Introduction Knee osteoarthritis (KOA) is a chronic disorder characterized by joint pain, increasing physical disability, and progressive cartilage degeneration, which can lead to total knee arthroplasty (TKA). Despite extensive research, the complex pathophysiology of KOA remains incompletely understood. As a result, no established disease‐modifying treatment exists today, and the management of KOA still relies on a combination of nonpharmacologic and pharmacologic modalities 1 primarily intended to relieve the symptoms of pain and loss of knee function. In clinical practice, acetaminophen and nonsteroidal antiinflammatory drugs (NSAIDs) are generally recommended to relieve pain and improve joint function 2, 3, 4, 5. However, due to their vascular and gastrointestinal toxicity 6, 7, the risk‐benefit ratio of these rapid‐acting drugs might not be favorable, particularly for long‐course therapies and within the aging population in which KOA is most prevalent 8. Intraarticular (IA) injections of corticosteroids such as triamcinolone hexacetonide and methylprednisolone acetate are also commonly prescribed. As a systemic absorption occurs following IA corticosteroid injection, systemic adverse events (AEs) can be expected and precautions should be observed in patients with concomitant diseases such as hypertension or diabetes mellitus 9, 10, 11, 12, 13. Box 1 Significance & Innovations Besides efficacy and safety, this review also addresses less frequently evaluated aspects of the intraarticular use of hyaluronic acid (HA) in the management of knee osteoarthritis (KOA) such as effectiveness and efficiency. This is the first time that a group of experts recommends systematic repeated intraarticular HA treatment in KOA patients who had a beneficial response to a previous cycle of treatment. IA hyaluronic acid (HA) is a local treatment modality devoid of the systemic AEs observed after IA corticosteroid injection or oral administration of analgesics and NSAIDs. Thus, IAHA represents an alternative to analgesics and NSAIDs in patients with comorbidities, as well as a secondary option in case of inadequate response to first‐line pharmacologic KOA treatments 14. Registered as medical devices in the US, IAHA preparations are currently approved for the symptomatic treatment of KOA, one of the most commonly affected joints. Despite being widely employed in the daily management of KOA for almost 20 years in the US (about 30 years in Europe and Japan), controversies persist regarding their efficacy and safety, as highlighted by discrepancies in the guidelines related to the use of IAHA in clinical practice 2, 3, 4, 5, 15, which largely reflect the divergent conclusions drawn by meta‐analyses on the topic 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30. For this purpose, the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis (ESCEO) convened an international task force of experts in OA and clinical research methodology. Seven members of this working group (CC, FR, PR, OB, GH‐B, AM, and DU) were entrusted with the task of preparing a review on diverse aspects of the use of IAHA in the management of KOA. Current knowledge on the mode of action, efficacy, effectiveness, safety, and cost‐effectiveness of this treatment modality was presented and discussed at a 1‐day meeting in January 2016 in Geneva (Switzerland). The objective of this review is to provide specialists and practicing physicians with clear, concise, and reasoned answers to questions they might have on the use of IAHA in the management of KOA. Efficacy Efficacy of IAHA on pain and joint function has been evaluated in numerous randomized controlled trials (RCTs). Consequently, 15 meta‐analyses/systematic reviews assessing the symptomatic effects of IAHA on KOA have been published to date 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30. Six of them concluded on a clear efficacy of IAHA 17, 20, 24, 27, 29, 30, 4 considered it as a marginally efficacious treatment modality 16, 19, 25, 26, and 2 meta‐analyses found no significant difference in efficacy