Advertisement

Intravoxel Incoherent Motion Diffusion-Weighted MR Imaging Findings of Infrapatellar Fat Pad Signal Abnormalities: Comparison Between Symptomatic and Asymptomatic Knee Osteoarthritis

Open AccessPublished:January 04, 2023DOI:https://doi.org/10.1016/j.acra.2022.11.010

      Rationale and objectives

      Infrapatellar fat pad (IPFP) proton density–weighted images (PdWI) hyperintense regions on MRI are an important imaging feature of knee osteoarthritis (KOA) and are thought to represent inflammation which may induce knee pain. The aim of the study was to compare the intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) findings of PdWI hyperintense regions of IPFP between symptomatic and asymptomatic KOA and to determine whether IVIM-DWI parameters can be used as an objective biomarker for symptomatic KOA.

      Materials and Methods

      In total, 84 patients with symptomatic KOA, 43 asymptomatic KOA persons, and 30 healthy controls with MRI were retrospectively reviewed. Demographic, IPFP-synovitis, Western Ontario and McMaster Osteoarthritis Index (WOMAC) pain sub-score, IPFP volume and depth and quantitative parameters of IVIM-DWI were collected. The chi-square test, Binary logistic regression and receiver operating characteristic curve (ROC) analysis were used for diagnostic performance comparison.

      Results

      The IPFP volume and depth were statistically significant differences between the non-KOA and sKOA groups (p<0.05). The IPFP PdWI hyperintense regions demonstrated significantly higher values of D and D* in the symptomatic KOA compared to those in the asymptomatic KOA (1.51±0.47 vs. 1.73±0.40 for D and 19.24±6.44 vs. 27.09±9.75 for D*) (both p<0.05). Multivariate logistic regression analyses showed that Higher D and D* values of IPFP hyperintense region were significantly associated with higher risks of knee pain (OR: 1.97; 95% CI: 1.21-3.19; p=0.006 for D and OR: 1.24; 95% CI: 1.09-1.41; p=0.001 for D*). Sensitivity and specificity of D value for symptomatic KOA were 80.28% and 83.33%, with an AUC of 0.78 (0.68-0.86). D* value had the sensitivity with 92.96% and a specificity of 58.33%, with an AUC of 0.82 (0.73-0.89) for symptomatic KOA.

      Conclusion

      IVIM-DWI can be used as an additional functional imaging technique to study IPFP with signal abnormalities on PdWI, and the D and D* values may have potential value to predict the symptom in mild-to-moderate KOA patients.

      Key Words

      Abbreviations:

      IVIM-DWI (Intravoxel incoherent motion diffusion-weighted imaging), KOA (Knee osteoarthritis), sKOA (symptomatic knee osteoarthritis), asKOA (asymptomatic knee osteoarthritis), IPFP (Infrapatellar fat pad), PdWI-HR (Proton density-weighted imaging hyperintense region), NSIR (Normal signal intensity region), WOMAC (the Western Ontario and McMaster Universities Index), KLG (Kellgren-Lawrence grade), BMI (Body mass index), VAS (Visual analogue scale), T1WI (T1-weighted image), 3D-DESS (3-dimensional double-echo steady-state), JSN (joint space narrowing), ROI (Region of interest), MOAKS (the MRI osteoarthritis knee score)

