Efficacy and Safety of Topical Diclofenac

PAINÒ 143 (2009) 238–245

www.elsevier.com/locate/pain

Efﬁcacy and safety of topical diclofenac containing dimethyl sulfoxide (DMSO) compared with those of topical placebo, DMSO vehicle and oral diclofenac for knee osteoarthritis

Lee S. Simon a, Lisa M. Grierson b, Zahid Naseer b, Arthur A.M. Bookman c, J. Zev Shainhouse b,* a SDG LLC Consulting, 124 Mt. Auburn Street, Cambridge, MA, USA b Nuvo Research Inc., 7560 Airport Road, Mississauga, ON, Canada L4T 4H4 c Division of Rheumatology, University Health Network, Western Division, University of Toronto, Toronto, ON, Canada article info abstract

Article history: While topical non-steroidal anti-inflammatory drugs are considered safe, their long-term efficacy for Received 1 October 2008 osteoarthritis has been suspect. We conducted a 12-week, double-blind, double-dummy, randomized Received in revised form 21 January 2009 controlled trial of topical diclofenac (TDiclo) in a vehicle solution containing dimethyl sulfoxide (DMSO) Accepted 9 March 2009 in 775 subjects with radiologically confirmed, symptomatic primary osteoarthritis of the knee. This 5- arm study compared TDiclo with a placebo solution, the DMSO vehicle, oral diclofenac (ODiclo) and Keywords: the combination of TDiclo + ODiclo for relieving the signs and symptoms of knee osteoarthritis. Subjects Topical diclofenac applied study solution, 40 drops four times daily, and took one study tablet daily for 12 weeks. Co-pri- NSAID mary efficacy variables were WOMAC pain and physical function and a patient overall health assessment. Osteoarthritis Secondary variables were WOMAC stiffness and patient global assessment (PGA) of the knee osteoarthri- Knee pain tis. TDiclo was superior to placebo for pain (À6.0 vs. À4.7, P = 0.015), physical function (À15.8 vs. À12.3, P = 0.034), overall health (À0.95 vs. À0.37, P < 0.0001), and PGA (À1.36 vs. À1.01, P = 0.016), and was superior to DMSO vehicle for all efficacy variables. No significant difference was observed between DMSO vehicle and placebo or between TDiclo and ODiclo. The commonest adverse event associated with TDiclo was dry skin (18.2%). Fewer digestive system and laboratory abnormalities were observed with TDiclo than with ODiclo. Addition of TDiclo to ODiclo did not increase the incidence of systemic adverse events. TDiclo in DMSO vehicle is an effective treatment option for knee osteoarthritis with efficacy similar to, but tolerability better than ODiclo. DMSO vehicle was no more efficacious than placebo. Ó 2009 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction meta-analysis of the same studies stressed that the various products showed symptom relief at 1 or 2 weeks but loss of benefit at When non-pharmacological measures are not sufficient to con- 4 weeks, and rejected them as sufficient evidence to justify a rec- trol the symptoms of osteoarthritis, current evidence-based guide- ommendation of long-term use. lines support pharmacological treatment with acetaminophen or Subsequently published randomized controlled studies have oral nonsteroidal anti-inflammatory drugs (NSAIDs) [3,39,56,57]. described a penetrating topical diclofenac solution (TDiclo) in a di- Acetaminophen has been linked with an increased risk of hepatic, methyl sulfoxide (DMSO)-containing vehicle as efficacious and safe hypertensive and cardiovascular adverse effects [14,19,20,42,52]. in relieving the symptoms of primary osteoarthritis of the knee NSAIDs are more effective [49] and carry more well-known gastro- over 4-, 6-, and 12-week treatment periods [5,12,43,50]. Recent intestinal and cardiovascular risk [4,14,28,37]. The attempt to min- topical NSAID reviews and meta-analyses [6,7,11,25,38,41,48,55] imize their risk of both morbidity and potential mortality by have evaluated the data from these TDiclo studies, and subse- prescribing COX-2 selective NSAIDs (‘coxibs’) has not achieved quently published national guidelines [15,39,47,56,57] have cited the goal [32,34]. these studies as evidence for the use of topical NSAIDs as first line Topical NSAIDs present a safer potential alternative [18,27] to therapy for osteoarthritis. oral therapy, with decreased systemic exposure to the active The efficacy of topical NSAIDs, such as TDiclo, is thought to be NSAID molecule. While previous reviews [36] had suggested that due to local action of the active NSAID molecule following its pen- topical NSAIDs are effective for osteoarthritis, Lin et al. [33] in their etration through the skin to the tissue sites of inflammation and pain. Diclofenac sodium is a member of the arylacanoic acid group * Corresponding author. Tel.: +1 905 673 6980. of NSAIDs with lipophilic properties that limit its percutaneous E-mail address: [email protected] (J.Z. Shainhouse). penetration [40]. The biological property of DMSO to enhance skin

