Entheseal bone remodeling and patient reported outcomes in osteoarthritis – A quantitative [18F]NaF-PET MRI study

Abstract

Objective:

Entheses have been discussed as potential sources of pain in osteoarthritis (OA). Entheseal bone remodeling quantified by [18F]NaF-PET could be a surrogate marker for tensile forces causing a painful tissue response. This study aimed to investigate the relationship between [18 F]NaF-uptake, MRI findings and knee-related symptoms.

Method:

Patients and healthy controls in this prospective observational study on patello-femoral OA (PFJOA) underwent simultaneous MR (at 3.0 T) and [18F]NaF PET imaging of both knees. Cartilage damage, subchondral marrow lesions, synovitis, and enthesophytes were semi-quantitatively assessed on MRI and entheseal tracer uptake was quantified in the anterior patella and tibial tuberosity. Symptoms were assessed with the Knee injury and Osteoarthritis Outcome Score (KOOS). Generalized estimating equations were used to investigate associations between tracer uptake and symptoms; analyses were additionally adjusted for MRI-detected changes (as alternative symptom drivers), age, sex and body mass index.

Results:

A total of 135 knees from 68 subjects were investigated (mean age 48.8 ± 10.4 years; 58.8% women). Higher [18F]NaF-uptake was significantly associated with worse outcomes for KOOS subscales sports (beta −2.7; p < 0.001), quality of live (beta 3.0; p < 0.001) and PFJ (beta 2.3; p=0.001), but not pain (beta −0.8; p=0.18), symptoms (beta −0.6; p=0.30) or activities of daily living (beta −1.1; p=0.06).

Conclusion:

Entheseal bone remodeling quantified by [18F]NaF-PET-MRI is associated with patient-reported clinical knee outcomes. These findings underscore the potential of entheseal biology to inform the understanding of symptom generation in degenerative disease of the musculoskeletal system.

Introduction

In osteoarthritis (OA), the limited correlation between imaging pathology and clinical symptoms has long been a topic of scientific discussion [1,2]. Among the imaging findings with the closest relationship with pain in OA are synovitis [3], including changes in Hoffa’s fat pad [4], bone marrow edema like lesions (BMELLs) [5], as well as cartilage lesions [6]. A tissue that has thus far received only limited attention is the attachment site of tendons and ligaments with bone, sometimes referred to as the ‘entheseal organ’. It is the primary site of inflammation in different rheumatic diseases (e.g. psoriatic arthritis (PsoA) and axial spondyloarthritis (axSpA) [7]), but also involved in mechanical enthesopathies, that are considered a local reaction to supraphysiological strain [8]. Somewhat distinct from the primarily inflammatory enthesitis, these mechanical conditions typically also involve the tendon itself, which appears thickened and edematous on imaging [9]. While some entheseal bone marrow edema is typically present, it is generally less pronounced than in inflammatory enthesopathy.

In osteoarthritis, a possible role of the entheses has been has been addressed in recent years: overlaps in imaging characteristics of OA and PsA have long been established [10], the collateral ligaments of the small finger joints play an important role in early OA of the hands [11], and the insertion of the posterior cruciate ligament has been implicated in the etiology of joint effusion in early knee OA [12]. Furthermore, Mattap et al. found an association between abnormalities of the tibial patella tendon insertion and pain in community dwelling older adults [13].

Current imaging techniques deployed in the investigation of entheseal pathology typically focus on the visualization of features of inflammatory response. High priority is generally given to hypervascularization, i.e. increased power-doppler signal on ultrasonography [14,15] and enhancement of gadolinium-based contrast agents on magnetic resonance imaging (MRI), or local osteitis, i.e. increased signal on fluid-sensitive MR sequences such as short-tau inversion-recovery (STIR) [16]. Somewhat more challenging than capturing this inflammatory response is the in-vivo capture of mechanical strain of the enthesis. A promising tool for the exploration of this feature is the quantification of bone remodeling at the enthesis, by imaging the uptake of [18 F]NaF on positron emission tomography (PET) [17]. Historically neglected in favor of ^{99 m}Tc, which is used in (low-resolution) bone scans [18], this tracer provides functional information on subtle changes in bone turnover and perfusion [19,20]. In this capacity it has recently gained traction in the imaging of musculoskeletal disorders, especially when used in the context of hybrid imaging in conjunction with MRI [21,22].