between IAHA and IA placebo 18, 21 (Table 1). Table 1 Meta‐analyses evaluating intraarticular hyaluronic acid in the symptomatic treatment of knee osteoarthritisa Reference Comparator Main outcome ES (95% CI) General outcome Lo et al, 2003 16 IA placebo Pain change from BL at M1–4 SMD = 0.32 (0.17, 0.47) Intermediate Wang et al, 2004 17 IA placebo Pain on movement MPD = 7.9 (4.1, 11.7) Positive Arrich et al, 2005 18 IA placebo Pain on movement at W2–6 WMD = 3.8 (−1.4, 9.1) I2 = 81% Negative (no clinical difference) Pain on movement at W10–14 WMD = 4.3 (0.9, 7.6) I2 = 0% Pain on movement at W22–30 WMD = 7.1 (2.4, 11.8) I2 = 0% Modawal et al, 2005 19 IA placebo Pain change at W5–7 MD = 17.6 (7.5, 28.0) Intermediate Pain change at W8–12 MD = 18.1 (6.3, 29.9) Pain change at W15–22 MD = 4.4 (−15.3, 24.1) Bellamy et al, 2006 20 IA placebo Pain/function at W5–13 WMD/SMD variable for different HA products Positive Medina et al, 2006 21 IA placebo Pain MPD = 2.4 (−0.6, 5.5) Negative (no statistical difference) Function MPD = 2.0 (−0.8, 4.9) Stiffness MPD = 5.5 (2.2, 8.8) Reichenbach et al, 2007 22 IA hylan Pain at end of followup SMD = −0.27 (−0.55, 0.01) I2 = 88% HA > hylan Bannuru et al, 2009 23 IA CS Pain change from BL at W2 SMDg = −0.39 (−0.65, −0.12) I2 = 47% Positive (HA > CS from W8 onward) Pain change from BL at W4 SMDg = −0.01 (−0.23, 0.21) I2 = 37% Pain change from BL at W8 SMDg = 0.22 (−0.05, 0.49) I2 = 47% Pain change from BL at W12 SMDg = 0.35 (0.03, 0.66) I2 = 49% Pain change from BL at W26 SMDg = 0.39 (0.18, 0.59) I2 = 0% Bannuru et al, 2011 24 IA placebo Pain change from BL at W4 SMDg = 0.31 (0.17, 0.45) I2 = 75% Positive Pain change from BL at W8 SMDg = 0.46 (0.28, 0.65) I2 = 75% Pain change from BL at W12 SMDg = 0.25 (0.15, 0.36) I2 = 60% Pain change from BL at W16 SMDg = 0.20 (0.11, 0.30) I2 = 7% Pain change from BL at W24 SMDg = 0.21 (0.10, 0.31) I2 = 32% Colen et al, 2012 25 IA placebo Pain change from BL at M3 vs. IA placebo: WMD = 10.2 (4.4, 16.0) I2 = 92% Intermediate vs hylan SMD = 0.07 (−0.10, 0.24) I2 = 72% Rutjes et al, 2012 26 Sham or no intervention Pain at end of followup SMD: 0.37 (0.28, 0.46) τ2 = 0.09 Intermediate Miller and Block, 2013 27 IA placebo Pain at W4–13 SMD = 0.43 (0.26, 0.60) I2 = 73% Positive Pain at W14–26 SMD = 0.38 (0.21, 0.55) I2 = 75% Function at W4–13 SMD = 0.34 (0.16, 0.51) I2 = 54% Function at W14–26 SMD = 0.32 (0.18, 0.45) I2 = 69% Bannuru et al, 2014 28 NSAIDs Pain change from BL at W4 SMDg = −0.01 (−0.18, 0.15) I2 = 0% Positive (HA = NSAIDs) Pain change from BL at W12 SMDg = 0.05 (−0.17, 0.28) I2 = 30% Pain change from BL at end of followup SMDg = 0.07 (−0.10, 0.24) I2 = 16% Bannuru et al, 2015 29 Other options Pain change from BL at M3 vs IA placebo: SMDg = 0.34 (0.26, 0.42)b Positive vs oral placebo: SMDg = 0.63 (0.39, 0.88)b Function change from BL at M3 vs IA placebo: SMDg = 0.30 (0.20, 0.40)b vs oral placebo: SMDg = 0.45 (0.08, 0.84)b Richette et al, 2015 30 IA placebo Pain at W12 SMD = 0.20 (0.12, 0.29) I2 = 32% Positive Function at W12 SMD = 0.12 (0.02, 0.22) I2 = 0% a Positive effect sizes (ES) favor intraarticular hyaluronic acid, negative ES favor comparator. 95% CI = 95% confidence interval; IA = intraarticular; BL = baseline; M = month; SMD = standardized mean difference; MPD = mean percent difference; W = week; WMD = weighted mean difference; MD = mean difference; HA = hyaluronic acid; hylan = hylan G‐F 20; CS = corticosteroid; SMDg = standardized mean difference adjusted for small sample size (Hedges' g); NSAIDs = nonsteroidal antiinflammatory drugs. b 95% credible interval. The reported effect sizes (ES) for pain favored IAHA over IA placebo and ranged from 0.20 (30) to 0.46 (24). Heterogeneity of outcomes between trials was relatively high in several meta‐analyses 16, 20, 24, 27. As shown by Bannuru et al, the benefits of IAHA vary over time, with a maximal effect on pain at 8 weeks (ES 0.46 [95% confidence interval (95% CI) 0.28, 0.65]) that is still observable at 24 weeks (ES 0.21 [95% CI 0.10, 0.31]) 24. This time dependency may partly contribute