To evaluate the efficacy and tolerability of a single intraarticular (IA) injection of hyaluronic acid (HA) for the treatment of hip osteoarthritis (OA).
To evaluate the efficacy and tolerability of a single intraarticular (IA) injection of hyaluronic acid (HA) for the treatment of hip osteoarthritis (OA).
A multicenter, randomized, parallel-group, placebo-controlled trial was conducted over 3 months. Patients (older than 30 years) with symptomatic hip OA (pain score of >40 mm on a visual analog scale [VAS]) and a Kellgren/Lawrence grade of 2 or 3 were randomly assigned to receive 1 fluoroscopically guided IA injection of HA (2.5 ml) or placebo (2.5 ml). Patients were followed up for 3 months. The main outcome measure was pain score on a VAS (100 mm) at month 3 compared with baseline. Secondary outcome measures were the proportion of responders defined by Osteoarthritis Research Society International criteria; Western Ontario and McMaster Universities Osteoarthritis Index subscores for pain, stiffness, and disability; and patient and physician global assessment. Randomization was computer generated. HA and placebo preparations were placed in numbered identical containers, and syringes were covered with masking tape. Physicians assessing outcomes were blinded with regard to group assignment.
Eighty-five patients were randomized to the HA group (n = 42) or placebo group (n = 43). Baseline characteristics were similar between the 2 groups. At 3 months, the decrease in pain score did not differ between the HA and placebo groups in the intent-to-treat analysis (mean ± SD decrease 7.8 ± 24.9 mm with HA versus 9.1 ± 27.4 mm with placebo; P = 0.98). The responder rates were 33.3% and 32.6% in the HA and placebo groups, respectively (P = 0.94). Other secondary end points did not differ between the groups, nor did use of rescue medication or frequency of adverse events.
Our findings indicate that a single IA injection of HA is no more effective than placebo in treating the symptoms of hip OA.
Osteoarthritis (OA) is the most common type of arthritis and the major cause of disability in elderly populations worldwide. Hip OA is the second most frequent form of OA affecting a large joint, and its prevalence ranges from 3% to 11% in populations older than 35 years (1–3).
Current therapies for hip OA include a combination of nonpharmacologic and pharmacologic treatments (4–6). Viscosupplementation is an intraarticular (IA) therapeutic modality that is based on the physiologic importance of hyaluronic acid (HA) in synovial joints. Its therapeutic goal is to restore the viscoelasticity of synovial fluid and the natural protective functions of HA in the joint. In vitro and in vivo studies have also suggested that HA could have protective effects on cartilage extracellular matrix and could reduce the production and activity of proinflammatory mediators and matrix metalloproteinases (7).
Viscosupplementation as therapy for knee OA has been the focus of numerous controlled trials and the subject of meta-analysis. In a recent review, Bellamy et al (8) concluded that viscosupplementation is effective for knee OA and has beneficial effects on pain and function. However, this efficacy is controversial (9, 10), and some authors do not recommend the use of IA injections of HA for the treatment of knee OA (11).
Data on the efficacy of viscosupplementation for hip OA are scarce, and there is even less evidence supporting the use of IA injections of HA for the treatment of hip OA. Results of several open-label trials have suggested that HA treatment could improve pain and function (for review, see refs. 12–14), although no definitive conclusions can be drawn from these studies because of their lack of a placebo group (12). The recent European League Against Rheumatism recommendations for the management of hip OA emphasized the need for a randomized controlled trial (RCT) of HA (5). Since then, Qvistgaard et al (15) have reported the results of a double-blind RCT involving 3 IA injections of a low molecular weight HA preparation in patients with hip OA. That trial did not demonstrate a significant beneficial effect of HA over placebo on pain or function (15). In view of these results, the efficacy of HA in hip OA still needs to be shown.
Repeating IA injections in a deep joint is problematic because of the need for fluoroscopy or ultrasound guidance and the theoretical increased risk of side effects such as flares or septic arthritis (16, 17). The results of a previous uncontrolled clinical trial suggested that 1 injection of HA could be an effective therapy for hip OA, because the proportion of responders, defined according to the Outcome Measures in Rheumatology Clinical Trials (OMERACT)–Osteoarthritis Research Society International (OARSI) criteria, was 53.6% (18). Thus, determining the efficacy of a single injection of HA as compared with placebo in patients with hip OA seems pertinent to clinical practice. For this purpose, we conducted a 3-month, randomized, placebo-controlled trial of a single injection of HA in patients in whom hip OA failed to respond to conventional pharmacologic therapy.