0304-3959/$36.00 Ó 2009 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.pain.2009.03.008 L.S. Simon et al. / PAINÒ 143 (2009) 238–245 239 penetration of both hydrophilic and lipophilic molecules is well oral diclofenac tablets. Subjects applied 40 drops of solution known [53], with recent research focusing on the mechanism of (approximately 1.2 mL) around the entire circumference of the enhancement [24]. The percutaneous absorption of diclofenac fol- study knee, without massage, four times daily, and took one study lowing a multidose regimen of TDiclo (in DMSO-containing vehi- tablet daily for up to 12 weeks. cle) was significantly enhanced compared with an aqueous Concomitant analgesic and anti-inflammatory medications, diclofenac formulation without DMSO [26]. Within the extensive including over-the-counter NSAIDs and other analgesics, were pro- literature on DMSO are unsubstantiated claims of therapeutic effi- hibited. Continuation of stable treatment with glucosamine, chon- cacy in the treatment of osteoarthritis; but no efficacy of DMSO droitin, anti-depressants or a proton pump inhibitor (previous vehicle was shown in a previous, 4-week randomized controlled 90 days), or low-dose (6325 mg/day) acetylsalicylic acid (previous study of TDiclo [12]. High dose exposure of non-primate species 30 days) was permitted. Acetaminophen was provided, and per- to DMSO has demonstrated some ocular abnormalities [44]; how- mitted (up to four 325-mg caplets per day) except during the ever no such changes have been observed in humans [13,23,46]. 3 days before each efficacy assessment. Other topical products on We describe here the safety and efficacy results of a confirma- the knee, including skin emollients, were prohibited. A patient tory 12-week, 5-arm randomized controlled study of TDiclo, with a gastrointestinal adverse event was allowed to start a proton including a placebo arm to establish its efficacy, a DMSO vehicle pump inhibitor. Compliance with the treatment regimen was as- arm to address DMSO claims of efficacy, an oral diclofenac arm sessed by weighing the solution bottles and counting study tablets to compare with TDiclo and a combination of TDiclo plus ODiclo at each clinic visit. to assess combined treatment. All study solutions appeared as identical clear, colorless liquids. It was expected that some subjects applying TDiclo or DMSO vehicle would report a garlic taste or odor from exhaling dimethyl sul- 2. Methods fide, a volatile DMSO metabolite; therefore, a token amount of DMSO (2.3%) was included in the placebo solution. Previous trials 2.1. Study subjects with TDiclo confirmed the success of this blinding procedure as the incidence of ‘taste perversion’ adverse events was no different This randomized, double-blind, double-dummy, placebo-, vehi- with placebo solution from TDiclo. Oral diclofenac and placebo tab- cle- and active-controlled study was conducted at 40 outpatient lets appeared identical (NovopharmÒ Inc.). Each study kit was centers in Canada and at 21 centers in the United States, from Feb- assembled according to a computer-generated randomization ruary 2004 to October 2005, following approval by the appropriate schedule created by an external statistician for each stratum using institutional review boards. Eligible subjects, after written in- a block size of 5. The randomization sequence was concealed from formed consent, included men and non-pregnant women aged investigators, subjects and the sponsor’s clinical research person- 40–85 with primary OA of the knee based on (i) standard radiolog- nel until after data lock. ical criteria for OA [2] on a recent (within 3 months) examination, (ii) pain, with regular use of a NSAID or other analgesic medication 2.3. Efficacy measures (at least 3 days a week in the previous month) and (iii) a flare of pain and minimum Likert pain score of 8 (40 on a scale normalized Each subject completed a full efficacy assessment questionnaire to 0–100; see below, Efficacy Measures) at baseline, following after randomization at baseline and at 4, 8 and 12 weeks, or at drop washout of that medication. [A flare was defined as an increase out. The co-primary efficacy variables [10] were defined a priori as in total Likert pain score of 25% and at least 2, and a score of at least WOMAC pain and physical function, measured by the 5-point Lik- moderate on one or more of the five items/questions of the WO- ert scale [9], and patient overall health assessment (POHA). The MAC LK3.1 pain subscale.] All knee films were reviewed by a single POHA (see Section 4) asked the question, ‘‘Considering all the ways radiologist and each compartment was scored (none = 1, mild = 2, your osteoarthritic (study) knee and its treatment affect you, moderate = 3, severe = 4). Only one knee was treated; where both including both positive and negative effects, how would you rate knees met all entry criteria, the more painful knee (or dominant your overall state of health in the past 48 hours?”. Secondary effi- knee if they scored the same) was selected. Standard exclusion cri- cacy variables were WOMAC stiffness and patient global assess- teria were employed, as described previously [43]. ment (PGA) of knee osteoarthritis. The PGA asked the question, ‘‘How has the osteoarthritis in your study joint been over the last 2.2. Study design 48 hours?”. The POHA and PGA were scored on a 5-point Likert scale. There was no assessment of the non-treated knee. Eligible subjects were stratified by the investigator at baseline into stratum 1 (no pain and normal radiological examination in 2.4. Safety measures the non-study knee) or stratum 2 (pain and/or abnormal radiological examination in the non-study knee), and were then random- At all visits, vital signs were measured, dermatological evalua- ized into the trial on receiving the next numbered study kit at tion of the study knee was done according to a standard numerical that clinic for that stratum. Each study kit contained topical solu- (0–4) scale [35] and adverse events were solicited using open- tion and oral tablets for one of the five treatment regimens: (i) ended questions. Adverse events were categorized according to ‘TDiclo’; topical diclofenac solution plus oral placebo tablets (topi- Coding Symbols for Thesaurus of Adverse Reaction Terms (COST- cal diclofenac solution is 1.5% w/w diclofenac sodium in a vehicle ART) [51]. Blood and urine samples were obtained for routine lab- solution containing 45.5% w/w DMSO and other excipients [Penn- oratory analysis at baseline, 4 and 12 weeks. Ocular examination saidÒ Topical Solution, Nuvo Research Inc.]), (ii) ‘DMSO vehicle’; (visual acuity test, slit lamp examination and lens assessment) vehicle solution plus oral placebo tablets (vehicle solution was was conducted at the baseline and final visit. the complete carrier, including 45.5% DMSO and other excipients, with no diclofenac), (iii) ‘placebo’; placebo solution plus oral pla- 2.5. Statistical analysis plan cebo tablets (placebo solution was a modified vehicle solution with only 2.3% DMSO, for blinding purposes, and no diclofenac), (iv) All planned analyses were specified a priori. Analysis of the Lik- ‘ODiclo’; placebo solution plus oral diclofenac tablets (100 mg slow ert efficacy data was by modified intent to treat (mITT), excluding release), or (v) ‘TDiclo + ODiclo’; topical diclofenac solution plus only those subjects who had no baseline data or did not administer 240 L.S. Simon et al. / PAINÒ 143 (2009) 238–245 at least one dose of both study solution and tablets [31]. Statistical (Fig. 1). Over 95% of patients had bilateral disease (pain or abnormal tests were two sided at the 5% level of significance. All efficacy radiological examination also in non-study knee) with pain in the analyses were done by analysis of covariance of the change in score contralateral knee in 90%. A total of 88% subjects met the modified from baseline to landmark final assessment, with baseline score as [29] American College of Rheumatology (ACR) criteria [1] for osteo- the covariate. Subjects in both randomization strata were com- arthritis, pain and osteophytes, and the remaining had pain with bined for all analyses. either joint space narrowing or subchondral sclerosis. Subjects in The primary efficacy comparison was TDiclo vs. placebo for the each treatment arm had similar demographic and baseline charac- primary efficacy variables, with no correction for analysis of multi- teristics (Table 1), duration of exposure and compliance (>89% of ple primary variables (regulatory design required superiority for expected use for topical solution and oral tablets). Withdrawals each primary variable). The sample size required to show the supe- for an adverse event were similar among treatment groups. With- riority of TDiclo over placebo was 142 subjects per arm, based on drawals for lack of effect were similar among TDiclo, placebo and an estimate from previous trials [5,12,43] of the difference (stan- DMSO vehicle arms, but fewer with the oral diclofenac arms (Fig. 1). dard deviation) between groups for the change in the score of 1.5 There were 772 subjects included in the mITT group. Three sub- (4.5) for pain, 5.0 (15) for physical function and 0.4 (1.2) for PGA, jects were excluded who did not take at least one dose of both top- power of 80% and Type I error rate of a = 0.052-tailed.Apost hoc sen- ical and oral medication, or had no data (Fig. 1). Individual subjects sitivity analysis of the primary efficacy data was performed by who omitted baseline assessment for a specific efficacy variable imputing no improvement (i.e., baseline score was carried forward were excluded from that analysis. and imputed as final score [BOCF]) to non-completers by reason of an adverse event or lack of effect and for subjects excluded from 3.2. Efficacy mITT. Secondary comparisons included TDiclo vs. placebo for the sec- The primary efficacy analyses, as shown in Table 2, show the ondary variables, and TDiclo vs. DMSO vehicle and DMSO vehicle superiority of TDiclo over placebo for the three co-primary vari- vs. placebo for all variables. A post hoc efficacy analysis compared ables – pain (P = 0.015), physical function (P = 0.034), and POHA TDiclo vs. ODiclo and TDiclo + ODiclo vs. ODiclo. For a missing final (P < 0.0001). Superiority was observed also for the secondary vari- score, last observation was carried forward. able PGA (P = 0.016), but not for stiffness. There was greater Safety analyses were performed on all subjects who received improvement with TDiclo (P < 0.05) than with DMSO vehicle for even one dose of either study medication. There was no imputation all five variables (Table 2). A post hoc BOCF sensitivity analysis of of missing safety data. the primary efficacy data of all 775 patients did not change any of the conclusions regarding the superiority of TDiclo over placebo 3. Results (pain, P = 0.031; physical function, P = 0.041; POHA, P = 0.0001). No significant efficacy advantage of the DMSO vehicle over pla- 3.1. Study subjects cebo was observed for the primary or secondary variables, except for the POHA (Table 2). A comparison of ODiclo vs. TDiclo found Of 1396 subjects screened, 775 were randomized to one of the no statistical difference for any of the five efficacy variables (Table five treatment arms and 527 subjects completed treatment 2). The combination of TDiclo + ODiclo was no better than ODiclo