to the differences in pain estimates that have been observed between meta‐analyses. Another possible explanation is the inclusion of clinical trials that widely differ in methodological quality. Thus, when the analysis was restricted to high‐quality trials, an ES of 0.34 (95% CI 0.02, 0.67) and of 0.20 (95% CI 0.03, 0.37) in favor of IAHA was obtained at 8 and 24 weeks, respectively 24. These results were confirmed by Richette et al, who examined clinical data obtained from IA placebo controlled trials with a low risk of bias only (n = 8) in order to reach the highest level of evidence. Based on an ES of 0.20 (95% CI 0.12, 0.29) for pain at 12 weeks, the authors concluded that IAHA provided a moderate but real effect on pain in patients with KOA 30. Since the estimates for pain mentioned above were based on the difference between the IA placebo and the IAHA effects, it is worth noting that evidence for a clinically significant response of the IA delivery method itself has recently been demonstrated in a meta‐analysis, with an ES of 0.29 (95% credible interval [95% CrI] 0.04, 0.54) in favor of IA placebo compared with oral placebo 29. In the same study, different treatment options of KOA were compared based on their ES versus oral placebo at 3 months. IAHA was found to be the most efficacious intervention with an ES of 0.63 (95% CrI 0.39, 0.88), followed by IA corticosteroids (ES 0.61 [95% CrI 0.32, 0.89]), diclofenac (ES 0.52 [95% CrI 0.34, 0.69]), ibuprofen (ES 0.44 [95% CrI 0.25, 0.63]), naproxen (ES 0.38 [95% CrI 0.27, 0.49]), celecoxib (ES 0.33 [95% CrI 0.25, 0.42]), and IA placebo. With an ES of 0.18 (95% CrI 0.04, 0.33), acetaminophen was considered as not superior to oral placebo in relieving pain 29. In direct comparison, IAHA was shown to be not significantly different from continuous oral NSAID treatment at 4 and 12 weeks regarding pain, function, and stiffness 28, and superior to IA corticosteroids from 8 to 26 weeks regarding pain 23. Effectiveness The extent to which IAHA achieves its intended effect in the “real‐life” clinical setting cannot be measured in RCTs, since patients in such studies are not representative of those seen in usual practice. Therefore, observational studies of real practice are better suited in evaluating effectiveness of IAHA. Petrella and Wakeford retrospectively assessed the effectiveness of IA crosslinked HA (hylan) using the Southwestern Ontario database, a Canadian real‐world cohort 31. For this purpose, they identified 1,263 patients with OA in 1 or both knees that received 2 consecutive series of IA hylan injections and no other prescribed OA medications. They compared them to a cohort of 3,318 demographically matched KOA patients who were never treated with IAHA. All patients were evaluated fully between 2006 and 2012. Results showed that in the group of patients who received repeated treatments of IA hylan, pain at rest and pain after a 6‐minute walk decreased by mean ± SD 3.7 ± 1.8 points and 5.6 ± 1.7 points on a 10‐point visual analog scale, respectively. In parallel, the distance walked in a 6‐minute walk test increased on average by 115 meters in this patient group. These improvements in pain and physical function were significantly greater than those achieved in KOA‐matched patients treated with other prescribed OA medications (intergroup comparison P 75 years, who are not recommended to take oral NSAIDs 4, 14. EULAR considers the use of IAHA for pain relief and knee functional improvement based on level 1B evidence 2. OARSI guidelines for the nonsurgical management of KOA probably raise the most considerable amount of misunderstanding. However, as clearly stated in the main part of the article, an “uncertain” classification was “…not intended to be a negative recommendation or preclude use of that therapy. Rather it indicates a role for physician‐patient interaction in determining whether this treatment may have merit in the context of its risk: benefit profile and the individual characteristics, co‐morbidities, and preferences of the patient” 3. In other