Introduction

The intra-articular insertion of preparations of hyaluronic acid (HA) into the knee is one of a relatively few options for the medical management of osteoarthritis.1 Despite the ubiquity of osteoarthritis, the take-up of this intervention in Australia has been low. There are a number of reasons why this has happened. Firstly, this treatment has been considered by the Federal Drug Administration (FDA) in America and subsequently by the Therapeutic Goods Administration in Australia as a 'medical device' rather than a pharmaceutical. In Australia this means that its cost is not subsidised through the Pharmaceutical Benefits Scheme. At the time of writing, a course of three injections of Synvisc (Genzyme, Ridgefield, NJ, USA) or a single injection of Synvisc One costs $A467 which must be bourne by the patient. The reasoning behind this decision perhaps lies in the prior development and use of HA as a vitreous replacement (Healon) during ophthalmic surgery.2 A second reason may be that initial thoughts about the mechanism of action of intra-articular HA in alleviating the symptoms of osteoarthritis emphasised the physico-chemical properties of high molecular weight HA rather than cell-mediated effects.3 Without a complete understanding of how intra-articular HA mediates its beneficial effect in osteoarthritis, this determination by the FDA may be premature. A third reason for poor uptake of this therapy by clinicians has been the lack of thoroughly convincing data about efficacy. Many trials have been conducted but most suffer from methodological difficulties. It is difficult to effectively blind a therapy that requires between three and five intra-articular injections at weekly intervals so many trials are non-blinded, some have not been randomised and most do not use the intention to treat criterion. Furthermore, inclusion criteria are often vague, there is variation in the time after injection that any benefit is measured, and most trials have small numbers. These methodological difficulties are best exemplified by Wang et al.,4 who in conducting their meta-analysis, identified 665 potentially relevant studies by electronically and hand searching the literature, only 20 of which satisfied preset criteria for inclusion. However, an important advance is that a single injection formulation has been developed and trialled5 and was placed on the market in late 2009. This, plus gathering evidence for efficacy, particularly an appreciation of the time course of response6 may allow intra-articular HA to take a larger role in the management of osteoarthritis.

Physiological role and properties of HA in synovial fluid

In outlining the rationale for viscosupplementation, Balazs and Denlinger3 outlined the ways in which the properties of synovial fluid may be changed when exogenous HA is added.

Synovial fluid can be considered to be a filtrate of plasma to which HA is added. HA is a very high molecular weight polysaccharide consisting of the alternating sugars D-glucuronic acid and N acetyl-glucosamine. In its natural state it is neither branched nor cross-linked. It is synthesised by the enzyme HA synthetase7 which is located on the inner surface of the cell membrane of type A synoviocytes and the nascent polysaccharide chain is extruded through the cell membrane into the extracellular space as it is synthesised. This synthetic mechanism allows the cell to deal with such a high molecular weight product. Its high molecular weight imparts to synovial fluid its high viscosity which is lost in both osteoarthritis but more so in inflammatory joint disease.8 Its functions within synovial fluid are not fully understood, and it lubricates but is not the sole lubricant.9 Its viscoelastic properties may assist in distributing load.3 The surface layers of articular cartilage receive nutrient from the synovial membrane rather than bone, and thus these nutrients must travel through the HA rich matrix of synovial fluid, which thus contributes to their facilitated transport.3 Also HA clearly has an effect on the function of a variety of cells, from inflammatory cells10 to feedback inhibition of it own synthesis which is mediated by the cell surface receptor CD44.11 Labelled studies have shown that HA circulates and exits from the joint via the lymphatics and is metabolised by cells of the reticuloendothelial system having a half-life of about 48 h.12 Artificially cross-linking HA has the effect of delaying its efflux from the joint, approximately doubling its half-life.3

Viscosupplementation

Viscosupplementation is a term introduced by Andre Balazs3 when he developed a method of chemically cross-linking HA derived from cockscomb in vitro producing a product with two components, 80% is polymer of molecular weight 6 x 10⁶ and 20% a cross-linked matrix of HA of higher molecular weight which forms a gel. This is known as hylan GF20 and is marketed as Synvisc. Other products have not used in vitro manipulation of the extracted HA. There are now at least five HA preparations available for use in the USA and two, Synvisc and Orthoartz (Seikagaku, Tokyo, Japan), are marketed in Australia (Table 1). A single injection preparation of hylan GF20 (Synvisc One) has recently come on the market.13