Assessed for eligibility (n=1396) Excluded (n=621) Insufficient flare of pain (n=266) Did not meet x-ray criteria (n=51) Failed other selection criteria (n=104) Withdrew consent (n=105) Exceeded 14-day washout window (n=33) Other (n=62) Randomized (n=775)

Allocated to topical Allocated to placebo Allocated to vehicle Allocated to oral Allocated to topical diclofenac solution (n=157) control (DMSO) diclofenac (ODiclo) diclofenac solution + (TDiclo) (n=154) Did not receive allocated (n=161) (n=151) oral diclofenac intervention (n=2; did not (TDiclo+ODiclo) take both topical and oral (n=152) medication)

Discontinued Discontinued Discontinued Discontinued Discontinued intervention (n=51) intervention (n=54) intervention (n=48) intervention (n=44) intervention (n=51) Adverse Event (n=16) Adverse Event (n=18) Adverse Event (n=12) Adverse Event (n=19) Adverse Event (n=23) Lack of Effect (n=16) Lack of Effect (n=18) Lack of Effect (n=17) Lack of Effect (n=5) Lack of Effect (n=9) Consent withdrawn (n=6) Consent withdrawn (n=6) Consent withdrawn (n=10) Consent withdrawn (n=8) Consent withdrawn (n=8) Lost to Follow-up (n=2) Lost to Follow-up (n=4) Lost to Follow-up (n=3) Lost to Follow-up (n=2) Lost to Follow-up (n=2) Other* (n=11) Other*(n=8) Other* (n=6) Other* (n=10) Other* (n=9)

Completed study Completed study Completed study Completed study Completed study (n=103) (n=103) (n=113) (n=107) (n=101)

Included in efficacy Included in efficacy Included in efficacy Included in efficacy Included in efficacy analysis** (n=154) analysis** (n=155) analysis** (n=161) analysis** (n=151) analysis** (n=151) Excluded (n=2; did not Excluded (n=1; no take at least one dose of data) both topical and oral medication)

Fig. 1. Flow of subjects through the trial. *‘Other’ reason for withdrawal was primarily a scheduling conﬂict. **Included in efﬁcacy analysis of pain variable. Additional subjects were excluded from the analysis of the variables physical function and patient overall health assessment due to missing baseline data (see Section 3). L.S. Simon et al. / PAINÒ 143 (2009) 238–245 241

