Article Text

Download PDFPDF

Imaging abnormalities that correlate with joint pain
  1. David T Felson1,2
  1. 1University of Manchester, Manchester, UK
  2. 2Boston University School of Medicine, Boston, Massachusetts, USA
  1. Correspondence to Professor David T Felson, Boston University School of Medicine, Suite 200, 650 Albany Street, Boston, MA 02118, USA; dfelson{at}


This article summarises evidence on structural pathology in the joint that causes joint pain. Evidence, mostly from studies using MRI scans in persons with and without joint pain has suggested that pain often originates in bone or synovium.

Statistics from

It has long been known that x-ray abnormalities of osteoarthritis often occur in the absence of pain in the joint and likewise that joint pain can be severe with little if any evidence of radiographic disease. One of the main reasons for this discordance between x-ray abnormalities and pain in the joint is that features that are pain sensitive within the joint are not visualised well or at all on the x-ray. MRI scans can identify abnormalities in many of the anatomical structures that probably cause pain. In particular, anatomical studies have suggested that ligaments, especially at their insertion sites: the outer rim of the meniscus in the knee, periosteal and subchrondral bone, synovium and soft tissues, are pain-sensitive tissues. Also, if the joint capsule gets distended by fluid, the capsule itself contains nocioceptive fibres within it that reflect this distension. Cartilage, on the other hand, is aneural and is probably therefore not a direct source of pain. As MRI can reveal structural abnormalities in almost all of the pain-sensitive tissue, it stands to reason that MRI scans might reveal the sources of joint pain. In particular, scans of those individuals with joint pain compared with those without joint pain might identify particular features that are associated with pain.

While histological studies have shown that these tissues are innervated, in-vivo human studies have corroborated the idea that these tissues, and not cartilage, are likely to be the source of pain in the joint in osteoarthritis. For example, in one study an orthopaedist underwent arthroscopy fully awake and with only the soft tissues around the joint anaesthetised. Probes inside the joint suggested1 that probes of cartilage were not painful, that probes of the insertion sites of ligaments were, and that the synovial reflection was consistently painful, along with Hoffa's fat pad. In this study the bone was not probed, although it is well known to be highly innervated.

In those with radiographic osteoarthritis only some of whom had pain, we embarked upon a series of studies to identify the findings on MRI that were associated with pain. The concept behind these studies was that if the MRI findings were unique to those with knee pain, they may be a source of this pain.

We found that among those with radiographic knee osteoarthritis, those with pain were more likely to have bone marrow lesions (35% of those with pain had bone marrow lesions vs only 2% of those without pain).2 They were also more likely to have synovial thickening on their MRI scans3 and effusions, especially large effusions that presumably stretched the articular capsule.3 Periarticular lesions such as anserine bursitis were also more prevalent in those in pain, but they were uncommon lesions on MRI.4 Bone marrow lesions are usually poorly circumscribed lesions seen on fat suppressed images (see figure 1) that occur in weight-bearing areas just underneath the cortical bone. Histological evaluations of these lesions suggest that they reflect bone trauma; in particular, lots of fibrosis and bone necrosis along with scattered osteoblasts and osteoclasts. There is actually little oedema present, at least in the histological studies that have been published.5 6 It is of note that bone bruises seen after injury in young people with knee problems, such as anterior cruciate ligament (ACL) tears, show very similar histology to these bone marrow lesions.

Figure 1

Bone marrow lesion (see arrow).

We discovered7 that bone marrow lesions were strikingly related to malalignment across the knee. For example, over 70% of knees that showed varus abnormalities of at least 7° showed a bone marrow lesion on the medial side of the knee compared with less than 20% in knees of those that were 0° or more valgus. Conversely, lateral bone marrow lesions were present in those with valgus and almost never in those with varus malalignment, suggesting that bone marrow lesions represent the structural equivalent of malalignment across the joint. Subsequent studies8 have shown that dynamic varus malalignment measured in a gait laboratory is also strongly associated with medial knee bone marrow lesions. Tibiofemoral alignment in the knee of at least 2° in either the varus or valgus direction is present in 82% of knees with clinical osteoarthritis.9 This suggests why bone marrow lesions are so prevalent in those with clinical knee osteoarthritis, that the malalignment that causes them is also highly prevalent.

While we found that synovitis was related to the presence of knee pain, the optimal examination of synovitis on MRI includes contrast enhancement with gadolinium, which permits a more comprehensive assessment of synovial proliferation and allows for distinguishing between synovial hypertrophy and fluid. Depending on whether histology, arthroscopy or MRI is used and depending on criteria, synovitis is present in 50% or more of persons with symptomatic knee osteoarthritis. In over 500 persons with or at risk of knee osteoarthritis in a study that included contrast-enhanced MRI, Baker and colleagues10 reported that synovitis of at least a mild degree was present in over 50% of persons with knee pain. Those with a lot or extensive synovitis constituted less than 20% of this sample, but those with a lot or extensive synovitis had 4.8 times the odds of mild pain (p=0.0017) and had a marked increased risk of moderate to severe knee pain (OR 9.2, p<0.0001) compared with those without synovitis. Even among persons without radiographic osteoarthritis (many of whom have some cartilage loss on their MRI suggesting early disease), extensive synovitis was present in a subsample and was associated with a marked increase in the risk of knee pain.

