ]. It also could offer a platform for investigating how genetic polymorphisms influence the analgesic effects of acupuncture across multi‐scale neural mechanisms. Based on the above information, this randomized controlled neuroimaging transcriptomics study aims to first analyze the impact of gene polymorphisms of key candidate genes (including GABRB3 rs4906902) on analgesic responses to acupuncture treatment in patients with KOA chronic knee pain. Second, this study focused on characterizing the brain gray matter structure in patients with chronic knee pain with key genotypes that influence their response to acupuncture analgesia. Finally, by integrating data from the Allen Human Brain Atlas, we explore the spatial coupling between the polymorphisms of key influencing genes and gene expression patterns in distinct brain regions, with their potential molecular mechanisms subsequently being examined. This study aimed to provide comprehensive evidence from clinical neuroimaging and molecular genetics for the precise and individualized treatment of acupuncture analgesia in clinical settings. 2. Methods The clinical and MRI data used in this study were sourced from two randomized controlled trials [ 56 , 57 ] conducted concurrently with consistent procedures at the same research center. The study protocols were registered with the Chinese Clinical Trial Registry (ChiCTR‐IOR‐17012364 and ChiCTR‐17012365) and approved by the Sichuan Provincial Regional Ethics Review Committee of Traditional Chinese Medicine (approval number: 2016KL‐017). Participants were recruited between October 2017 and September 2021 at the following primary recruitment sites: the Hospital of Chengdu University of Traditional Chinese Medicine, the Third Affiliated Hospital of Chengdu University of Traditional Chinese Medicine, and the outpatient clinics of Sichuan Provincial Orthopedics Hospital. Supplementary recruitment methods included community posters, online WeChat platforms, and paper leaflets. The workflow of this study is illustrated in Figure 1 . FIGURE 1. Open in a new tab Study and data analysis flow chart. AG, acupuncture group (VA + SA); PB, placebo group; SA, sham acupuncture group; SC, celecoxib group; VA, verum acupuncture group; WL, waiting list group. 2.1. Participants Patients with chronic knee pain who met the diagnostic criteria for KOA were recruited. The diagnosis of KOA was established based on the American College of Rheumatology (ACR) 1991 revised criteria [ 58 ]. The inclusion criteria were as follows: (1) meeting the diagnostic criteria for KOA; (2) aged 40–60 years; (3) right‐handed; (4) persistent knee pain for the past 3 months, with an average Visual Analog Scale (VAS) pain score of ≥ 3 points recorded within 2 weeks prior to enrollment [ 59 ]; (5) a Kellgren‐Lawrence grade of 0‐II, confirmed by X‐ray examination; (6) no history of relevant therapy, including the use of KOA‐related therapeutic drugs within 2 weeks prior to enrollment, acupuncture received within the past 1 month, or intra‐articular sodium hyaluronate injection within the past 6 months; (6) signed informed consent obtained from the patient or their legal representatives. The exclusion criteria were as follows: (1) any history of severe somatic diseases (e.g., major primary neurological or cardiovascular diseases), rheumatoid arthritis, or other chronic pain conditions (including craniocerebral trauma accompanied by impaired consciousness); (2) history of substance abuse (e.g., alcohol or drug abuse) or psychiatric disorders; (3) contraindications to MRI scan or acupuncture, or celecoxib allergy; (4) pregnancy or lactation; (5) current participation in any other clinical trials. 2.2. Interventions This study used a centralized randomization design for a 2‐week intervention preceded by a 2‐week run‐in period for participant screening. Using SAS 9.2 software, 180 patients were randomly assigned to five groups in a 1:1:1:1:1 ratio ( n = 36 in each group); verum acupuncture (VA), sham acupuncture (SA), celecoxib (SC), placebo (PB), and waiting list (WL) groups. The allocation sequence was generated using an independent central randomization center. Allocation concealment was ensured using sequentially numbered, opaque, sealed envelopes, and allocation was executed sequentially by a telephone randomization administrator. The blinding strategy involved participant‐investigator double‐blinding for the SC/PB group and single‐blinding for the VA/SA group. The trial quality was further enhanced through triple assurance, encompassing standardized participant operational procedures, data collection by independent evaluators, and third‐party statistical analysis. All participants underwent the intervention under ethical supervision. All genetic information and MRI data were collected at baseline, while clinical evaluations were performed before and after the intervention. The intervention protocols were as follows: the VA group received acupuncture treatment at four standard acupoints on the knee joint, whereas the SA group received treatment at four non‐acupoints on the knee joint (Figure S1 ) [ 56 ]. In both the SA and VA groups, standardized procedures were performed by experienced physicians using disposable sterile needles (0.25 × 40 mm), with a depth of 12–38 mm, requiring needle manipulation for Deqi sensation induction [ 60 ]. During the observation period, the SC group received oral celecoxib capsules (200 mg/day, once daily), the PB group received a matched placebo (once daily), and the WL group received no therapeutic intervention. All patients were instructed to refrain from using other therapeutic interventions during the study period. Sustained‐release ibuprofen capsules (300 mg/dose, with an interval of at least 8 h, and a daily limit of 600 mg) are advised for emergencies [ 61 ]. All medications administered in this study were recorded. 2.3. Clinical Outcome Measures Clinical efficacy was primarily evaluated using the VAS [ 62 ] and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [ 63 ]. The visual analog scale (VAS) was used to assess the patients' average knee pain scores over the preceding 2 weeks, with scores ranging from 0 (no pain) to 10 (most severe pain imaginable). Conversely, the WOMAC evaluated the functional status of the patients' knee joints during the same two‐week period. Additionally, the Short‐Form McGill Pain Questionnaire (SF‐MPQ) [ 64 ], which includes the sensory and affective components of pain, and the 12‐item Short Form Survey (SF‐12) [ 65 ] were used to assess patients' quality of life. All scales were administered before and after 2‐week interventions. The criteria for determining responders and non‐responders to acupuncture treatment were as follows: Responders and non‐responders were identified based on previous literature and the minimal clinically important difference (MCID) of the VAS and WOMAC scores obtained before and after treatment [ 66 ]. The response rate was defined as the proportion of MCID responders, where the |improvement value| ≥ MCID. Responders had to simultaneously meet the following criteria: (1) an absolute improvement in VAS ≥ 2 points, or a relative improvement in VAS ≥ 40%; and (2) an absolute improvement in WOMAC function ≥ 6 points, or a relative improvement in WOMAC function ≥ 26%. Patients who did not meet both criteria simultaneously were classified as treatment non‐responders [ 12 , 67 , 68 ]. 2.4. Biospecimen Procurement and Genotyping Biospecimen procurement and genotyping were conducted by Shanghai Jingzhou Gene Technology Co. Ltd. Genotyping analysis was conducted for four SNP loci (GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, and BDNF rs6265) using human whole blood gDNA samples, at a concentration of 20–100 ng/μL. The TaqMan SNP Genotyping System, employing a 384‐well plate reaction system, was used. Amplification was performed using an ABI 9700 PCR instrument and the data were subsequently read on an ABI 7900 platform. The detection success rate for all samples was 100%, while the genotype distribution for each SNP locus was as follows: the GG genotype was predominant for COMT rs4680 (50.72%); for BDNF rs6265, the CT genotype was most frequently observed (50.72%); the AG genotype showed the highest proportion of OPRM1 rs1799971 (44.93%); and the AA genotype was predominant in GABRB3 rs4906902 (43.48%). Hardy–Weinberg equilibrium testing showed that all loci were in genetic equilibrium (all p > 0.05). Rigorous quality control measures were implemented, including negative (H 2 O) and positive controls. DNA quantification and quality control, conducted using Qubit and Nanodrop, yielded a high genotyping success rate of 99.4%. Among the investigated SNPs, the minor allele frequencies (MAF) for GABRB3 rs4906902 was 36.92% ( p = 0.99), for OPRM1 rs1799971 was 42.31% ( p = 0.79), for COMT rs4680 was 27.69% ( p = 0.99), and for BDNF rs6265 was 49.23% ( p = 0.83), respectively. No significant deviation from the Hardy–Weinberg equilibrium was observed for any of these loci. Blood and saliva samples were also collected. DNA was extracted from these samples using the Qiagen QIAamp and PrepIT‐L2P purification reagents. Genotyping was then performed by integrating the MassArray iPLEX platform with TaqMan technology, which collectively ensured the reliability of the experimental results. As a candidate gene association study without genome‐wide SNP genotyping, we implemented two measures to mitigate potential population stratification. First, all participants were non‐kinship relationship Han Chinese with long‐term residence (> 10 years) in Chengdu city, suggesting a relatively homogeneous ancestral background. Second, genotype distributions of all four candidate SNPs (GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, and BDNF rs6265) were consistent with Hardy–Weinberg equilibrium ( p > 0.05), supporting genotyping quality and genetic stability. Thus, the confounding effect of population stratification on our results was considered minimal. 