Table 1 Demographic and baseline characteristics.a

Characteristic TDiclo (n = 154) Placebo (n = 155) DMSO (n = 161) ODiclo (n = 151) TDiclo + ODiclo (n = 151) Age, mean (SD), yr 61.7 (9.8) 61.7 (10.0) 62.1 (9.3) 62.0 (10.5) 60.6 (10.0) Gender Male 50 (32.5) 65 (41.9) 71 (44.1) 56 (37.1) 50 (33.1) Female 104 (67.5) 90 (58.1) 90 (55.9) 95 (62.9) 101 (66.9) Race/ethnicity Caucasian 120 (77.9) 120 (77.4) 123 (76.4) 119 (78.8) 116 (76.8) Black 11 (7.1) 9 (5.8) 7 (4.3) 6 (4.0) 8 (5.3) Asian 15 (9.7) 20 (12.9) 14 (8.7) 13 (8.6) 8 (5.3) Hispanic 6 (3.9) 5 (3.2) 11 (6.8) 11 (7.3) 12 (7.9) Other 2 (1.3) 1 (0.6) 6 (3.7) 2 (1.3) 7 (4.6) Body mass index, kg/m2 Underweight (<18.5) 1 (0.6) 0 0 1 (0.7) 0 Normal (18.5–25) 21 (13.6) 19 (12.3) 21 (13.0) 8 (5.3) 17 (11.3) Overweight (25–30) 46 (29.9) 46 (29.7) 46 (28.6) 39 (25.8) 47 (31.3) Obese (>30) 85 (55.2) 89 (57.4) 93 (57.8) 102 (67.5) 84 (56.0) Incidence of hypertensionb 5 (3.3) 3 (1.9) 7 (4.4) 6 (4.0) 4 (2.7) Total study knee radiographic score, mean (SD)c 16.3 (3.4) 16.4 (2.9) 16.6 (3.3) 16.5 (2.9) 16.5 (3.0) Number of knees with OA 1 Knee 2 (1.3) 7 (4.5) 7 (4.3) 2 (1.3) 4 (2.6) 2 Knee 152 (98.7) 148 (95.5) 154 (95.7) 149 (98.7) 147 (97.4)

Table 2 Efﬁcacy variable scoresa.

TDiclob placebob DMSOb ODiclob TDiclo+ODiclob P value TDiclo vs. Placebo TDiclo vs. DMSO TDiclo vs. ODiclo WOMAC Pain (n = 154) (n = 155) (n = 161) (n = 151) (n = 151) Baseline score 13.2 (3.4) 12.9 (3.3) 13.0 (3.2) 13.2 (3.0) 13.2 (3.4) Change in scorec À6.0 (4.5) À4.7 (4.4) À4.7 (4.3) À6.4 (4.1) À7.0 (4.8) 0.015 0.009 0.429 WOMAC Physical Function (n = 154) (n = 153) (n = 161) (n = 151) (n = 150) Baseline score 41.7 (12.8) 41.6 (11.7) 41.4 (11.4) 42.1 (12.0) 41.0 (11.2) Change in scorec À15.8 (15.1) À12.3 (14.7) À12.1 (14.6) À17.5 (14.3) À18.7 (14.0) 0.034 0.026 0.319 POHA (n = 154) (n = 152) (n = 160) (n = 150) (n = 148) Baseline score 2.34 (1.02) 2.22 (1.03) 2.30 (1.14) 2.23 (1.12) 2.19 (1.04) Change in scorec À0.95 (1.30) À0.37 (1.04) À0.65 (1.12) À0.88 (1.31) À0.95 (1.21) <0.0001 0.016 0.956 PGA (n = 154) (n = 153) (n = 161) (n = 151) (n = 150) Baseline score 3.12 (0.78) 3.04 (0.82) 3.13 (0.74) 3.04 (0.87) 3.08 (0.81) Change in scorec À1.36 (1.19) À1.01 (1.18) À1.07 (1.10) À1.42 (1.29) À1.53 (1.27) 0.016 0.018 0.439 WOMAC Stiffness (n = 154) (n = 153) (n = 161) (n = 151) (n = 150) Baseline score 5.14 (1.63) 5.01 (1.70) 5.12 (1.61) 5.21 (1.72) 5.07 (1.53) Change in scorec À1.93 (2.01) À1.52 (2.05) À1.48 (2.07) À2.07 (2.02) À2.30 (2.00) 0.112 0.035 0.596

Abbreviations: TDiclo, topical diclofenac solution plus oral placebo; Placebo, topical placebo plus oral placebo; DMSO, topical dimethyl sulfoxide-containing vehicle plus oral placebo; ODiclo, oral diclofenac plus topical placebo; TDiclo+ODiclo, topical diclofenac solution plus oral diclofenac; WOMAC, Western Ontario McMaster Universities LK3.1 Osteoarthritis Index; POHA, patient overall health assessment; PGA, patient global assessment of the study knee. a Data are presented as mean (SD). Maximum score for pain, 20; physical function, 68; POHA and PGA, 4; stiffness, 8. b The number of subjects (n) varied by efﬁcacy parameter because individual subjects did not submit a full baseline assessment. c Final score minus baseline score. alone for all variables (pain, P = 0.30; physical function, P = 0.33; The rate with DMSO vehicle was intermediate between TDiclo POHA, P = 0.43; stiffness, P = 0.16; PGA, P = 0.50). and placebo. Only ﬁve (3.2%) subjects in TDiclo withdrew due to Mean [SD] acetaminophen use with TDiclo (0.64 [0.83] caplets an application site reaction. per day) was lower than that with placebo (0.95 [1.14], The incidence of adverse events of the digestive system with P = 0.005) and DMSO vehicle (0.99 [1.11], P = 0.002), and was not TDiclo was no greater than that with placebo and was much lower different from that with ODiclo (0.55 [0.82], P = 0.41) or TDi- than with ODiclo and TDiclo + ODiclo. Withdrawal due to an ad- clo + ODiclo (0.46 [0.70]; P = 0.10). verse event of the digestive system was higher in ODiclo (11 [7.3%]) vs. TDiclo (4 [2.6%]) and placebo (3 [1.9%]). The combina- 3.3. Safety tion of TDiclo + ODiclo did not increase the incidence of events of the digestive system over ODiclo alone (Table 3). Skin adverse events predominated with TDiclo, most being dry No difference between treatment groups was observed for car- skin (Table 3). The overall incidence of skin adverse events was diovascular events, which were <2% in each treatment group. The similar in TDiclo + ODiclo, and lower in placebo, and ODiclo arms. incidence of hypertension was similar for all groups (1.3% for TDi- 242 L.S. Simon et al. / PAINÒ 143 (2009) 238–245