      INTRODUCTION

      Knee osteoarthritis (KOA) represents one of the most common chronic degenerative diseases in the elderly, with a prevalence estimated more than 50% in individuals over 65 years old (
      • Hunter DJ
      • Bierma-Zeinstra S.
      Osteoarthritis.
      ). Knee pain is the most common and disabling symptom, and is typically mainly weight-bearing (mechanical) pain, with an incidence of 36.8-60.7%, resulting in the declined functional status and quality of life of patients (
      • van der Heijden RA
      • de Vries BA
      • Poot DHJ
      • et al.
      Quantitative volume and dynamic contrast-enhanced MRI derived perfusion of the infrapatellar fat pad in patellofemoral pain.
      ,
      • Arendt-Nielsen L.
      Pain sensitisation in osteoarthritis.
      ). The infrapatellar fat pad (IPFP) is an intra-capsular, yet extra-synovial, adipose tissue in the anterior part of the knee joint between the lower edge of the patella, femoral condyle, tibial plateau, and patellar tendon (
      • Ballegaard C
      • Riis R
      • Bliddal H
      • et al.
      Knee pain and inflammation in the infrapatellar fat pad estimated by conventional and dynamic contrast-enhanced magnetic resonance imaging in obese patients with osteoarthritis: a cross-sectional study.
      ,
      • Zeng N
      • Yan ZP
      • Chen XY
      • et al.
      Infrapatellar fat pad and knee osteoarthritis.
      ).Recent studies have confirmed that the IPFP may have a more specific function due to its rich arterial blood supply and sensory innervations (
      • Nemschak G
      • Pretterklieber M.
      The patellar arterial supply via the infrapatellar fat pad (of Hoffa): a combined anatomical and angiographical analysis.
      ,
      • Leese J
      • Davies D.
      An investigation of the anatomy of the infrapatellar fat pad and its possible involvement in anterior pain syndrome: a cadaveric study.
      ).Reports have shown that the IPFP can be infiltrated by immune cells and secrete inflammatory mediators, with microscopic changes, such as thickening of interlobular septa, fibrotic changes and vascularization (
      • Favero M
      • El-Hadi H
      • Belluzzi E
      • et al.
      Infrapatellar fat pad features in osteoarthritis: a histopathological and molecular study.
      ,
      • Macchi V
      • Stocco E
      • Stecco C
      • et al.
      The infrapatellar fat pad and the synovial membrane: an anatomo-functional unit.
      ,
      • An JS
      • Tsuji K
      • Onuma H
      • et al.
      Inhibition of fibrotic changes in infrapatellar fat pad alleviates persistent pain and articular cartilage degeneration in monoiodoacetic acid-induced rat arthritis model.
      ).Furthermore, MRI has indicated that the IPFP with higher blood perfusion is a source of knee pain (
      • van der Heijden RA
      • de Vries BA
      • Poot DHJ
      • et al.
      Quantitative volume and dynamic contrast-enhanced MRI derived perfusion of the infrapatellar fat pad in patellofemoral pain.
      ).
      MRI is widely used to assess the pathological changes of KOA, including the cartilage, bone marrow, ligaments, and meniscus (
      • Hunter DJ
      • Bierma-Zeinstra S.
      Osteoarthritis.
      ,
      • Solivetti F
      • Guerrisi A
      • Salducca N
      • et al.
      Appropriateness of knee MRI prescriptions: clinical, economic and technical issues.
      ,
      • Su X
      • Zhang Y
      • Gao Q
      • et al.
      Preliminary study on the assessment of early cartilage degeneration by quantitative ultrashort echo time magnetic resonance imaging in vivo.
      ).The most common feature of the IPFP on MRI is the presence of proton density–weighted image hyperintense region (PdWI-HR), and the increased signal intensity in the IPFP which is thought to be a manifestation of inflammation is associated with knee structural abnormalities in osteoarthritis (
      • Han W
      • Aitken D
      • Zheng S
      • et al.
      Association between quantitatively measured infrapatellar fat pad high signal-intensity alteration and magnetic resonance imaging-assessed progression of knee osteoarthritis.
      ,
      • Klein-Wieringa I
      • Bd Lange-Brokaar
      • Yusuf E
      • et al.
      Inflammatory cells in patients with endstage knee osteoarthritis: a comparison between the synovium and the infrapatellar fat pad.
      ).To date, studies have mainly concentrated on the morphological description or semi-quantitative assessment of the IPFP (
      • Cai J
      • Xu J
      • Wang K
      • et al.
      Association between infrapatellar fat pad volume and knee structural changes in patients with knee osteoarthritis.
      ,
      • Okita Y
      • Oba H
      • Miura R
      • et al.
      Movement and volume of infrapatellar fat pad and knee kinematics during quasi-static knee extension at 30 and 0° flexion in young healthy individuals.
      ).A previous study showed that Individuals with patellofemoral joint osteoarthritis had a larger IPFP than controls, and IPFP volume was directly related to patellofemoral joint OA pain (
      • Cowan S
      • Hart HF
      • Warden SJ
      • et al.
      Infrapatellar fat pad volume is greater in individuals with patellofemoral joint osteoarthritis and associated with pain.
      ). In contrast, a recent cross-sectional study showed that no volume variation of IPFP in KOA (
      • He J
      • Ba H
      • Feng J
      • et al.
      Increased signal intensity, not volume variation of infrapatellar fat pad in knee osteoarthritis: a cross-sectional study based on high-resolution magnetic resonance imaging.
      ).In the MRI osteoarthritis knee score (MOAKS), according to the presence and size of hyperintense signals within the IPFP, semi-quantitative scores on PdWI could be obtained (
      • Hunter DJ
      • Guermazi A
      • Lo GH
      • et al.
      Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score).
      ). Intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) can yield a variety of quantitative data, where D is the water diffusion coefficient in the tissue(true diffusion coefficient), D* is the water diffusion coefficient in blood (pseudo-diffusion coefficient), and f is the flowing blood fraction (perfusion fraction), which can provide information on tissue microcirculation and blood flow in addition to information on tissue microstructure from diffusion MRI without the need for exogenous contrast agents (
      • Le Bihan D
      What can we see with IVIM MRI?.
      ). IVIM-DWI is thus widely used to investigate diseases affecting various systems in the body (
      • Cao J
      • Gao S
      • Zhang C
      • et al.
      Differentiating atypical hemangiomas and vertebral metastases: a field-of-view (FOV) and FOCUS intravoxel incoherent motion (IVIM) diffusion-weighted imaging (DWI) study.
      ). However, it is still unclear how these signal alterations in IPFP are related to perfusion and/or diffusion changes.
      Thus, the present study aimed to quantitatively evaluate differences in IVIM-DWI parameters between the PdWI-HR and adjacent normal signal intensity region (PdWI-NSIR) in patients with symptomatic and asymptomatic KOA. Our hypothesis was that PdWI-hyperintense region demonstrated a higher degree of diffusion and/or perfusion in symptomatic KOA.

      MATERIALS AND METHODS

      This retrospective study was approved by the local ethics committee, in accordance with the Declaration of Helsinki (No. SZFYIEC-PJ-2019[3+]), and all patients signed the informed consent forms.