2.5. Structural Neuroimaging MRI Scan Structural MRI was conducted at the Biomedical Engineering and Imaging Center of the University of Electronic Science and Technology of China. All brain imaging data were acquired using a 3.0T GE Discovery MR750 (GE Healthcare) system, equipped with an 8‐channel head coil. The 3D‐T1 weighted imaging scan parameters were as follows: repetition time (TR) = 6.008 ms, echo time (TE) = 1.7 ms, data matrix 256 × 256, field of view (FOV) 256 × 256 mm 2 , voxel size 1 × 1 × 1 mm 3 , and scan duration 4 min. During MRI scanning, foam padding was used to minimize head motion, while earplugs were provided to reduce ambient noise. All participants were instructed to remain awake, maintain their heads still, and close their eyes during the scan. 2.6. Data Analysis 2.6.1. Clinical Data Analysis All statistical analyses were conducted using SPSS software (version 25.0; SPSS Inc., Chicago, IL, USA). Data normality was evaluated using both histogram visualization and the Shapiro–Wilk test. Parametric tests were applied for normally distributed data. Intergroup comparisons of continuous variables were conducted using an independent sample t ‐test for two‐category grouping and one‐way analysis of variance (ANOVA) for multi‐category grouping. Post hoc pairwise comparisons were conducted using the LSD or SNK methods. For categorical variables, Pearson's chi‐square tests were used to analyze associations. Non‐normally distributed data were analyzed using non‐parametric tests. Intergroup comparisons of continuous variables were conducted using the Mann–Whitney U test for two‐category grouping or the Kruskal–Wallis H test for multi‐category grouping. The association between categorical variables was assessed using the chi‐squared test (with Fisher's exact test applied when theoretical frequencies were insufficient) and binary logistic regression. Quantitative data were presented as the mean ± standard deviation (x ± s) for normally distributed variables and as the median (interquartile range: P25, P75) for non‐normally distributed variables. Categorical data are presented as the frequencies (percentages, n %). All multiple comparisons were controlled for Type I errors using Bonferroni correction. All hypothesis tests were two‐sided, and the significance threshold was set at α = 0.05. 2.6.2. Gene Polymorphism Analysis To preliminarily evaluate the association of each gene locus (GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, and BDNF rs6265) with the analgesic response in each intervention group, we initially compared the proportion of treatment responders among patients across different genotypes using cross‐tabulation and chi‐square tests. When the expected frequency in any cell was less than five, the Fisher's exact test was used. Subsequently, for each gene locus, a separate logistic regression model was applied to analyze its association with efficacy, with the results expressed as odds ratios (OR) and 95% confidence intervals (CI). To evaluate the effect of minor alleles (G/T) on efficacy, we constructed three genetic models: (1) additive model: genotypes were treated as an ordinal variable (AA/CC vs. AG/CT vs. GG/TT), assuming that the effects of AG/CT and AA/CC increased by n‐fold and 2n‐fold, respectively, relative to GG/TT; (2) dominant model: genotypes carrying at least one major allele (A/C) (AA/CC + AG/CT) were combined and analyzed as a dichotomous variable against GG/TT, where the two genotypes carrying one or two A/C alleles had similar response rates; (3) recessive model: genotypes carrying two major alleles (AA/CC) were compared with the remaining genotypes (AG/CT + GG/TT), assuming that only homozygous (AA/CC) genotypes affected the response. To examine whether the association between the different interventions and treatment efficacy varied by genotype, models analogous to the additive, dominant, and recessive models were fitted. These models included interaction terms between the intervention groups and the four previously mentioned genetic loci. Furthermore, multivariate logistic regression models were employed to evaluate the independent association between each intervention group, genetic locus, and efficacy. Adjustments were made for differences in baseline covariates. The results are expressed as OR (95% CI). All statistical tests were conducted at a significance level of p < 0.05. 2.6.3. Structural Neuroimaging Data Analysis Structural MRI data were analyzed using voxel‐based morphometry (VBM) method. First, raw neuroimaging DICOM data were converted into the NIFTI format using MRICRON software ( https://www.nitrc.org/projects/mricron/ ). Second, structural image analysis was conducted using the MATLAB R2022b platform employing the CAT12 toolbox ( http://dbm.neuro.uni‐jena.de/cat/ ) within the SPM12 environment ( http://www.fil.ion.ucl.ac.uk/spm/software/spm12/ ) [ 69 ]. The processing pipeline strictly followed the standard protocol (outlined at: https://neuro‐jena.github.io/cat12‐help/ ), which involved dual quality control: (1) visual inspection for artifacts before preprocessing; (2) statistical quality control after segmentation (the CAT12 “Check Homogeneity” module assessed intra‐group homogeneity), with all images demonstrating a correlation > 0.75 and no exclusions made. Following bias field correction, T1‐weighted images were registered using linear (12‐parameter affine) and nonlinear transformations. These were then spatially normalized using the DARTEL algorithm [ 53 ], and segmented into the gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF). The total intracranial volume (TIV), calculated as GM + WM + CSF, was subsequently determined. The segmented tissue images were further normalized using the high‐dimensional DARTEL [ 70 , 71 ]. These images were then modulated to preserve the original tissue volume [ 72 ], and the resulting GM images were smoothed using a 6 mm FWHM Gaussian kernel. Building upon the findings of previous clinical scale and genetic association analyses, VBM was employed to compare the differences in gray matter volumes in associated genotypes between the responder and non‐responder groups. This analysis was implemented in SPM12. The statistical model included age, gender, and TIV as covariates, and the significance threshold was set at a voxel‐level p < 0.005 (uncorrected) and a cluster‐level p < 0.05 (FWE corrected). 2.6.4. Transcriptome‐Neuroimaging Analysis 2.6.4.1. Gene Expression Dataset and Preprocessing Gene expression data from the Allen Human Brain Atlas (AHBA), obtained from six healthy adult donors, were integrated into this study [ 73 ]. Standardized preprocessing of AHBA gene expression data was conducted using the Abagen toolbox ( https://www.github.com/netneurolab/abagen ) [ 74 ]. The specific steps were as follows: (1) probe re‐annotation and filtering; probes with signal intensities below 50% background noise were removed. For multiple probes targeting the same gene, the probe with the lowest coefficient of variation in expression across donors was selected as representative; (2) spatial normalization: tissue sample coordinates were mapped to the MNI152 standard space; (3) brain region mapping: using the Brainnetome Atlas (BN‐246) [ 75 ], samples were assigned to their nearest neighboring brain region; (4) expression value normalization: a Robust Sigmoid (SRS) function was applied to normalize the gene expression values for each donor; (5) brain region average expression calculation: the average expression value of all matching samples within each brain region was calculated. A final expression matrix (238 × 15 633) comprising 238 brain regions (eight regions were excluded due to a lack of sample matching) and 15 633 genes was obtained for subsequent association analysis. This processing pipeline adhered to the Abagen standard protocol, thereby ensuring data comparability and reproducibility (Figure 2 ). FIGURE 2. Open in a new tab Neuroimaging transcriptomics analysis flow chart. (a) Gene expression analysis. Gene expression data were obtained from the Allen Human Brain Atlas (AHBA) and underwent comprehensive preprocessing. Whole‐brain analysis was performed using the Brainnetome Atlas, ultimately generating a regional gene expression matrix. (b) Neuroimaging transcriptional analysis. Pearson correlation analysis was employed to establish associations between gray matter volume changes in the caudate nucleus (derived from VBM in knee osteoarthritis patients) and gene expression data. Subsequently, enrichment analysis was conducted on the list of significant genes to reveal relevant molecular functions (MF), biological processes (BP), and cellular components (CC). 