While the connection of inflammatory imaging findings and functional disability and pain in enthesopathy is well established [23], less is known about the mechanical aspect. Thus, the aim of this proof-of-concept study was to investigate the association of bone remodeling at the entheseal regions of the anterior patella and tibia and clinical MR imaging features of entheseal stress and knee-related clinical symptoms.

Materials and methods

Study subjects

This investigation constitutes an ancillary analysis of an ongoing prospective observational study on isolated OA of the patellofemoral joint (PFJOA), with recruitment between 2022 and 2024 in a single tertiary medical center in the United States, details of which have been reported previously [24]; for this analysis, we only used baseline data and excluded subjects with incomplete or unusable imaging (see Fig. 1). In this study, patients were recruited based on MRI-determined OA of the patella-femoral joint (PFJOA) on clinical scans as determined by an experienced musculoskeletal radiologist and then invited for further imaging assessments. Additionally, volunteers without symptoms or imaging findings of PFJOA were recruited, with intended matching with PFJOA patients for age, sex and body mass index (BMI). Inclusion criteria for all subjects were skeletal maturity, meeting physical requirements of scanner equipment (BMI below 38 kg/m² and knee circumference ≤ 49 cm). For recruitment into the patient group, subjects were required to have both evidence of OA in the PFJ on MRI and pain during stair ascend or descend at the time of the screening interview. Exclusion criteria were history of traumatic knee injury, tendon or ligament injuries, complex meniscal tears, known patellar tendinosis (clinical diagnosis or imaging findings), inflammatory arthritis (rheumatoid arthritis, Spondyloarthritis), history of vestibular conditions affecting gait patterns, contraindications to MR scanning, history of recent corticosteroid injection to the knee, or using investigational pharmaceuticals. All participants gave written informed consent before enrolment and the internal review board approved of this study prior to commencement (IRB 21–34763).

Fig. 1.

Patient flow and demographics. Participant demographics are given per subject, whereas patient reported outcomes are given per knee, as mean, SD and range. *=at time of analysis, recruitment ongoing. +=excluded for movement artifacts (n=2) and degraded image quality from coil malfunction (n=1).

Patient reported outcomes

Via self-administered questionnaire, study participants provided the Knee injury and Osteoarthritis Outcomes Score (KOOS) [25], which details symptoms and function as well as quality of life over the last 7 days and was developed and validated for functionally mobile individuals with OA; a separate questionnaire was provided for each knee. In addition to the common the subscales of pain (KOOS_PAIN), symptoms (KOOS_SYMP), sporting activities (KOOS_SPORTS), activities of daily living (KOOS_ADL), and quality of life (KOOS_QOL), a specific subscale for PFJ-related pain (KOOS_PFJ) [26] was collected. The KOOS scale ranges from 100 (indicating no impairment) to 0 (indicating maximum impairment), with scores lower than 70 indicating moderate and scores lower than 50 indicating severe impairment [25]. Furthermore, participants rated pain during stair ascent and descent for each knee separately on a visual analogue scale (VAS_ASC, VAS_DESC), where 0 indicates no pain, and 10 the highest imaginable level of pain.

Image acquisition

All study subjects received simultaneous imaging of both knees in a 3.0 T whole-body hybrid PET-MRI scanner (Signa PET-MR, GE Healthcare) and details on image acquisition have been published previously [24]. As clinical MR sequences, a three-dimensional proton-density fat-saturated fast-spin-echo (3D PD fs FSE, i.e. CUBE) of both knees was acquired (Field of view: 15 x 15 cm, slice thickness 0.6 mm; TR 1200 ms, TE 26 ms). For time-of-flight (TOF) PET imaging, [18 F]NaF was used as a radiotracer, sourced from the institutional cyclotron facility, using current good manufacturing practice guidelines [27]. Subjects received an average dose of 112.7 MBq (SD 3.0 MBq; range 107.5 – 121.0 MBq) [18 F]NaF; the static-PET scan was acquired 60–80 min after injection, and a conversion coefficient of 0.024 mSv/MBq [28] was used to calculate radiation exposure to each subject.