      Study Population

      From October 2019 to October 2021, a total of 157 participants were enrolled in the study. Inclusion criteria: (
      • Hunter DJ
      • Bierma-Zeinstra S.
      Osteoarthritis.
      ) clinical diagnosis performed following the standard for KOA by the European League Against Rheumatism (
      • Zhang W
      • Doherty M
      • Peat G
      • et al.
      EULAR evidence-based recommendations for the diagnosis of knee osteoarthritis.
      ), (
      • van der Heijden RA
      • de Vries BA
      • Poot DHJ
      • et al.
      Quantitative volume and dynamic contrast-enhanced MRI derived perfusion of the infrapatellar fat pad in patellofemoral pain.
      ) Kellgren-Lawrence grade (KLG) 2-3, (
      • Arendt-Nielsen L.
      Pain sensitisation in osteoarthritis.
      ) age ≥40 years old, and (
      • Ballegaard C
      • Riis R
      • Bliddal H
      • et al.
      Knee pain and inflammation in the infrapatellar fat pad estimated by conventional and dynamic contrast-enhanced magnetic resonance imaging in obese patients with osteoarthritis: a cross-sectional study.
      ) BMI<30 kg/m2.
      Exclusion criteria: (
      • Hunter DJ
      • Bierma-Zeinstra S.
      Osteoarthritis.
      ) a history of knee trauma (<6 months), malignant tumor, and drug treatment or intra-articular injection (<6 months) or even knee surgery, (
      • van der Heijden RA
      • de Vries BA
      • Poot DHJ
      • et al.
      Quantitative volume and dynamic contrast-enhanced MRI derived perfusion of the infrapatellar fat pad in patellofemoral pain.
      ) any inflammatory arthritis (tuberculosis, suppurative arthritis, gouty arthritis, rheumatoid arthritis, etc.), (
      • Arendt-Nielsen L.
      Pain sensitisation in osteoarthritis.
      ) hyperthyroidism or hypothyroidism, acute or chronic renal failure, (
      • Ballegaard C
      • Riis R
      • Bliddal H
      • et al.
      Knee pain and inflammation in the infrapatellar fat pad estimated by conventional and dynamic contrast-enhanced magnetic resonance imaging in obese patients with osteoarthritis: a cross-sectional study.
      ) any history of continuous intake of arthritis causing medications, (
      • Zeng N
      • Yan ZP
      • Chen XY
      • et al.
      Infrapatellar fat pad and knee osteoarthritis.
      ) any family history or hereditary cases of osteoarthritis, (
      • Nemschak G
      • Pretterklieber M.
      The patellar arterial supply via the infrapatellar fat pad (of Hoffa): a combined anatomical and angiographical analysis.
      )contraindication for MRI examination. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) was acquired by questionnaires to assess the knee pain (comprised five questions), which was evaluated by the Visual Analog Scale (VAS) score (0-10 points) (
      • Salaffi F
      • Leardini G
      • Canesi B
      • et al.
      Reliability and validity of the western ontario and McMaster universities (WOMAC) osteoarthritis index in italian patients with osteoarthritis of the knee.
      ). The symptomatic KOA was defined as the presence of unilateral or bilateral knee pain of any subscale score ≥4 points (knee discomfort for at least one month in the past 12 months). Finally, 84 patients with symptomatic KOA (sKOA group), 43 asymptomatic KOA persons (asKOA group), and 30 healthy volunteers were allocated to the non-KOA group. A flowchart for the study population is presented in Figure 1.
      Figure 1
      Figure 1Flowchart of patients’ selection process. KLG, Kellgren-Lawrence Grade, WOMAC, the Western Ontario and McMaster Universities Index, OA, osteoarthritis.
      Age, sex, and the affected limb of all participants were recorded. Height and weight were measured in light clothing without shoes, and BMI (weight/height2, kg/m2) was calculated. Measurements were made according to the Western Ontario and McMaster Universities Index (WOMAC), which evaluates pain, stiffness, and physical dysfunction by using a 0-10 points’ visual analogue scale (VAS). The scale comprised24 questions, including five on pain evaluation, two on joint stiffness, and 17 on physical function; each question ranged from 0 point (no pain, joint stiffness, or physical dysfunction) to 10 points (severe pain, joint stiffness, or limitation of physical function). Finally, the total WOMAC score was 0-240, including pain (0-50), stiffness (0-20) and physical function (0-170) (
      • Salaffi F
      • Leardini G
      • Canesi B
      • et al.
      Reliability and validity of the western ontario and McMaster universities (WOMAC) osteoarthritis index in italian patients with osteoarthritis of the knee.
      ).

      Imaging Acquisition

      All subjects underwent weight-bearing anteroposterior and lateral knee radiography. The affected knees were fully extended for weight-bearing anteroposterior radiography, aligning the centerline with the lower edge of the patella. Patients stood with the knees flexed to 30° for lateral radiographs and aligning the centerline to the midpoint of the line between the lower edge of the patella and the popliteal fossa. The focal film's distance was 110 cm, and the auto-exposure mode was adopted.
      All subjects underwent MRI on a 3-T MRI scanner (MAGNETOM Skyra; Siemens AG, Erlangen, Germany) and using an identical MRI protocol with a dedicated eight-channel transmit-receive knee coil. All participants were placed with the knee in about 15° flexion in the supine position when scanned and were asked to sit and rest for 30 minutes prior to the scan. Conventional T1-weighted image (T1WI), three-dimensional double-echo steady-state (3D-DESS) sequence, and fat-suppressed proton density-weighted fast spin echo (FS-PdWI-FSE) sequence were first obtained for anatomical and morphological assessments of the knee joint. IVIM-DWI sequence was obtained using the single-shot spin-echo echo-planar imaging (SS-SE-EPI), with gradient reversal fat suppression. The MRI sequences and parameters are summarized in Table 1.
      TABLE 1The Scanning Parameters for MRI Sequences
      ParametersT1WI3D-DESSPdWIIVIM
      AcquisitionSagittalSagittalSagittalSagittal
      Contrast agentN/AN/AN/AN/A
      Repetition time (ms)36614.823403700
      Echo time (ms)8.754175
      b value (s/mm2)N/AN/AN/A0, 50, 100, 150, 200, 400, 600, 800
      Averages2121,1,1,3,3,3,5,5
      DirectionsN/AN/AN/A4
      Flip angle120°25°120°180°
      Slice thickness (mm)4.50.64.54.5
      Matrix size269 × 384240 × 256208 × 32096 × 96
      Voxel size0.5 × 0.5 × 4.50.6 × 0.6 × 0.60.6 × 0.6 × 4.51.5 × 1.5 × 4.5
      Field of view (mm)180 × 180180 × 180180 × 180180 × 180
      Slices181601818
      Scan time1min39s3 min43s1min52s7min18s
      T1WI, T1-weighted imaging; 3D-DESS, 3-dimensional double-echo steady-state; PdWI, proton density-weighted; IVIM, Intravoxel incoherent motion.

