Corticosteroid injections in osteoarthritis: should we care about predicting response?

Towards a better understanding of the drivers of peripheral pain This editorial refers to Predictors of response to intra-articular steroid injections in knee osteoarthritis—a systematic review, by Nasimah Maricar et al. Rheumatology 2013; doi:10.1093/rheumatology/kes368, on pages 1022–1032. In this issue of Rheumatology, Maricar et al. [1] present a thorough systematic literature review that examines the predictors of response to intra-articular corticosteroid (IACS) injections in OA of the knee. This article highlights the paucity and poor quality of studies in this area. Most reports have had too few patients to examine any predictors, used different steroid preparations and doses, used different outcome measures and timing of their application and examined a diverse range of predictor variables. To the practising clinician with injection skills, a reasonable thing might be to question why we need such predictors. IACS therapy is long established, it has a moderate analgesic effect size unlike most OA therapies [2], is generally well tolerated by patients, is inexpensive and associated with very low rates of side effects. Many patients do not get adequate pain relief with other OA therapies or cannot tolerate them, so alternative therapeutic options are very limited. As highlighted by the authors of this review, many injections miss the joint cavity unless guided by imaging, a facility still often not feasible in many clinical situations. So why should we explore this issue? The real importance underlying the search for prediction of analgesic response is to determine the key peripheral drivers of nociceptive pain in OA, to optimize the use of existing therapies and to aid in the rational development of new therapies. Modern imaging studies of knee OA have highlighted pathology of the synovium and bone, both richly innervated structures, as the main tissue structures associated with pain [3]. Although the strengths of these structure–pain associations are never strong, when one considers the individual patient pain experience, the range of pain assessment tools (which may measure different constructs, for example, a numeric rating scale measuring global knee pain may measure something different from the WOMAC scale where the pain questions are activity related) and the rather elementary quantification of imaging-detected pathology (often using semi-quantitative scores assessed at a variable number of intra-articular sites), it is probably surprising that any associations have been demonstrated. Corticosteroids do seem to provide a good clinical model to explore analgesic response. A modest beneficial effect size of IACS in OA knee is reported, although the benefits are short-lived [2]. The route and dose of administration appear to be crucial for efficacy. Recently a 4-week randomized controlled trial (RCT) of low-dose oral prednisolone (5 mg daily) failed to demonstrate analgesic benefit for OA of the hand [4]. That this represents a failure of adequate dosing is suggested by the preliminary report of a 6-week RCT of a higher dose of prednisolone (7.5 mg) for OA of the knee that showed a positive analgesic benefit [5]. That corticosteroids work as analgesics through anti-inflammatory-related mechanisms can be reasonably assumed. Inflammation and synovitis are integral to the OA process [6]. Synovitis is present from the earliest stages of chondropathy, is to some degree proportional to the amount of cartilage loss and (when studied using the most sensitive modern imaging technique, contrast-enhanced MRI) is present in approximately 90% of symptomatic OA knees [7, 8]. Osteoarthritis and not osteoarthrosis is indeed the correct name for the process underpinning the familiar clinical syndrome [6]. The frequency of inflammation in OA makes it unlikely that merely detecting the presence of inflammation will hold the key to understanding analgesic response. However, accurately assessing inflammation in the joint may help distinguish different tissue sources of pain (e.g. synovitis vs bone pathology) within a given patient. It is also critical to examine whether extent or severity of synovitis predicts response to anti-inflammatory therapies, but it is important to realize that many different definitions of synovitis are used and these vary widely across imaging studies. US and MRI can accurately assess morphological measures of inflammation: be that synovial hypertrophy, certain intrasynovial features of inflammation (as detected with contrast-enhanced MRI and US power Doppler signal) or effusions. The studies that used US in this article did not suggest US synovial predictors but they had certain limitations and neither attempted quantification of inflammation [9, 10]; in one study, only greyscale and not power Doppler was assessed [9]. Some recent work supports the concept that inflammatory characteristics of synovial pathology will provide insights into corticosteroid analgesic response. In a study of 35 people with OA knee where approximately half were given IACS, more patients with reduced power Doppler signal at 1 week reported reduced pain, and changes in pain and PD signal were weakly correlated (H. I. Keen et al., unpublished results). The preliminary report from a study of 13 knee OA patients imaged with dynamic MRI at baseline and 2 weeks after IACS suggested that a marker of synovial vascularity (early enhancement rate) was associated with significant reduction in pain [11]. That markers of vascularity may provide a good predictor of response is supported by the growing understanding of increased angiogenesis in OA and its relationship to nerve growth and pain [12]. It is unlikely that simple bedside predictors such as pain severity (a complex construct to measure) or examination findings (given the known insensitivity of clinical examination for detecting structural OA pathology compared with modern imaging) will provide strong prediction of response to IACS. Imaging biomarkers may provide more insight and they are in a state of rapid development; however, more information is still needed to understand their responsiveness and optimal application in well-designed trials if they are to help address issues of therapeutic response. However, we should keep in mind that a better understanding of peripheral pain sensitizers and tissue-level or molecular markers of these may be required to truly predict analgesic response. Disclosure statement: The author has declared no conflicts of interest.