words, OARSI privileges the use of IAHA for specific clinical phenotypes, which should be defined by the prescribing physician. Finally, as recently highlighted by Altman et al 50, clinical practice guidelines, and a fortiori recommendations for the use of IAHA in the management of KOA, are not intended to create uniformity and suppress treating physicians' self‐analysis of the patient situation; they are designed to provide the best evidence‐based information available to help physicians in making a treatment decision. Mode of action In the US, from a purely regulatory point of view, the symptomatic effects of most of the IAHA preparations are considered to primarily result from the unique properties of HA in solution; when injected intraarticularly, exogenous HA is able to compensate for the drop in HA concentration and chain length that has been observed during the progression of OA, thus restoring the elastic and viscous properties of the synovial fluid, which are responsible for its resistance to compression and its lubricating effect, respectively 51, 52. However, these direct mechanical effects of exogenous HA cannot account for its long‐term benefits observed in clinical trials 24, as it is cleared from the joint within a few days, depending on the IAHA preparation. Over the past 30 years, several possible pharmacologic mechanisms of action explaining how the clinical effects of IAHA could persist for several months have been proposed. A recent review of the preclinical basic science literature performed by Altman et al 53 highlighted the major role of HA binding to cluster of differentiation 44 receptors in this complex mechanism, as numerous of its mediated effects (e.g., inhibition of interleukin [IL]–1β, IL‐6, and matrix metalloproteinase [MMP] expression, and reduction in prostaglandin E2 synthesis) contribute to the chondroprotection, proteoglycan/glycosaminoglycan synthesis, antiinflammatory, and subchondral effects as observed in vitro. Alternative pathways involving HA binding to intercellular adhesion molecule 1 and modulation of transient receptor potential vanilloid channel 1 activity 54 have also been described. In addition, the toll‐like receptor (TLR) signaling pathway has been suggested to contribute to the antiinflammatory and anticatabolic (inhibition of inflammation‐induced activation of MMPs) effects of IAHA in joint tissues 55, 56. Indeed, there is evidence that HA fragments that are highly expressed in the OA joint can transduce inflammatory signals through TLR2, TLR4, or both. When exogenous HA is injected in the joint, the proportion of these fragments is reduced, thus modulating TLR‐mediated innate immune responses. The ESCEO group is not convinced that one mode of action is sufficient to explain the therapeutic trajectory of IAHA and privilege the hypothesis that several mechanisms overlap and interact to relieve OA pain. Furthermore, in vitro results obtained on chondrocytes may not be the same as what happens clinically (i.e., in an anaerobic medium). Systematically repeated treatment Is it useful and acceptable to propose to re‐inject patients with a low level of pain? This very relevant issue in daily clinical practice has been addressed in the AMELIA (Osteoarthritis Modifying Effects of Long‐Term Intra‐Articular Adant) study 44, a long‐term, randomized, placebo‐controlled trial carried out in 306 patients with symptomatic KOA who received 4 consecutive cycles (5 weekly IA injections) of either noncrosslinked HA of biofermentative origin or placebo. Followup visits were conducted at 6 months after the first and second cycles and at 12 months after the third and fourth cycles, resulting in a total study duration of 40 months. At the end of the study, the responder rate according to the Outcome Measures in Rheumatology/OARSI response criteria 57 was significantly higher in the IAHA group than in the control group (80.5% versus 65.8%; RR 1.22 [95% CI 1.07, 1.41], P = 0.004). More interestingly, the number of responders to IAHA progressively increased after each treatment cycle, while