Sexual Dimorphism in Clinical Manifestations of Knee Osteoarthritis.

Sexual dimorphism in clinical manifestations of knee osteoarthritis - 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 Med Sci Sports Exerc . Author manuscript; available in PMC: 2026 Apr 19. Published in final edited form as: Med Sci Sports Exerc. 2026 Jan 26;58(6):1159–1171. doi: 10.1249/MSS.0000000000003944 Search in PMC Search in PubMed View in NLM Catalog Add to search Sexual dimorphism in clinical manifestations of knee osteoarthritis Atsushi Hoki Atsushi Hoki , PT 1 Department of Physical Therapy, Matsuda Orthopedic Clinic, Saitama, Japan Find articles by Atsushi Hoki 1 , Tsubasa Iwasaki Tsubasa Iwasaki , MSc, PT 1 Department of Physical Therapy, Matsuda Orthopedic Clinic, Saitama, Japan Find articles by Tsubasa Iwasaki 1 , Yoshikazu Matsuda Yoshikazu Matsuda , MD 2 Department of Orthopedic Surgery, Matsuda Orthopedic Clinic, Saitama, Japan 3 Department of Pharmacy, Josai University, Saitama, Japan Find articles by Yoshikazu Matsuda 2, 3 , Fabrisia Ambrosio Fabrisia Ambrosio , PhD, MPT 4 Discovery Center for Musculoskeletal Recovery, Schoen Adams Research Institute at Spaulding, Charlestown, MA 5 Department of Physical Medicine & Rehabilitation, Harvard Medical School, Boston, MA 6 Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, MA Find articles by Fabrisia Ambrosio 4, 5, 6 , Hirotaka Iijima Hirotaka Iijima , PhD, PT 4 Discovery Center for Musculoskeletal Recovery, Schoen Adams Research Institute at Spaulding, Charlestown, MA 5 Department of Physical Medicine & Rehabilitation, Harvard Medical School, Boston, MA 6 Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, MA Find articles by Hirotaka Iijima 4, 5, 6, * Author information Article notes Copyright and License information 1 Department of Physical Therapy, Matsuda Orthopedic Clinic, Saitama, Japan 2 Department of Orthopedic Surgery, Matsuda Orthopedic Clinic, Saitama, Japan 3 Department of Pharmacy, Josai University, Saitama, Japan 4 Discovery Center for Musculoskeletal Recovery, Schoen Adams Research Institute at Spaulding, Charlestown, MA 5 Department of Physical Medicine & Rehabilitation, Harvard Medical School, Boston, MA 6 Department of Physical Medicine & Rehabilitation, Spaulding Rehabilitation Hospital, Charlestown, MA Contributors (1) AH and HI were the main contributors to the conception and design of the study; AH, TI, and YM were the main contributors to the acquisition of data; AH, FA, and HI contributed to the analysis and interpretation of data. (2) AH and HI drafted the manuscript, and AH, ED, FA, and HI critically revised the manuscript. (3) All authors have provided final approval for the version to be submitted. * Corresponding author: Hirotaka Iijima, PhD, PT, Discovery Center for Musculoskeletal Recovery, Schoen Adams Research Institute at Spaulding. Department of Physical Medicine & Rehabilitation, Harvard Medical School. 149 13th Street, Room 5.106, Charlestown, MA 02129. [email protected] Issue date 2026 Jun 1. PMC Copyright notice PMCID: PMC13091494  NIHMSID: NIHMS2161331  PMID: 41593828 The publisher's version of this article is available at Med Sci Sports Exerc Abstract Objectives: While sexual dimorphism of knee osteoarthritis (KOA) is well established, sex-specific clinical manifestations—particularly involving peri-articular tissues undetectable by radiography—remain underexplored. This study aimed to define female-specific alterations in joint integrity, peri-articular muscle quality, symptom presentation, and the transcriptomic landscape of peri-articular muscles, with the goal of uncovering the mechanistic contributions of each to KOA pathophysiology. Methods: Forty-nine participants (32 females, 17 males; Kellgren-Lawrence grade 1–2) underwent clinical assessment, including (1) quantitative ultrasound assessment of the vastus medialis (VM) and rectus femoris (RF) muscles; (2) MRI to assess joint integrity; and (3) patient-reported outcomes. Principal component analysis (PCA) followed by receiver operating characteristic curve analysis were conducted to identify discriminative sex-specific imaging and symptom features. Correlation-based network analysis examined sex-specific interdependencies among clinical variables. Publicly available transcriptomic datasets were analyzed to identify molecular drivers underlying female-specific muscle quality changes. Results: Despite similar radiographic severity and symptom presentation across the sexes, female individuals exhibited greater cartilage degeneration and higher fatty infiltration in the VM and RF. These features were central to sex separation in the PCA, with both features identified as network hubs in female individuals, indicating interconnected muscle-joint degeneration. Transcriptomic analysis revealed enrichment of adipogenic reprogramming in female individuals, suggesting aberrant intramuscular fat programming. Discussions: Our findings uncover a distinct female-specific musculoskeletal phenotype in early-stage KOA, characterized by muscle degeneration and cartilage deterioration undetectable by radiography. These female-specific clinical manifestations may be due, at least partly, to aberrant adipogenic programming in muscle. These findings provide mechanistic and clinical insight into sexual dimorphism in KOA. Keywords: Knee osteoarthritis, sex differences, intramuscular fat, cartilage degeneration, network analysis, transcriptomic analysis INTRODUCTION The 2019 Global Burden of Disease Study highlights a paradox in aging trajectories, which demonstrated that although females, on average, have a longer lifespan than males, females typically experience worse health outcomes than males as they age.( 1 ) One prominent example is knee osteoarthritis (KOA), a leading cause of disability among older adults. Postmenopausal females have a higher incidence of KOA and present with more severe disease progression over time.( 2 ) Despite the well-documented sexual dimorphism of KOA in disease progression, we know little about sex differences in clinical presentation. This is particularly true at an earlier stage of the disease, a critical window of opportunity to restore joint health before the onset of debilitating pain and functional limitations.( 3 ) This knowledge gap has arguably obstructed the development of effective rehabilitative strategies to prevent, attenuate, or reverse functional declines in the knee joint health of female aging population. Recent research has begun to shift attention from joint-level pathology alone to also include the surrounding periarticular tissues, including the quadriceps muscles, which are essential for maintaining joint stability and regulating load distribution.( 4 ) Age-related decline in skeletal muscle quality, characterized by increased fat deposition within muscle tissue, typically begins in the fourth decade of life and is exacerbated by immobilization.( 5 ) Fatty infiltration and degeneration of these muscles have emerged as key features associated with KOA progression.( 6 ) However, the extent to which these changes differ between sexes remains unclear. Furthermore, although transcriptomic analyses have identified pathways related to muscle degeneration and adipogenesis,( 7 ) few studies have directly connected these molecular profiles to sex-specific musculoskeletal phenotypes in KOA. To address this knowledge gap, the current study integrates multimodal assessments—including radiography, magnetic resonance imaging (MRI), musculoskeletal ultrasound, and transcriptomic profiling—to examine sex differences in intra- and periarticular tissue integrity among individuals with mild KOA. The objective of this investigation was to identify key structural and muscular features that distinguish male and female individuals. To this end, principal component analysis (PCA) and network analysis were applied. Additionally, we leveraged publicly available RNA-seq data to determine whether female-predominant muscle degeneration is driven by distinct molecular programs of activity in fibro adipogenic progenitors (FAP).( 8 ) Our findings offer new insights into the biological underpinnings of sex disparities in early KOA and point toward potential targets for sex-specific interventions. METHODS Subjects Individuals with KOA were recruited at the outpatient department of the Matsuda Orthopedic Clinic in Saitama, Japan, from June 2021 to October 2021. All recruited participants were previously seen in the orthopedic clinic for pain in one or both knees and had a diagnosis of KOA within 3 months. The presence of KOA was confirmed by an experienced orthopedic surgeon using the radiographically defined Kellgren and Lawrence grading scale (KL grade).