Image assessment

Structural knee joint lesions

Semi-quantitative and qualitative assessments of clinical MRI scans were performed by trained board-certified radiologists, blinded to PET-imaging, clinical symptoms and patient demographics. 3D CUBE images were utilized in a dedicated DICOM-viewer, allowing for dynamic multiplanar reconstructions. Assessment of articular lesions was performed using the cartilage and bone marrow edema-like lesion (BMELL) categories of a simplified Whole Organ Magnetic Resonance imaging score (WORMS) [29] by experienced radiologists (Z.A. and V.K., with 9 and 6 years of experience, respectively) as part of the study protocol. This score assesses cartilage lesions on a scale of 0 (=no abnormalities) to 6 (=diffuse (> 75%) full-thickness cartilage loss), in a total of six locations: patella, trochlea, medial and lateral tibia and femur. BMELLs, i.e. subchondral increased fluid signal within bone, were assessed in the same six locations: 0=no lesion; 1=lesion < 5 mm; 2=lesion 5–20 mm; 3=lesion > 20 mm. Imaging examples are given in Fig. 2.

Fig. 2.

Structural imaging findings. Axial (row a) and sagittal (row b) reconstructions of 3D CUBE images. Column 1 show structural knee joint damage assessed using WORMS score (Whole Organ Magnetic Resonance imaging Score). 1a=note the widespread full-thickness cartilage loss in the lateral facet of the patella (white arrowheads; WORMS cartilage score 6) and accompanying effusion (black arrow; effusion score 2). 1b=large bone marrow edema-like lesion (BMELL) of the femoral trochlea (white arrow; WORMS bone marrow score 3). Column 2 shows synovitis. 2a=marked synovial thickening (limits of thickened synovial membrane indicated by black arrowheads; synovial proliferation score 2). 2b=increased fluid signal in Hoffa’s fat pad (black arrow; Hoffa’s synovitis score of 2). Column 3 shows enthesophytes. 3a=bony protrusions of the anterior patella (black arrow; enthesopathy score 1). 3b=enthesophyte of the cranial patella (white arrow; enthesopathy score 1).

Synovitis

As an important imaging correlate of pain, effusion/synovitis was also assessed, using established scoring tools [30]: effusion was assessed as absent (=0), mild (=1), moderate (=2) or severe (=3); synovial proliferation was graded as absent (=0), mild synovial thickening (=1) or marked thickening with synovial bands (2); Hoffa’s synovitis (i.e. pathological fluid signal in Hoffa’s fat pad) was assessed as absent (=0), affecting less than 33% (=1), 33–66% (=2) or more than 66% of the volume of Hoffa’s fat pad (=3). Example images are provided in Fig. 2.

Enthesopathy

Furthermore, MR imaging signs of enthesopathy were assessed as absent (=0), small bony irregularities (=1), or definite enthesophyte (=2) (see Fig. 2). Elevation of fluid signal (i.e. bone marrow edema, BME) in the enthesis-adjacent bone was graded as absent or present. Assessments for entheseal MRI lesions were performed for the separately for the cranial, middle, and caudal anterior patella and the anterior tibial tuberosity. All assessments were performed by a radiologist with expertise in arthritis imaging (K.Z., 8 years of experience in MSK imaging), blinded to clinical information and PET imaging at the time of assessment. To show internal validity and reproducibility of these assessments, a sample of 20 right knees was re-scored in random order by the first radiologist after a wash-out period of four weeks and additionally scored by a second radiologist (V.K.).