      Image Analysis

      All images were independently interpreted by two radiologists (WB and DH) with 3 and 8 years of experience in musculoskeletal imaging, respectively, who were blinded to patients and controls’ data.
      The knee score was determined by consensus of the two readers with reference to the atlas according to KLG: grade 0, normal; grade 1, no joint space narrowing(JSN) and suspicious osteophytes; grade 2, suspicious JSN and mild osteophytes; grade 3, definite JSN, moderate osteophytes, and/or subchondral bone sclerosis; grade 4, marked JSN, large osteophytes, and/or severe subchondral bone sclerosis (
      • Kellgren J
      • Lawrence J.
      Radiological assessment of osteo-arthrosis.
      ).
      The 3D-DESS images were imported into the ITK-SNAP software (version 3.8.0, http://www.itksnap.org/). The regions of interest (ROIs) were manually drawn the IPFP boundary layer by layer using ITK-SANP software by one radiologist (TH) with 7 years of experience in musculoskeletal imaging until all IPFP was included. IPFP volume was computed by the software program. The maximal extension of the IPFP from anterior to posterior was calculated as the IPFP depth (Fig 2). The segmentation results were then confirmed by an expert radiologist specializing in musculoskeletal imaging diagnosis. In cases of disagreement, consensus was reached through consultation to avoid marking beyond the edge of the IPFP.
      Figure 2
      Figure 2Manual segmentation of the IPFP in sagittal 3-D double-echo steady-state sequence images. a IPFP depth was defined as the maximal extension of the IPFP from anterior to posterior. b The volume of the IPFP was measured by manually drawing contours around the IPFP on a section-by-section 3D-DESS. c 3-dimensional reconstruction of a complete IPFP.
      According to the MOAKS(19), IPFP synovitis, defined as discrete areas of increased signal within IPFP, was scored semi-quantitatively and assessed for magnitude as follows: grade 0 = normal (none); grade 1 = mild (<10% of the region); grade 2 = moderate (10-20% of the region); and grade 3 = severe (>20% of the region).
      IVIM-DWI images were imported into the post-processing software (Body Diffusion Toolbox; Siemens AG, Germany). Two ROIs were delineated in the IPFP on IVIM-DWI images, one within the PdWI-HR and the other in an adjacent area within the PdWI-normal signal intensity region (PdWI-NSIR). ROIs were put manually in IPFP on b=0 DWI map which had a better SNR, then ROIs were copied automatically to the maps of D, D* and f maps (Fig 3). Finally, the D, D* and f values were computed by the software program. When multiple hyperintense regions were found in the IPFP, the ROI was only placed in the largest region, avoiding cystic degeneration, loose bodies, and normal high signal vessels in the IPFP. One ROI was drawn in the IPFP of whom without PdWI hyperintensity. Intra- and inter-observer reproducibility was assessed in randomly selected 30 subjects (15 right and 15 left knees) who were not included in this study. Two experienced observers performed the image analyses for inter-observer variability. One observer repeated the measurements after 1 month to test intra-observer variability.
      Figure 3
      Figure 3A 63-year-old male patient with KOA, KLG = 3, and WOMAC-pain score =38. The abnormal signal intensity change is evident in the anterosuperior portion of the IPFP on transverse and sagittal PdWI (a and b). Two ROIs were drawn in the PdWI-hyperintense region and adjacent normal signal region respectively. ROIs were put manually in IPFP on b=0 DWI map which had a better SNR (3c), then ROIs were copied automatically to the maps of D, D* and f maps (3d–f). The D, D*, and f value of PdWI-HR and adjacent PdWI-NSIR were1.93 × 10−3 mm2/s vs. 0.92 × 10−3 mm2/s, 26.33 × 10−3 mm2/s vs. 20.09 × 10−3 mm2/s, and 0.34 vs. 0.29, respectively.

      Statistical Analysis

      The Kolmogorov–Smirnov test was used to assess the normality of continuous variables. The independent-samples t-test was used to compare patients’ characteristics. Intraclass correlation coefficient (ICC) was used to assess the intra- and inter-observer agreement of IVIM parameters. Differences in IVIM parameters among the three groups were compared using analysis of covariance (ANCOVA) followed by the least-significant difference (LSD) post-hoc test. Univariate and multivariate binary logistic regression analyses were used to assess the associations between IPFP volume, IPFP depth, and IVIM parameters (independent variables) and knee pain (dependent variable) before and after adjustment for age, BMI, MOAKS-IPFP synovitis, and KLG. The receiver operating characteristic (ROC) curve was used to analyze the area under the curve (AUC), sensitivity, and specificity of quantitative values for diagnosing sKOA using knee pain as a reference. Statistical analysis was performed with SPSS 26.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism 6.0 (GraphPad Software Inc., San Diego, CA, USA) software. p<0.05 was considered statistically significant.

      RESULTS

      Patients’ Characteristics

      Eighty-four symptomatic KOA patients (30 males and 54 females, mean age 58.3 ± 7.6 years), 43 asymptomatic KOA persons (18 males and 25 females, mean age 56.6 ± 6.3 years) and 30 healthy controls (15 males and 15 females, mean age 45.5 ± 6.7 years) were involved in this study. The prevalence of the PdWI-hyperintense IPFP regions were different between the groups: 71 of 84 (85%) symptomatic KOA patients and 24 of 43 (56%) asymptomatic KOA patients. There were significant differences in KLG and MOAKS-IPFP synovitis between the sKOA and asKOA groups (p<0.05). Differences between pain, stiffness, physical function, and total score of WOMAC between the sKOA and asKOA groups were statistically significant (p<0.05).All characteristics of the subjects are presented in Table 2.
      TABLE 2Demographic and Clinical Data
      CharacteristicNon-KOA (n=30)asKOA (n=43)sKOA (n=84)p
      Sex (M/F)14/1618/2530/540.536
      Age (yrs)45.5±6.756.6±6.358.3±7.6p1<0.001

      p2<0.001

      p3=0.207
      Height (m)1.68±0.0681.66±0.0711.64±0.067p1=0.207

      p2=0.002

      p3=0.062
      Weight (kg)63.01±7.2670.83±11.3470.62±9.87p1=0.001

      p2<0.001

      p3=0.908
      BMI (kg/m2)22.30±2.0525.64±3.1726.34±2.78p1<0.001

      p2<0.001

      p3=0.175
      Lateral (R/L)15/1523/2045/390.941
      Symptom duration (yrs)N/AN/A11.6±5.3N/A
      WOMAC
       PainN/A9.70±3.7518.65±4.900.000
       StiffnessN/A4.88±2.867.15±3.400.000
       Physical functionN/A43.26±17.1556.27±20.140.000
       Total scoreN/A57.84±20.3182.08±26.480.000
      KLG
       KLG-0 (%)11 (36.67)0 (0)0 (0)0.001
       KLG-1 (%)19 (63.33)0 (0)0 (0)
       KLG-2 (%)0 (0)31 (72.09)35 (41.67)
       KLG-3 (%)0 (0)12 (27.91)49 (58.33)
      MOAKSIPFP- synovitis
       grade 0 (%)30 (100)19 (44.19)13 (15.48)0.000
       grade 1 (%)015 (34.88)14 (16.67)
       grade 2 (%)05 (11.63)29 (34.52)
       grade 3 (%)04 (9.30)28 (33.33)
      Means ± SD are indicated, the significance level was p <0.05.p1 = Non-KOA vs asKOA; p2 = Non-KOA vs sKOA; p3= asKOA vs sKOA KOA, knee osteoarthritis; BMI, body mass index; WOMAC, the Western Ontario and McMaster Universities Index; KLG, Kellgren-Lawrence grading; MOAKS, MRI Osteoarthritis Knee Score. p<0.05 was considered statistically significant.

