Hypointense signals in the infrapatellar fat pad assessed by magnetic resonance imaging are associated with knee symptoms and structure in older adults: a cohort study

Osteoarthritis (OA) is the most prevalent chronic joint disorder, characterized by pain and progressive deterioration of joint structures, and is strongly associated with risk factors such as age, female sex and obesity [1]. The most commonly affected joint is the knee, and the whole knee joint structures including articular cartilage, subchondral bone, synovium, ligaments, meniscus and periarticular fat pad can be affected in the course of knee OA [2]. Imaging biomarkers, especially from magnetic resonance imaging (MRI), have been used in OA research for over a decade [3]. Because of its advantage in direct visualization of morphology and integrity of the whole joint, MRI has been considered as a sensitive and accurate tool to assess cartilage loss, subchondral bone abnormalities, synovitis, and ligament and meniscal lesions [4‐6]. Quantitative or semiquantitative scoring systems have been developed for evaluating these structural changes in OA [4‐6].

The infrapatellar fat pad (IPFP), the local fat around the knee joint, may play an important role in the initiation and progression of knee OA [7, 8]. Biomechanically, it can promote efficient lubrication, reduce impact loading and absorb forces generated through the knee joint, which may be protective against OA [9]. Biochemically, it can produce various pro-inflammatory cytokines and adipokines, which may be deleterious to the knee joint [10‐13]. Pathological examination of IPFP obtained from patients with end-stage OA found that vascular neoformations, fibrosis, and chronic inflammation were present in these specimens [14]. Dragoo et al. have suggested that sagittal MRI can be used to assess abnormal IPFP quality, including fibrosis, inflammation, oedema and mass-like lesions [15].

So far, there are few clinical and epidemiological studies reporting the association between abnormal changes within the IPFP and knee osteoarthritic changes. Higher signal intensity change around the IPFP assessed by T2-weighted MRI has been considered as a surrogate for peripatellar synovitis [16, 17]. Our previous study reported that high signal intensity alteration within the IPFP was associated with knee symptoms and structural changes in older adults [18]. While hyperintense signals within the IPFP on T2-weighted MRI can indicate inflammation, acute haemorrhage and/or oedema, hypointense signals within the IPFP on T2-weighted MRI may indicate fibrosis [15]. So far, there are no studies reporting whether hypointense signals within the IPFP are associated with symptoms and structures in knee OA. The aim of this study was to describe whether hypointense signals within the IPFP measured by T2-weighted MRI are associated with symptoms or joint structural abnormalities cross-sectionally and longitudinally in older adults.

A total of 874 subjects between 50 and 80 years of age (mean, 62.1 years) took part in the present study. There were no significant differences in demographic factors (age, sex, and BMI) between these participants and those excluded (n = 226) (data not shown). Over 2.6 years, 104 subjects were lost to follow-up study due to: 25 deceased, 18 moved to other states or overseas, 12 had joint replacement, 24 physically unable, and 25 no reason specified. The remaining 770 subjects completed the follow-up study and the first 357 had the second MRI scans but not the others as the MRI machine in the hospital was decommissioned. There were no significant differences between these subjects and those without follow-up MRI, as previously described [28].

Table 1 describes characteristics of the study population. There was no significant difference in age, patellar cartilage volume and knee pain between subjects with and without IPFP hypointense signals; but the group with IPFP hypointense signals had a greater proportion of men, and higher prevalence of JSN, osteophytes, BMLs, cartilage defects, as well as higher BMI. Additionally, these subjects had greater tibial cartilage volume and tibial bone area.

IPFP hypointense signals were significantly and positively associated with baseline (data not shown) and increases in (Fig. 2) cartilage defects at all compartments in unadjusted analyses. They were significantly and positively associated with all cartilage defects after adjustment for age, sex, BMI, and radiographic OA cross-sectionally and longitudinally (Table 2). These associations remained significant after further adjustment for BMLs, except that the longitudinal association at the patellar site decreased in magnitude and became of borderline significance (Table 2).