Quantitative [18F]NaF-PET analysis

Quantitative assessment of [18F]NaF uptake on PET was performed manually on fusion images of PET and 3D Cube: a radiologist (K.Z., 1 year of experience with quantitative PET imaging) marked the area of highest uptake in the entheseal anterior part of the patella and tibia separately with a region of interest (ROI) tool, using dedicated software (Visage, Version 7.1.18, Visage Imaging Enterprise) (see Fig. 3). From these ROI measurements, maximum standardized uptake values (SUV_MAX), i.e. a normalized (semi-quantitative) measure of the maximum radiotracer concentration within the specified anatomic region [31], were noted. In a sample of 20 right knees, assessments were repeated after two weeks by the first radiologist and also performed by a second radiologist (V.K., 1 year of experience with quantitative PET imaging) to calculate intra- and inter-reader agreement.

Fig. 3.

Quantitative [18F]NaF-PET measurements. 3D Cube images in axial and sagittal reconstruction, overlayed with [18F]NaF-PET images.

1=measurement in the anterior patella, with region of interest (ROI) fitted to the anatomical region of the enthesis. 2=measurement in the tibial tuberosity. 3=sagittal image for reference, line with small dots indicating height of patella measurement, line with dashes indicating tibial measurement.

Statistical analysis

Unless stated otherwise, descriptive results are reported as means with standard deviations. Semi-quantitative imaging findings for BMELL (WORMS_BMELL) and synovitis were summarized into separate sum scores per knee, while cartilage lesions were captured in a binary fashion, with a score of 2 or greater in either patella or trochlea indicating PFJ-OA. Inter-reader and intra-reader agreement of morphological and quantitative assessments were analyzed using weighted Cohen’s kappa (with linear weights and bootstrapped calculation confidence intervals) and intra-class correlation coefficients (ICC), respectively. We employed generalized estimating equations (GEE) using a linear regression model with a Gaussian family and an independence working correlation structure to account for clustering by subject. Firstly, SUV_MAX values were used as outcomes and enthesopathy on MRI (ordinal: 0–3) was used as predictor, adjusted for age (in years, continuous), sex (binary) and BMI (continuous). Secondly, KOOS scores were used as outcomes and SUV_MAX values were used as predictors, with additional adjustment for synovitis (continuous sum score), BMELL (continuous sum score), PFJOA status (binary), age, sex and BMI. Exploratory interaction analyses were performed to elucidate whether the association between entheseal bone remodeling and patient related outcomes differed according to PFJ-OA status. Residual diagnostics, including the residuals vs. fitted plot and histogram, suggested mild heteroscedasticity and a slight deviation from normality, consistent with the presence of outliers and skewed outcome distribution. These patterns were not extreme, and the use of GEE with robust standard errors mitigates their impact on inference. No major violations of model assumptions were identified. All analyses were performed using Python Version 3.11.8, setting statistical significance to p < 0.050. To account for multiple testing across six primary outcomes, we applied a Bonferroni correction to our significance threshold (p < 0.008). Confidence intervals are presented at the 95% level for interpretive consistency and transparency but are not used as the criterion for statistical significance.

Results

Study subjects and patient reported outcomes

A total of 135 knees from 68 participants, including 20 knees from 10 healthy control subjects were included in the study. An overview of the subject flow and demographic characteristics of the study population is provided in Fig. 1. A slight majority of included individuals were women (58.8%, 40/68; of these 16 self-identified as postmenopausal), the mean age was 48.8 (SD10.4) years, the mean BMI was 25.6 (SD 4.1) kg/m². The mean ionizing radiation exposure from [18F]NaF injection was 2.7 mSv (SD 0.07 mSV; range: 2.6 – 2.9 mSv) per subject, which is lower than the natural annual radiation exposure in the United States, which is estimated at 3.1 mSv [32].

PROs, assessed separately for each knee, showed lowest morbidity in the subscale KOOS_ADL with a mean score of 93.1 (SD 9.7), indicating almost no impairment for activities of daily living, and higher morbidity for both KOOS_SPORT (mean 76.3, SD 23.4) and KOOS_QOL (mean 68.7, SD 25.4), indicating moderate impairment for sporting and recreation activities and knee-related quality of life, respectively. Pain during stair ascending and descending (VAS 0–10) was very limited, with a mean score of 0.3 (SD 0.7) and 0.3 (SD 0.8) for ascending and descending, respectively.