We conducted this multicenter, prospective, randomized, double-blind, placebo-controlled study in France over 12 weeks, between January 2005 and March 2006, in accordance with the ethics principles of the Declaration of Helsinki. The length of the inclusion period was planned to be 14 months because April 2006 corresponded to the use-by date of the placebo preparations. This study was approved by the research ethics committee of Henri-Mondor Hospital. Written informed consent was obtained from patients before inclusion. This report complies with the guidelines of the Consolidated Standards of Reporting Trials Statement (19).
Outpatients fulfilling the American College of Rheumatology criteria for the diagnosis of hip OA (20) were recruited from 26 rheumatology departments in France. To be included in the study, patients had to be ages 30–80 years, have radiographically confirmed hip OA, have a Kellgren/Lawrence grade (21) of 2 or 3, and have had symptoms of hip OA for at least 1 month, defined as daily pain score between 40 and 80 mm on a 100-mm visual analog scale (VAS) despite treatment with acetaminophen (4 gm/day) and/or nonsteroidal antiinflammatory drugs (NSAIDs) taken regularly in adequate doses. Exclusion criteria included pregnancy, a Kellgren/Lawrence grade of 1 or 4, major acetabular dysplasia of the target joint, inflammatory joint disease, chondrocalcinosis of the hip, history of allergy or intolerance to HA, skin changes in the injection area with risk of infection, surgery on the target hip within the last 6 months, intermittent claudication, current anticoagulant therapy or viscosupplementation within the last 6 months, and oral corticosteroid treatment or IA corticosteroid injection into the hip or knee joint within the last month. Radiographic evidence of hip OA and other radiographic eligibility criteria were checked by a central reader (TC).
Patients were randomly assigned in a 1:1 ratio to receive a single injection of 2.5 ml of HA (Adant; Daiichi Sankyo, Reuil Malmaison, France) or 2.5 ml of saline water (placebo). Adant is a biotechnically obtained HA (Streptococcus fermentation) with an average molecular weight of 900,000 daltons.
Patients were assigned to the treatment or placebo group according to a preestablished computer-generated global randomization list. To preserve allocation concealment, sealed containers containing the treatment (HA or placebo) were centrally prepared and administered serially to participants.
Under fluoroscopic guidance, a needle was directed into the hip joint, and arthrocentesis was performed before each injection. Trained physicians performed all injections. In the absence of synovial fluid, the IA positioning of the needle was verified by the injection of 0.1–1 ml sodium and meglumine ioxaglate (Hexabrix) before the injection of HA or saline. No bed rest was required after the injection, but a low level of activity was recommended for the remainder of the day. If analgesics or NSAIDs were taken before entering the trial, the dosage was not modified until the inclusion visit. Acetaminophen was allowed throughout the study. Use of NSAIDs or step 2 analgesics for the affected hip was permitted only if the symptoms did not respond to optimal doses of acetaminophen (4 gm/day). Washout of concomitant analgesics or NSAIDs was not performed before each assessment. However, treatment with any of these drugs (including acetaminophen) was recorded in a daily logbook. Drugs prescribed for a disorder other than OA were reported in the case report form at each clinic visit throughout the trial.
Syringes containing HA or placebo were covered with masking tape to ensure that the physicians administering the intervention, patients, and care providers were blinded with regard to treatment group. In addition, the physicians collecting the outcome data were not those administering the intervention.
After the inclusion visit, patients were examined on the day of injection, on day 7, and then monthly for 3 months. The primary end point for assessment of efficacy was set at 3 months postinjection. The primary outcome measure was mean variation in pain score compared with the score at baseline. The severity of pain was evaluated at each visit by use of a continuous 100-mm VAS assessing the global level of pain in the target hip, regardless of the circumstances over the previous 48 hours. Secondary outcomes were the proportion of responders, defined according to the OARSI criteria (22); the Western Ontario and McMaster Universities Osteoarthritis Index global score and subscores for pain, stiffness, and disability (23); the patient and physician global assessment of the severity of hip OA measured on a 100-mm VAS; and treatment with analgesics and NSAIDs. Outcomes were measured at inclusion and at baseline (immediately before injection). During each visit, the investigators evaluated the frequency of local and systemic adverse events (AEs).