Cross-sectionally, IPFP hypointense signals were not significantly associated with medial and lateral tibial cartilage volume, but significantly and negatively associated with patellar cartilage volume after adjustment for age, sex, BMI, radiographic OA and tibial bone area (Additional file 1: Table S1). This significant association disappeared after further adjustment for patellar cartilage defects but remained unchanged after further adjustment for patellar BMLs (Additional file 1: Table S1). Longitudinally, IPFP hypointense signals were negatively and significantly associated with change in lateral tibial cartilage volume, but not with changes in medial tibial and patellar cartilage volume, after adjustment for age, sex, BMI, radiographic OA, tibial bone area and baseline cartilage volume, and this association remained after further adjustment for cartilage defects or BMLs (Additional file 1: Table S1).

In cross-sectional analyses, IPFP hypointense signals were significantly and positively associated with any BMLs and tibiofemoral BMLs after adjustment for age, sex, BMI and radiographic OA. The associations decreased in magnitude and became non-significant after further adjustment for cartilage defects (Table 3). Longitudinally, IPFP hypointense signals were significantly and positively associated with increases in BMLs at all sites before (Fig. 2) and after adjustment for age, sex, BMI and radiographic OA (Table 3). After further adjustment for cartilage defects, the associations decreased in magnitude and became non-significant at lateral tibiofemoral and patellar sites (Table 3).

IPFP hypointense signals were significantly and positively associated with total knee pain, pain when going up/down stairs, at night while in bed, and when sitting/lying after adjustment for age, sex, BMI, and radiographic OA in cross-sectional analyses, but these significant associations disappeared after further adjustment for cartilage defects or BMLs except for pain when at night while in bed (Table 4). Longitudinally, IPFP hypointense signals were significantly associated with an increase in total knee pain, pain when walking on a flat surface, pain when going up/down stairs and when standing (Table 4), but these became non-significant after further adjusting for cartilage defects. The associations between IPFP hypointense signals and an increase in knee pain decreased in magnitude but became non-significant for all knee pain subscales (except pain when going up/down stairs) after further adjustment for BMLs (Table 4).

In cross-sectional analyses, IPFP hypointense signals were significantly and positively associated with ROA (OR: 2.91, p < 0.001), tibiofemoral joint space narrowing (OR: 2.72 and 1.59, respectively, for medial and lateral compartments; both p < 0.01), and tibiofemoral osteophytes (OR: 4.25 and 3.34, respectively, for medial and lateral compartments; both p < 0.001), after adjustment for age, sex and BMI.

Higher grade of IPFP hypointense signals was significantly associated with smaller IPFP maximal area after adjustment for age, gender, BMI, and total tibial bone area (Additional file 2: Figure S1). The associations of IPFP hypointense signals with the above outcome measures remained unchanged after further adjustment for IPFP maximal area (data not shown).

This study is the first study to investigate the association of IPFP hypointense signals with knee structural and symptom changes in older adults. We found that IPFP hypointense signals were cross-sectionally associated with increased knee symptoms, cartilage defects, BMLs, and radiographic OA, and with reduced patellar cartilage volume and IPFP maximal area. Longitudinally, these signal intensity changes predicted increases in cartilage defects at all sites and BMLs, loss of lateral tibial cartilage volume and increases in knee symptoms. The associations with knee cartilage defects remained significant after adjustment for BMLs, but the associations with BMLs and knee pain were weakened after adjustment for cartilage defects. This suggests that IPFP hypointense signals are associated with increased knee cartilage defects primarily and with knee BMLs and pain secondarily in older adults.

IPFP, an intracapsular but extrasynovial structure [29], is situated in the knee under the patella, between the patellar tendon, femoral condyle and tibial plateau [7], and is structurally similar to subcutaneous adipose tissue [30]. As a deformable soft tissue within the anterior compartment of the knee joint, it can adapt to change contours of the articular surface and is able to distribute synovial fluid within the joint cavity to reduce the articular surface cushion, besides supporting the feasibility of intra-articular ligaments [9]. Composed of a fibrous scaffold with adipose tissue and synovial recesses within it, the IPFP can secrete cytokines, adipokines, and lipid mediators [10‐13, 31]. Therefore, the IPFP may play a biphasic role in the pathologic progression of knee abnormalities.