Descriptive imaging findings

An overview of descriptive qualitative and quantitative imaging findings is provided as Table 1. Presence of bony abnormalities of the tibial tuberosity was less common (3.7%, 5/135), than in the patella (65.9%, 89/135). No instances of entheseal BME were noted. As expected, median scores of clinical imaging findings indicated significantly more severe findings in PFJ-OA cases vs. controls for synovitis (2 (IQR 1–3) vs. 1 (IQR 0–2)), BMELL (2 (IQR 1–3) vs. 0 (IQR 0–0)), and cartilage sum scores (5.75 (IQR 5–7.5) vs. 1 (1–2)). Representative examples of these imaging abnormalities are given in Fig. 2. Mean SUV_MAX in the patella was 2.0 (SD 1.9; range 0.3–14.2) and 2.6 (SD 1.5; range 0.4–7.9) in the tibial tuberosity; for neither location, differences in mean values were significant between cases and controls. Paired comparisons of bone turnover between index and control knees in the subset of participants with asymmetrical disease showed considerable individual variability, with a mix of increases and decreases across participants (see Supplement 1).

Table 1.

Descriptive imaging findings.

	All (n=135)	PFJ-OA		p
		Cases (n=50)	Controls (n=85)
SUVMAX patella [mean, SD]	2.01 (1.92)	2.21 (1.93)	1.86 (1.91)	0.30
SUVMAX tibia [mean, SD]	2.59 (1.45)	2.62 (1.62)	2.56 (1.31)	0.83
Enthesopathy patella [%, n]	65.9% (89/135)	64.4% (38/50)	67.1% (51/85)	0.89
Enthesopathy tibial tuberosity [%, n]	3.7% (5/135)	1.7% (1/50)	5.3% (4/85)	0.53
Synovitis [median sum score out of 8, IQR]	1 (1–1.25)	2 (1–3) *	1 (0–2)	0.017
WORMS BMELL [median sum score out of 6, IQR]	0 (0–2)	2 (1–3) *	0 (0–0)	< 0.001
WORMS Cartilage PFJ [median sum score out of 12, IQR]	3 (1–5)	5.75 (5–7.5) *	1 (1–2)	< 0.001

SUV_MAX= maximum standardized uptake value. IQR=Interquartile range. BMELL=bone marrow edema like lesion (sum score). PFJ-OA=status of MRI detected PFJ-OA, based on extent of cartilage damage (WORMS cartilage score > 2 in either patella or trochlea). Significantly higher values in case-control comparisons (p < 0.05) are printed in bold and marked with an asterisk (*).

Intra- and inter-reader agreement

Inter-reader agreement for enthesopathy was good to excellent, with Cohen’s kappa of 0.77 (95%CI 0.60; 1.00) for the cranial, 0.89 (95%CI 0.62; 1.00) for the mid, 0.82 (95%CI 0.46; 1.00) for the caudal patella and 0.63 (95%CI 0.41; 0.78) for the tibial tuberosity. Intra-reader agreement was also high, with kappa of 0.73 (95%CI 0.36; 1.00), 0.89 (95%CI 0.71; 1.00), 0.83 (95%CI 0.46; 1.00) and 0.64 (95%CI 0.60; 0.66) for the respective regions. As no instance of enthesis-associated BME was observed in any reading session, agreement was not calculated. For quantitative assessments of tracer uptake, intra- and inter-reader agreement, expressed as ICCs, were 0.88 (95%CI 0.62;0.97) and 0.70 (95%CI 0.39; 0.87) for the patella and 0.98 (95% CI 0.96; 0.99) and 0.70 (95%CI 0.39; 0.87) for the tibia, respectively.