A sample size of 122 patients (61 patients in each group) was needed to demonstrate a difference between the HA and placebo groups for at least a 20-mm change in pain score between baseline and the 3-month end point, assuming a standard deviation of 15 mm on the VAS pain scale with an alpha error of 5% and a power of 90% (2-sided test). Assuming a withdrawal rate of 8%, the estimated necessary sample size was 130 patients. The sample size calculation was based on a 20-mm difference in pain level between groups, a difference that was considered clinically relevant given the invasiveness of such IA treatment. Changes in variables were compared between groups by analysis of covariance, with baseline values used as a covariate for quantitative variables. Proportions were compared using the chi-square test for nominal variables. Analysis was based on the intent-to-treat principle with the last observation carried forward to replace missing data. A per-protocol analysis was also conducted. Statistical analysis was performed using SAS, version 8.2 (SAS Institute, Cary, NC). P values less than 0.05 (2-tailed) were considered significant. Results are expressed as the mean ± SD or number and percentage.
Ninety-seven patients were screened. Eighty-five patients were randomized (42 to the HA group and 43 to the placebo group) and their data analyzed; 5 patients did not complete the 12-week followup (Figure 1). Table 1 summarizes the main baseline characteristics of the 85 patients. At baseline, the demographic and clinical characteristics did not differ between the 2 groups.
|HA (n = 42)||Placebo (n = 43)|
|Age, years||60.8 ± 10.2||59.5 ± 12.6|
|Sex, % female||64||53|
|Weight, kg||73.7 ± 12.3||74.8 ± 17.7|
|Height, cm||166.2 ± 8.5||167.6 ± 10.2|
|Body mass index, kg/m2||26.7 ± 4.2||26.4 ± 4.2|
|Duration of disease, years||4.2 ± 4.7||4.6 ± 5.9|
|Kellgren/Lawrence grade 2, % of patients||16.7||9.3|
|Kellgren/Lawrence grade 3, % of patients||83.3||90.7|
|Pain score (100-mm VAS)||58.4 ± 11.8||60.4 ± 10.2|
|Patient global assessment of severity of hip OA (100-mm VAS)||58.6 ± 16.4||57.6 ± 12.8|
|Physician global assessment of severity of hip OA (100-mm VAS)||55.5 ± 14.2||53.1 ± 11.6|
|WOMAC pain score||49.8 ± 15.6||51.4 ± 13|
|WOMAC stiffness score||52.3 ± 20.1||57.6 ± 14.4|
|WOMAC function score||51.3 ± 16.8||49.7 ± 13.4|
|WOMAC global score||49.9 ± 16.4||50.6 ± 11.7|
Symptom outcomes showed mild improvement from baseline through the 12-week assessment in both groups. Figure 2 shows the change in pain score in patients treated with HA and in patients receiving placebo. Between baseline and the 3-month end point, the mean ± SD change in pain score (on the VAS) in the intent-to-treat population was −7.8 ± 24.9 mm in the HA group and −9.1 ± 27.4 mm in the placebo group (P = 0.98). The difference between the treatment groups, computed as the least squares mean in the HA group minus the least squares mean in the placebo group, was −0.15 (95% confidence interval [95% CI] −11.04, 10.74). The OARSI responder rate was 33.3% in the HA group and 32.6% in the placebo group (P = 0.94) (Table 2). The groups did not differ significantly in other secondary end points. Moreover, analysis of the logbooks did not reveal any difference in concomitant treatment between the 2 groups. Analgesics and/or NSAIDs were taken on 81% of the study days by patients in the placebo group and on 88% by patients in the HA group (P = 0.38). Overall, the same trends were found in the per-protocol analysis.