MRI has been used to assess signal alterations within or around the IPFP, and the high signal intensity alteration was mainly considered as synovial inflammation or Hoffa synovitis [16, 32]. In addition, hypointense signals closing to the synovium within or around the IPFP was regarding as chronic synovitis [33, 34]. Hoffa synovitis is recognised as a key imaging biomarker for knee OA [6], and can predict the progression of knee OA [32]. Our previous study also suggested that high signal intensity alteration within the IPFP was associated with knee structural and symptomatic changes in older adults [18]. As IPFP signal intensity alteration may reflect different pathological changes, more studies are required to assess the roles of IPFP signal intensity changes in knee OA.

So far, there have been no studies reporting the clinical significance of IPFP hypointense signals observed on T2-weighted fat-saturated MRI images. Low signal intensity changes within the IPFP may represent fibrosis or postoperative scarring [29, 35], chronic inflammation progressing from acute inflammation of the synovium due to microtrauma of this tissue [36], or synovial thickening or fibrosis [33, 37]. A study compared low signal intensity changes in plantar fat pad on MRI with histological changes and reported that these signal changes corresponded to fibrosis [38]. This fibrosis can be induced by chronic inflammation in the synovium [36], or periarticular surgeries or trauma around knees [29]. Although the roles of synovitis, surgical history and trauma in knee OA have been identified [39, 40], there has been no evidence showing that low signal intensity within IPFP assessed by MRI is associated with symptoms and knee structural changes in the knee.

In this study, we found that in older adults, IPFP hypointense signals were consistently associated with knee cartilage defects cross-sectionally and longitudinally, independent of factors such as BMLs. They were also associated with BMLs and reduced cartilage volume cross-sectionally and longitudinally, but these associations were largely dependent of cartilage defects. Further, there was significant association between IPFP hypointense signals and knee pain, but again, these associations were largely dependent of cartilage defects rather than BMLs. These suggest that IPFP hypointense signals may induce knee structural changes and symptoms starting from cartilage defects.

The mechanisms underlying the associations between IPFP low signal intensity and knee OA measures are largely unknown. These hypointense signals within the synovial membrane have been considered as synovial fibrosis and were corresponding to chronic synovitis [33, 34]. This chronic synovitis can contribute to cartilage degradations and knee pain [41]. Other pathological changes such as adipocyte necrosis or adipose fibrosis may be observed as hypointense signals within the IPFP in MR imagines. Fibrosis is an abnormal tissue healing process that occurs sequentially from an inflammatory response to surgery or injury of the knee and severe fibrosis was found in 33 % of IPFP biopsy specimens resected from patients with end-stage knee OA [14]. Fibrosis was also found in monoiodoacetate-induced OA models and was associated with knee pain [36]. In addition, a study focusing on the effect of strenuous running on IPFP histological changes in a rat OA model reported that fibrosis within IPFP was associated with excess physical activities and related to knee pain [42]. Fibrosis within IPFP may increase cartilage contact pressures and decrease the ability of absorbing the shock through the knee, and thus induce the degradation of neighbouring knee structures including cartilage and subchondral bone. Furthermore, our current study found that this abnormal signal was negatively associated with IPFP maximum area, suggesting it may decrease this absorbing ability through reducing the IPFP size.

The main strength of this cohort study lies in a large sample size with the comprehensive MRI structural measurements. There are several potential limitations. First, the response rate at baseline was 57 %, possibly due to the extensive protocol, which did leave the possibility open for selection bias. However, there were no significant differences in age, gender and BMI between those who responded and those did not. We also had high rates of retention (82 %) to offset this. Second, we did not have radiographic OA measurements at the follow-up because X-ray is insensitive for change over this short period, so we are unable to determine the association between IPFP quality and change in radiographic OA. Third, measurement error may influence results. However, all measures were highly reproducible suggesting this is unlikely. Lastly, histological examinations were not able to be performed in this community-based study, so the pathological changes associated with IPFP low signal intensity are unknown.

In conclusion, hypointense signals in the IPFP were associated primarily with increased knee cartilage defects and also with BMLs and knee symptoms in cross-sectional and longitudinal analyses, suggesting the abnormality represented by this signal has a potentially important role in osteoarthritis progression.