response to IA placebo remained fairly stable, with a statistically significant between‐group difference from 1 year onward (P < 0.05). No increase in AEs occurred with repeated cycles of IAHA. Based on these results, the ESCEO task force encourages the use of repeated cycles of IAHA in patients who responded to the first injection, starting a new treatment cycle as soon as the first symptoms appear. Molecular weight There are more than 80 marketed IAHA preparations worldwide. They differ in many characteristics, including origin (animal versus biofermentation), mean MW (500–6,000 kDa) and MW distribution, molecular structure (linear, crosslinked, and a mix of both), method of crosslinking, concentration (0.8–30 mg/ml), volume of injection (0.5–6.0 ml), and posology. Some of the preparations include different concentrations of additives, such as mannitol, sorbitol, or chondroitin sulfate. While each of these parameters may theoretically have an impact on the effect of the IAHA treatment, research has mostly focused on the potential differences resulting from the size of HA. For this purpose, the exogenous HA available for IA injections are divided into 3 MW categories: low (500–730 kDa), intermediate (800–2,000 kDa), and high (2,000–6,000 kDa), including crosslinked formulations of HA 58. The rationale for an influence of the MW results from research on the mode of action of IAHA. Indeed, as summarized in the review by Altman et al 53, basic preclinical science showed that higher MW HAs may provide superior chondroprotective, proteoglycan/glycosaminoglycan synthesis, antiinflammatory, mechanical, and analgesic effects. Whether these differences observed in vitro translate into clinical evidence is not clearly established yet. Two meta‐analyses comparing different IAHA preparations have been published to date (Table 1). The first one compared hylan with lower‐MW IAHA and found no clinically relevant benefits in terms of efficacy of either type of preparation. However, the risks for local AEs (RR 1.91 [95% CI 1.04, 3.49]; I2 = 28%) and postinjection flares (RR 2.04 [95% CI 1.18, 3.53]; I2 = 0%) were observed to be twice as high with hylan than with low‐ or intermediate‐MW HA 22. In addition, a higher number of pseudosepsis cases were reported with hylan than with other IAHA preparations, whose risk might increase with subsequent courses 40. In the second meta‐analysis, comparing different commercially available IAHA products, the authors were not able to conclude that one brand had a better efficacy than another due to the heterogeneity of the studies and outcomes 25. In addition, most head‐to‐head RCTs performed to date have found noninferiority with respect to symptomatic efficacy between the various HA preparations evaluated 59, 60, 61, 62, 63, 64. To our knowledge, only 1 RCT was able to demonstrate a statistically significant difference between 2 IAHA preparations varying in MW regarding symptomatic efficacy. Indeed, this study conducted in 400 patients with KOA showed that an intermediate‐MW HA provided statistically superior pain relief at 6 months than a low‐MW HA. Secondary end points confirmed the primary finding 58. Based on the above, the ESCEO task force considers that there is currently no clinical evidence supporting an advantage in efficacy of one product over another. Furthermore, if certain intrinsic properties (e.g., MW) of particular IAHA preparations provide beneficial results in comparison with other IAHA products, characteristics related to the patient are likely to prevail. Efficiency Only 9 pharmacoeconomic studies evaluating the cost‐effectiveness of IAHA have been published to date (Table 3) 65, 66, 67, 68, 69, 70, 71, 72, 73. Most of them were conducted in patients with KOA who received IA injections of hylan 65, 66, 67, 71. Table 3 Pharmacoeconomic studies evaluating the cost‐effectiveness of IAHA in the symptomatic treatment of knee osteoarthritisa Reference Source of data Country Duration Product vs. comparator Cost‐effectiveness ICER Waddell et al, 2001 (65) Cohortb US 3 years Synvisc + AC vs. AC NA Dominant Torrance et al, 2002 (66) RCT 74 Canada 12 months Synvisc + AC vs. AC NA $10,000 Canadian per QALY Kahan et al, 2003 (67) RCT 67 France 9 months Synvisc vs. CC €829 for 32% improvement in Lequesne index Dominant Yen et al, 2004 (68) Cohortb Taiwan 6 months Artz vs. naproxen $1,538 US per QALY $42,000 US per QALY Mazières et al, 2007 (69) Cohort 69 France 6 months Suplasyn vs. CC (BL) €528 for 27% improvement in Lequesne index Dominant Turajane et al, 2007 (70) Retrospective 75 Thailand 2 years Hyalgan + AC vs. AC NA NA Chou et al, 2009 (71) Cohort 71 Taiwan 6 months Synvisc vs. CC (BL) $299,456 NT per QALYc NA Artz vs. CC (BL) $394,021 NT per QALY NA Miller and Block 2014 (72) Cohort 76 US 2 years Registered IAHA vs. CCb $12,800 US per QALY < $23,400 US per QALY Hatoum et al, 2014 (73) RCTs 77, 78 US 12 months Euflexxa vs. CC (BL) $21,281 US per QALY Dominant Hatoum et al, 2014 (73) RCTs 74, 77, 78 US 12 months Euflexxa vs. AC $8,816 US per QALY $38,741 US per QALY a ICER = incremental cost‐effectiveness ratio; AC = appropriate care (e.g., nonpharmacologic therapy, nonsteroidal antiinflammatory drugs [NSAIDs], analgesics, corticosteroids, total knee arthroplasty); NA = not available; RCT = randomized controlled trial; QALY = quality‐adjusted life year; CC = conventional care (i.e., nonpharmacologic therapy, NSAIDs, analgesics); BL = baseline. b Hypothetical. c $1 NT = $0.0315 US (2008 exchange). Although pharmacoeconomic data are overall in favor of IAHA, only a few evaluations included sensitivity analyses or were performed based on RCTs. Furthermore, some costs were not included, such as those associated with potential AEs. Therefore, the ESCEO task force recommends investigating more in the identification, measurement, and comparison of the costs, risks, and benefits of IAHA in the management of KOA. Conclusion This review addresses major aspects related to the use of IAHA in the management of KOA. It emphasizes its good safety profile and its moderate but real efficacy on symptoms, which is in the same range as other pharmacologic treatment modalities (e.g., NSAIDs) used in this indication. The effectiveness of IAHA has also been highlighted based on real‐life clinical data and should therefore be helpful to clinicians when making care decisions tailored to individual patient needs. The ESCEO working group is aware that IAHA is not a panacea for treating a heterogeneous disease such as KOA. However, KOA is known to mostly affect the elderly and individuals with significant comorbidities, thereby constraining the use of several conventional therapeutic options such as acetaminophen, NSAIDs, opioids, duloxetine, or TKA. Therefore, the experts convened by the ESCEO feel that the medical community cannot afford to neglect the use of IAHA, particularly now that systematic repetitive treatment cycles have been shown to yield positive results in terms of both efficacy and safety. Indeed, they consider that IAHA is a reasonable alternative in patients with KOA who have not sufficiently responded to previous pharmacologic treatments, and a key weapon in the therapeutic armamentarium for managing KOA in patients with contraindications to conventional interventions. The ESCEO task force calls for additional well‐conducted clinical trials (RCTs, cohort studies) to define the predictive factors (i.e., patient phenotypes, treatment characteristics) associated with an optimal risk‐benefit ratio. Such research may further help in determining candidates that may take most advantage of IAHA. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Cooper had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design Cooper, Bruyère, Reginster. Acquisition of data Cooper, Rannou, Richette, Bruyère, Herrero‐Beaumont, Migliore, Uebelhart, Reginster. Analysis and interpretation of data Cooper, Rannou, Richette, Bruyère, Al‐Daghri, Altman, Brandi, Collaud Basset, Herrero‐Beaumont, Migliore, Pavelka, Uebelhart, Reginster. ADDITIONAL DISCLOSURE Author Collaud Basset is an employee of TRB Chemedica International SA.