|HA (n = 42)||Placebo (n = 43)|
|Pain score (100-mm VAS)||−7.8 ± 24.9||−9.1 ± 27.4|
|OARSI responders at 3 months, % of patients||33.3||32.6|
|Patient global assessment of severity of hip OA (100-mm VAS)||−7.0 ± 24.9||−5.4 ± 27.2|
|Physician global assessment of severity of hip OA (100-mm VAS)||−5.5 ± 25.4||−3.0 ± 26.3|
|WOMAC pain score||−8.6 ± 22.3||−7.5 ± 24.6|
|WOMAC stiffness score||−2.8 ± 25.2||−11.1 ± 26.0|
|WOMAC function score||−6.7 ± 22.7||−5.7 ± 19.9|
|WOMAC global score||−6.2 ± 21.3||−6.5 ± 20.2|
The proportion of patients who experienced at least 1 AE was similar between the 2 groups (40.5% and 34.9% in the HA and placebo groups, respectively; P = 0.59). Two patients in the placebo group and 5 in the HA group had an AE that was possibly or probably related to treatment (P = 0.26). The AEs in the placebo group were pain during injection in 1 patient and exacerbation of hip pain in 1 patient, and the AEs in the HA group were pruritus in 1 patient, hematoma at the injection site in 1 patient, and pain flares in 3 patients. Most of these AEs were mild to moderate in intensity. One patient in the HA group underwent total hip replacement because of rapidly destructive hip OA.
This multicenter, randomized, placebo-controlled study failed to demonstrate a statistically significant effect of 1 IA injection of HA compared with placebo in patients with hip OA, during a 3-month observation period. The present study is, to our knowledge, the first RCT to evaluate the efficacy of a single injection of HA for the treatment of hip OA.
We planned a study of only a single IA injection for the following reasons. First, repeated injections might lead to an increased risk of local side effects, which could be difficult to manage in a deep joint such as the hip, in particular when analysis of synovial fluid is needed. Second, previous open-label trials of a single injection of HA for the treatment of hip OA suggested promising results (18, 24). Third, the technical difficulty of the injection procedure, which involves the use of fluoroscopic or ultrasound guidance, may limit the number of injections and could cause discomfort for patients.
Several hypotheses could explain the lack of efficacy of a potentially active treatment in an RCT. These include a high placebo effect, the design of the study, and the lack of efficacy of the drug itself (25). IA procedure, notably with injections of HA into the knee, is commonly associated with a high placebo response that might limit the ability to detect benefits of the treatment (8, 22, 25). However, in the present study we did not observe a high placebo response that may have skewed the results observed in the treatment group. The mean ± SD change in pain score was −9.1 ± 27.4 mm in the placebo group and −7.8 ± 24.9 mm in the HA group (P = 0.98). A similar, low placebo response was observed in 2 recent RCTs of hip OA. Qvistgaard et al (15) found that the mean (range) within-group difference in pain score (on walking) on a 100-mm VAS on day 90 after saline injection was −5 (−13, 2) (15). In another RCT that compared the efficacy of a single injection of a steroid with a single injection of placebo, the proportion of responders, defined according to the OMERACT-OARSI criteria, was 4.8% in the placebo group 3 months after treatment (26). In view of these results, the extent of the placebo response following IA injections could be lower in hip OA than in knee OA, as was also suggested by a recent meta-analysis that evaluated the placebo effect and its determinants in OA (27).
It is unlikely that the negative results of our trial were due to the HA preparation we used (Adant). Two comparative studies with different HA formulations have been conducted in patients with hip OA. One compared a single injection of Adant (n = 91) with a single injection of Synocrom (Cromo Pharma, Loebendorf, Austria) (n = 20) or a single injection of hylan G-F 20 (Synvisc; Genzyme, Boston, MA) (n = 15) (13), and the other compared 3 weekly injections of Synvisc (n = 22) with 3 weekly injections of Ostenil (Chemedica, Archamps, France) (n = 26) (28). Neither of these studies, which included limited numbers of patients and had methodologic weaknesses, demonstrated any significant difference between groups treated with HA products differing in molecular weight.