Entheseal bone turnover and patient specific factors

Factors of influence on entheseal bone turnover were investigated using GEE, with bone turnover (SUV_MAX) as the outcome and presence of enthesophytes, age, sex and BMI as predictors. Enthesopathy was not significantly associated with bone turnover in either the patella (beta 0.58, 95%CI −0.08, 1.12) or the tibia (beta −0.07, 95%CI −0.55, 0.42), but age showed a significant positive association with bone turnover in both regions, with a beta of 0.03 (95%CI 0.01–0.06) for both location. Female sex did not exhibit a significant association with bone turnover (patella: −0.61 (95%CI −1.46, 0.23); tibia: −0.22 (95%CI −0.77, 0.33). BMI showed a significant positive association with tibial (0.08; 95%CI 0.00, 0.15) but not patellar bone turnover (0.08; 95%CI −0.01, 0.17).

Entheseal bone turnover and pain

Associations between KOOS subscales and tracer uptake in the patella are given in Table 2. We found significant associations between SUV_MAX in the anterior patella and severity of patient reported outcomes, specifically in the KOOS subscales sports (−2.68, 95%CI −4.14; −1.21), quality of life (−2.97 95%CI −4.59; −1.35) and PFJ-related symptoms (−2.28, 95%CI −3.58; −0.98). Notably, the association between symptoms and imaging findings were stronger for entheseal bone turnover than BMELL (beta range −0.91–1.03, see Table 2), which is generally accepted as an imaging correlate of pain in OA. Strength of association for synovitis was similar to entheseal bone turnover (beta range −0.91 to −2.81, see Table 2), but missed statistical significance after multiplicity correction. There was no significant association between tracer uptake in the tibial tuberosity and any of the clinical outcomes (full results given as Supplement 2). Similarly, there was no significant association between presence of enthesophytes and clinical outcomes. A clinical imaging example is given in Fig. 4 and scatterplots for entheseal bone turnover and KOOS subscales is provided as Supplement 3. As formal analyses did not yield a significant statistical interaction between PFJOA status and the association between bone turnover and clinical outcomes, we refrained from analyses stratified by disease severity.

Table 2.

Association of symptoms and tracer uptake and MRI lesions.

	Association (beta, 95% CI)
	Pain	Symptoms	Activities of daily living	Sports	Quality of life	Patello-femoral joint
SUVMAX patella	−0.78 (−1.92; 0.35) p=0.18	−0.64 (−1.85; 0.58) p=0.30	−1.11 (−2.25; 0.03) p=0.06	−2.68* (−4.14; −1.21) p < 0.001	−2.97* (−4.59; −1.35) p < 0.001	−2.28* (−3.58; −0.98) p=0.001
Synovitis	−1.05 (−2.46; 0.36) p=0.14	−0.91 (−2.58; 0.77) p=0.29	−0.70 (−2.08; 0.68) p=0.32	−2.80 (−5.57; −0.04) p=0.047	−2.20 (−4.63; 0.23) p=0.08	−2.21 (−4.57; 0.15) p=0.07
BMELL	1.03 (−1.15; 3.22) p=0.36	−0.91 (−4.35; 2.53) p=0.60	0.08 (−1.95; 2.11) p=0.94	−0.19 (−3.44; 3.81) p=0.92	−0.01 (−4.30; 4.28) p > 0.99	−0.20 (−3.66; 3.27) p=0.91
PFJ-OA	−8.41* (−13.77; −3.05) p=0.002	−7.95 (−13.86; −0.24) p=0.042	−4.76 (−9.06; −0.47) p=0.030	−8.53 (−18.67; 1.62) p=0.10	−13.79 (−25.02; 2.58) p=0.016	−8.01 (−17.08; 1.07) p=0.08
Age	0.24 (0.03; 0.44) p=0.026	0.32 (0.04; 0.60) p=0.024	0.11 (−0.14; 0.35) p=0.38	0.33 (−0.21; 0.87) p=0.23	0.64 (0.15; 1.14) p=0.011	0.41 (−0.03; 0.84) p=0.07
Female sex	2.91 (−1.64; 7.46) p=0.21	−1.14 (−6.00; 3.73) p=0.65	2.76 (−1.46; 6.98) p=0.20	−1.93 (−11.07; 7.20) p=0.68	6.25 (−2.45; 14.95) p=0.16	0.14 (−7.48; 7.76) p=0.97
BMI	−0.05 (−0.81; 0.72) p=0.91	−0.57 (−1.33; 0.19) p=0.14	−0.13 (−0.76; 0.49) p=0.68	−0.85 (−2.20; 0.50) p=0.22	−0.40 (−2.05; 1.25) p=0.64	−0.60 (−1.80; 0.60) p=0.33