2.6.4.2. Identifying Genes Expressions Associated With VBM Results We subsequently employed Partial least squares (PLS) regression analysis to investigate the spatial association between whole‐brain gene expression patterns and VBM differences based on GABRB3 rs4906902 genotyping (AG/GG vs. AA). This method, which is widely used in transcriptome analysis, identifies weighted expression patterns of a set of genes (independent variables) [ 76 , 77 ], the spatial distribution of which optimally predicts the spatial patterns of VBM differences between two groups ( t ‐statistic, dependent variable). The significance of the variance explained by PLS1 was assessed using 10 000 permutation tests of the dependent variable, and the variability in the contribution of each gene to PLS1 was estimated using the bootstrap method. The ratio of gene weights to bootstrap standard errors was calculated as a Z ‐score, based on which gene contributions were ranked. Genes with strong PLS1 weights ( Z > 3 or Z < −3, p < 0.05, FDR‐corrected) were identified as key genes explaining gray matter volume changes (see Figure 2 ). 2.6.4.3. Functional Enrichment Analysis Functional enrichment analysis was conducted using the Gene Ontology (GO) framework to explore the biological significance of relevant genes. This analysis was conducted using the ToppGene platform ( https://toppgene.cchmc.org/ ), which integrates various knowledge databases to reveal biological themes and identify overrepresented GO terms [ 78 ] from an input gene list. GO terms were categorized into three classes: biological processes, molecular functions, and cellular components. All human genes were used as the background set, with a significance threshold established at FDR‐corrected p < 0.05. Subsequently, GO enrichment analysis was conducted separately for the PLS (partial least squares) + gene set (genes with PLS1 weight Z ‐score > 3), PLS− gene set (genes with PLS1 weight Z ‐score < −3), and GABRB3 (see Figure 2 ). To gain a deeper insight into the interactions among the different genes associated with neuroimaging changes, a protein–protein interaction (PPI) network [ 79 ] was constructed using the STRING database ( https://string‐db.org/ ). This network was subsequently employed to identify key hub genes within the PLS+ and PLS− gene sets, thereby elucidating the interactions between the proteins encoded by these genes. For network construction, the medium confidence threshold was set to 0.90, with all other parameters maintained at the default settings of the platform, and the free nodes were subsequently removed. In the network topological analysis, genes with a degree ranking in the top 10% were defined as hub genes. Finally, the spatiotemporal expression trajectories of the representative hub genes were visualized using Cytoscape software (Figure 2 ). 3. Results This study enrolled 180 patients with chronic knee pain due to KOA, of whom 36 each were randomly allocated to the VA, SA, SC, PB, and WL groups. In total, 138 patients completed the study and were included in the final analysis. Based on the exclusion criteria, 8, 10, 8, 7, and 9 patients were excluded from the VA, SA, SC, PB, and WL groups, respectively. The major reasons for exclusion included excessive head motion (12 cases), lack of genetic testing (11 cases), and withdrawal of consent (10 cases) (see Figure 1 ). All adverse events (e.g., acupuncture‐related pain, bleeding, fainting, and celecoxib‐related gastrointestinal reactions) were managed promptly and documented in case report forms. No concomitant medications were used during the study period. 3.1. Baseline Characteristics No significant differences were found between the VA, SA, SC, PB, and WL groups in terms of sex; age; BMI; disease duration; or the VAS, WOMAC, SF‐MPQ, or SF‐12P scores ( p > 0.05). Moreover, the distributions of the genetic polymorphisms GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, and BDNF rs6265 were not significantly different between the verum and sham acupuncture groups or between the acupuncture and non‐acupuncture intervention groups (Table S1 and Figure S2 ). 3.2. Clinical Outcomes Compared with their respective baseline values, patients in the VA, SA, SC, and PB groups showed improvements in the VAS, WOMAC, SF‐MPQ, and SF‐12 scores (Table S2 and Figure S3 ). However, no statistical significance was observed in the VA and SA groups for any outcome measures (VAS, SF‐MPQ, WOMAC, and SF‐12), and there was no significant difference in the treatment response rates between the two acupuncture groups (Table S3 and Figure S4 ). These findings are consistent with previous RCTs and meta‐analyses suggesting that the specific effect of puncturing verum acupoints over puncturing non‐acupoints is minimal to moderate in patients with KOA knee pain patients [ 80 , 81 , 82 ]. Given the lack of significant differences, the VA and SA groups were pooled into a single acupuncture group (AG) to enhance statistical power. Using Bonferroni‐corrected Mann–Whitney U tests, comparisons were subsequently made between the AG and SC, AG and PB, and AG and WL groups. Consistent with earlier studies indicating comparable analgesic effects between acupuncture and NSAIDs [ 83 ], no significant differences were detected between AG and SC. However, significant differences were found between AG and PB in terms of changes in VAS and SF‐MPQ scores, but not in the WOMAC or SF‐12 scores. Between the AG and WL groups, a significant difference was observed only in the SF‐MPQ changes, with no differences in the VAS, WOMAC, or SF‐12 scores (Table S4 and Figure S4 ). 3.3. Gene Polymorphism's Effect on Therapeutic Efficacy of Different Interventions Initial chi‐square tests revealed that only GABRB3 rs4906902 was significantly associated with treatment response in the AG group (recessive model, p = 0.022; additive model, p = 0.031). In contrast, no significant associations were observed for any of the genetic models (additive, dominant, or recessive) of GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, or BDNF rs6265 in the SC, PB, and WL groups (Table 1 ). TABLE 1. The gene type distribution of intervention responders in different intervention groups. Characteristics AG SC PB WL n Responder p n Responder p n Responder p n Responder p GABRB3 rs4906902 (AA/AG vs. GG) 46/8 27/4 0.943 25/3 19/2 1.000 27/2 8/1 0.532 21/6 9/4 0.385 GABRB3 rs4906902 (AG/GG vs. AA) 33/21 23/8 0.022 15/13 9/12 0.084 11/18 3/6 1.000 19/8 10/3 0.678 GABRB3 rs4906902 (AA vs. AG vs. GG) 21/25/8 8/19/4 0.031 13/12/3 12/7/2 0.116 18/9/2 6/2/1 0.701 8/13/6 3/6/4 0.695 OPRM1 rs1799971 (AA/AG vs. GG) 45/9 12/13/6 0.820 26/2 10/9/2 0.810 26/3 2/6/1 0.728 25/2 6/6/1 1.000 OPRM1 rs1799971 (AG/GG vs. AA) 32/22 19/12 0.724 13/15 11/10 0.396 19/10 7/2 0.431 14/13 7/6 1.000 OPRM1 rs1799971 (AA vs. AG vs. GG) 22/23/9 25/6 0.806 15/11/2 19/2 1.000 10/16/3 8/1 1.000 13/12/2 12/1 1.000 COMT rs4680 (AA/AG vs. GG) 29/25 19/12 0.194 14/14 10/11 1.000 15/14 4/5 0.700 10/17 4/9 0.695 COMT rs4680 (AG/GG vs. AA) 49/5 28/3 1.000 26/2 19/2 1.000 27/2 8/1 0.532 24/3 12/1 1.000 COMT rs4680 (AA vs. AG vs. GG) 5/24/25 3/16/12 0.415 2/12/14 2/8/11 0.814 2/13/14 1/3/5 0.586 3/7/17 1/3/9 1.000 BDNF rs6265 (CC/CT vs. TT) 37/17 20/11 0.462 23/5 17/4 1.000 22/7 8/1 0.382 21/6 10/3 1.000 BDNF rs6265 (CT/TT vs. CC) 44/10 25/6 1.000 16/12 11/10 0.662 23/6 6/3 0.339 18/9 9/4 1.000 BDNF rs6265 (CC vs. CT vs. TT) 10/27/17 6/14/11 0.691 12/11/5 10/7/4 0.630 6/16/7 3/5/1 0.415 9/12/6 4/6/3 1.000 Open in a new tab Note: Additive model, n (CC vs. CT vs. TT)/ n (AA vs. AG vs. GG); Dominant model, n (CC/CT vs. TT)/ n (AA/AG vs. GG); Recessive model, n (CT/TT vs. CC)/ n (AG/GG vs. AA). Categorical data were presented as frequency ( n ). Chi‐square test and Fishers exact test was used to compare the distribution of responder proportions among the AG, SC, PB, and WL groups. A p value of < 0.05 was considered statistically significant. Red color values indicate P 〈 0.05. Abbreviations: AG, acupuncture group (VA + SA); PB, placebo group; SC, celecoxib group; WL, waiting list group. Further univariate logistic regression, adjusted for baseline clinical covariables (Figure S5 ), confirmed a specific association in the AG group between the acupuncture treatment response and the recessive model of GABRB3 rs4906902 (AG/GG genotype; OR = 3.74, 95% CI: 1.18–11.83, p = 0.025). Carriers of the G allele had a 2.7‐fold higher acupuncture analgesic response rate. No significant associations were detected between any other intervention group (SC, PB, and WL) and the target genetic models (all p > 0.05, Figure 3a ), consistent with the initial results. FIGURE 3. Open in a new tab Logistic regression results of gene polymorphisms in different intervention groups. AG, acupuncture group (VA + SA); PB, placebo group; SA, sham acupuncture group; VA, verum acupuncture group; SC, celecoxib group; WL, waiting list group; Additive model, n (CC vs. CT vs. TT)/ n (AA vs. AG vs. GG); Dominant model, n (CC/CT vs. TT)/ n (AA/AG vs. GG); Recessive model, n (CT/TT vs. CC)/ n (AG/GG vs. AA). Gene polymorphisms: GABRB3 rs4906902, OPRM1 rs1799971, COMT rs4680, and BDNF rs6265 genes. Homozygotes were combined with heterozygotes in the recessive model according to the presence of one or two G alleles. Unadjusted/adjusted OR, 95% CI, and p values are shown for patients carrying one or two G alleles relative to patients carrying none. A significant positive association was observed exclusively between AG‐responders and the recessive model (AG/GG genotype) of GABRB3 rs4906902 (OR = 3.737, 95% CI: 1.18–11.83, p = 0.025). No significant correlations were found for SC, PB, or WL with any genotype or genetic model (all p > 0.05). 