Preparations of hyaluronic acid available for clinical use

Efficacy

There are at least six meta-analyses4,14,15,16,17,18 (Table 2) published on the therapeutic efficacy of intra-articular HA preparations including a Cochrane review but their results have been conflicting. The large number of meta-analyses reflects variation in protocols and outcomes reported in the published trials. Positive results were reported in two analyses.4,16 A small effect was reported in two,14,15 including the Cochrane review and two others which reported that intra-articular HA was not more effective than placebo.17,18 A meta-analysis that analysed studies which compared intra-articular HA with intra-articular corticosteroids5 has also recently been published and may shed some light on this variability. They identified seven eligible trials that included 606 participants and were able to compare at multiple time points after injection. Intra-articular corticosteroids were significantly better using standard outcome measures for osteoarthritis compared to intra-articular HA at 2 weeks. The treatments were of equal efficacy at 4 weeks, but at 8 and 12 weeks post treatment intra-articular HA was superior to intra-articular corticosteroid, and this benefit extended out to 26 weeks. This variation in time course of response and the importance of adequate follow up has implications for anyone designing a clinical trial with these agents. Another variation that may affect outcome is the preparation used. In the trials analysed by Chevalier et al.5 Hyalgan was used except for one which used Orthovisc and Synvisc. This introduces another variation which may affect the outcome.

Meta-analysis of trials of intra-articular hyaluronic acid preparations

Possible modes of action

The viscoelastic properties of hylan GF20 are similar to those of knee synovial fluid of healthy young adults aged 18-27 years. Thus restoration of synovial fluid elasto-viscosity is achieved, but the HA turnover studies tell us that this is a temporary effect. That clinical benefits extend out up to 6 months tells us that other mechanisms are likely to be at play.5 These include cellular effects that may be mediated by an altered pericellular environment or by specific cell receptors. Those postulated are, promotion of endogenous synoviocyte HA production,20 stimulation of chondrocyte matrix component synthesis21 and inhibition of chondrocyte matrix metalloproteinase synthesis.22 This contention is supported by preliminary data showing improved patella cartilage quality at 8 weeks post injection compared to pre-injection using magnetic resonance imaging scans.23

Changes in intra-articular HA concentration, and the change in the improved viscoelastic properties of synovial fluid from treated knees also stretch out to 3 months. Anti-inflammatory effects of HA have been identified and a reduction in intrasynovial nitrous oxide has been measured in treated knees. Finally, Gomis et al.24 have suggested that bathing nociceptors, presumably in the synovium and possibly bone, may modulate their function.

Side-effect profile

Clinicians' choice of preparation will be influenced by the side-effect profiles of the different preparations. Adverse events with HA products are not common and the large majority are local. Even so, when analysing side-effects among trials, difficulties occur because definitions and reporting of local adverse events are not uniform from trial to trial. Furthermore, there is now a suggestion that side-effects are more common with the chemically cross-linked preparations. Reichenbach et al.19 have recently conducted a meta-analysis of trials that directly compare straight chain preparations of HA with hylan GF20 and have showed a relative risk of 1.91 for side-effects with hylan GF20 compared to the straight chain preparations. They estimated an overall occurrence rate for side-effects of 77 local adverse events per 1000 treated patients over 1 year. They also concluded that there was no difference in efficacy between the two preparations. In order to conduct their meta-analysis, they had to stratify the side-effects observed in the various trials into three categories: local pain, flare (i.e. pain and swelling or warming) and severe inflammatory reaction of the treated knee. The latter includes the occurrence of 'pseudosepsis' a side-effect sometimes due to calcium pyrophosphate crystal shedding which may require hospitalisation, re-aspiration and work-up for septic arthritis. True sepsis following injection appears no more commonly with HA preparations (including hylan GF20) than with placebo injections.

Other joints

These trials and the vast majority of the data gathered have looked at the use of these agents in treating osteoarthritis of the knee. There have, however, been a increasing number of reports using this therapy in a variety of other joints including the hip,25 the lumbar facet joints26 and the first metatarsophalangeal joint.27 The numbers of patients treated are small and the trials in general are not worthy of meta-analysis.

Conclusion

There exists no broadly recognised disease modifying therapy for osteoarthritis, that is, a therapy that substantially alters the natural history of the disease. In the last decade, a cloud has been drawn over the use of non-steroidal anti-inflammatory agents (NSAIDs), particularly the COX-2 specific NSAIDs due to their being linked with increased coronary and cerebrovascular risk.28 Therefore, therapies for osteoarthritis that do not share this risk, such as intra-articular HA preparations, need to be re-appraised. However, the evidence by which one can estimate the potential benefit of intra-articular HA therapy has been compromised by a number of factors. Most obvious is the recurrently invasive mode of administration which has limited trial size. There has also been a lack of understanding of which of the physiological effects may be mediating the observed clinical benefit and in particular at what time(s) after treatment any benefit should be measured. This has led to a multiplicity of small trials and variation in the interpretation of the meta-analyses performed on the trials. In Australia, a further disincentive is the high cost which must be completely bourne by the patient on the basis that the therapy is a medical device rather than a pharmaceutical. Signs that this situation may be changing is the recent introduction of a single injection preparation of cross-linked HA, which unfortunately has no bearing on cost but will make administration simpler and easier to study. Finally, there is an appreciation that the benefits of intra-articular HA may be delayed but more sustained when compared to intra-articular corticosteroids, which is currently the most common form of intra-articular therapy for osteoarthritis.

References