Table 3 Incidence of adverse events.a

Adverse event TDiclo (n = 154) Placebo (n = 157) DMSO (n = 161) ODiclo (n = 151) TDiclo + ODiclo (n = 152) Any adverse event 96 (62.3) 90 (57.3) 97 (60.2) 94 (62.3) 98 (64.5) Abnormal taste sensation or odor 0 (0.0) 1 (0.6) 1 (0.6) 0 (0.0) 1 (0.7) Any digestive system event 10 (6.5) 15 (9.6) 18 (11.2) 36 (23.8) 39 (25.7) Abdominal pain 5 (3.2) 1 (0.6) 5 (3.1) 11 (7.3) 3 (2.0) Dyspepsia 4 (2.6) 6 (3.8) 6 (3.7) 6 (4.0) 5 (3.3) Diarrhea 2 (1.3) 3 (1.9) 2 (1.2) 7 (4.6) 12 (7.9) Liver function tests abnormal 3 (1.9) 1 (0.6) 6 (3.7) 12 (7.9) 11 (7.2) Rectal hemorrhage 1 (0.6) 0 (0.0) 0 (0.0) 0 (0.0) 5 (3.3) Nausea 0 (0.0) 0 (0.0) 1 (0.6) 3 (2.0) 5 (3.3) Any skin/appendages event 41 (26.6) 12 (7.6) 27 (16.8) 11 (7.3) 47 (30.9) Dry skin (application site) 28 (18.2) 5 (3.2) 18 (11.2) 4 (2.6) 30 (19.7) Contact dermatitis (application site) 4 (2.6) 1 (0.6) 5 (3.1) 1 (0.7) 12 (7.9) Rash 4 (2.6) 0 (0.0) 2 (1.2) 0 (0.0) 0 (0.0) Contact dermatitis with vesicles (application site) 3 (1.9) 0 (0.0) 0 (0.0) 1 (0.7) 6 (3.9) Pruritis (application site) 2 (1.3) 0 (0.0) 0 (0.0) 0 (0.0) 1 (0.7) Other system eventsb Headache 27 (17.5) 18 (11.5) 21 (13.0) 26 (17.2) 21 (13.8) Back pain 15 (9.7) 10 (6.4) 15 (9.3) 11 (7.3) 4 (2.6) Arthralgia 14 (9.1) 15 (9.6) 25 (15.5) 12 (7.9) 7 (4.6) Pain 7 (4.5) 5 (3.2) 11 (6.8) 8 (5.3) 1 (0.7) Respiratory disorder 5 (3.2) 6 (3.8) 4 (2.5) 8 (5.3) 7 (4.6) Accidental injury 4 (2.6) 6 (3.8) 7 (4.3) 4 (2.6) 6 (3.9) Abnormal vision 4 (2.6) 5 (3.2) 4 (2.5) 4 (2.6) 1 (0.7) Conjunctivitis 4 (2.6) 1 (0.6) 0 (0.0) 3 (2.0) 0 (0.0)

clo, ODiclo and TDiclo + ODiclo; 1.2% for DMSO vehicle; 0.6% for Changes in key NSAID-related laboratory parameters are shown placebo). There was no difference among the groups for reports in Table 4. Overall, the mean change in the laboratory value and of an abnormal taste sensation or odor, conﬁrming the success of the number of subjects developing an abnormality showed no dif- the blinding procedure for DMSO. ference between TDiclo and placebo or DMSO vehicle, but a greater There was no serious adverse event in the TDiclo arm, four in mean change and higher incidence of abnormality occurred with placebo (one anemia, one cerebrovascular accident, one fractured the ODiclo arms. With ODiclo compared with TDiclo, there was a hip, one dislocated prosthetic hip), one in DMSO vehicle (acute greater mean change in hemoglobin, alanine aminotransferase, enteritis), one in ODiclo (post-polypectomy lower gastrointestinal gamma-glutamyl transpeptidase, creatinine and creatinine clear- bleed, 8 days after withdrawal of study medication) and three in ance, and a greater number of subjects developing an abnormality TDiclo + ODiclo (one leg cellulitis, one unstable angina, one tran- for these parameters (Table 4). Development of abnormal labora- sient ischemic attack). tory parameters was generally below clinically relevant levels.

Table 4 Analysis of changea in laboratory parameters.

TDiclo (n = 145) Placebo (n = 142) DMSO (n = 150) ODiclo (n = 138) TDiclo + ODiclo (n = 141) AST Mean change, U/L À0.9 (10.3) À0.6 (5.2) 0.2 (8.5) 2.5 (10.9) 4.1 (29.6) Normal to abnormalb, N (%) 10 (6.9) 5 (3.5) 8 (5.3) 27 (19.6) 20 (14.2) ALT Mean change, U/L À1.0 (11.7) À0.3 (9.9) À0.6 (9.8) 7.2 (25.3) 8.2 (68.9) Normal to abnormal, N (%) 6 (4.1) 4 (2.8) 2 (1.3) 26 (18.8) 24 (17.0) Hemoglobin Mean change, g/L À1.7 (6.2) À0.8 (6.2) À0.4 (6.5) À3.8 (7.1) À4.8 (6.8) Normal to abnormal, N (%) 3 (2.1) 7 (4.9) 5 (3.3) 8 (5.8) 18 (12.6) Creatinine Mean change, lmol/L À0.4 (10.5) 0.8 (9.0) 0.3 (10.3) 3.1 (11.0) 4.4 (11.2) Normal to abnormal, N (%) 4 (2.8) 8 (5.6) 6 (4.0) 10 (7.2) 15 (10.6) Creatinine clearancec Mean change, mL/min 0.4 (10.3) À0.5 (8.6) À0.5 (8.3) À2.4 (8.7) À3.3 (9.7) Normal to abnormal, N (%) 11 (7.6) 8 (5.7) 9 (6.0) 10 (7.2) 16 (11.4)