3.4. GABRB3 rs4906902 Gene Polymorphism's Effect on Acupuncture Analgesia Efficacy Baseline analysis revealed no significant differences in demographic or clinical indicators among the four genotype groups in the AG cohort, except for the pretreatment VAS scores (Tables S5–S7 ). Therefore, a multivariate logistic regression analysis was conducted with adjustments for baseline VAS scores and genetic covariates. Initial chi‐square tests under a recessive model revealed a significantly higher analgesic response rate in G allele carriers (AG/GG genotypes) than in AA homozygotes (69.67% vs. 38.10%, p = 0.022; Table 1 ). Univariate logistic regression analysis confirmed a significant association between the recessive model and treatment response (odds ratio [OR] = 3.74, 95% CI: 1.18–11.83, p = 0.025). After multivariate adjustment (for baseline VAS and other genes), the association was further strengthened (OR = 3.85, 95% CI: 1.17–12.61, p = 0.026). This effect was strictly model‐dependent: neither the dominant model (OR = 1.42, 95% CI: 0.32–6.40, p = 0.647) nor the additive model (OR = 0.567, 95% CI: 0.250–1.27, p = 0.567) was significant. Furthermore, no significant associations were found for OPRM1 rs1799971, COMT rs4680, or BDNF rs6265 in any genetic model ( p > 0.05, Figure 3b ), indicating that the enhancement of acupuncture analgesic efficacy by the G allele was both gene‐specific and model‐dependent. The AG group included 54 patients (31 responders and 23 non‐responders). Based on the MCID [ 66 ], the clinical response rate to acupuncture was 57.41% (31/54). Key analyses showed: (1) no significant difference in acupoint allocation between responders and non‐responders ( p > 0.05, Table S8 ), indicating that efficacy differences were not due to acupoints selection; (2) comparable baseline demographics (gender, age, BMI, and disease duration) and disease‐related scores (VAS, SF‐MPQ, WOMAC, and SF‐12P) between groups (all p > 0.05, Table S9 ); and (3) significantly greater improvement in pain scales (VAS and MPQ) in responders versus non‐responders ( p < 0.05, Table S10 ). These results excluded baseline bias and acupoint effects, confirming that outcome differences stem from the acupuncture itself and further supporting GABRB3 rs4906902 as a reliable predictor of acupuncture analgesia efficacy. 3.5. VBM Results (GABRB3 rs4906902 AG/GG Carriers vs. AA Carriers) VBM analysis based on the GABRB3 rs4906902 genotype showed that, among patients receiving acupuncture treatment, those carrying the AG/GG genotype had significantly reduced gray matter volume in the brain regions of the bilateral head caudate nucleus, left putamen, and left ventral striatum, compared to AA homozygotes (at a voxel‐level p < 0.005 (uncorrected) and a cluster‐level p < 0.05 (FWE corrected)) (Table 2 and Figure 4 ). TABLE 2. Baseline gray matter VBM comparisons between AG/GG and AA GABRB3 rs4906902 genotype in KOA knee pain patients in acupuncture group. Contrast L/R Voxels Brain regions MNI (x, y, z) Z AA > AG/GG L 1061 Caudate/Putamen/Ventral Striatum −9, −6, 16.5 4.51 R 587 Caudate 7.5, 10.5, 16.5 3.72 AA < AG/GG No result Open in a new tab Note: A threshold of a voxel‐wise p < 0.005 uncorrected and p < 0.05 family wise error (FWE) correction at cluster level was applied for all analysis. Abbreviations: KOA, knee osteoarthritis; L, left; MNI, Montreal Neurological Institute atlas; R, right; VBM, voxel‐based morphometry. FIGURE 4. Open in a new tab Baseline gray matter VBM comparisons between AG/GG and AA GABRB3 rs4906902 genotype in KOA knee pain patients in acupuncture group. KOA, knee osteoarthritis; L, left; R, right; VBM, voxel‐based morphometry. A threshold of a voxel‐wise p < 0.005 uncorrected and p < 0.05 family wise error (FWE) correction at cluster level was applied for all analysis. 3.6. Neuroimaging Transcriptomics Results We conducted a transcriptome‐wide spatial correlation analysis using AHBA to investigate the molecular associations underlying VBM alterations. Partial least‐squares (PLS) regression identified genes highly expressed in the head caudate nucleus, putamen, and ventral striatum. The first PLS component (PLS1) was extracted as it accounted for the largest proportion of covariance (23.7%) in the spatial VBM patterns (Figure 5a ). FIGURE 5. Open in a new tab Gene expression profiles associated with GABRB3 genotype differences. (a) The PLS1 component accounted for 23.7% of the significant covariance in the spatial pattern of gray matter volume (GMV) differences between the GABRB3 AA genotype and AG/GG genotype knee pain patients in the acupuncture group. The scatter plot revealed a significant positive correlation ( r = 0.4868, p < 0.001) between regional PLS1 scores (representing weighted gene expression patterns) and regional GMV differences ( t ‐values for AA vs. AG/GG contrast). (b) Genes that positively and negatively weighted PLS1 values for the subsequent GO enrichment. The GABRB3 gene itself exhibited a significant negative weight ( Z = −6.32) in PLS1 analysis. Among the top‐weighted genes in PLS1, the highest positive‐weight gene EPN3 showed positive correlation with gray matter volume (GMV) differences between GABRB3 AG/GG and AA genotype groups, while the highest negative‐weight gene PSIP1 demonstrated inverse correlation with the same GMV contrast. We subsequently identified 3279 genes significantly associated with VBM changes (Table S11 ), including 1624 positively correlated ( Z > 3; designated PLS+ gene set) and 1655 negatively correlated ( Z < −3; PLS− gene set) genes. The PLS+ set showed high expression in the brain regions where gray matter volume is larger in AA carriers, while the PLS− set is enriched in brain regions where gray matter volume is smaller in AG/GG carriers, indicating spatial concordance between gene expression and genotype‐dependent structural differences. Notably, GABRB3 itself showed a strong negative weight on PLS1 ( Z = −6.32), belonging to the PLS− gene set, indicating higher expression in the head caudate nucleus, putamen, and ventral striatum with relatively increased gray matter volume in AA carriers (or reduced in AG/GG individuals), reflecting a role in enhancing GABAergic inhibition during pain modulation. EPN3 (encoding clathrin assembly protein 3), which is linked to synaptic dysfunction in neurodegenerative disorders, exhibited the highest positive weight. The most negatively weighted gene, PSIP1 (PC4 and SFRS1‐interacting protein 1), was associated with the activation of neuronal activity‐dependent gene networks (Figure 5b ). Gene Ontology (GO) and protein–protein interaction (PPI) network enrichment analyses revealed significant functional enrichment across multiple biological processes after removing outlier clusters (PBH‐FDR < 0.05). GABRB3‐associated genes were significantly enriched in GABAergic signaling pathways, including “GABA‐A receptor activity”(MF), “GABA‐gated chloride ion channel activity”(MF), and “GABA‐A receptor complex”(CC). At the biological process level, GABRB3 was enriched in terms such as “response to anesthesia” and “regulation of REM sleep,” reflecting the critical role of the GABAergic system in higher neural functions (Table 3 , Table S12 ). The PLS+ gene set showed significant enrichment in mitochondrial function (Table S13 ), with hub genes (NDUFS1, NDUFS2, and CYC1) implicated in energy homeostasis. In contrast, the PLS set was enriched in synaptic organization (Table S14 ), with key regulators (DLG4, PRKACB, and H4C6) involved in glutamatergic synaptic plasticity (Figure 6 and Figure S6 ). TABLE 3. GO analysis results of GABRB3 gene. Category Term GO: Molecular function GABA‐gated chloride ion channel activity GABA‐A receptor activity AP‐2 adaptor complex binding GABA receptor activity Ligand‐gated monoatomic anion channel activity GO: Biological process Response to anesthetic Circadian sleep/wake cycle, REM sleep Cellular response to histamine Hard palate development Response to histamine GO: Cellular component GABA‐A receptor complex GABA receptor complex Inhibitory synapse Chloride channel complex Terminal bouton Open in a new tab Note: The top 15 enriched pathways from Gene Ontology (GO) biological process‐based enrichment analysis of the GABRB3 gene set. FIGURE 6. Open in a new tab GO terms for GABRB3 and other significant genes. (a) GO terms for GABRB3. (b) GO terms for PLS+ genes. (c) GO terms for PLS− genes. The size of the circle represents the number of genes involved in the specific term, and the color represents the corrected p values ( p < 0.05, FDR‐corrected). 