      Observer Reproducibility

      D, D*, and f derived from IVIM were all in excellent intra- and inter-observer agreement (all ICCs ≥ 0.75) (Table 3).
      Table 3Intra- and Inter-Observer ICCs for the Measurements Derived From IVIM Images of IPFP
      ParametersInter-Observer ICCs (95% CI)Intra-Observer ICCs(95% CI)
      D0.885 (0.764-0.941)0.893 (0.756, 0.967)
      D
      D, pseudo-diffusion coefficient; f, perfusion fraction; IVIM, intravoxel incoherent m
      0. 810 (0.662-0.921)0.848 (0.766, 0.930)
      F0. 765 (0.600-0.911)0.773 (0.675, 0.871)
      ICCs, intraclass correlation coefficients; CI, confidence interval; D, diffusion coefficient;
      low asterisk D, pseudo-diffusion coefficient; f, perfusion fraction; IVIM, intravoxel incoherent m

      Differences in IPFP Volume, IPFP Depth, and IVIM Parameters Among the Three Groups

      In the sKOA group, the IPFP PdWI-HR demonstrated significantly higher values of D and D* compared to those in the asKOA group (p๤0.05). Compared to the non-KOA group, D value of the PdWI-NSIR in the sKOA and asKOA groups was higher, and there were statistically significant differences among the three groups (p=0.013). The IPFP volume was statistically significant differences between the non-KOA and sKOA groups (p=0.030). IPFP depth in the sKOA and asKOA groups was deeper than that in non-KOA group (p=0.026) (Fig 4).
      Figure 4
      Figure 4Morphometric characteristics and IVIM parameters of IPFP evaluated in non-KOA, asOA, and sKOA group. IPFP volume (4a) and IPFP depth (4b) were evaluated in the three groups. Comparison of the D value (4c), D* value (4d), and f (4e) of PdWI-HR and PdWI-NSIR in non-KOA, asOA, and sKOA group. Data are expressed as mean ± standard deviation. Data are expressed as mean ± standard deviation. D, true diffusion coefficient; D*, pseudo-diffusion coefficient; f, perfusion fraction; asKOA, asymptomatic KOA; sKOA, symptomatic KOA; PdWI, proton density-weighted imaging; HR, hyperintense region; NSIR, normal signal intensity region.

      Associations of IPFP Volume, IPFP Depth, and IVIM Parameters with Knee Pain

      Higher D and D* values of IPFP hyperintense region were significantly associated with higher risks of knee pain in both univariate and multivariate logistic regression analyses. However, there were no correlation between D value and D* value with total WOMAC score, stiffness, and physical function subscales (p>0.05). Furthermore, IPFP volume, IPFP depth, and f value were all not associated with WOMAC (Fig 5).
      Figure 5
      Figure 5Forest plot shows the associations of IPFP volume, IPFP depth, and IVIM parameters with knee pain. The D and D* values showed significant associations with knee pain in both univariate and multivariate logistic regression analyses. Dependent variable: knee pain (yes vs. no); independent variables: IPFP volume, IPFP depth, and IVIM parameters. Statistically significant associations are shown in bold.
      #Adjusted for age, BMI, MOAKS-IPFP synovitis, and KLG.
      IVIM, intravoxel incoherent motion, OR, odds ratio, CI, confidence interval, IPFP, infrapatellar fat pad.

      Diagnostic Performance of D and D* Values in Assessing the Symptomatic and Asymptomatic KOA

      The sensitivity and specificity of D value for symptomatic KOA were 80.28% and 83.33%, with an AUC of 0.78 (0.68-0.86). D* value showed a sensitivity with 92.96% and a specificity of 58.33%, with an AUC of 0.82 (0.73-0.89) for symptomatic KOA (Fig 6).
      Figure 6
      Figure 6The ROC curves of D and D* values were shown to evaluate the performance of diagnosing symptomatic KOA. To differentiate asymptomatic KOA from symptomatic KOA, the AUCs for D and D* values were 0.78 and 0.82, respectively.