SUV_MAX= maximum standardized uptake value (in the anterior patella). BMELL=bone marrow edema like lesion (sum score). PFJ-OA=status of MRI detected PFJ-OA, based on extent of cartilage damage (WORMS cartilage score > 2 in either patella or trochlea). Beta coefficients with 95% CIs (in brackets) and p-values, derived from GEE, treating both limbs of one subject as related observations. Lower KOOS scores indicate higher disability. Significant results (p < 0.008) are printed in bold and marked with an asterisk (*).

Fig. 4.

Clinical imaging examples. Fusion images of 3D CUBE (3 mm maximum intensity projections, anatomical sagittal reconstruction) and [18F]NaF PET (1a and 2a) with magnifications of anterior patella without PET overlay. Below: patient reported outcomes. 1: 45-year-old male with substantial impairment. Clinical MRI revealed no structural abnormalities or evidence of synovitis in the knee joint. Note the enthesophytes (white arrow) and bony irregularities across the anterior patella (black arrow) alongside the marked tracer uptake (SUV_MAX 8.0). 2: 36-year-old female without notable pain, imaging abnormalities or visible tracer uptake (SUV_MAX 0.2).

Discussion

To the best of our knowledge, this is the first quantitative study linking increased entheseal bone turnover quantified by [18F]NaF PET to patient-reported outcomes. We found associations between bone turnover in the patellar enthesis and patient reported outcomes, specifically for the KOOS subscales of sports (−2.68, 95%CI −4.14; −1.21), quality of life (−2.97 95%CI −4.59; −1.35) and PFJ (−2.28, 95%CI −3.58; −0.98). By controlling for established imaging markers of joint damage and pain we were able to show that entheseal bone remodeling assessed by PET offers complementary information for understanding symptom generation in degenerative disease.

Interestingly, the associations between bone turnover and patient-reported outcomes were not significant for the tibial enthesis. To explain this finding, it is instructive to refer to the clinically described entity of jumper’s knee, a tendinopathy of the patella observed mainly in athletes who perform jumping sports, which also favors the proximal over the distal patella enthesis in terms of imaging findings [33]. This finding, as well as the limited association between demographic factors and entheseal bone turnover illustrate the complexity of entheseal biology, warranting further studies, including specific biomechanics.

Entheseal tracer uptake in our study population was not high, with average SUV_MAX values of 2 and 2.5 for patella and tibia respectively, and values above 4.0 in only 30 out of 135 knees. Notably, these values are lower than those seen in the subchondral bone in advanced OA [24] and much lower than those reported in a previous study of [18F]NaF-PET CT in axSpA, which showed SUV_MAX values as high as 21.4 in sacroiliac joints with erosions [34] and likely incident bony repair. These findings underscore the importance of quantitative over qualitative approaches in this specific line of research, as otherwise subtle differences and nuanced information may not be captured.

Out of the investigated demographic covariables, only age showed a (limited) association with the extend of entheseal tracer uptake in the anterior patella. This is plausible, as the elasticity of tendon decreases with age [35], thus increasing the tensile forces acting on the entheses. Somewhat surprising is the sparsity of differences between men and women, considering the well-established sex differences in bone remodeling [36]. A possible explanation could be that less than half our female subjects were postmenopausal, as the many important bone-related metabolic differences between men and women take effect after menopause [37].

Apart from patient-reported outcomes we also investigated the relationship between established imaging markers of entheseal pathology and bone turnover; partly due to the applied imaging protocol, we mainly focused on bony enthesophytes [38]. Micro injury at the entheses triggers the differentiation of resident mesenchymal cells to osteoblasts and thus leads to new bone formation in a process similar to fracture repair [39,40]. There are certain limitations to the use of such imaging markers to assess entheseal strain, however. Firstly, little is known about the longitudinal behavior of these lesions, and it is plausible to hypothesize, that once the bony protrusion is formed, it does not recede, even if the mechanical strain normalizes. Secondly, it is likely the last part of the local reaction after the tissue undergoes strain and may thus be considered only the tip of the iceberg. In our cohort, presence of bony irregularities was not associated with any of the PROs, which is in line with previous findings in patients with psoriatic arthritis [41].