( 9 ) Participants also underwent MRI of their symptomatic knee(s). The eligibility criteria were as follows: (1) at risk or with mild KOA (KL grade 1-2); (2) absence of osteonecrosis of the medial femoral condyle; (3) no history of surgery to the extremities; (4) ability to walk without an assistive device; (5) no cognitive dysfunction; (6) no neuromuscular disease; and (7) no trauma to bilateral lower extremities. All these criteria were assessed based on information in the medical records. Given that radiographically defined KL grade 1, but not grade 0, predicts development of KOA to a grade 2 or higher within 3–5 years, this study included individuals with KL grades 1 but excluded grade 0.( 10 , 11 ) This study was approved by the Ethics Committee of the Medical Corporation Nagomi (No. 20210427). All participants provided signed informed consent to participation in the studies included and publication of study results. The current cross-sectional study is reported following the “Strengthening the Reporting of Observational Studies in Epidemiology” (STROBE).( 12 ) Measurements Muscle quality of the RF and VM was quantified by ultrasound, and structural joint (e.g. articular cartilage, subchondral bone, and synovium) abnormalities were identified using MRI. These assessments were limited to the symptomatic knee(s), as defined by the presence of knee pain and/or functional limitations from medical examination by physician. Demographic characteristics, body mass index (BMI), ultrasound-based muscle thickness, and self-reported measures of knee pain and disability were also assessed as participant characteristics and/or covariates. Ultrasound imaging of RF and VM muscles Ultrasound images of the symptomatic knee(s) were acquired using a B-mode imaging device that was equipped with an 8-MHz linear-array probe (SNiBLE, Konica Minolta Ltd., Japan). Owing to the lack of evidence for specific ultrasound parameters in the assessment of muscle quality in the RF and VM muscles, ultrasound settings were optimized to visually discriminate the fascia of the RF and VM from the muscle fibers and other surrounding tissues (frequency: 9 MHz, gain: 25 dB, dynamic range: 60), in accordance with a previous study. These ultrasound settings were kept consistent across all participants, and images were taken in a relaxed supine position with the arms and legs extended. The RF was evaluated on the transverse line halfway between the anterior superior iliac spine and proximal end of patella while the VM was evaluated at the 5cm medial side from the 20% distal point of anterior superior iliac spine and proximal end of patella.( 13 ) The head of the probe was maintained perpendicular to the muscles examined. A water-soluble transmission gel (PROJELLY, Jex Inc., Japan) was applied to the skin to enhance acoustic coupling. EI of the captured ultrasound images was then assessed using ImageJ software (version 1.53, National Institutes of Health, USA).( 14 ) Regions of interest were selected at depths of 1.0-4.0 cm from the surface of skin on the screen display. Echo intensity normalized by subcutaneous fat thickness The EI of the analysis region was calculated using grayscale values from 0 (black) to 255 (white), with a higher mean pixel intensity value indicating fatty infiltration and/or fibrosis tissue (i.e., poor muscle quality).( 15 , 16 ) All ultrasound imaging and ImageJ analyses were performed by a trained investigator (AH). To assess the reliability of the imaging procedure, the same investigator performed another assessment more than six months after the first assessment, revealing an excellent intraclass correlation coefficient (ICC 1,1 : 0.948, 95% confidence interval [CI]: 0.912–0.970).( 17 ) To account for the effect of subcutaneous fat thickness on echo measurements, a correction factor was applied to echo measurements,( 15 ) as recommended in several studies.( 18 , 19 ) The formula for the corrected EI was as follows: y 2 = y 1 + ( x * c f ) with y 2 = corrected echo intensity; y 1 = raw echo intensity; x = subcutaneous fat thickness; and cf = correction factor of 40.5278. Subcutaneous fat thickness was measured from the superior border of the superficial aponeurosis to the inferior border of the dermis layer, per the established protocol validated by MRI and computed tomography.( 15 , 20 ) In addition, as a covariate, muscle thickness (cm) was measured from the cortex of the femur to the superior border of the superficial aponeurosis of the RF and VM muscles. Assessment of structural abnormalities in the knee joint by MRI Structural abnormalities in the knee joint were identified using MRI (1.5 Tesla whole-body MR system, Canon Medical Systems, Japan). Images were obtained as sagittal proton-density-weighted images (3600 msec TR (Repetition Time), 30 msec TE (Echo Time), 256×384 matrix, 3.5 mm slice thickness, 0.3 mm inter-slice gap, 180 mm×180 mm field of view, TSE fact (Turbo Spin Echo fact): 15, scan time: 3 min, frequency encoding: anterior-posterior). We used a semi-quantitative scoring method of the whole-organ magnetic resonance imaging score (WORMS),( 21 ) which has been used previously for individuals with early KOA.( 22 ) WORMS (0 – 332 points) includes the following categories: (1) cartilage (0 – 84 points), (2) bone marrow lesion (0 – 45 points), (3) subarticular cysts (0 – 45 points), (4) subarticular bone attrition (0 – 42 points), (5) osteophytes (0 – 98 points), (6) medial / lateral meniscal integrity (0 – 12 points), (7) anterior / posterior cruciate ligament and medial / lateral ligament integrity (0 – 3 points), and (8) synovitis (0 – 3 points). The whole knee joint was divided into the medial tibiofemoral joint, lateral tibiofemoral joint, and patellofemoral joint. The femoral articular surface, medial tibial plateau and lateral tibial plateau were divided into three sub-regions (anterior, central, posterior). The tibia has an additional subspinous sub-region comprised of the non-articulating portion of the tibial plateau beneath the tibial spines. The assessment was performed by the same investigator (AH) who had undergone three weeks of formal training by an experienced orthopedic surgeon and radiologist prior to the start of the study. The investigator read MRI images in a random order with a blinded subject ID. At the second evaluation of WORMS by the same investigator, more than six months after the initial evaluation, the intraclass correlation coefficient was good (ICC 1,1 : 0.849, 95%CI: 0.752, 0.910).( 17 ) Participant characteristics and covariates Data on participant age, sex, and height were self-reported in the electronic medical records. Body mass was measured on a digital scale with the participants dressed but not wearing shoes. BMI was calculated by dividing body mass (kg) by height (m 2 ). A knee radiograph of the symptomatic knee in the anteroposterior view in the weight-bearing position was obtained for all subjects. Using this image, we assessed the femur-tibia angle (FTA) as a measure of the anatomical axis (higher values indicate more varus alignment). The center of the FTA was defined from three points: a point at the base of the tibial spines, at bisecting the femur and tibia, originating 10 cm from the knee joint surface.( 23 ) FTA was then corrected to account for sex differences in knee alignment, which was calculated by the addition of 3.5° for women and 6.4° for men, respectively.( 24 ) This corrected FTA was used for the subsequent statistical analyses as an alternative measure of full-limb radiograph in weight-bearing position.( 25 ) To quantify knee symptoms, all subjects completed the Knee Injury and Osteoarthritis Outcome Score (KOOS).( 26 ) The KOOS has the following subcategories: pain, symptoms, activities of daily living (ADL), sport and recreation function, and knee-related quality of life, with higher scores (0-100) representing better function. This study used the KOOS pain and ADL subscales for descriptive analysis. Principal component analysis (PCA) Considering the complexity and interrelationships of various outcomes of individuals with KOA, we performed principal component analysis (PCA) to identify the main patterns. Each clinical outcome was standardized by setting the mean to 0 and the standard deviation to 1. Principal components were extracted from the standardized values of 23 outcomes, and those explaining up to 80% of the cumulative variance were retained. The resulting principal component scores were then compared between males and females to assess potential sex differences.