4. Discussion Overall, this study identified a specific association between the GABRB3 rs4906902 polymorphism and the analgesic response to acupuncture treatment in KOA patients with chronic knee pain. Specifically, AG/GG carriers showed a significantly higher response rate to acupuncture treatment than AA homozygotes (69.67% vs. 38.10%, respectively). This genotype‐dependent pattern was not observed in the SC, PB, or WL groups. Neuroimaging analysis further revealed that AG/GG carriers in the acupuncture group exhibited reduced gray matter volume in the head caudate nucleus, putamen, and ventral striatum compared with AA carriers. Spatial transcriptomic and molecular network analyses further demonstrated that these structural differences were correlated with GABRB3 expression patterns and functional pathways. To the best of our knowledge, this is the first study to illustrate the effect of gene polymorphisms and the CNS mechanism of individualized analgesic effects of acupuncture in patients with chronic pain. 4.1. GABRB3 rs4906902 Gene Polymorphism Affects Acupuncture Analgesia Responses The clinical findings of this study indicate that the GABRB3 rs4906902 polymorphism was significantly associated with differences in the analgesic effect of acupuncture treatment in patients with KOA chronic knee pain, with AG/GG genotype carriers exhibiting better analgesic efficacy. Although direct clinical evidence of the association between GABRB3 variants and the efficacy of acupuncture analgesia remains limited, substantial preclinical evidence supports the central role of the GABAergic system in acupuncture analgesia. For example, acupuncture has been shown to upregulate GABA_A receptor expression and enhance GABAergic neuronal activity, thereby inhibiting central pain transmission in rodent models [ 84 , 85 ]. Genetic studies have also indicated that variations in GABA‐related genes may be associated with pain sensitivity and the analgesic response [ 86 ]. Our results are consistent with these earlier findings and provide further evidence that the GABRB3 genotype may modulate individual analgesic responses to acupuncture treatment in patients with chronic pain. Notably, this study identified no significant effects of OPRM1, COMT, or BDNF polymorphisms on the analgesic response to acupuncture in patients with chronic knee pain. Although prior studies have indicated that these genes might influence the effects of acupuncture mainly through the opioid, dopaminergic, or neurotrophic pathways [ 87 , 88 , 89 ], the negative results observed in this study may be attributed to the limited sample size or the different mechanisms and circuit‐specific nature of the genetic modulation underlying acupuncture efficacy in chronic musculoskeletal pain. 4.2. GABRB3 rs4906902 Gene Polymorphism Affects Pain‐Modulating Brain Regions' Gray Matter Volume in Acupuncture Analgesia Responders Neuroimaging results showed that AG/GG carriers in the acupuncture group exhibited significantly reduced gray matter volume in the basal ganglia structures (including the head caudate nucleus, putamen, and ventral striatum) compared to AA homozygotes. These regions serve as critical hubs for gating and integrating pain‐related information [ 84 , 90 ]. The head and caudate integrate cognitive, attentional, executive, and nociceptive processes, receiving input from the prefrontal cortex and projecting to the medial dorsal thalamic nucleus, thus forming a recurrent cortico‐basal‐thalamocortical circuit [ 91 , 92 ]. This loop is thought to modulate the cognitive‐evaluative, attentional, and emotional aspects of pain and may contribute to central sensitization and pain chronicity. The putamen integrates motor, sensory, and partial affective information through inputs from the motor and somatosensory cortices. Its efferent neural projections to the globus pallidus and substantia nigra modulate thalamic and cortical motor regions that are thought to contribute to pain avoidance and motor responses [ 93 ]. The ventral striatum further receives inputs from the limbic areas (including the orbitofrontal cortex, anterior cingulate, hippocampus, and amygdala) and projects to the ventral pallidum and ventral tegmental area (VTA). At the core of the motivation–emotion–reward circuit, it is involved in processing pain‐related aversion, loss of motivation, and reward upon pain relief [ 90 , 94 ]. Collectively, these structures form the core output nuclei of the basal ganglia. Their efferent projections are primarily mediated by GABAergic neurons through the direct and indirect pathways, and act as a functional gate for nociceptive signals relayed from the thalamus to the cortex, and further modulate descending pain control pathways such as the periaqueductal gray [ 91 , 92 , 95 ]. Through extensive connections with the thalamus, cortex, and limbic system, the head caudate (cognitive‐affective gating), putamen (sensorimotor integration), and ventral striatum (motivation–emotion–reward) collectively regulate perceptual, emotional, and behavioral dimensions of pain, forming an integrated control hub highly responsive to treatment interventions such as acupuncture [ 84 , 90 ]. Notably, AG/GG carriers (who exhibited better acupuncture analgesic responses) showed reduced gray matter volume in these brain regions. We propose that this may reflect adaptive neuroplastic remodeling driven by enhanced GABAergic inhibition (e.g., synaptic pruning and circuit refinement [ 96 , 97 ]) rather than neurodegeneration, potentially reflecting a more efficient neural architecture despite lower volume. This “less volume, greater efficiency” remodeling parallels the “hub susceptibility” model in chronic pain [ 98 ], thereby suggesting a potential principle whereby structural refinement may promote functional optimization. 4.3. GABAergic Transcriptional Regulation Mediates Genotype‐Dependent Acupuncture Analgesia Integrated neuroimaging and transcriptomic analyses revealed increased GABRB3 expression in the head caudate, putamen, and ventral striatum of AG/GG carriers in the AG. These brain regions spatially corresponded to PLS1‐negative weighting and gray matter volume reduction, and showed functional enrichment for GABA_A receptor activity. The AG/GG genotype may upregulate GABRB3 expression by enhancing transcription factor binding or chromatin accessibility in intronic regions [ 99 , 100 ], thereby potentially increasing the density and stability of β3 subunits in postsynaptic membranes. Elevated β3 subunit levels might enhance the precision and efficiency of inhibitory control by PV+ and SST+ interneurons over medium spiny neurons (MSNs) [ 97 ]. This could strengthen local GABAergic inhibition and may suppress excessive glutamatergic inputs from prefrontal and limbic areas, and restore thalamic gating and excitatory‐inhibitory (E/I) balance [ 91 , 92 , 95 ]. Whole‐picture enrichment analysis showed that the PLS+ and PLS− gene sets were most prominently associated with mitochondrial energy metabolism and glutamatergic synaptic regulation, respectively. Key hub molecules—including DLG4 (PSD95), regulating glutamatergic synaptic structure; PRKACB, enhancing GABA_A receptor function through β3 subunit phosphorylation; histone variants such as H4C6 involved in epigenetic modulation; and mitochondrial genes (NDUFS1/2, CYC1) providing energy support for high‐energy‐consuming inhibitory processes—collectively form a coordinated regulatory network. This network thereby provides essential molecular support and bioenergetic resources for enhanced GABAergic inhibition and neuroplastic remodeling driven by GABRB3 overexpression. Notably, these genotype–phenotype–functional correlations were significant only in the acupuncture group, indicating that the core mechanism underlying individualized acupuncture analgesia may primarily depend on GABAergic plasticity. In summary, we propose that the AG/GG genotype of GABRB3 is associated with increased expression of the β3 subunit and enhanced GABAergic inhibition in the brain of chronic knee pain patients, which may contribute to elevated basal inhibitory tone within the basal ganglia‐thalamic circuit and a potential shift in excitatory/inhibitory (E/I) balance toward inhibition. This suggests the formation of a modulable state primed for intervention. When acupuncture activates descending inhibitory pathways, such as the serotonergic system originating from the raphe nucleus, this preconditioned state may enable more efficient amplification of GABAergic inhibition and suppression of glutamatergic transmission in key brain regions, such as the head caudate nucleus. This could promote rapid restoration and maintenance of an optimal GABA/glutamate balance, potentially being associated with more pronounced analgesic effects. 4.4. Acupuncture Versus NSAIDs Analgesic Mechanisms While both acupuncture and celecoxib demonstrated comparable clinical efficacy in our study, the underlying analgesic mechanisms may differ. Celecoxib, a selective cyclooxygenase‐2 (COX‐2) inhibitor, primarily exerts its analgesic effects through peripheral mechanisms by reducing prostaglandin synthesis at sites of inflammation in the knee joint [ 57 ]. In contrast, acupuncture is thought to modulate pain predominantly through central mechanisms, including the activation of descending inhibitory pathways, regulation of endogenous opioid release, and restoration of central excitatory/inhibitory balance [ 16 , 17 , 84 ]. The absence of association between GABRB3 rs4906902 polymorphism and treatment response in the celecoxib group suggests that genetic variants influencing GABAergic neurotransmission may specifically predict responses to centrally acting analgesic interventions rather than peripherally acting NSAIDs. This distinction has important implications for multimodal pain therapy. Patients carrying the AG/GG genotype, who exhibit enhanced GABAergic tone in pain‐modulating brain regions, may be more likely to benefit from acupuncture or other neuromodulatory therapies targeting central pain processing. Conversely, the comparable efficacy of celecoxib across genotypes suggests that peripheral anti‐inflammatory mechanisms may be less dependent on individual GABAergic genetic variation. These findings suggest the potential for a precision medicine approach to chronic pain management, wherein genetic profiling could help guide the selection between central (acupuncture) and peripheral (NSAIDs) analgesic strategies. Future studies directly comparing genetically stratified responses to acupuncture versus first‐line pharmacotherapies would provide important evidence for implementing such personalized treatment algorithms in clinical practice. 