No subject developed a clinically significant change in hemoglobin greater for PGA than for POHA for both TDiclo (À1.36 vs. or creatinine. An increase in any liver enzyme to three times the À0.95) and ODiclo (À1.42 vs. À0.88), perhaps reflecting a penalty upper limit of normal occurred in one subject with placebo, one for adverse effects imposed by the subject on efficacy. Whereas with DMSO vehicle, two with ODiclo and three with TDiclo + ODi- ODiclo scored a slightly higher improvement on PGA, TDiclo clo, but none with TDiclo. scored slightly higher on POHA perhaps reflecting a greater pen- Ocular examination at baseline and final revealed no change in alty for the adverse effects of ODiclo. visual acuity (data not shown) and no difference in the develop- The commonest adverse events with TDiclo therapy were ment of lens abnormality (cataract) with placebo (4 [2.6%]) vs. TDi- application site skin reactions and the majority entailed only clo (2 [1.3%]) or DMSO vehicle (6 [3.8%]). skin dryness. The incidence was similar for DMSO vehicle sug- gesting that it is due primarily to the vehicle. DMSO may dis- solve surface lipids, leaving the skin dry. To detect the 4. Discussion maximum potential skin adverse event profile, subjects were not allowed to apply any skin emollient to counter the antici- The results of this study clearly show the effectiveness of TDi- pated drying effect of the vehicle carrier. There was no apparent clo, a solution of diclofenac in a DMSO vehicle, applied topically difference between TDiclo and placebo in NSAID-associated gas- to treat the symptoms of osteoarthritis of the knee. TDiclo was trointestinal adverse events (the incidence was actually lower superior to both topical comparator groups (placebo and DMSO with TDiclo). There is a potential for bias if investigators or sub- vehicle) for all three primary efficacy variables, pain, function jects presume that diclofenac applied topically will not cause and patient overall health. Efficacy was confirmed by a conserva- gastrointestinal or other systemic events and ignore such ad- tive BOCF sensitivity analysis. This study followed a typical 12- verse events. Inclusion of the oral placebo countered this possi- week oral NSAID trial design – primary osteoarthritis of the knee bility but likely contributed inordinately to the adverse events of with pain and abnormal radiological findings. Although most sub- the digestive system reported in even 9.6% of subjects using pla- jects had bilateral osteoarthritis, only one painful knee was treated cebo (Table 3). with topical solution. Inasmuch as outcome measures of physical Previous reports in non-primate species (dogs, rabbits, pigs) function and overall patient health assessment are likely to be neg- identified ocular abnormalities with prolonged high dose exposure atively influenced by symptoms in the non-treated knee, this trial to DMSO [44]. Clinical studies in humans where DMSO was applied design biased against TDiclo. In any oral NSAID trial, and in this directly to the eye or to the skin at extremely high doses (greater study’s ODiclo arms, subjects automatically treat both knees, than 25 times daily dose of DMSO as in TDiclo application), showed avoiding these factors. The inclusion of oral therapy for all groups no deleterious ocular changes [13,23,46]. In this study, no eye lens added a second placebo effect to the topical comparator arms, abnormalities were observed with DMSO vehicle or TDiclo imposing a yet higher burden to prove the efficacy of TDiclo over treatment. placebo. Despite these elements in trial design the efficacy of TDi- The combination of TDiclo + ODiclo showed no increase in ad- clo was robustly established. verse events relative to TDiclo or ODiclo alone. The blood level of The response of patients in the ODiclo group in this study (40– diclofenac following topical application as TDiclo is only about 48% improvement in variable score) was the same as seen in other 12 ng/mL [30] and the incremental increase with the combination oral NSAID trials [8,54], providing a powerful external audit on the would be negligible compared with the predicted level of 1500 ng/ validity of the trial design and conduct, and further confirmation of mL that is reported with oral diclofenac [17]. Although combined the sustained efficacy of TDiclo. TDiclo was found to be as effective treatment with TDiclo + ODiclo did not provide a statistical advan- as ODiclo at relieving the symptoms of knee osteoarthritis with tage over ODiclo alone, this regimen could be a reasonable treat- less NSAID-related systemic toxicity than ODiclo. These observa- ment paradigm in an individual with persistent or breakthrough tions support the safety and efficacy results of a previous equiva- knee pain despite an oral NSAID. lence study of this TDiclo vs. ODiclo [50]. The cumulative data In conclusion, we have shown in this study that this formu- with TDiclo sustain the recent recommendations advocating the lation of TDiclo provides durable improvement in the symptoms use of topical NSAIDs as first-line therapy for the treatment of of osteoarthritis of the knee, and that relief is not contributed osteoarthritis [15,39,47,56,57]. by the DMSO carrier. Efficacy of TDiclo was comparable to that Claims of therapeutic efficacy for DMSO in osteoarthritis were achieved with ODiclo. Dry skin was frequent but systemic discounted in an earlier, 4-week trial with TDiclo [12] and are fur- NSAID class-related clinical and laboratory adverse events oc- ther disproven by this 12-week study. The DMSO-containing vehi- curred less frequently than with ODiclo. For the patient initiat- cle likely retains a role in TDiclo by facilitating percutaneous ing pharmacological therapy for relief of symptoms associated penetration of the active diclofenac sodium [26]. As such, general- with osteoarthritis of the knee based on current treatment ization of the results with TDiclo to other topicals is cautioned. The guidelines, this TDiclo formulation is a viable, evidence-based various NSAID formulations differ not only in their active medica- treatment option. tion but also in the components of the vehicle, the nature of the formulation (lotion, gel, solution, patch, etc.) and the presence of Acknowledgements penetration enhancers [21,22,45]. Lin et al. in their critical meta- analysis of topical NSAIDs observed also that efficacy might be The authors acknowledge the commitment of all study drug specific [33]. investigators and support staff, without whom the present The patient overall health assessment (POHA) was chosen as a study would not have been possible. The sponsor, Nuvo Re- primary variable and the more common patient global assess- search Inc., provided medication and funding for this study, ment (PGA) as a secondary variable, at the request of the FDA. and managed and analyzed the data. All study authors contrib- The POHA is a more holistic question which affords the subject uted to interpretation of data and critical review of the manu- the opportunity to intuitively integrate the overall efficacy of script, and agreed upon the final version independently of the the treatment (as assessed also by the PGA) with the adverse ef- sponsor. Dr. Simon is a consultant and Dr. Bookman has chaired fects it causes (or is perceived to cause). The PGA question was advisory boards for the sponsor. Dr. Grierson, Mr. Naseer and the same as that used in the previous trials of TDiclo Dr. Shainhouse own stock in and are employees of Nuvo Re- [5,12,43,50]. It is interesting to note that the improvement was search Inc. 244 L.S. Simon et al. / PAINÒ 143 (2009) 238–245