( 27 , 28 ) Differential co-expression network analysis Differential co-expression network analysis is a systems-level approach used to compare co-expression relationships between two or more groups. In this study, we applied this method to examine sex-specific patterns in clinical outcomes of individuals with KOA. To construct the networks, we used the dcanr R package to compute differential co-expression scores between males and females based on Spearman correlations.( 27 ) Changes in the strength of correlations between variables across groups were statistically tested and corrected for multiple comparisons using the false discovery rate (FDR), with a threshold of 0.05 applied to retain only statistically significant connections. Variables with many significant differential connections, referred to as network hubs, may represent key features that are more interconnected in one group than the other. More specifically, examining each variable independently is like assessing isolated joint movements—such as hip extension or knee flexion—without observing how they coordinate during gait. In contrast, network analysis captures how these elements interact as an integrated system, revealing patterns of coordination that might underlie disease mechanism.( 28 ) In this context, a network hub can be thought of as a joint that coordinates multiple movements to maintain smooth gait patterns; however, these hubs reflect statistical centrality rather than proven causality. In this study, we further related these network-level changes to clinical variables using sex-stratified linear regression analyses, including interaction terms to formally test whether the associations differed between males and females. Together, this pipeline provides a detailed view of how clinical feature relationships differ by sex, potentially uncovering novel therapeutic targets or stratification markers in knee osteoarthritis. RNA-sequencing data processing and analysis We accessed publicly available transcriptomic data c Data preprocessing and differential gene expression were performed in accordance with the established workflow.( 7 , 29 ) Raw count data were normalized by counts per million (CPM), and genes with low expression were removed using the filterByExpr function from edgeR with a minimum count of 10, ensuring inclusion of genes with sufficient biological and statistical relevance.( 30 ) To focus on protein-coding genes, we retrieved gene annotation from the Ensembl database and selected only protein-coding genes located on autosomal chromosomes ( 1 – 22 ), reducing potential confounding from sex chromosome dosage while acknowledging that this excludes sex-linked transcriptional effects. The remaining genes were further normalized using Trimmed Mean of M-values (TMM) and transformed with voom to model the mean-variance relationship for linear modeling. Differential expression analysis between female and male subjects was performed using the limma R package,( 31 ) with significance defined as the Benjamini–Hochberg FDR < 0.05 and |log2 fold change| > 1. To identify biological processes associated with differentially expressed genes, we performed gene ontology (GO) biological process enrichment analysis. Gene symbols were converted to ENTREZ IDs using the org.Hs.eg.db database. Enrichment analysis was performed using the enrichGO function from the clusterProfiler package, with Benjamini–Hochberg correction and a q-value cutoff of 0.05. Gene set enrichment analysis (GSEA) was performed using the GSEA web tool provided by the Broad Institute ( https://www.gsea-msigdb.org/gsea/index.jsp ). For input, genes were ranked by t values derived from the limma differential expression analysis of RNA-sequencing data from skeletal muscle across individuals with KOA and age-matched healthy adults ( GSE242202 ). The genes associated with FAP adipogenesis, we originally defined,( 7 ) were used as a gene set. The minimum and maximum criteria for selection of gene sets from the collection were 10 and 500 genes, respectively. The transcriptomic responses of female elderly skeletal muscle to the immobilization protocol (i.e., 5-day bed rest) relative to baseline control ( GSE113165 ) were analyzed in the same manner using GSEA, following the procedures described in our previous work.( 7 ) Statistical analysis To assess whether the main principal components reflected sex differences, we performed receiver operating characteristic (ROC) analysis and calculated the area under the curve (AUC). A sex difference was considered clinically meaningful if the AUC exceeded 0.75.( 32 ) The minimum sample size required to detect significance was calculated from pilot data of individuals with KOA (n = 20) using Power and Sample Size Program, version 3.1.6 (Vanderbilt University Medical Center, USA).( 33 ) From preliminary data with a power of 0.9 and type I error probability of 0.05, we identified that 40 subjects would be needed to detect a significant relationship in the PC1 score between female and male. After considering the potential 10% dropout rate due to the exclusion criteria and missing data, the required sample size of this study was determined to be 45 participants. To assess the robustness of the PCA, as a post-hoc analysis, Leave-One-Out-Cross-Validation (LOOCV) was performed. In LOOCV, models were repeatedly built after the removal of one subject at a time.( 34 ) We assessed the variability in sex classification performance, measured by the AUC, during LOOCV. Additionally, to further support the robustness of the PCA results given the limited sample size, we conducted sparse principal component analysis (sPCA).( 35 ) Unlike conventional PCA, which minimizes least square error, sPCA introduces L1 regularization (Lasso regression) to reduce dimensionality by shrinking small loadings to zero. This approach allows for the selection of the most stable and interpretable variables contributing to each principal component, thereby enhancing the reliability of the analysis under limited sample conditions.( 36 ) Also, we conducted down-sampling analysis for sample size imbalance between female and male individuals. The validation result was calculated by repeating down-sampling of the majority class 1,000 times. This method can reduce the risk of overfitting due to sample size imbalance.( 37 ) All statistical analyses were performed using EZR, version 1.54 (Saitama Medical Center, Jichi Medical University, Japan), which is a graphical user interface for R, version 4.0.3 (The R Foundation for Statistical Computing, Austria), a modified version of R commander designed to add statistical functions frequently used in biostatistics. Statistical significance was set at p <0.05. RESULTS Females exhibited more severe muscle-joint abnormalities, despite similar radiographic disease presentation A total of 51 participants were initially recruited for this study, but two individuals were excluded due to moderate to severe KOA (i.e., KL grade 3). The final cohort comprised 49 individuals (52 knees) who underwent comprehensive assessments including radiography, MRI, and ultrasound. Participants ranged in age from 44 to 78 years, with females comprising 64.3% of the sample. All individuals exhibited either unilateral or bilateral at-risk or mild KOA. Table 1 summarizes the participants’ characteristics. Of the 49 participants with mild KOA, 30 (61%) had a normal BMI (BMI 18.5 to 24.9 kg/m 2 ), 14 (29%) were pre-obese (BMI 25.0 to 29.9 kg/m 2 ), and 5 (10%) were obese (BMI <30 kg/m 2 ). Of the 52 knees, 39 (75%) had varus alignment (corrected FTA >181°), 12 (23%) had neutral alignment (178° < corrected FTA <181°), and 1 (2%) had valgus alignment (corrected FTA <178°), according to the alignment categories previously established.