4.5. Clinical Applicability and Future Perspectives This study provides preliminary neuroimaging‐genetic evidence for individualized acupuncture treatment in knee osteoarthritis. In clinical practice, genotyping of GABRB3 and related variants may enable physicians to identify patients with superior analgesic responses to acupuncture prior to treatment initiation. Such a “pre‐selection” strategy could enhance therapeutic success rates while avoiding unnecessary medical expenditures, offering potential health economic benefits. However, translation into routine clinical application requires validation in larger, more diverse independent cohorts, with refinement of predictive accuracy through polygenic risk models. Future direct comparisons between genetically guided acupuncture protocols and first‐line pharmacotherapies such as celecoxib will provide more robust evidence‐based support for its role in precision pain management. 5. Limitations This study had several limitations. Firstly, the specific molecular mechanism by which rs4906902 regulates GABRB3 expression lacks functional validation, such as through reporter gene assays and ChIP‐seq experiments. Second, the dynamic monitoring of GABA/glutamate concentration changes in the caudate nucleus before and after acupuncture, which could directly test the neurotransmitter rebalancing hypothesis, was not performed. Third, gene expression profiles from the AHBA are based on limited samples and may not fully represent the molecular environment in the brains of patients with chronic KOA. Fourth, transcriptome‐imaging correlations are inherently spatial and do not establish causal relationships. Finally, differences in spatial resolution between neuroimaging and transcriptomic data restrict more refined regional analyses. 6. Conclusion Overall, this study showed that genetic polymorphisms at GABRB3 rs4906902 could influence the analgesic effect of acupuncture treatment in patients with KOA chronic knee pain, with AG/GG genotype carriers exhibiting a superior analgesic effect. This finding might be associated with pain‐modulating gray matter volume reduction in brain regions and upregulation of GABRB3 gene expression. AG/GG carriers may possess enhanced GABA receptor efficiency and improved regulation of the GABA/glutamate balance in pain‐modulating brain regions, thus having a better analgesic effect. Author Contributions Z.L. and F.Z. are the corresponding authors. S.W., W.W., and J.Z. contributed equally to this article. Study protocol and design: F.L., Z.L., and F.Z.; acquisition of data: J.Z., S.C., X.D., C.T., W.H., and Y.C.; analysis and interpretation of data: S.W., W.W., and Z.L.; drafting of the manuscript: S.W., W.W., and Z.L.; proofread and revise the manuscript: all authors. Funding This study was supported by funds from the National Natural Science Foundation of China (82474657, 81973958, and 81774400) and Distinguished Young Scholars Project of Science and Technology Department of Sichuan Province (No. 2025NSFJQ0056). Ethics Statement This study was approved by the supervision of the Sichuan Regional Ethics Review Committee on Traditional Chinese Medicine (ethical approval number 2016KL‐017). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Data S1: fsb271779‐sup‐0001‐DataS1.zip. FSB2-40-e71779-s001.zip (6.8MB, zip) Acknowledgments We are grateful to Professor Yong Huang and Dr. Xue‐fei Qin for their kind help in the participant enrollment. Contributor Information Fanrong Liang, Email: [email protected]. Zhengjie Li, Email: [email protected]. Data Availability Statement The datasets used during this study are available from the corresponding authors on reasonable request. References 1. Felson D. T., Lawrence R. C., Dieppe P. A., et al., “Osteoarthritis: New Insights. Part 1: The Disease and Its Risk Factors,” Annals of Internal Medicine 133 (2000): 635–646. [ DOI ] [ PubMed ] [ Google Scholar ] 2. Loeser R. F., Goldring S. R., Scanzello C. R., and Goldring M. B., “Osteoarthritis: A Disease of the Joint as an Organ,” Arthritis and Rheumatism 64 (2012): 1697–1707. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Yang G., Wang J., Liu Y., et al., “Burden of Knee Osteoarthritis in 204 Countries and Territories, 1990–2019: Results From the Global Burden of Disease Study 2019,” Arthritis Care & Research 75, no. 12 (2023): 2489–2500. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Mohan V., Mokdad A. H., Mozaffarian D., et al., “Global, Regional, and National Incidence, Prevalence, and Years Lived With Disability for 354 Diseases and Injuries for 195 Countries and Territories, 1990–2017: A Systematic Analysis for the Global Burden of Disease Study 2017,” Lancet 392, no. 10159 (2018): 1789–1858. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Vos T., Flaxman A. D., Naghavi M., et al., “Years Lived With Disability (YLDs) for 1160 Sequelae of 289 Diseases and Injuries 1990–2010: A Systematic Analysis for the Global Burden of Disease Study 2010,” Lancet 380, no. 9859 (2012): 2163–2196. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Hiligsmann M. L., Cooper C., Arden N., et al., “Health Economics in the Field of Osteoarthritis:An Expert's Consensus Paper From the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis(ESCEO),” Seminars in Arthritis and Rheumatism 43, no. 3 (2013): 303–313. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Richmond J., Hunter D., Irrgang J., et al., “American Academy of Orthopaedic Surgeons Clinical Practice Guideline on the Treatment of Osteoarthritis (OA) of the Knee,” Journal of Bone and Joint Surgery. American Volume 92 (2010): 990–993. [ DOI ] [ PubMed ] [ Google Scholar ] 8. Bannuru R. R., Osani M. C., Vaysbrot E. E., et al., “OARSI Guidelines for the Non‐Surgical Management of Knee, Hip, and Polyarticular Osteoarthritis,” Osteoarthritis and Cartilage 27 (2019): 1578–1589. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Mäntyselkä P., Kumpusalo E., Ahonen R., Kumpusalo A., Kauhanen J., and Viinamäki H., “Pain as a Reason to Visit the Doctor: A Study in Finnish Primary Health Care,” Pain 89 (2001): 175–180. [ DOI ] [ PubMed ] [ Google Scholar ] 10. McAlindon T. E., Bannuru R. R., Sullivan M. C., et al., “OARSI Guidelines for the Non‐Surgical Management of Knee Osteoarthritis,” Osteoarthritis and Cartilage 22, no. 3 (2014): 363–388. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Kolasinski S. L., Neogi T., Hochberg M. C., et al., “2019 American College of Rheumatology/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee,” Arthritis Care & Research 72, no. 2 (2020): 220–233. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Tu J. F., Yang J. W., Shi G. X., et al., “Efficacy of Intensive Acupuncture Versus Sham Acupuncture in Knee Osteoarthritis: A Randomized Controlled Trial,” Arthritis and Rhematology 73 (2021): 448–458. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Mavrommatis C. I., Argyra E., Vadalouka A., and Vasilakos D. G., “Acupuncture as an Adjunctive Therapy to Pharmacological Treatment in Patients With Chronic Pain due to Osteoarthritis of the Knee: A 3‐Armed, Randomized, Placebo‐Controlled Trial,” Pain 153 (2012): 1720–1726. [ DOI ] [ PubMed ] [ Google Scholar ] 14. Berman B. M., Lao L., Langenberg P., Lee W. L., Gilpin A. M. K., and Hochberg M. C., “Effectiveness of Acupuncture as Adjunctive Therapy in Osteoarthritis of the Knee: A Randomized, Controlled Trial,” Annals of Internal Medicine 141 (2004): 901–910. [ DOI ] [ PubMed ] [ Google Scholar ] 15. Vickers A. J., Cronin A. M., Maschino A. C., et al., “Acupuncture for Chronic Pain: Individual Patient Data Meta‐Analysis,” Archives of Internal Medicine 172, no. 19 (2012): 1444–1453. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Liu S., Wang Z., Su Y., et al., “A Neuroanatomical Basis for Electroacupuncture to Drive the Vagal‐Adrenal Axis,” Nature 598 (2021): 641–645. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Zhou J., Zeng F., Cheng S., et al., “Modulation Effects of Different Treatments on Periaqueductal Gray Resting State Functional Connectivity in Knee Osteoarthritis Knee Pain Patients,” CNS Neuroscience & Therapeutics 29 (2023): 1965–1980, 10.1111/cns.14153. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Yang M., Baser R. E., Khanin R., et al., “COMT Val158Met Affects the Analgesic Response to Acupuncture Among Cancer Survivors With Chronic Pain,” Journal of Pain 24, no. 9 (2023): 1721–1730. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Luo Y., Kusay A. S., Jiang T., Chebib M., and Balle T., “Delta‐Containing GABA_A Receptors in Pain Management: Promising Targets for Novel Analgesics,” Neuropharmacology 195, no. 1 (2021): 108675, 10.1016/j.neuropharm.2021.108675. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Ma Y., Pan H., and Fields H. L., “Contribution of Brainstem GABA(A) Synaptic Transmission to Morphine Analgesic Tolerance,” Pain 122, no. 1–2 (2006): 163–173. [ DOI ] [ PubMed ] [ Google Scholar ] 21. Tan X., Zheng D. H., Lin Q., et al., “Confirmation of Pain‐Related Neuromodulation Mechanism of Bushen Zhuangjin Decoction on Knee Osteoarthritis,” Journal of Ethnopharmacology 324 (2024): 117772. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Smith S. B., Maixner D. W., Fillingim R. B., et al., “Large Candidate Gene Association Study Reveals Genetic Risk Factors and Therapeutic Targets for Fibromyalgia,” Arthritis and Rheumatism 64, no. 2 (2012): 584–593, 10.1002/art.33338. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Qian X., Zhao X. Y., Yu L., et al., “Current Status of GABA Receptor Subtypes in Analgesia,” Biomedicine & Pharmacotherapy 168 (2023): 115800. [ DOI ] [ PubMed ] [ Google Scholar ] 24. Lin J.‐G. and Chen W.‐L., “Acupuncture Analgesia: A Review of Its Mechanisms of Actions,” American Journal of Chinese Medicine 36, no. 4 (2008): 635–645. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Mague S. D. and Blendy J. A., “OPRM1 SNP (A118G): Involvement in Disease Development, Treatment Response, and Animal Models,” Drug and Alcohol Dependence 108, no. 3 (2010): 172–182. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Gonçalves F. d. T., Pacheco‐Barrios K., Rebello‐Sanchez I., et al., “Association of mu Opioid Receptor (A118G) and BDNF (G196A) Polymorphisms With Rehabilitation‐Induced Cortical Inhibition and Analgesic Response in Chronic Osteoarthritis Pain,” International Journal of Chronic Pain 34 (2022): 100330. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Govil M., Mukhopadhyay N., Holwerda T., et al., “Effects of Genotype on TENS Effectiveness in Controlling Knee Pain in Persons With Mild to Moderate Osteoarthritis,” European Journal of Pain 24, no. 2 (2020): 398–412. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Suchting R., Teixeira A. L., Ahn B., et al., “Changes in Brain‐Derived Neurotrophic Factor From Active and Sham Transcranial Direct Current Stimulation in Older Adults With Knee Osteoarthritis,” Clinical Journal of Pain 37, no. 12 (2021): 898–903. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Simao A. P., Mendonca V. A., de Oliveira A. T., et al., “Involvement of BDNF in Knee Osteoarthritis: The Relationship With Inflammation and Clinical Parameters,” Rheumatology International 34, no. 8 (2014): 1153–1157. [ DOI ] [ PubMed ] [ Google Scholar ] 30. Chen C. H., Huang C. C., Cheng M. C., et al., “Genetic Analysis of GABRB3 as a Candidate Gene of Autism Spectrum Disorders,” Molecular Autism 5, no. 1 (2014): 36. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Johannesen K. M., Iqbal S., Guazzi M., et al., “Structural Mapping of GABRB3 Variants Reveals Genotype‐Phenotype Correlations,” Genetics in Medicine 24, no. 3 (2022): 681–693. [ DOI ] [ PubMed ] [ Google Scholar ] 32. Shi Y. W., Zhang Q., Cai K., et al., “Synaptic Clustering Differences due to Different GABRB3 Mutations Cause Variable Epilepsy Syndromes,” Brain 142, no. 10 (2019): 3028–3044, 10.1093/brain/awz250. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Urak L., Feucht M., Fathi N., et al., “A GABRB3 Promoter Haplotype Associated With Childhood Absence Epilepsy Impairs Transcriptional Activity,” Human Molecular Genetics 15 (2006): 2533Y41. [ DOI ] [ PubMed ] [ Google Scholar ] 34. Zhang J. H., Sato M., and Tohyama M., “Region‐Specific Expression of the mRNAs Encoding Beta Subunits (Beta 1, Beta 2, and Beta 3) of GABAA Receptor in the Rat Brain,” Journal of Comparative Neurology 303, no. 4 (1991): 637–657. [ DOI ] [ PubMed ] [ Google Scholar ] 35. Akagawa K., Saji M., Aoki N., and Kanno T., “Muscimol Suppresses rCGU Increase in the Substantia Nigra After Destruction of the Caudate Nucleus,” Neuroreport 3, no. 5 (1992): 441–444. [ DOI ] [ PubMed ] [ Google Scholar ] 36. Weiss S. A., “Chloride Ion Dysregulation in Epileptogenic Neuronal Networks,” Neurobiology of Disease 177 (2023): 106000. [ DOI ] [ PubMed ] [ Google Scholar ] 37. Pernhorst K., Raabe A., Niehusmann P., et al., “Promoter Variants Determine γ‐Aminobutyric Acid Homeostasis‐Related Gene Transcription in Human Epileptic Hippocampi,” Neurology 77, no. 19 (2011): 1769–1776. [ DOI ] [ PubMed ] [ Google Scholar ] 38. Lu Y., Zheng J., Xiong L., Zimmermann M., and Yang J., “Spinal Cord Injury‐Induced Attenuation of GABAergic Inhibition in Spinal Dorsal Horn Circuits Is Associated With Down‐Regulation of the Chloride Transporter KCC2 in Rat,” Journal of Physiology 586, no. 23 (2008): 5701–5715. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Jasmin L., Rabkin S. D., Granato A., Boudah A., and Ohara P. T., “Analgesia and Hyperalgesia From GABA‐Mediated Modulation of the Cerebral Cortex,” Nature 424, no. 6946 (2003): 316–320. [ DOI ] [ PubMed ] [ Google Scholar ] 40. Shen Y., Shen Y., Wang M., et al., “A Spatial Imaging‐Transcriptomics Paradigm for Deciphering the Molecular Basis of Microscopic MRI in the Normal Brain and Alzheimer's Disease,” Cell Reports 44, no. 8 (2025): 116073. [ DOI ] [ PubMed ] [ Google Scholar ] 41. Martins D., Giacomel A., Williams S. C. R., et al., “Imaging Transcriptomics: Convergent Cellular, Transcriptomic, and Molecular Neuroimaging Signatures in the Healthy Adult Human Brain,” Cell Reports 37, no. 12 (2021): 110173. [ DOI ] [ PubMed ] [ Google Scholar ] 42. Jia J., Yu Y., Deng J. H., et al., “A Review of Omics Research in Acupuncture: The Relevance and Future Prospects for Understanding the Nature of Meridians and Acupoints,” Journal of Ethnopharmacology 140 (2012): 594–603. [ DOI ] [ PubMed ] [ Google Scholar ] 43. Chen P., Zhao K., Zhang H., et al., “Altered Global Signal Topography in Alzheimer's Disease,” eBioMedicine 89 (2023): 104455, 10.1016/j.ebiom.2023.104455. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Arnatkeviciute A., Fulcher B. D., and Fornito A., “A Practical Guide to Linking Brain‐Wide Gene Expression and Neuroimaging Data,” NeuroImage 189 (2019): 353–367. [ DOI ] [ PubMed ] [ Google Scholar ] 45. Martins D., Dipasquale O., Veronese M., et al., “Transcriptional and Cellular Signatures of Cortical Morphometric Remodelling in Chronic Pain,” Pain 163, no. 6 (2021): e759–e773. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Zheng J., Womer F. Y., Tang L., et al., “Integrative Omics Analysis Reveals Epigenomic and Transcriptomic Signatures Underlying Brain Structural Deficits in Major Depressive Disorder,” Translational Psychiatry 14, no. 1 (2024): 17. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Ji Y., Zhang X., Wang Z., et al., “Genes Associated With Gray Matter Volume Alterations in Schizophrenia,” NeuroImage 225 (2020): 117526. [ DOI ] [ PubMed ] [ Google Scholar ] 48. Yu L., Yang L., and Chen X., “Cerebral Blood Flow Changes and Their Spatial Correlations With GABAa and Dopamine‐D1 Receptor Explaining Individual Differences in Chronic Insomnia and the Therapeutic Effects of Acupuncture,” Human Brain Mapping 46, no. 4 (2025): e70183. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Li J., Peng C., and He K., “The Central Mechanisms of Electroacupuncture at LR3 in the Treatment of Spontaneous Hypertension: A PET and mRNA Transcriptome Study,” Frontiers in Cardiovascular Medicine 11 (2024): 1358426. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. Qiao H. F., “New Technologies Serve as Accelerators for Breakthroughs in Modern Acupuncture Research,” Zhen Ci Yan Jiu 50, no. 5 (2025): 526–530. [ DOI ] [ PubMed ] [ Google Scholar ] 51. Liu T., Wang S., Tang Y., et al., “Structural and Functional Alterations in MRI‐Negative Drug‐Resistant Epilepsy and Associated Gene Expression Features,” NeuroImage 302 (2024): 120908. [ DOI ] [ PubMed ] [ Google Scholar ] 52. Gong X., Li W., Luo C., et al., “Neurochemical and Molecular Characteristics of Altered Brain Functional Activity in the Anti‐NMDAR Encephalitis,” Neurobiology of Disease 215 (2025): 107050. [ DOI ] [ PubMed ] [ Google Scholar ] 53. Teng Y., Luo W., Zhou J., et al., “Clinical Research on Tai Chi: A Review of Health Benefits,” Acupuncture and Herbal Medicine 4 (2024): 320–345. [ Google Scholar ] 54. Adewale Q., Khan A. F., Lin S. J., et al., “Patient‐Centered Brain Transcriptomic and Multimodal Imaging Determinants of Clinical Progression, Physical Activity, and Treatment Needs in Parkinson's Disease,” Npj Parkinson's Disease 11, no. 1 (2025): 29. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Fox F. A. U., Diers K., Lee H., et al., “Association Between Accelerometer‐Derived Physical Activity Measurements and Brain Structure: A Population‐Based Cohort Study,” Neurology 99, no. 11 (2022): e1202–e1215. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 56. Guo J., Chen Y., Li Z., et al., “The Cerebral Mechanism of Acupuncture for Treating Knee Osteoarthritis: Study Protocol for a Randomized Controlled Trial,” Trials 20 (2019): 126. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Tang C., Dong X., He W., et al., “Cerebral Mechanism of Celecoxib for Treating Knee Pain: Study Protocol for a Randomized Controlled Parallel Trial,” Trials 20 (2019): 58. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Altman R. D., “Classification of Disease: Osteoarthritis,” Seminars in Arthritis and Rheumatism 20 (1991): 40–47. [ DOI ] [ PubMed ] [ Google Scholar ] 59. Kohn M. D., Sassoon A. A., and Fernando N. D., “Classifications in Brief: Kellgren‐Lawrence Classification of Osteoarthritis,” Clinical Orthopaedics and Related Research 474 (2016): 1886–1893. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 60. Kong J., Gollub R., Huang T., et al., “Acupuncture de Qi, From Qualitative History to Quantitative Measurement,” Journal of Alternative and Complementary Medicine 13, no. 10 (2007): 1059–1070. [ DOI ] [ PubMed ] [ Google Scholar ] 61. Argoff C. E., “Topical Analgesics in the Management of Acute and Chronic Pain,” Mayo Clinic Proceedings 88, no. 2 (2013): 195–205, 10.1016/j.mayocp.2012.11.015. [ DOI ] [ PubMed ] [ Google Scholar ] 62. Wewers M. E. and Lowe N., “K.A Critical Review of Visual Analogue Scales in the Measurement of Clinical Phenomena.Research in Nursing &,” Health 13, no. 4 (1990): 227–236. [ DOI ] [ PubMed ] [ Google Scholar ] 63. Bellamy N., Buchanan W. W., Goldsmith C. H., Campbell J., and Stitt L. W., “Validation Study of WOMAC:A Health Status Instrument for Measuring Clinically Important Patient Relevant Outcomes to Antirheumatic Drug Therapy in Patients With Osteoarthritis of the Hip or Knee,” Journal of Rheumatology 15, no. 12 (1988): 1833–1840. [ PubMed ] [ Google Scholar ] 64. Melzack R., “The Short‐Form McGill Pain Questionnaire,” Pain 30, no. 2 (1987): 191–197, 10.1016/0304-3959(87)91074-8. [ DOI ] [ PubMed ] [ Google Scholar ] 65. Gandek B., Ware J., Aaronson N. K., et al., “Cross‐Validation of Item Selection and Scoring for the SF‐12 Health Survey in Nine Countries:Results From the IQOLA Project,” Journal of Clinical Epidemiology 51, no. 11 (1998): 1171–1178. [ DOI ] [ PubMed ] [ Google Scholar ] 66. Beaton D. E., Bombardier C., Katz J. N., and Beaton D., “A Taxonomy for Responsiveness,” Journal of Clinical Epidemiology 54, no. 12 (2001): 1204–1217. [ DOI ] [ PubMed ] [ Google Scholar ] 67. Tubach , “Evaluation of Clinically Relevant Changes in Patient Reported Outcomes in Knee and Hip Osteoarthritis:The Minimal Clinically Important Improvement,” Annals of the Rheumatic Diseases 64, no. 1 (2005): 29–33. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 68. Lin L. L., Tu J. F., Wang L. Q., et al., “Acupuncture of Different Treatment Frequencies in Knee Osteoarthritis:A Pilot Randomised Controlled Trial,” Pain 161, no. 11 (2020): 2532–2538. [ DOI ] [ PubMed ] [ Google Scholar ] 69. Besteher B., Gaser C., Langbein K., Dietzek M., Sauer H., and Nenadic I., “Effects of Subclinical Depression, Anxiety and Somatization on Brain Structure in Healthy Participants,” Journal of Affective Disorders 215, no. 9 (2017): 111–117. [ DOI ] [ PubMed ] [ Google Scholar ] 70. Ashburner J., “A Fast Diffeomorphic Image Registration Algorithm,” NeuroImage 38 (2007): 95–113. [ DOI ] [ PubMed ] [ Google Scholar ] 71. Gaser C., Dahnke R., Thompson P. M., Kurth F., and Luders E., “CAT ‐ A Computational Anatomy Toolbox for the Analysis of Structural MRI Data,” Human Brain Mapping 37, no. 3 (2016): 336–348, 10.1002/hbm.23024. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Good C. D., Johnsrude I. S., Ashburner J., Henson R. N. A., Friston K. J., and Frackowiak R. S. J., “A Voxel‐Based Morphometric Study of Ageing in 465 Normal Adult Human Brains,” NeuroImage 14 (2001): 21–36. [ DOI ] [ PubMed ] [ Google Scholar ] 73. Hawrylycz M. J., Lein E. S., Guillozet‐Bongaarts A. L., et al., “An Anatomically Comprehensive Atlas of the Adult Human Brain Transcriptome,” Nature 489, no. 7416 (2012): 391–399. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 74. Markello R. D., Arnatkeviciute A., Poline J. B., Fulcher B. D., Fornito A., and Misic B., “Standardizing Workflows in Imaging Transcriptomics With the Abagen Toolbox,” eLife 10 (2021): 10. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Fan L., Li H., Zhuo J., et al., “The Human Brainnetome Atlas: A New Brain Atlas Based on Connectional Architecture,” Cerebral Cortex 26, no. 8 (2016): 3508–3526. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 76. Morgan S. E., Seidlitz J., Whitaker K. J., et al., “Cortical Patterning of Abnormal Morphometric Similarity in Psychosis Is Associated With Brain Expression of Schizophrenia‐Related Genes,” Proceedings of the National Academy of Sciences of the United States of America 116, no. 19 (2019): 9604–9609. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 77. Li L., Wei Y., Zhang J., et al., “Gene Expression Associated With Individual Variability in Intrinsic Functional Connectivity,” NeuroImage 245 (2021): 118743. [ DOI ] [ PubMed ] [ Google Scholar ] 78. Chen J., Bardes E. E., Aronow B. J., and Jegga A. G., “ToppGene Suite for Gene List Enrichment Analysis and Candidate Gene Prioritization,” Nucleic Acids Research 37 (2009): W305–W311. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 79. Szklarczyk D., Gable A. L., Lyon D., et al., “STRING v11: Protein‐Protein Association Networks With Increased Coverage, Supporting Functional Discovery in Genome‐Wide Experimental Datasets,” Nucleic Acids Research 47, no. D1 (2019): D607–D613. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Witt C., Brinkhaus B., Jena S., et al., “Acupuncture in Patients With Osteoarthritis of the Knee: A Randomised Trial,” Lancet 366 (2005): 136–143. [ DOI ] [ PubMed ] [ Google Scholar ] 81. Hinman R. S., McCrory P., Pirotta M., et al., “Acupuncture for Chronic Knee Pain: A Randomized Clinical Trial,” Journal of the American Medical Association 312 (2014): 1313–1322. [ DOI ] [ PubMed ] [ Google Scholar ] 82. Gollub R. L., Kirsch I., Maleki N., et al., “A Functional Neuroimaging Study of Expectancy Effects on Pain Response in Patients With Knee Osteoarthritis,” Journal of Pain 19 (2018): 515–527. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 83. Sun Z., Qu X., Wang T., Liu F., and Li X., “Effects of Warm Acupuncture Combined With Meloxicam and Comprehensive Nursing on Pain Im Provement and Joint Function in Patients With Knee Osteoarthritis,” Journal of Healthcare Engineering 31 (2022): 9167956. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 84. Lu W., Zhang T., Li M., et al., “Clinical Efficacy and Potential Mechanisms of Acupuncture for Parkinson's Disease: The Role of GABAergic Signaling,” Frontiers in Neuroscience 19 (2025): 1525486. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 85. Kang X., Chen S., Ma L., Kong Y., Huang Y., and Shi S., “Regulatory Effects of Head Acupuncture on the GABAergic System and Wnt7a/β‐Catenin Pathway in Post‐Stroke Spasticity Rats,” Journal of Molecular Histology 56, no. 4 (2025): 228. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 86. Zhu H., Xiang H. C., Li H. P., et al., “Inhibition of GABAergic Neurons and Excitation of Glutamatergic Neurons in the Ventrolateral Periaqueductal Gray Participate in Electroacupuncture Analgesia Mediated by Cannabinoid Receptor,” Frontiers in Neuroscience 13 (2019): 484. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 87. Bach P., Vollsta Dt‐Klein S., Kirsch M., et al., “Increased Mesolimbic Cue‐Reactivity in Carriers of the Mu‐Opioid‐Receptor Gene OPRM1 A118G Polymorphism Predicts Drinking Outcome: A Functional Imaging Study in Alcohol Dependent Subjects,” European Neuropsychopharmacology 25, no. 8 (2015): 1128–1135. [ DOI ] [ PubMed ] [ Google Scholar ] 88. Zhang X., Lee M. R., Salmeron B. J., et al., “Prefrontal White Matter Impairment in Substance Users Depends Upon the Catechol‐o‐Methyl Transferase (COMT) Val158Met Polymorphism,” NeuroImage 69 (2012): 62–69. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 89. Moffat J. J., Sakhai S. A., Hoisington Z. W., Ehinger Y., and Ron D., “The BDNF Val68Met Polymorphism Causes a Sex Specific Alcohol Preference Over Social Interaction and Also Acute Tolerance to the Anxiolytic Effects of Alcohol, a Phenotype Driven by Malfunction of BDNF in the Ventral Hippocampus of Male Mice,” Psychopharmacology 240, no. 2 (2023): 303–317. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 90. Mercer Lindsay N., Chen C., Gilam G., Mackey S., and Scherrer G., “Brain Circuits for Pain and Its Treatment,” Science Translational Medicine 13, no. 619 (2021): eabj7360. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 91. Haber S. N. and Knutson B., “The Reward Circuit: Linking Primate Anatomy and Human Imaging,” Neuropsychopharmacology 35, no. 1 (2010): 4–26. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 92. Chen G., Jiang C., Li S., et al., “The Role of the Human Caudate Nucleus in Acupuncture Analgesia,” Acupuncture & Electro‐Therapeutics Research: International Journal of Integrated Medicine 7, no. 4 (1982): 255–265. [ DOI ] [ PubMed ] [ Google Scholar ] 93. Stocco A., “A Biologically Plausible Action Selection System for Cognitive Architectures: Implications of Basal Ganglia Anatomy for Learning and Decision‐Making Models,” Cognitive Science 42, no. 2 (2017): 457–490. [ DOI ] [ PubMed ] [ Google Scholar ] 94. Takahashi Y. K., Langdon A. J., Niv Y., and Schoenbaum G., “Temporal Specificity of Reward Prediction Errors Signaled by Putative Dopamine Neurons in Rat VTA Depends on Ventral Striatum,” Neuron 91, no. 1 (2016): 182–193. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 95. Heinricher M. M., Tavares I., Leith J. L., et al., “Descending Control of Nociception: Specificity, Recruitment and Plasticity,” Brain Research Reviews 60, no. 1 (2008): 214–225. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 96. Gerdeman G. L., Partridge J. G., Lupica C. R., and Lovinger D. M., “It Could Be Habit Forming: Drugs of Abuse and Striatal Synaptic Plasticity,” Trends in Neurosciences 26, no. 4 (2003): 184–192. [ DOI ] [ PubMed ] [ Google Scholar ] 97. Di Cristo G., Pizzorusso T., Cancedda L., et al., “GABAergic Circuit Development and Its Implication for CNS Disorders,” Neural Plasticity 2011 (2011): 623705. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 98. Martins D., Dipasquale O., Veronese M., et al., “Transcriptional and Cellular Signatures of Cortical Morphometric Remodelling Inchronic Pain,” Pain 163, no. 6 (2021): e759–e773. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 99. Poulter M. O., Du L., Weaver I. C. G., et al., “GABAA Receptor Promoter Hypermethylation in Suicide Brain: Implications for the Involvement of Epigenetic Processes,” Biological Psychiatry 64, no. 8 (2008): 645–652. [ DOI ] [ PubMed ] [ Google Scholar ] 100. Samaco R. C., Hogart A., and LaSalle J. M., “Epigenetic Overlap in Autism‐Spectrum Neurodevelopmental Disorders: MECP2 Deficiency Causes Reduced Expression of UBE3A and GABRB3,” Human Molecular Genetics 14, no. 4 (2004): 483–492. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Data S1: fsb271779‐sup‐0001‐DataS1.zip. FSB2-40-e71779-s001.zip (6.8MB, zip) Data Availability Statement The datasets used during this study are available from the corresponding authors on reasonable request. Articles from The FASEB Journal are provided here courtesy of Wiley ACTIONS View on publisher site PDF (3.2 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top