      DISCUSSION

      To our knowledge, this is the first study to quantitatively evaluate differences in IVIM-DWI parameters between the PdWI hyperintense regions and adjacent normal signal intensity region in patients with symptomatic and asymptomatic KOA. Our major findings were that the IPFP PdWI-HR of symptomatic KOA demonstrated significantly higher values of D and D* compared to asymptomatic KOA, and the KOA patients had a higher D value than normal controls, irrespective of the presence of a PdWI-HR. Furthermore, our results also showed that the higher D and D* values of IPFP hyperintense region were significantly associated with higher risks of knee pain in both univariate and multivariate logistic regression analyses. These associations suggesting IVIM parameters (D and D* value) of IPFP hyperintense regions may have a predictive role on symptoms of KOA.
      We found a progressive increase of IPFP volume and depth in the sKOA and asKOA groups compared to the non-KOA group, while higher IPFP volume and IPFP depth were not associated with the increase in knee pain. Inconsistent with a previous study, 35 symptomatic KOA patients and 11 asymptomatic controls and reported that symptomatic KOA patients had a more voluminous IPFP than asymptomatic controls, and a larger IPFP volume was associated with worse pain (
      • Cowan S
      • Hart HF
      • Warden SJ
      • et al.
      Infrapatellar fat pad volume is greater in individuals with patellofemoral joint osteoarthritis and associated with pain.
      ). Although the IPFP is a known potent source of knee pain, the causality between OA and IPFP volume remains controversial, and further longitudinal studies are required to confirm these findings. Furthermore, there were significant differences in age and BMI between non-KOA and KOA groups and with the IPFP volume increasing with higher BMI/age. A previous study (38 normal weight (BMI 18.5-25), 38 pre-obese (BMI 25-30), 38 obese I (BMI 30-35) and 38 obese II (BMI 35-40)) show a significant association between IPFP volume and BMI between groups, and a ceiling effect of IPFP volume occurred at a BMI of approximately 30 kg/m2 (
      • Burda B
      • Steidle-Kloc E
      • Dannhauer T
      • et al.
      Variance in infra-patellar fat pad volume: does the body mass index matter?-Data from osteoarthritis initiative participants without symptoms or signs of knee disease.
      ). In this study, although the participants with BMIage kg/m2 were excluded, we also obtained the similar results. This explains why obesity is a major risk factor for osteoarthritis. Another intriguing cohort study of 15 controls and 15 knee osteoarthritis showed that IPFP volume was correlated with age in the osteoarthritic group (
      • Chuckpaiwong B
      • Charles HC
      • Kraus VB
      • et al.
      Age-associated increases in the size of the infrapatellar fat pad in knee osteoarthritis as measured by 3T MRI.
      ). Presumably, the factors related to KOA (including age and BMI) may also induce enlargement of the IPFP, suggesting the age/BMI had a potential important causal mechanism in OA pain.
      Our study showed that 85% of symptomatic KOA patients and 56% of asymptomatic KOA patients had high signal intensities on PdWI, and this was in line with recent study (
      • BAd Vries
      • RAvd Heijden
      • Poot DHJ
      • et al.
      Quantitative DCE-MRI demonstrates increased blood perfusion in Hoffa's fat pad signal abnormalities in knee osteoarthritis, but not in patellofemoral pain.
      ). The IPFP is also known as the “Hoffa's fat pad”, and contains an abundant vascular network (
      • Draghi F
      • Ferrozzi G
      • Urciuoli L
      • et al.
      Hoffa's fat pad abnormalities, knee pain and magnetic resonance imaging in daily practice.
      ). Studies have shown that high signal intensity changes in the IPFP on PdWI are quite common, which are considered as manifestations of knee joint inflammation, edema, and angiogenesis (
      • Rvd Heijden
      • Jd Kanter
      • Bierma-Zeinstra S
      • et al.
      Structural abnormalities on magnetic resonance imaging in patients with patellofemoral pain: a cross-sectional case-control study.
      ,
      • Wang K
      • Ding C
      • Hannon M
      • et al.
      Quantitative signal intensity alteration in infrapatellar fat pad predicts incident radiographic osteoarthritis: the osteoarthritis initiative.
      ). Previous studies have confirmed that all IPFP measures, including mean value and standard deviation of the whole IPFP signal intensity were positively associated with OA-related biochemical biomarkers and chronic inflammation occurrence (
      • Dragoo J
      • Johnson C
      • McConnell J.
      Evaluation and treatment of disorders of the infrapatellar fat pad.
      ,
      • Cen H
      • Yan Q
      • Han W
      • et al.
      Longitudinal association of infrapatellar fat pad signal intensity alteration with biochemical biomarkers in knee osteoarthritis.
      ). In our study, the IPFP PdWI-HR demonstrated significantly higher values of D and D* compared to asymptomatic KOA patients, which suggested inflammatory pathogenesis of this region in KOA patients. Based on the IVIM theory, the D value represents a slow compartment linked to pure molecular diffusion. The increase of D value in the high signal area may be due to the long-term biomechanical effects of the knee joint, as well as the edema caused by vasodilation resulting from chronic mechanical friction or impact. Furthermore, our findings revealed that KOA patients generally had a higher D value than healthy controls, irrespective of the presence of a PdWI-HR. Recent studies have shown that the IPFP constituting a complex, essential, and highly active metabolic and endocrine organ (
      • Felimban R
      • Ye K
      • Traianedes K
      • et al.
      Differentiation of stem cells from human infrapatellar fat pad: characterization of cells undergoing chondrogenesis.
      ,
      • Klein-Wieringa I
      • Kloppenburg M
      • Bastiaansen-Jenniskens Y
      • et al.
      The infrapatellar fat pad of patients with osteoarthritis has an inflammatory phenotype.
      ). Favero et al. demonstrated that the levels of monocyte chemoattractant protein-1 (MCP-1) and interleukin-6 (IL-6) related to inflammatory infiltration in the IPFP were significantly elevated in patients with OA, which induced a series of microscopic changes, including interlobular septum thickening and vascularization, resulting in changes in microcirculation perfusion in IPFP(8). Consistent with these studies, our results prompted that the adjacent region of IPFP hyperintensity might have been infiltrated by inflammation cell, which needed further histopathological evidence. Of great interest, significant difference of D* value between asymptomatic and symptomatic KOA patients were also found. D* value represents a fast compartment linked to perfusion-related diffusion (pseudo-diffusion coefficient). Presumably, the interlobular septum thickening and vascularization in IPFP might be responsible for the increased perfusion. In addition, there was no significant difference in f value between the three groups, which could be due to the large proportion of fat in the IPFP, and the lipid composition could lead to the underestimation of perfusion fraction. Collectively, these findings support to the potential role of IPFP signal abnormality in the induction of symptom of OA, and which might be quantified by IVIM technique.
      The IPFP is rich in nerve fiber endings and plays an important role in knee pain perception, thus, it is considered as one of the main sources of knee pain in OA patients (
      • Clockaerts S
      • Bastiaansen-Jenniskens Y
      • Runhaar J
      • et al.
      The infrapatellar fat pad should be considered as an active osteoarthritic joint tissue: a narrative review.
      ,
      • Greene M
      • Loeser R.
      Aging-related inflammation in osteoarthritis.
      ). The IPFP fibrosis led to the new angiogenesis and sensory nerve fiber endings, which were associated with long-term pain (
      • Onuma H
      • Tsuji K
      • Hoshino T
      • et al.
      Fibrotic changes in the infrapatellar fat pad induce new vessel formation and sensory nerve fiber endings that associate prolonged pain.
      ). Knee pain was enhanced more than two times after IPFP resection, therefore, it was suggested that the IPFP should be cautiously removed or at least the “vascular-concentrated area” should be preserved in OA patients undergoing total knee arthroplasty (
      • Nisar S
      • Lamb J
      • Somashekar N
      • et al.
      Preservation vs. resection of the infrapatellar fat pad during total knee arthroplasty part II: a systematic review of published evidence.
      ).Our results showed that the higher D and D* values of IPFP hyperintense region were significantly associated with higher risks of knee pain in both univariate and multivariate logistic regression analyses. Furthermore, the D value with a cut-off of 0.15 × 10−3 mm2/s showed sensitivity of 80.28% and specificity of 83.33% for diagnosing symptomatic KOA. The D* value showed a sensitivity with 92.96% and a specificity of 58.33% for diagnosing symptomatic KOA when the cut-off was 23.8 × 10−3 mm2/s. A previous cohort study on 1100 individuals revealed that the maximum sagittal area of female IPFP was negatively correlated with the WOMAC pain score of the knee joint (
      • Pan F
      • Han W
      • Wang X
      • et al.
      A longitudinal study of the association between infrapatellar fat pad maximal area and changes in knee symptoms and structure in older adults.
      ). However, in our research, there were no correlations between the IPFP MRI morphometric characteristics (IPFP volume and depth) and knee pain. This controversial result might be that the severe KOA (KLG=4) patients were excluded in our study. Therefore, the most important findings in our study are the significant associations between IVIM parameters (D and D* value) of IPFP and symptoms of patients with knee osteoarthritis, and the D and D* values were the independent predictors of symptomatic KOA.
      Our study has its limitations. First, our study was a single center, retrospective, cross-sectional design, and the correlation between symptomatic progression and IVIM parameters in KOA patients could not be longitudinally analyzed. Second, the study only included patients with mild-to-moderate KOA, and further research with severe KOA is needed to elucidate the role of IVIM in IPFP. Third, the imaging quality of the IVIM needs to be improved, due to the low signal-to-noise ratio of EPI. Some of the foci were very small, which was difficult to address on the IVIM images, and lead to poor evaluation of small ROIs. Even if we only chose the largest lesion in each case, there is still a pitfall with our method.
      In conclusion, IVIM-DWI can be used as an additional functional imaging technique to study IPFP with signal abnormalities on PdWI, and the D and D* values may have potential value to predict the symptom in mild-to-moderate KOA patients.