( 24 ) Compared to males, females demonstrated less pronounced varus alignment but exhibited more advanced structural joint abnormalities, particularly in cartilage ( Table 1 ). Additionally, the periarticular muscles (VM and RF) showed higher echo intensity values (i.e., greater fatty infiltration) in females after adjusting for subcutaneous fat thickness. In contrast, muscle thickness was greater in males, though this difference was observed only in the RF ( Table 1 ). Table 1. Participants characteristics (49 participants with 52 symptomatic knees) Variable Total Female (n = 34) Male (n = 18) p-value Person-level characteristics Age, years 64.15 ± 8.87 64.47 ± 8.66 63.56 ± 9.48 0.727 BMI, kg/m 2 24.70 ± 3.67 24.81 ± 4.18 24.51 ± 2.61 0.783 The period from the initial visit to assessment Number of days, days 33 (0-832) 33 (0-832) 34 (0-404) 0.729 Frequency of visits, times 5 (1-51) 5 (1-51) 5 (1-27) 0.884 Knee-level characteristics KL grade (grade 1 - 4) Grade 1 (n) 13 8 (23.5%) 5 (27.8%) 0.747 Grade 2 (n) 39 26 (76.5%) 13 (72.2%) Cor.FTA, degree 182.68 ± 2.58 181.43 ± 1.97 185.03 ± 1.85 <0.001 § KOOS, points Pain (0 -100) 39.28 ± 20.94 36.31 ± 17.42 44.92 ± 25.98 0.170 ǂ Symptoms (0 -100) 42.05 ± 23.69 40.80 ± 22.81 44.43 ± 25.77 0.627 ǂ ADL (0 -100) 32.87 ± 26.36 30.19 ± 23.33 37.94 ± 31.39 0.324 ǂ Sport/Recreation (0 -100) 46.54 ± 23.02 46.91 ± 23.96 45.83 ± 21.78 0.847 ǂ QOL (0 -100) 52.50 ± 14.31 52.89 ± 15.59 51.75 ± 11.89 0.649 ǂ WORMS, points Total (0 - 332) 19.76 ±9.63 22.18 ± 10.24 15.19 ± 6.41 0.008 ǂ Cartilage (0 - 84) 9.61 ± 5.03 10.68 ± 4.75 7.58 ± 5.06 0.032 ǂ BML (0 - 45) 1.90 ± 3.66 2.47 ± 4.26 0.83 ± 1.76 0.136 ǂ Bone cyst (0 - 45) 0.79 ± 1.09 0.91 ±1.11 0.56 ± 1.04 0.178 ǂ Osteophyte (0 - 98) 1.94 ± 2.82 2.27 ± 3.0 1.33 ± 2.40 0.192 ǂ Menisci (0 - 12) 2.63 ± 0.95 2.68 ± 1.04 2.56 ± 0.78 0.677 ǂ Bone attrition (0 - 42) 0.96 ± 1.34 0.61 ± 0.92 1.15 ± 1.50 0.187 ǂ Ligaments (0 - 3) 0 ± 0 0 0 - Synovitis (0 - 3) 1.94 ± 1.05 2.06 ± 1.10 1.72 ± 0.96 0.278 ǂ Cor. Echo Intensity, VM (0 - 255) 138.30 ± 29.18 150.01 ± 27.73 116.17 ± 16.4 <0.001 ǂ Arbitrary Units (A.U.) RF (0 - 255) 140.56 ± 26.66 151.10 ± 24.05 120.65 ± 19.20 <0.001 ǂ Muscle thickness, cm VM 1.77 ± 0.44 1.73 ± 0.42 1.86 ± 0.50 0.328 ǂ RF 1.54 ± 0.30 1.47 ± 0.28 1.67 ± 0.30 0.023 ǂ Subcutaneous fat thickness, cm VM 1.29 ± 0.63 1.56 ± 0.58 0.78 ± 0.31 <0.001 ǂ RF 1.26 ± 0.57 1.48 ± 0.55 0.87 ± 0.34 <0.001 ǂ Open in a new tab These values are presented as mean ± SD, while the median period from the initial visit to assessment is presented (min-max). Score of KOOS pain and KOOS ADL are percentage scores from 0 to 100. For KOOS pain, a score of 100 represents no pain, and pain increases as scores approach 0. KOOS ADL with 100 representing no disability and 0 representing maximum disability. p -value of KL grade and the period from the initial visit to assessment were calculated by Fisher’s exact test and Wilcoxon rank sum test, respectively. p -values of age and BMI obtained by t-test. p -value of corrected FTA and the other were obtained by analysis of covariance (ANCOVA) adjusted for KL grade ( § ) and age ( ǂ ), respectively. Bold text presents statistical significance. Abbreviations: ADL, activities of daily living; BMI, body mass index; BML, bone marrow lesions; Cor.echo intensity, corrected echo intensity; Cor.FTA, corrected femur-tibia angle; KL grade, Kellgren, and Lawrence grading scale; KOOS, Knee Injury and Osteoarthritis Outcome Score; RF, rectus femoris; SD, standard deviation. VM, vastus medialis; WORMS, whole-organ magnetic resonance imaging score. To interrogate the multidimensional interplay between structural, muscular, and symptomatic features, PCA was performed. PCA is a dimensionality reduction technique that transforms a large set of correlated variables into a smaller set of uncorrelated variables called principal components (PCs). This approach allows us to identify patterns in the data and highlight the variables that contribute most significantly to the observed differences between sexes. In this study, we standardized values from 23 outcomes related to joint structure, muscle characteristics, and clinical symptoms. Of note, WORMS ligament sub-score was excluded from the analysis due to zero value for all participants. PCA demonstrated that the first two PCs accounted for 41% of the total variance in the dataset. Notably, PC1 alone explained a substantial portion of the variance and revealed a clear separation between males and females with KOA ( Figure 1A ), indicating prominent sex differences along this axis. To further assess the discriminatory power of PC1 and PC2 regarding sex differences, we performed ROC analysis. PC1 showed a high area under the curve (AUC) of 0.84 (95% CI: 0.73–0.95), confirming its strong capacity to differentiate between males and females. In contrast, PC2 had a modest AUC of 0.61 (95% CI: 0.45–0.78), suggesting limited discriminatory power ( Figure 1B ). PC1 predominantly captures degenerative alterations in intra-articular and periarticular tissues, and these high PC1 contributors were distributed differently between males and females ( Figure 1C ). In contrast, PC2 was driven primarily by subjective symptom metrics such as KOOS pain, ADL, and symptom sub-scores, indicating its alignment with patient-reported outcome measures ( Figure 1D ). Figure 1. Sex differences in clinical presentation in individuals with KOA. Open in a new tab A , PCA from 23 outcome variables showing the distinct clinical presentation between females (n = 34) and males (n = 18). B , ROC analysis to classify females and males with KOA based on both PC1 and PC2. AUCs and their 95% confidence intervals are provided. Sensitivity, p (c), and specificity, q (c), in Youden index points are also provided. The lower one-side 95% CI of PC1 values was equivalent to 0.750 which is considered as clinically acceptable.( 65 ) C , Heatmap of the PC1 scores, displaying the variables in descending order of their contribution to PC1. The highest contributions to PC1 are shown at the top, with “VM: Cor.EI” having the greatest contribution to PC1. The top six variables were highlighted to emphasize their dominant influence on PC1. D , Biplot of PC1 and PC2 component loadings, showing the relationship between the principal components and their respective variables. PC1, which demonstrated significant sex differences, was primarily driven by factors related to disease (i.e., cartilage degeneration) and periarticular joint tissue integrity. In contrast, PC2, which did not show sex differences, was predominantly influenced by subjective symptoms and disability measures. The component loading values of PC1 and PC2 are expressed in absolute values in the plot. The top five to six variables contributing to each component were highlighted to facilitate interpretation. Abbreviations: Abs, absolute value; ADL, activities of daily life; BMI, body mass index; FTA, femoral tibia angle, KL grade, kellgren-lawrence grade; KOA, knee osteoarthritis; KOOS, Knee Injury and Osteoarthritis Outcome Score; MRI, magnetic resonance imaging; PCA, principal component analysis; PC, principal component; WORMS, whole-organ magnetic resonance imaging score. To ensure the robustness of PCA results, we used LOOCV to investigate how much the sex classification performance, measured by AUC, fluctuates when samples are excluded. The results showed minimal fluctuation of AUC, and the classification performance based on PC1 remained stable (mean AUC = 0.84, range of AUC = 0.82 – 0.85, SD = 0.004). This suggests that PC1 may have a consistent ability to separate groups. After examining the variables in PC1 of LOOCV that had a large mean component loading, it was also observed that they were the same variables identified during PCA (corrected EI of VM; 0.34, WORMS total; 0.33, Subcutaneous fat thickness of VM; 0.32, corrected EI of RF; 0.29, Subcutaneous fat thickness of RF; 0.27). Additionally, we complemented the analysis with sPCA to compensate for the small sample size, which applies L1 regularization to identify the most stable and interpretable variables contributing to each principal component. Using this approach, the sPCA-derived PC1 (sPC1) consisted of 21 variables with nonzero loadings, all of which were consistently selected across 500 bootstrap iterations (stability ≥ 0.9), demonstrating high stability. The sPC1 score effectively discriminated between males and females with KOA, achieving an area under the curve (AUC) of 0.838 (permutation p <0.001). Importantly, the high-loading variables identified by sPCA—such as WORMS total, corrected echo intensity of VM, and subcutaneous fat thickness of VM for sPC1, and KOOS Pain, KOOS Symptom, and KOOS ADL for sPC2—closely matched those identified by conventional PCA, supporting the robustness and interpretability of the PCA-derived structure. Furthermore, to