Cross-sectional evaluations of MRI findings related to knee pain are challenging to interpret because in clinical knee osteoarthritis the MRI often shows multiple findings that co-occur, all of which could cause pain (see figure 2). Bone marrow lesions are often present along with synovitis, effusions and other features (see below) such as bone attrition, which have been related in recent studies to pain. With such a complex picture in persons with knee osteoarthritis, it is often difficult to figure out whether there is a single source or multiple sources of pain and which of these features are causing the pain.

Figure 2

Typical osteoarthritis knee with several pathological features that could cause pain: bone marrow lesions (especially in anterior femur (thin arrow), synovitis (large white block arrow) and bone attrition (arrowhead).

With this problem in mind, our studies have increasingly examined changes in these features on MRI and whether these changes correlate with changes in pain. As in the studies above, we have focused on knee problems in which there is a known history of pain fluctuation. MRI (see figure 3) have also shown at least over a 1–2 year period of time that there is often fluctuation in these lesions. Post-traumatic bone marrow lesions, such as occur after ACL tears or other internal derangements of the knee, also resolve with time. In fact, data suggest that there is fluctuation in all of the structural findings seen on MRI that have been associated with pain.

Figure 3

Bone marrow lesions appear and enlarge over 30 months in a malaligned knee.

A series of studies has suggested that change in the MRI lesions that have been associated cross-sectionally with pain is associated in the expected direction with change in pain. For example, in a longitudinal study, Hill et al11 found that change in synovial thickening was modestly associated (r=0.3, p=0.0005) with change in pain severity, such that clinical osteoarthritis patients whose pain got worse tended to have a worsening of their synovitis. Felson et al12 reported from a longitudinal study in which two MRI were acquired at an interval of 15 months that persons with new or enlarging bone marrow lesions tended to have new-onset knee pain, and that those with knee pain at baseline tended to have worse knee pain if their bone marrow lesions got bigger. Osteoarthritis structurally tends to worsen over time. A worsening lesion may occur in the context of a deteriorating joint, and does not necessarily mean that the particular lesion is associated with worsening pain as these lesions may reflect a global worsening of disease. Recent studies have gone beyond even this two time point evaluation to look at fluctuations over multiple time periods and their relation to fluctuations in pain. In the first such study, Zhang et al13 had shown that fluctuations in bone marrow lesions, effusions and synovitis all relate to fluctuations in pain. In what may be the most important findings relevant to treatment opportunity, which after all is targeting reduction in pain, these authors showed that a reduction in the size of bone marrow lesions and in the size of the effusions was associated independently with reductions in pain. This provides strongly suggestive evidence that targeting these structural findings, if one could, might result the alleviation of pain.

A number of other insights have arisen from studies evaluating MRI features and their relation to pain. First, there are some features that are either inconsistently or not especially associated with pain in a number of studies. Englund et al14 have reported that mensical tears are associated with structural osteoarthritis but not independently with pain. The meniscal tears being studied were chronic tears such as occur in the presence of osteoarthritis, not necessarily acute tears occurring in younger persons that often present with acute swelling and pain. Tears of the ACL are quite common in osteoarthritis and are not especially associated with the occurrence of pain15 and popliteal (Baker's) cysts are not associated with knee pain.3

The relation of cartilage loss with knee pain is complicated. A number of studies have reported an association of decreased knee cartilage with knee pain.16 17 As noted earlier, cartilage contains no pain fibres and therefore would be an unlikely source of pain. However, cartilage loss releases debris into the joint fluid, which is then ingested or phagocytosed by the synovium, often generating an inflammatory response within the synovium such as lining cell hyperplasia. Synovitis is associated with pain, and cartilage loss indirectly is likely to be related to pain by dint of its effect on synovitis.

Second, as noted earlier, many lesions often coexist in patients with knee osteoarthritis, lesions that could each be associated with or cause pain. Bone marrow lesions have been the focus of a number of studies testing their association with pain, but other bone lesions appear to be related to the occurrence and perhaps severity of pain, including bone attrition,18 which is often present with bone marrow lesions but appears to synergise with bone marrow lesions in causing pain. Another lesion recently described as being related to pain19 is when cartilage is denuded to bone. In this circumstance, the bony protection afforded by cartilage, which protects bone from being damaged by loading is stripped away, leaving bone more vulnerable to damage. It is thus likely that when cartilage is completely lost, bone is more vulnerable to injury, and this situation is likely to be more strongly associated with pain than when cartilage is preserved.

In conclusion, MRI has been an extremely valuable tool for providing insights for factors into the joint that are likely to be associated with the occurrence of pain in the joint. Most of the studies so far have been performed in the knee. Periarticular lesions such as those reflecting bursitis outside the joint can also be related to knee pain and should be evaluated clinically. Studies of structures within the knee have suggested that bone marrow lesions, bone attrition, synovitis and effusions are related to the occurrence of knee pain and that their fluctuation directly relates to the fluctuation in knee pain. These findings suggest new targets for treatment in osteoarthritis; one could either focus on making these lesions resolve or diminish in size or could focus on their cause such as malalignment across the joint. Either way one might alleviate pain and stabilise disease progression.


The author greatly appreciates the technical assistance of Laura Heathers in manuscript preparation.



  • Competing interests None.

  • Funding This study was supported by a grant from NIH and Arthritis Research UK.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.