      FUNDING

      This project has been funded with support from General Special Scientific Research Program of the Department of Education of Shaanxi Province (No.22JK0356) and Subject Innovation Team of Shaanxi University of Chinese Medicine (No.2019-QN09) and the Key Research and Development Program of Shaanxi Provincial Science and Technology Department (No.2018ZDXM-SF-009).

      REFERENCEs

        • Hunter DJ
        • Bierma-Zeinstra S.
        Osteoarthritis.
        Lancet (London, England). 2019; 393: 1745-1759https://doi.org/10.1016/s0140-6736(19)30417-9
        • van der Heijden RA
        • de Vries BA
        • Poot DHJ
        • et al.
        Quantitative volume and dynamic contrast-enhanced MRI derived perfusion of the infrapatellar fat pad in patellofemoral pain.
        Quant imaging med surg. 2021; 11: 133-142https://doi.org/10.21037/qims-20-441
        • Arendt-Nielsen L.
        Pain sensitisation in osteoarthritis.
        Clin Exp Rheumatol. 2017; 107 (35 Suppl): 68-74
        • Ballegaard C
        • Riis R
        • Bliddal H
        • et al.
        Knee pain and inflammation in the infrapatellar fat pad estimated by conventional and dynamic contrast-enhanced magnetic resonance imaging in obese patients with osteoarthritis: a cross-sectional study.
        Osteoarthritis cartilage. 2014; 22: 933-940https://doi.org/10.1016/j.joca.2014.04.018
        • Zeng N
        • Yan ZP
        • Chen XY
        • et al.
        Infrapatellar fat pad and knee osteoarthritis.
        Aging Dis. 2020; 11: 1317-1328https://doi.org/10.14336/ad.2019.1116
        • Nemschak G
        • Pretterklieber M.
        The patellar arterial supply via the infrapatellar fat pad (of Hoffa): a combined anatomical and angiographical analysis.
        Anat res int. 2012; 2012713838https://doi.org/10.1155/2012/713838
        • Leese J
        • Davies D.
        An investigation of the anatomy of the infrapatellar fat pad and its possible involvement in anterior pain syndrome: a cadaveric study.
        J anat. 2020; 237: 20-28https://doi.org/10.1111/joa.13177
        • Favero M
        • El-Hadi H
        • Belluzzi E
        • et al.
        Infrapatellar fat pad features in osteoarthritis: a histopathological and molecular study.
        Rheumatology (Oxford). 2017; 56: 1784-1793https://doi.org/10.1093/rheumatology/kex287
        • Macchi V
        • Stocco E
        • Stecco C
        • et al.
        The infrapatellar fat pad and the synovial membrane: an anatomo-functional unit.
        J anat. 2018; 233: 146-154https://doi.org/10.1111/joa.12820
        • An JS
        • Tsuji K
        • Onuma H
        • et al.
        Inhibition of fibrotic changes in infrapatellar fat pad alleviates persistent pain and articular cartilage degeneration in monoiodoacetic acid-induced rat arthritis model.
        Osteoarthritis cartilage. 2021; 29: 380-388https://doi.org/10.1016/j.joca.2020.12.014
        • Solivetti F
        • Guerrisi A
        • Salducca N
        • et al.
        Appropriateness of knee MRI prescriptions: clinical, economic and technical issues.
        La Radiologia medica. 2016; 121: 315-322https://doi.org/10.1007/s11547-015-0606-1
        • Su X
        • Zhang Y
        • Gao Q
        • et al.
        Preliminary study on the assessment of early cartilage degeneration by quantitative ultrashort echo time magnetic resonance imaging in vivo.
        Quant imaging med surg. 2022; 12: 3803-3812https://doi.org/10.21037/qims-21-1181
        • Han W
        • Aitken D
        • Zheng S
        • et al.
        Association between quantitatively measured infrapatellar fat pad high signal-intensity alteration and magnetic resonance imaging-assessed progression of knee osteoarthritis.
        Arthritis care res. 2019; 71: 638-646https://doi.org/10.1002/acr.23713
        • Klein-Wieringa I
        • Bd Lange-Brokaar
        • Yusuf E
        • et al.
        Inflammatory cells in patients with endstage knee osteoarthritis: a comparison between the synovium and the infrapatellar fat pad.
        J rheumatol. 2016; 43: 771-778https://doi.org/10.3899/jrheum.151068
        • Cai J
        • Xu J
        • Wang K
        • et al.
        Association between infrapatellar fat pad volume and knee structural changes in patients with knee osteoarthritis.
        J rheumatol. 2015; 42: 1878-1884https://doi.org/10.3899/jrheum.150175
        • Okita Y
        • Oba H
        • Miura R
        • et al.
        Movement and volume of infrapatellar fat pad and knee kinematics during quasi-static knee extension at 30 and 0° flexion in young healthy individuals.
        The Knee. 2020; 27: 71-80https://doi.org/10.1016/j.knee.2019.10.019
        • Cowan S
        • Hart HF
        • Warden SJ
        • et al.
        Infrapatellar fat pad volume is greater in individuals with patellofemoral joint osteoarthritis and associated with pain.
        Rheumatology international. 2015; 35: 1439-1442https://doi.org/10.1007/s00296-015-3250-0
        • He J
        • Ba H
        • Feng J
        • et al.
        Increased signal intensity, not volume variation of infrapatellar fat pad in knee osteoarthritis: a cross-sectional study based on high-resolution magnetic resonance imaging.
        J orthop surg (Hong Kong). 2022; 3010225536221092215https://doi.org/10.1177/10225536221092215
        • Hunter DJ
        • Guermazi A
        • Lo GH
        • et al.
        Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score).