address the sample size imbalance (female in 34 knees, male in 18 knees), we performed down-sampling analysis by randomly matching the group size (18 vs. 18) 1,000 times from sPCA and AUC estimation for each iteration. The mean AUC was calculated as 0.84, which is equal to the original result (PCA;0.84, sPCA;0.84). This result showed the discriminative performance of PC1 was not driven by sample imbalance. Network analysis indicated periarticular muscle as a central to the sex differences Next, to develop targeted and effective early intervention strategies for females—who are known to be at a higher risk of KOA progression—it is crucial to identify which of these sex-specific features are most strongly linked to disease development. While prior analyses revealed sex-based differences in individual structural and muscular features, such findings alone do not capture how these features interact within the broader musculoskeletal system. To address this gap, we aimed to uncover sex-specific patterns of inter-variable relationships, with the goal of identifying potential drivers of disease progression unique to females. For this purpose, we employed differential co-expression network analysis, a method that allows for the comparison of correlation patterns between groups—in this case, between males and females. Unlike traditional correlation analyses that assess the strength of association between two variables across all individuals, differential co-expression focuses on whether the relationships between variables change depending on individual characteristics (e.g., sex). This method constructs networks where nodes represent variables (e.g., joint structure, muscle quality, symptoms) and edges represent the strength and direction of associations between them. By contrasting these networks between sexes, we can identify which connections are gained, lost, or inverted, thereby revealing sex-specific interaction patterns within the musculoskeletal system ( Figure 2A ). Such patterns may provide mechanistic insights and help prioritize targets for intervention. Figure 2. Network paradigm to map sex-dependent inter-relationships among clinical outcomes. Open in a new tab A , Schematic representation of the network paradigm, differential clinical manifestation network , to map distinct inter-relationships among clinical outcomes between females and males. B , RF and VM were identified as key hubs in the network. Of note, the correlation network does not imply causal relationships between variables. Instead, it identifies undirected associations, meaning that the edges only represent correlations without indicating the direction of influence. Each node represents a variable derived from WORMS and ultrasound measurements. C , Female-specific relationship between corrected EI in the RF and VM. Linear regression lines are drawn using simple regression analysis. D , Female-specific relationship between age and corrected EI in the VM. Linear regression lines are drawn using simple regression analysis. Abbreviations: Corr.EI, corrected echo intensity; RF, rectus femoris; VM, vastus medialis WORMS, whole-organ magnetic resonance imaging score. Notably, both the corrected echo intensities of RF and VM were identified as network hubs in the differential co-expression network, along with the WORMS total score, suggesting that degeneration in these muscles and overall joint structural abnormalities may be centrally involved in mediating sex-specific differences in early KOA ( Figure 2B ). In particular, the corrected echo intensity of RF exhibited the highest centrality (betweenness = 51.5), which quantifies how frequently a variable serves as a bridge linking other components of the network. This finding implies that the RF may act as a central connector, mediating the interplay between muscle and joint changes that characterize the early degenerative process. To further explore this, we examined the relationship between RF and VM and how this association varies by sex. Using a linear regression model with VM as the outcome and RF as the predictor, we found a significant positive association between RF and VM in females ( Figure 2C ), whereas no significant association was observed in males ( Figure 2C ). The stronger coupling between RF and VM degeneration in females suggests potential targets for sex-specific interventions aimed at mitigating disease progression. In the co-expression network, various factors showed edges to corrected echo intensity, reflecting potential sex-dependent associations with muscle quality. As an example, age was also connected to the corrected echo intensity of VM in the network. To examine whether age-related changes in corrected echo intensity of VM are sex-dependent, we again conducted a linear regression analysis using VM as the outcome and age as the predictor. This analysis revealed a significant age-associated increase in corrected echo intensity in females, but not in males ( Figure 2D ). These findings suggest a female-specific vulnerability of VM to age-related degenerative changes, which may further contribute to the elevated risk of KOA progression in females. Female quadriceps muscle exhibited a distinct transcriptomic response related to adipogenesis of fibroadipogenic progenitor similar to that seen after immobilization To investigate whether female-predominant fatty infiltration (i.e., increased EI) in the quadriceps muscle is associated with sex-specific transcriptomic responses, we performed a differential gene expression analysis using publicly available RNA-seq data derived from the vastus lateralis muscle of individuals with KOA ( GSE242202 ; n = 37; mean age = 72 years; 86.5% female).( 38 ) Raw expression data were background-corrected, quantile-normalized, and log2-transformed. Genes with low expression across samples were filtered out. We focused on protein-coding genes and excluded those located on sex chromosomes to avoid sex-linked expression bias. After preprocessing, differential expression analysis between males and females was conducted, yielding 8,533 genes for further analyses ( Figure 3A ). Using thresholds of adjusted p <0.05 and |log 2 fold change| >1, we identified a subset of genes that were differentially expressed between females and males ( Figure 3B ). Notably, Prg4 , Ncam1 , and Olfm1 were among the significantly downregulated genes in females, suggesting their potential involvement in sex-specific molecular responses in quadriceps muscle in individuals with KOA. To gain functional insights into the sex-biased transcriptomic profile, we next performed GO enrichment analysis using the differentially expressed genes. Among the top enriched terms, mesenchymal cell differentiation emerged as a prominent category ( Figure 3C ). This finding is particularly relevant given the role of mesenchymal progenitor cells in skeletal muscle remodeling and fatty infiltration. Figure 3. Transcriptomic signatures reveal sex-related differences in KOA and links to FAP adipogenesis. Open in a new tab A , Schematic representation of analytical flow for bulk RNA-seq from symptomatic KOA (GSE GSE242202 ). B , Volcano plot highlighting the transcriptomic shift in females compared to male counterparts. C , GO enrichment analysis of the differentially expressed genes (adjusted p <0.05 and |log 2 fold change| >1). D , GSEA to investigate the relationship between sex-related transcriptomic response and FAP adipogenesis genes. The impact of immobilization protocol on the FAP adipogenesis genes is also provided as a reference line (orange dotted line). E , Graphical abstract summarizing sex-specific periarticular muscle and joint degeneration in KOA. Females exhibit greater fatty infiltration and adipogenic transcription in quadriceps muscles, along with increased cartilage damage, compared with males, despite similar radiographic disease severity. Abbreviations: FAPs, fibro-adipogenic progenitors; FC, fold change; FDR, false discovery rate; GSEA, gene set enrichment analysis; NES, normalized enrichment score. Over the last decade, numerous studies have provided important insights into the cellular origin of fatty infiltration in skeletal muscle.( 39 – 41 ) Fatty infiltration is driven, at least in part, by muscle-resident fibro-adipogenic progenitors (FAPs), a type of mesenchymal progenitor cells within skeletal muscle.( 39 – 41 ) While FAPs are key regulators of muscle regeneration and homeostasis, they also contribute to chronic inflammation, fibrosis, and fat deposition when dysregulated.