References [28] Hippisley-Cox J, Coupland C, Logan R. Risk of adverse gastrointestinal outcomes in patients taking cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs: population based nested case–control [1] Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, Christy W, Cooke TD, analysis. BMJ 2005;331:1310–6. Greenwald R, Hochberg M, Howell D, Kaplan D, Koopman W, Longley S III, [29] Hochberg MC, Altman RD, Brandt KD, Clark BM, Dieppe PA, Griffin MR, Mankin H, McShane DJ, Medsger Jr T, Meenan R, Mikkelsen W, Moskowitz R, Moskowitz RW, Schnitzer TJ. Guidelines for the medical management of Murphy W, Rothschild B, Segal M, Sokoloff L, Wolfe F. Development of criteria osteoarthritis. Part II. Osteoarthritis of the knee. Arthritis Rheum for the classification and reporting of osteoarthritis: classification of 1995;38:1541–6. osteoarthritis of the knee. Arthritis Rheum 1986;29:1039–49. [30] Hui X, Hewitt PG, Poblete N, Maibach HI, Shainhouse JZ, Wester RC. In vivo [2] Altman RD, Hochberg M, Murphy Jr WA, Wolfe F, Lequesne M. Atlas of bioavailability and metabolism of topical diclofenac lotion in human individual radiographic features in osteoarthritis. Osteoarthritis Cartilage volunteers. Pharm Res 1998;15:1589–95. 1995;3(Suppl. A):3–70. [31] ICH Harmonised Tripartite Guideline. E9: Statistical Principles for Clinical [3] American College of Rheumatology Subcommittee on Osteoarthritis Trials. International Conference on Harmonisation of Technical Requirements Guidelines. Recommendations for the medical management of osteoarthritis for Registration of Pharmaceuticals for Human Use. Geneva, Switzerland, of the hip and knee: 2000 update. Arthritis Rheum 2000;43:1905–15. 1998. Available from: http://www.ich.org/UrlGrpServer.jser?@_ID=276&@_ [4] Antman EM, Bennett JS, Daugherty A, Furberg C, Roberts H, Taubert KA. Use of TEMPLATE=254. nonsteroidal anti-inflammatory drugs: an update for clinicians: a scientific [32] Kearney PM, Baigent C, Godwin J, Halls H, Emberson JR, Patrono C. Do selective statement from the American Heart Association. Circulation cyclo-oxygenase-2 inhibitors and traditional non-steroidal anti-inflammatory 2007;115:1634–42. drugs increase the risk of atherothrombosis? Meta-analysis of randomized [5] Baer PA, Thomas LM, Shainhouse Z. Treatment of osteoarthritis of the knee trials. BMJ 2006;332:1302–8. with a topical diclofenac solution: a randomised controlled, 6-week trial [33] Lin J, Zhang W, Jones A, Doherty M. Efficacy of topical non-steroidal anti- [ISRCTN53366886]. BMC Musculoskelet Disord 2005;6:44. inflammatory drugs in the treatment of osteoarthritis: meta-analysis of [6] Banning M. The use of topical diclofenac for pain in osteoarthritis of the knee: a randomised controlled trials. BMJ 2004;329:324–6. review. Br J Community Nurs 2006;11:487–92. [34] Mamdani M, Juurlink DN, Kopp A, Naglie G, Austin PC, Laupacis A. [7] Banning M. Topical diclofenac: clinical effectiveness and current uses in Gastrointestinal bleeding after the introduction of COX 2 inhibitors: osteoarthritis of the knee and soft tissue injuries. Expert Opin Pharmacother ecological study. BMJ 2004;328:1415–6 [Epub 2004 May 11]. 2008;9:2921–9. [35] Marzulli FN, Maibach HI. Contact allergy: predictive testing in humans. In: [8] Bellamy N, Buchanan WW, Chalmers A, Ford PM, Kean WF, Kragg GR, Gerecz- Marzulli FN, Maibach HI, editors. Dermatotoxicology and pharmacology. Simon E, Campbell J. A multicenter study of tenoxicam and diclofenac in Advances in modern toxicology. Washington (DC): Hemisphere Publishing; patients with osteoarthritis of the knee. J Rheumatol 1993;20:999–1004. 1977. p. 353–72. [9] Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation [36] Mason L, Moore RA, Edwards JE, Derry S, McQuay HJ. Topical NSAIDs for study of WOMAC: a health status instrument for measuring clinically chronic musculoskeletal pain: systematic review and meta-analysis. BMC important patient relevant outcomes to antirheumatic drug therapy in Musculoskelet Disord 2004;5:28. patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833–40. [37] Moore RA, Derry S, McQuay HJ. Cyclo-oxygenase-2 selective inhibitors and [10] Bellamy N, Kirwan J, Boers M, Brooks P, Strand V, Tugwell P, Altman R, Brandt nonsteroidal anti-inflammatory drugs: balancing gastrointestinal and K, Dougados M, Lequesne M. Recommendations for a core set of outcome cardiovascular risk. BMC Musculoskelet Disord 2007;8:73. measures for future phase III clinical trials in knee, hip and hand osteoarthritis. [38] Moore RA, Derry S, McQuay HJ. Topical agents in the treatment of rheumatic Consensus development at OMERACT III. J Rheumatol 1997;24:799–802. pain. Rheum Dis Clin N Am 2008;34:415–32. [11] Biswal S, Medhi B, Pandhi P. Longterm efficacy of topical nonsteroidal anti- [39] National Collaborating Centre for Chronic Conditions. Osteoarthritis: National inflammatory drugs in knee osteoarthritis: metaanalysis of randomized clinical guideline for care and management in adults. London: Royal College placebo controlled clinical trials. J Rheumatol 2006;33:1841–4. Physicians; 2008. Available from: http://www.nice.org.uk/nicemedia/pdf/ [12] Bookman AA, Williams KS, Shainhouse JZ. Effect of a topical diclofenac solution CG059FullGuideline.pdf. for relieving symptoms of primary osteoarthritis of the knee: a randomized [40] Nishihata T, Kotera K, Nakano Y, Yamamazaki M. Rat percutaneous transport of controlled trial. CMAJ 2004;171:333–8. diclofenac and influence of hydrogenated soya lecitin. Chem Pharm Bull [13] Brobyn RD. The human toxicology of dimethyl sulfoxide. Ann NY Acad Sci 1987;35:3807–12. 