        Osteoarthritis cartilage. 2011; 19: 990-1002https://doi.org/10.1016/j.joca.2011.05.004
        • Le Bihan D
        What can we see with IVIM MRI?.
        NeuroImage. 2019; 187: 56-67https://doi.org/10.1016/j.neuroimage.2017.12.062
        • Cao J
        • Gao S
        • Zhang C
        • et al.
        Differentiating atypical hemangiomas and vertebral metastases: a field-of-view (FOV) and FOCUS intravoxel incoherent motion (IVIM) diffusion-weighted imaging (DWI) study.
        Eur spine J. 2020; 29: 3187-3193https://doi.org/10.1007/s00586-020-06632-z
        • Zhang W
        • Doherty M
        • Peat G
        • et al.
        EULAR evidence-based recommendations for the diagnosis of knee osteoarthritis.
        Ann rheum. 2010; 69: 483-489https://doi.org/10.1136/ard.2009.113100
        • Salaffi F
        • Leardini G
        • Canesi B
        • et al.
        Reliability and validity of the western ontario and McMaster universities (WOMAC) osteoarthritis index in italian patients with osteoarthritis of the knee.
        Osteoarthritis cartilage. 2003; 11: 551-560https://doi.org/10.1016/s1063-4584(03)00089-x
        • Kellgren J
        • Lawrence J.
        Radiological assessment of osteo-arthrosis.
        Ann rheum. 1957; 16: 494-502https://doi.org/10.1136/ard.16.4.494
        • Burda B
        • Steidle-Kloc E
        • Dannhauer T
        • et al.
        Variance in infra-patellar fat pad volume: does the body mass index matter?-Data from osteoarthritis initiative participants without symptoms or signs of knee disease.
        Annals anat. 2017; 213: 19-24https://doi.org/10.1016/j.aanat.2017.04.004
        • Chuckpaiwong B
        • Charles HC
        • Kraus VB
        • et al.
        Age-associated increases in the size of the infrapatellar fat pad in knee osteoarthritis as measured by 3T MRI.
        J orthop res. 2010; 28: 1149-1154https://doi.org/10.1002/jor.21125
        • BAd Vries
        • RAvd Heijden
        • Poot DHJ
        • et al.
        Quantitative DCE-MRI demonstrates increased blood perfusion in Hoffa's fat pad signal abnormalities in knee osteoarthritis, but not in patellofemoral pain.
        Eur radiol. 2020; 30: 3401-3408https://doi.org/10.1007/s00330-020-06671-6
        • Draghi F
        • Ferrozzi G
        • Urciuoli L
        • et al.
        Hoffa's fat pad abnormalities, knee pain and magnetic resonance imaging in daily practice.
        Insights imaging. 2016; 7: 373-383https://doi.org/10.1007/s13244-016-0483-8
        • Rvd Heijden
        • Jd Kanter
        • Bierma-Zeinstra S
        • et al.
        Structural abnormalities on magnetic resonance imaging in patients with patellofemoral pain: a cross-sectional case-control study.
        Am J sports med. 2016; 44: 2339-2346https://doi.org/10.1177/0363546516646107
        • Wang K
        • Ding C
        • Hannon M
        • et al.
        Quantitative signal intensity alteration in infrapatellar fat pad predicts incident radiographic osteoarthritis: the osteoarthritis initiative.
        Arthritis care res. 2019; 71: 30-38https://doi.org/10.1002/acr.23577
        • Dragoo J
        • Johnson C
        • McConnell J.
        Evaluation and treatment of disorders of the infrapatellar fat pad.
        Sports medicine. 2012; 42: 51-67https://doi.org/10.2165/11595680-000000000-00000
        • Cen H
        • Yan Q
        • Han W
        • et al.
        Longitudinal association of infrapatellar fat pad signal intensity alteration with biochemical biomarkers in knee osteoarthritis.
        Rheumatol. 2022; https://doi.org/10.1093/rheumatology/keac214
        • Felimban R
        • Ye K
        • Traianedes K
        • et al.
        Differentiation of stem cells from human infrapatellar fat pad: characterization of cells undergoing chondrogenesis.
        Tissue eng Part A. 2014; 20: 2213-2223https://doi.org/10.1089/ten.tea.2013.0657
        • Klein-Wieringa I
        • Kloppenburg M
        • Bastiaansen-Jenniskens Y
        • et al.
        The infrapatellar fat pad of patients with osteoarthritis has an inflammatory phenotype.
        Ann rheum dis. 2011; 70: 851-857https://doi.org/10.1136/ard.2010.140046
        • Clockaerts S
        • Bastiaansen-Jenniskens Y
        • Runhaar J
        • et al.
        The infrapatellar fat pad should be considered as an active osteoarthritic joint tissue: a narrative review.
        Osteoarthritis cartilage. 2010; 18: 876-882https://doi.org/10.1016/j.joca.2010.03.014
        • Greene M
        • Loeser R.
        Aging-related inflammation in osteoarthritis.
        Osteoarthritis cartilage. 2015; 23: 1966-1971https://doi.org/10.1016/j.joca.2015.01.008
        • Onuma H
        • Tsuji K
        • Hoshino T
        • et al.
        Fibrotic changes in the infrapatellar fat pad induce new vessel formation and sensory nerve fiber endings that associate prolonged pain.
        J orthop res. 2020; 38: 1296-1306https://doi.org/10.1002/jor.24580
        • Nisar S
        • Lamb J
        • Somashekar N
        • et al.
        Preservation vs. resection of the infrapatellar fat pad during total knee arthroplasty part II: a systematic review of published evidence.
        The Knee. 2019; 26: 422-426https://doi.org/10.1016/j.knee.2019.01.007
        • Pan F
        • Han W
        • Wang X
        • et al.
        A longitudinal study of the association between infrapatellar fat pad maximal area and changes in knee symptoms and structure in older adults.
        Ann rheum dis. 2015; 74: 1818-1824https://doi.org/10.1136/annrheumdis-2013-205108