( 39 – 41 ) Damage to skeletal muscle triggers a transient phase of FAP proliferation that promotes muscle stem cell differentiation, followed by a return to baseline numbers as a result of apoptosis and subsequent phagocytic clearance.( 42 , 43 ) However, in the setting of chronic muscle inflammation, as is the case with aged muscle,( 41 , 44 ) FAP proliferation is maintained at high levels owing to impaired apoptosis and removal of dead cells via phagocytosis.( 41 ) The result is persistence and adipogenic differentiation of FAPs, thereby inducing the buildup of intra- and inter-muscular adipose tissue. ( 39 – 41 ) Although no study to our knowledge has identified genes that drive FAP adipogenic differentiation, Wosczyna et al showed that miR-206 mimicry in vivo limits intra-muscle fat accumulation, while miR-206 inhibition by siRNA increased the adipogenic differentiation of FAPs in vitro .( 45 ) Given that these previous experiments were designed to elucidate the fate switch from FAPs to adipocytes, investigators evaluated gene expression before mature adipocytes were formed.( 45 ) Accordingly, our own processing of the RNA-seq data by Wosczyna et al identified 161 differentially expressed genes (false discovery rate <0.05) that may drive adipocytic differentiation of FAPs.( 7 ) Using these 161 genes defined, we then performed gene set enrichment analysis (GSEA) to mechanistically determine the link between female skeletal muscle and FAP adipogenesis.( 46 ) GSEA is a computational tool that provides insight into biological processes or pathways underlying a given phenotype.( 47 ) We found that female and male participants displayed distinct transcriptomic responses related to FAP adipogenesis ( Figure 3D ). Among the leading-edge genes—a subset of genes that contributes most to the enrichment score and drives the association between the gene set and phenotype—Prg4 emerged as the top-ranked contributor to the enrichment observed in female skeletal muscle. This is particularly compelling given that previous work has identified a Prg4 + FAP subpopulation with potent adipogenic potential,( 48 ) suggesting that sex differences in FAP behavior may reflect differential expansion or activation of Prg4 + FAPs. Intriguingly, the impact of female sex on FAP adipogenesis had a similar effect comparable to that seen after immobilization protocol, a known contributor of fatty infiltration ( Figure 3D ).( 49 , 50 ) Together, the findings suggest that the FAP adipogenic program is promoted in female quadriceps muscle, which is in line with the ultrasound imaging as documented by increased corrected EI. DISCUSSION This study demonstrated that females at risk for or with mild KOA exhibit more severe joint structural abnormalities and greater fatty infiltration of periarticular muscles than males, despite similar radiographic disease severity (see Graphical Abstract in Figure 3E ). PCA revealed clear sex differences, with females scoring higher on components reflecting intra- and periarticular degeneration. Differential co-expression network analysis identified corrected EI of the RF and VM as key hubs associated with sex-specific differences, showing stronger associations in females. Regression analyses further revealed that VM EI increased with age in females but not in males, indicating a sex-specific vulnerability. Transcriptomic analysis of quadriceps muscle supported these findings, showing a female-biased gene expression profile enriched for pathways involved in adipogenic differentiation of FAPs—a pattern resembling that induced by immobilization. Importantly, this study integrates a series of clinical functional and imaging outcomes with transcriptomics to uncover a female-specific trajectory of muscle and joint degeneration in earlier stage of KOA. Our findings highlight a previously unrecognized mechanistic link between periarticular muscle quality decline and joint vulnerability in females, providing new insight into the sexual dimorphism of KOA pathogenesis. This work advances our understanding of clinical manifestations in a sex-dependent manner and offers a potential biomarker-guided framework for developing targeted, sex-specific intervention strategies. While radiographic findings appeared similar between males and females with early-stage KOA, our multimodal assessment revealed that females exhibited more advanced degenerative changes in both intra-articular joint structures and periarticular muscles. This discrepancy suggests that standard radiographs may fail to capture early tissue-level differences, particularly in females. In clinical practice, radiography remains the most commonly used diagnostic tool, yet it often overlooks subtle soft tissue changes that occur in the early phases of KOA. By incorporating ultrasound assessments, we identified greater fatty infiltration in the quadriceps muscles in females. These changes were not evident on radiographs but likely contribute to functional decline. Our findings are consistent with epidemiological evidence showing that females are at higher risk of KOA progression, especially after menopause.( 51 ) While prior studies have focused primarily on cartilage degradation, symptom severity, or muscle weakness,( 52 – 54 ) our results suggest that intra-muscular fat accumulation may emerge earlier and may be a key factor driving sex-specific differences in disease trajectory. Our network-based approach provided novel systems-level insights that extend beyond what traditional group comparisons can reveal. While PCA helped identify that PC1—largely defined by imaging markers of joint and muscle degeneration—clearly separated males and females, the network approach revealed how these markers are interconnected. Specifically, we found that, in females, echo intensity values of the RF and VM formed densely connected hubs, exhibiting stronger correlations with both structural joint damage and functional outcomes than in males. These findings suggest that RF and VM degeneration may act as integrative nodes, linking joint pathology with muscle quality in a sex-specific manner. Unlike prior studies that considered muscle and joint metrics in parallel,( 55 , 56 ) our findings highlight that their interrelationships—not just their individual values—may be key to understanding early sex differences in KOA. These findings highlight the potential of network-based methodologies, offering a powerful framework to identify critical nodes for targeted intervention. Our transcriptomic analysis adds a new molecular dimension to understanding sexual dimorphism in early-stage KOA, revealing a female-specific gene expression profile in quadriceps muscle that is particularly enriched for pathways related to FAP adipogenesis. While previous studies have identified chronic inflammation as a driver of FAP proliferation and fat accumulation in aging muscles,( 8 , 57 ) few have explored how these processes differ between sexes. Notably, our gene set enrichment analysis revealed that the FAP adipogenic program was more upregulated in female muscle, to a degree similar to what is observed in immobilization—a known inducer of fatty infiltration.( 50 ) Although the RNA-seq data were obtained from a public database distinct from the cohort undergoing ultrasound assessment, both cohorts share key characteristics of symptomatic KOA patients, particularly older age (60–75 years old) and female predominance. Therefore, the transcriptomic findings can be interpreted in the context of clinical data, while acknowledging that information regarding other factors, such as comorbidities and medication use, was not available and may confound outcomes. This finding not only supports our ultrasound data, showing greater fatty infiltration in females, but also introduces a novel potential molecular mechanism by which sex may influence early KOA progression through maladaptive muscle remodeling. Several limitations of our study warrant consideration. First, cross-sectional design limits our ability to draw causal inferences about the relationship between muscle quality, joint degeneration, and functional outcomes. Previous longitudinal study reported the quadriceps adiposity associated with greater odds of MRI degeneration such as cartilage loss or bone marrow lesions over 3 years.