An imaging marker of arguably larger importance is bone marrow of the enthesis – in our relatively healthy population, we did not observe this imaging pathology. Bone marrow edema, i.e. increase in fluid signal within the bone marrow, depicted on fat-suppressed, fluid-sensitive MR sequences, is a non-specific finding, and may be observed with the widest array of pathologies, including trauma, tumor and infection [42]. The relative lack of bone marrow edema, even in cases of markedly increased bone turnover, is at first surprising but may be explained from different angles. Firstly, it is well known from the imaging of traumatic bone injury, that compressive trauma is associated with much more intense edema than distractive trauma – in fact, avulsion fractures as the most extreme form of distractive trauma can be observed without any edema, while compressive injuries without discernible fracture may show intense edema, sometime termed a bone bruise [43,44]. Thus, the absence of edema does not exclude substantial tensile forces acting on these regions. Secondly, it has been argued that the observed entheseal bone marrow edema is a sign of local inflammation, with vasodilation and locally altered membrane permeability [39]. While the mechanical stress at the entheses can doubtlessly trigger such an inflammatory reaction, our findings indicate that an inflammatory reaction, of the extent that it causes macroscopic imaging changes in the adjacent bone, is not mandatory for patient discomfort.

Due to (potentially prohibitively) high costs, the widespread adoption of [18F]NaF PET MRI in clinical routine imaging is unlikely. Thus far, [18F]NaF PET has been used on its own or in combination with CT in rheumatological research, showing increased bone turnover in PsA patients in whole body imaging [45], in a separate study in the enthesis-nail complex [46], and in comparison to [68Ga] fibroblast activation protein inhibitor (FAPI), with the latter showing a closer association to clinical inflammatory activity in PsoA [47]. Expanding on our findings in non-inflammatory disease, this imaging technique may be used to develop imaging biomarkers from routine clinical MRI or ultrasound or aid in the validation of bone texture derived biomarkers from quantitative CT [48] or advanced MR imaging techniques such as ultra-short/zero echo time imaging (UTE/ZTE) [49]. Such advanced imaging markers of entheseal strain could be of great value in a wide array of both inflammatory and degenerative diseases, including axSpA and PsA but also non-specific low back pain, diffuse idiopathic skeletal hyperostosis and osteoarthritis. Furthermore, imaging biomarkers as objective markers of pain may become valuable tools in future clinical trials on disease modifying osteoarthritis drugs (DMOADs).

Our study had limitations that warrant critical discussion. The imaging protocol applied in this study was targeted at cartilage pathology, and thus not optimized to image pathology of tendons, entheses or bone marrow – it is for example possible, that very subtle bone marrow edema and smaller bony irregularities may have been missed. Furthermore, in inflammatory conditions of the enthesis, enhancement after contrast injection is considered a sensitive marker [50] and knowledge of the relationship between perfusion and bone turnover would have provided a more comprehensive picture of the entheseal pathobiology. Also, our method only assessed the maximum observed tracer concentration, without a more in-depth analysis of uptake patterns, treating focal hot spots and increased tracer uptake across the whole entheseal region the same; such analyses remain as an objective for future work. Lastly, the limited sample size prevented further subgroup and sensitivity analyses, such as a comparison between pre- and postmenopausal women.

In conclusion, [18F]NaF PET as a surrogate marker for bone remodeling provides novel insights into entheseal biology, furthering the understanding of the incomplete concordance of imaging findings and patient-reported outcomes in degenerative disease. Using these methods, advanced imaging biomarkers for entheseal stress may be developed, that could potentially help in parsing mechanical from inflammatory triggers of pain in different rheumatic diseases, including osteoarthritis and SpA, by supplying information that is complementary to currently used imaging techniques.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.joca.2025.09.008.