Our study was designed and conducted according to international guidelines (29, 30). It was a 2-arm, randomized, double-blind, controlled trial, with recommended outcome variables. However, the number of patients enrolled in our study did not reach the a priori sample size we calculated, due to delayed recruitment and the use-by date of the preparations, which limited our period of inclusion to 14 months. Although our study was underpowered, the absence of any differences between treatment arms likely reflects the lack of clinical effect of this treatment. The sample size calculation was based on a 20-mm difference in pain level between groups. This difference was considered clinically relevant given the invasiveness of such IA treatment and the results of previous open-label studies suggesting that HA injections were effective in patients with hip OA (18, 31). The negative results of this trial are unlikely to be related to a difference in the use of analgesics or NSAIDs, because their rate of intake did not differ between groups.
A recent review of 7 trials of HA injection in patients with hip OA (14) included 2 trials (1 uncontrolled study  and 1 RCT ), in which the proportion of responders was analyzed as an outcome measure (32). In the open-label trial, following a single injection of hylan G-F 20, the proportion of responders 3 months after treatment was 53.6% (18), and in the study by Qvistgaard et al (15), the proportion of responders on day 28 was 53% in the HA group as compared with 44% in the placebo group; responder proportion was not calculated at month 3. Although trials with different designs may not be comparable, these proportions are within the same range as our results.
In the only previously published report of a placebo-controlled trial of HA in hip OA, patients were injected 3 times at 2-week intervals with sodium hyaluronate (Hyalgan; Expanscience, Paris, France) (n = 33), placebo (n = 36), or methylprednisolone (n = 32) (15). The effect size of HA compared with placebo was 0.4 (95% CI −0.1, 0.9) (P = 0.13), which was higher than that observed in our trial (−0.15 mm [95% CI −11.0, 10.7], P = 0.98). In the former study, the effect of HA in the subgroup of patients without joint effusion was comparable with the significant effect observed following corticosteroid injections in the same subgroup, which suggests that HA could be effective in this condition. In the present study, we could not explore the effect of a single injection of HA in patients with joint effusion versus those without joint effusion since synovial effusions were present in only 11.7% of the patients in the placebo group and 9.6% of the patients in the HA group.
Finally, the lack of demonstration of an effect of Adant on OA symptoms might be explained by a real lack of efficacy of a single injection of HA in hip OA. One plausible explanation could be the known rapid clearance of HA from the synovial fluid compartment (10), which suggests that a single IA injection of HA might be insufficient to have an effect on OA symptoms.
Overall, the IA injection of HA into the hip joint was well tolerated by the OA patients in our study. No infections or other serious AEs related to treatment occurred during the study period.
Well-designed clinical trials to evaluate the efficacy of HA in hip OA are still scarce and include limited numbers of patients. In that sense, our trial represents a step forward in this investigation. In conclusion, this first 2-arm, placebo-controlled study to evaluate the effect of HA in hip OA failed to demonstrate any benefit of a single injection of HA compared with placebo. Future studies are warranted to confirm our results and to evaluate the effects of repeated HA injections in hip OA.
We are grateful to Catherine Koch for safe conduct of the trial and to all of the physicians who performed the injections. We are grateful to the patients and to the following physicians (all in France) who contributed patients: Dr. Ziza (Paris), Dr. Couret (Valence), Professor Lechevalier (St Mandé), Professor Goupille (Tours), Professor Saraux (Brest), Professor Blotman (Montpellier), Dr. Thomas (Thionville), Dr. Balblanc (Belfort), Professor Wendling (Besançon), Professor Duquesnoy (Lille), Dr. Zarnitsky (Le Havre), Dr. Degieux (Paris), Professor Hardy (Boulogne), and Professor Lafforgue (Marseille). We also thank the Association Rhumatisme et Travail (Centre Viggo Petersen, Hôpital Lariboisière, Paris, France), which funded the copyediting of the original manuscript.
Dr. Chevalier 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 design. Richette, Ravaud, Clerson, Chevalier.
Acquisition of data. Chevalier.
Analysis and interpretation of data. Richette, Ravaud, Clerson, Chevalier.
Manuscript preparation. Richette, Ravaud, Conrozier, Euller-Ziegler, Mazières, Maugars, Mulleman, Clerson, Chevalier.
Statistical analysis. Ravaud, Clerson.
Daiichi Sankyo was involved in the study design, data collection, and data analysis and agreed to submit the manuscript for publication in its present form.