( 58 ) However, as these studies did not focus on sex differences, further longitudinal research that takes sex differences into account is necessary to determine whether early degeneration in the quadriceps, particularly in the RF and VM, precedes or accelerates KOA progression in female individuals. Second, while we incorporated multiple imaging modalities and transcriptomic analyses, our relatively modest sample size—especially for the molecular data—may limit the generalizability of our findings. Future studies should aim to validate these results in larger, more diverse cohorts, including postmenopausal women and individuals at various stages of KOA. Third, while our network-based approach illuminated inter-tissue relationships, it does not account for potential systemic factors such as hormonal fluctuations, metabolic status, or inflammatory cytokines, all of which could mediate sex differences in musculoskeletal degeneration.( 59 – 61 ) Lastly, it is important to note that we focused our analysis on the RF and VM muscles of the thigh, as these have been commonly measured in previous studies.( 62 ) However, we cannot ignore the possibility that other thigh muscles might also play significant roles in muscle remodeling and degeneration in a sex-specific manner. Further exploration of muscle-specific roles in other regions of the thigh would help clarify these potential relationships. CONCLUSION In summary, our multi-modal, network-driven analysis revealed that females with early-stage KOA exhibit interconnected degeneration of periarticular muscles and joint structures compared to males, with the RF and VM emerging as central nodes in this pathological network. These imaging-based findings were further supported by a female-specific molecular signature consistent with FAP-driven adipogenesis in quadriceps muscle. Together, these results suggest that early maladaptive muscle remodeling may contribute to the heightened vulnerability of females to KOA progression. While differences in skeletal muscle mass and muscle activation are typically considered the main contributors to sex differences in muscle strength,( 60 , 63 ) our findings suggest that fatty infiltration in the quadriceps, which is linked to FAP-driven adipogenesis and resembles changes observed during immobilization, also contributes to sex differences in KOA progression. Early monitoring muscle degeneration in female could expand opportunities for timely interventions such as nutritional management and exercise.( 7 , 64 ) By uncovering sex-specific network patterns and molecular pathways, this study provides a foundation for developing targeted, sex-informed rehabilitation strategies that address not only joint preservation but also muscle integrity in the early stages of KOA. Acknowledgments We thank the members of the Physical Therapy and Radiology Department of Matsuda Orthopedic Clinic for extending support in data collection. The figures were created using BioRender.com . Footnotes Competing interests The authors have no conflicts of interest to declare. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine. References 1. Global Burden of Disease Collaborative Network. Global Burden of Disease Study 2019 (GBD; 2019). Available from: https://vizhub.healthdata.org/gbd-results/ . [ Google Scholar ] 2. Hame SL, Alexander RA. Knee osteoarthritis in women. Curr Rev Musculoskelet Med. 2013;6(2):182–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 3. Mahmoudian A, Lohmander LS, Mobasheri A, Englund M, Luyten FP. Early-stage symptomatic osteoarthritis of the knee — time for action. Nature Reviews Rheumatology. 2021;17(10):621–32. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Øiestad BE, Juhl CB, Eitzen I, Thorlund JB. Knee extensor muscle weakness is a risk factor for development of knee osteoarthritis. A systematic review and meta-analysis. Osteoarthritis Cartilage. 2015;23(2):171–7. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Nair KS. Aging muscle. Am J Clin Nutr. 2005;81(5):953–63. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Pedroso MG, Castilho De Almeida A, Jéssica ·, et al. Fatty infiltration in the thigh muscles in knee osteoarthritis: a systematic review and meta-analysis. Rheumatol Int. 2019;39:627–35. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Iijima H, Ambrosio F, Matsui Y. Network-based systematic dissection of exercise-induced inhibition of myosteatosis in older individuals. J Physiol. 2025;603(1):45–67. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Chen W, You W, Valencak TG, Shan T. Bidirectional roles of skeletal muscle fibro-adipogenic progenitors in homeostasis and disease. Ageing Res Rev. 2022;80:101682. [ DOI ] [ PubMed ] [ Google Scholar ] 9. KELLGREN JH, LAWRENCE JS. Radiological Assessment of Osteo-Arthrosis. Ann Rheum Dis. 1957;16(4):494. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Hart DJ, Spector TD. Kellgren & Lawrence grade 1 osteophytes in the knee—doubtful or definite? Osteoarthritis Cartilage. 2003;11(2):149–50. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Cibere J, Sayre EC, Guermazi A, et al. Natural history of cartilage damage and osteoarthritis progression on magnetic resonance imaging in a population-based cohort with knee pain. Osteoarthritis Cartilage. 2011;19(6):683–8. [ DOI ] [ PubMed ] [ Google Scholar ] 12. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: Guidelines for reporting observational studies. International Journal of Surgery. 2014;12(12):1495–9. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Taniguchi M, Fukumoto Y, Kobayashi M, et al. Quantity and Quality of the Lower Extremity Muscles in Women with Knee Osteoarthritis. Ultrasound Med Biol. 2015;41(10):2567–74. [ DOI ] [ PubMed ] [ Google Scholar ] 14. Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source platform for biological-image analysis. Nature Methods. 2012;9(7):676–82. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Young H-J, Jenkins NT, Zhao Q, Mccully KK. Measurement of intramuscular fat by muscle echo intensity. Muscle Nerve. 2015;52(6):963–71. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Pillen S, Tak RO, Zwarts MJ, et al. Skeletal Muscle Ultrasound: Correlation Between Fibrous Tissue and Echo Intensity. Ultrasound Med Biol. 2009;35(3):443–6. [ DOI ] [ PubMed ] [ Google Scholar ] 17. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med. 2016;15(2):155–63. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Oranchuk DJ, Stock MS, Nelson AR, Storey AG, Cronin JB. Variability of regional quadriceps echo intensity in active young men with and without subcutaneous fat correction. Applied Physiology, Nutrition, and Metabolism. 2020;45(7):745–52. [ Google Scholar ] 19. Burton AM, Stock MS. Consistency of novel ultrasound equations for estimating percent intramuscular fat. Clin Physiol Funct Imaging. 2018;38(6):1062–6. [ Google Scholar ] 20. Watanabe Y, Ikenaga M, Yoshimura E, Yamada Y, Kimura M. Association between echo intensity and attenuation of skeletal muscle in young and older adults: A comparison between ultrasonography and computed tomography. Clin Interv Aging. 2018;13:1871–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Peterfy CG, Guermazi A, Zaim S, et al. Whole-Organ Magnetic Resonance Imaging Score (WORMS) of the knee in osteoarthritis. Osteoarthritis Cartilage. 2004;12(3):177–90. [ DOI ] [ PubMed ] [ Google Scholar ] 22. Hada S, Kaneko H, Sadatsuki R, et al. The degeneration and destruction of femoral articular cartilage shows a greater degree of deterioration than that of the tibial and patellar articular cartilage in early stage knee osteoarthritis: a cross-sectional study. Osteoarthritis Cartilage. 2014;22(10):1583–9. [ DOI ] [ PubMed ] [ Google Scholar ] 23. McDaniel G, Mitchell KL, Charles C, Kraus VB. A comparison of five approaches to measurement of anatomic knee alignment from radiographs. Osteoarthritis Cartilage. 2010;18(2):273–7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Ohi H, Iijima H, Aoyama T, Kaneda E, Ohi K, Abe K. Association of frontal plane knee alignment with foot posture in patients with medial knee osteoarthritis. BMC Musculoskelet Disord. 2017;18(1):1–10. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Kraus VB, Vail TP, Worrell T, McDaniel G. A comparative assessment of alignment angle of the knee by radiographic and physical examination methods. Arthritis Rheum. 2005;52(6):1730–5. [ DOI ] [ PubMed ] [ Google Scholar ] 26. Roos EM, Lohmander LS. The Knee injury and Osteoarthritis Outcome Score (KOOS): From joint injury to osteoarthritis. Health Qual Life Outcomes. 2003;1(1):1–8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. Bhuva DD, Cursons J, Smyth GK, Davis MJ. Differential co-expression-based detection of conditional relationships in transcriptional data: Comparative analysis and application to breast cancer. Genome Biol. 2019;20(1):1–21. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Iijima H, Ambrosio F. Network Medicine as a Tool to Enhance Regenerative Rehabilitation Practice. Am J Phys Med Rehabil. 2025;104(3):271–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Iijima H, Zhang F, Ambrosio F, Matsui Y. Network-based cytokine inference implicates Oncostatin M as a driver of an inflammation phenotype in knee osteoarthritis. Aging Cell. 2024;23(2). [ Google Scholar ] 30. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26(1):139–40. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Ritchie ME, Phipson B, Wu D, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 32. Fan J, Upadhye S, Worster A. Understanding receiver operating characteristic (ROC) curves. Canadian Journal of Emergency Medicine. 2006;8(1):19–20. [ DOI ] [ PubMed ] [ Google Scholar ] 33. Dupont WD, Plummer WD. Power and Sample Size Calculations for Studies Involving Linear Regression. Control Clin Trials. 1998;19(6):589–601. [ DOI ] [ PubMed ] [ Google Scholar ] 34. Vehtari A, Gelman A, Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Stat Comput. 2017;27(5):1413–32. [ Google Scholar ] 35. Zou H, Hastie T, Tibshirani R. Sparse Principal Component Analysis. Journal of Computational and Graphical Statistics. 2006;15(2):265–86. [ Google Scholar ] 36. Greenacre M, Groenen PJF, Hastie T, D’Enza AI, Markos A, Tuzhilina E. Principal component analysis. Nature Reviews Methods Primers. 2022;2(1):1–21. [ Google Scholar ] 37. ElRafey A, Wojtusiak J. Recent advances in scaling-down sampling methods in machine learning. Wiley Interdiscip Rev Comput Stat. 2017;9(6):e1414. [ Google Scholar ] 38. Kurochkina NS, Orlova MA, Vigovskiy MA, et al. Age-related changes in human skeletal muscle transcriptome and proteome are more affected by chronic inflammation and physical inactivity than primary aging. Aging Cell. 2024;23(4):e14098. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Uezumi A, Fukada SI, Yamamoto N, Takeda S, Tsuchida K. Mesenchymal progenitors distinct from satellite cells contribute to ectopic fat cell formation in skeletal muscle. Nat Cell Biol. 2010;12(2):143–52. [ DOI ] [ PubMed ] [ Google Scholar ] 40. Heredia JE, Mukundan L, Chen FM, et al. Type 2 innate signals stimulate fibro/adipogenic progenitors to facilitate muscle regeneration. Cell. 2013;153(2):376–88. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Uezumi A, Ito T, Morikawa D, et al. Fibrosis and adipogenesis originate from a common mesenchymal progenitor in skeletal muscle. J Cell Sci. 2011;124(Pt 21):3654–64. [ DOI ] [ PubMed ] [ Google Scholar ] 42. Mozzetta C, Consalvi S, Saccone V, et al. Fibroadipogenic progenitors mediate the ability of HDAC inhibitors to promote regeneration in dystrophic muscles of young, but not old Mdx mice. EMBO Mol Med. 2013;5(4):626–39. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Lemos DR, Babaeijandaghi F, Low M, et al. Nilotinib reduces muscle fibrosis in chronic muscle injury by promoting TNF-mediated apoptosis of fibro/adipogenic progenitors. Nat Med. 2015;21(7):786–94. [ DOI ] [ PubMed ] [ Google Scholar ] 44. Wang J, Leung KS, Chow SKH, Cheung WH. Inflammation and age-associated skeletal muscle deterioration (sarcopaenia). J Orthop Translat. 2017;10:94–101. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Wosczyna MN, Perez Carbajal EE, Wagner MW, et al. Targeting microRNA-mediated gene repression limits adipogenic conversion of skeletal muscle mesenchymal stromal cells. Cell Stem Cell. 2021;28(7):1323–1334.e8. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Barbie DA, Tamayo P, Boehm JS, et al. Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1. Nature. 2009;462(7269):108–12. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545–50. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. Fitzgerald G, Turiel G, Gorski T, et al. MME+ fibro-adipogenic progenitors are the dominant adipogenic population during fatty infiltration in human skeletal muscle. Communications Biology. 2023;6(1):1–21. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Belavý DL, Möhlig M, Pfeiffer A, Felsenberg D, Armbrecht G. Preferential deposition of visceral adipose tissue occurs due to physical inactivity. Int J Obes (Lond). 2014;38(11):1478–80. [ DOI ] [ PubMed ] [ Google Scholar ] 50. Manini TM, Clark BC, Nalls MA, Goodpaster BH, Ploutz-Snyder LL, Harris TB. Reduced physical activity increases intermuscular adipose tissue in healthy young adults. Am J Clin Nutr. 2007;85(2):377–84. [ DOI ] [ PubMed ] [ Google Scholar ] 51. Srikanth VK, Fryer JL, Zhai G, Winzenberg TM, Hosmer D, Jones G. A meta-analysis of sex differences prevalence, incidence and severity of osteoarthritis. Osteoarthritis Cartilage. 2005;13(9):769–81. [ DOI ] [ PubMed ] [ Google Scholar ] 52. Tschon M, Contartese D, Pagani S, Borsari V, Fini M. Gender and Sex Are Key Determinants in Osteoarthritis Not Only Confounding Variables. A Systematic Review of Clinical Data. Journal of Clinical Medicine. 2021;10(14):3178. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Glass N, Segal NA, Sluka KA, et al. Examining sex differences in knee pain: the Multicenter Osteoarthritis Study. Osteoarthritis Cartilage. 2014;22(8):1100–6. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Ding C, Cicuttini F, Blizzard L, Scott F, Jones G. A longitudinal study of the effect of sex and age on rate of change in knee cartilage volume in adults. Rheumatology. 2007;46(2):273–9. [ DOI ] [ PubMed ] [ Google Scholar ] 55. Cai G, Jiang M, Cicuttini F, Jones G. Association of age, sex and BMI with the rate of change in tibial cartilage volume: A 10.7-year longitudinal cohort study. Arthritis Res Ther. 2019;21(1):1–11. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 56. Øiestad BE, Juhl CB, Culvenor AG, Berg B, Thorlund JB. Knee extensor muscle weakness is a risk factor for the development of knee osteoarthritis: an updated systematic review and meta-analysis including 46 819 men and women. Br J Sports Med. 2022;56(6):349–55. [ DOI ] [ PubMed ] [ Google Scholar ] 57. Molina T, Fabre P, Dumont NA. Fibro-adipogenic progenitors in skeletal muscle homeostasis, regeneration and diseases. Open Biol. 2021;11(12). [ Google Scholar ] 58. Kumar D, Link TM, Jafarzadeh SR, LaValley MP, Majumdar S, Souza RB. Association of Quadriceps Adiposity With an Increase in Knee Cartilage, Meniscus, or Bone Marrow Lesions Over Three Years. Arthritis Care Res (Hoboken). 2021;73(8):1134–9. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Tanamas SK, Wijethilake P, Wluka AE, et al. Sex hormones and structural changes in osteoarthritis: A systematic review. Maturitas. 2011;69(2):141–56. [ DOI ] [ PubMed ] [ Google Scholar ] 60. Szilagyi IA, Waarsing JH, Van Meurs JBJ, Bierma-Zeinstra SMA, Schiphof D. A systematic review of the sex differences in risk factors for knee osteoarthritis. Rheumatology. 2023;62(6):2037–47. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. Perruccio A V, Badley EM, Power JD, et al. Sex differences in the relationship between individual systemic markers of inflammation and pain in knee osteoarthritis. Osteoarthr Cartil Open. 2019;1(1–2):100004. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 62. Hoki A, Amico ED’, Ambrosio F, Iwasaki T, Matsuda Y, Iijima H. Fat infiltration in the vastus medialis implicates joint structural abnormalities in early-stage symptomatic knee osteoarthritis. medRxiv. 2023;2022.05.27.22275636. [ Google Scholar ] 63. Segal NA, Nilges JM, Oo WM. Sex differences in osteoarthritis prevalence, pain perception, physical function and therapeutics. Osteoarthritis Cartilage. 2024;32(9):1045–53. [ DOI ] [ PubMed ] [ Google Scholar ] 64. Gong Y, Yang Y, Zhang X, Tong L. Comparative efficacy of exercise, nutrition, and combined exercise and nutritional interventions in older adults with sarcopenic obesity: a protocol for systematic review and network meta-analysis. Syst Rev. 2025;14(1):77. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Fan J, Upadhye S, Worster A. Understanding receiver operating characteristic (ROC) curves. CJEM. 2006;8(1):19–20. [ DOI ] [ PubMed ] [ Google Scholar ] ACTIONS View on publisher site PDF (1.6 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