1975;243:497–506. [41] Ozguney I. An alternative topical treatment of osteoarthritis of the knee [14] Chan AT, Manson JE, Albert CM, Chae CU, Rexrode KM, Curhan GC, Rimm EB, with cutaneous diclofenac solution. Expert Opin Pharmacother 2008;9: Willett WC, Fuchs CS. Nonsteroidal anti-inflammatory drugs, acetaminophen, 1805–16. and the risk of cardiovascular events. Circulation 2006;113:1578–87. [42] Pincus T, Koch G, Lei H, Mangal B, Sokka T, Moskowitz R, Wolfe F, [15] Chou R, Helfand M, Peterson K, Dana T, Roberts C. Comparative effectiveness Gibofsky A, Simon L, Zlotnick S, Fort JG. Patient Preference for Placebo, and safety of analgesics for osteoarthritis. Comparative Effectiveness Review Acetaminophen (paracetamol) or Celecoxib Efficacy Studies (PACES): two No. 4 (Prepared by the Oregon Evidence-based Practice Center under Contract randomised, double blind, placebo controlled, crossover clinical trials in No. 290-02-0024.). Rockville (MD): Agency for Healthcare Research and patients with knee or hip osteoarthritis. Ann Rheum Dis 2004;63: Quality; September 2006. Available from: www.effectivehealthcare.ahrq.gov/ 931–9. reports/final.cfm. [43] Roth SH, Shainhouse JZ. Efficacy and safety of a topical diclofenac solution [16] Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum (Pennsaid) in the treatment of primary osteoarthritis of the knee: a creatinine. Nephron 1976;16:31–41. randomized, double-blind, vehicle-controlled clinical trial. Arch Intern Med [17] Editorial Staff. Anti-inflammatory drugs, nonsteroidal (systemic). In: Klasco 2004;164:2017–23. RK, editor. USP DIÒ Drug Information for the Health Care Professional. [44] Rubin LF. Toxicologic update of dimethyl sulphoxide. Ann NY Acad Sci Micromedex, Inc., Englewood, Colorado, 20th ed.; 2000. p. 435 [Table 1]. 1983;411:6–10. [18] Evans JMM, McMahon AD, McGilchrist MM, White G, Murray FE, McDevitt DG, [45] Shainhouse JZ. OARSI guidelines for hip and knee OA: deciphering the topical MacDonald TM. Topical non-steroidal anti-inflammatory drugs and admission drug mélange. Osteoarthritis Cartilage 2008;16:1586–7. to hospital for upper gastrointestinal bleeding and perforation: a record [46] Shirley HH, Lundergan MK, Williams HJ, Spruance SL. Lack of ocular changes linkage case–control study. BMJ 1995;311:22–6. with dimethyl sulfoxide therapy of scleroderma. Pharmacotherapy [19] Forman JP, Rimm EB, Curhan GC. Frequency of analgesic use and risk of 1989;9:165–8. hypertension among men. Arch Intern Med 2007;167:394–9. [47] Tannenbaum H, Bombardier C, Davis P, Russell AS. An evidenced-based [20] Forman JP, Stampfer MJ, Curhan GC. Non-narcotic analgesic dose and risk of approach to prescribing nonsteroidal anti-inflammatory drugs. Third Canadian incident hypertension in US women. Hypertension 2005;46:500–7. Consensus Conference. J Rheumatol 2006;33:140–57. [21] Galer BS. All topical NSAIDs are not created equal – understanding topical [48] Towheed TE. PennsaidÒ therapy for osteoarthritis of the knee: a systematic analgesic drug formulations. Pain 2008;139:237–8. doi:10.1016/ review and metaanalysis of randomized controlled trials. J Rheumatol j.pain.2008.08.009. 2006;33:567–73. [22] Galer BS. Topical drugs for the treatment of pain. In: Loeser JD, editor. Bonica’s [49] Towheed TE, Maxwell L, Judd MG, Catton M, Hochberg MC, Wells G. management of pain. Philadelphia (PA): Lippincott Williams & Wilkins; 2001. Acetaminophen for osteoarthritis. Cochrane Database Syst Rev 2006 (1). p. 1736–41. 10.1002/14651858.CD004257.pub2 [Art. No. CD004257]. [23] Garcia CA. Ocular toxicology of dimethyl sulfoxide and effects on retinitis [50] Tugwell PS, Wells GA, Shainhouse JZ. Equivalence study of a topical diclofenac pigmentosa. Ann NY Acad Sci 1983;411:48–51. solution (PennsaidÒ) compared with oral diclofenac in symptomatic treatment [24] Gurtovenko AA, Anwar J. Modulating the structure and properties of cell of osteoarthritis of the knee: a randomized controlled trial. J Rheumatol membranes: the molecular mechanism of action of dimethyl sulfoxide. J Phys 2004;31:2002–12. Chem B 2007;111:10453–60. [51] U.S. Food and Drug Administration. COSTART: Coding Symbols for Thesaurus [25] Haynes S, Gemmell H. Topical treatments for osteoarthritis of the knee. Clin of Adverse Reaction Terms. 4th ed. Springfield (VA): National Technical Chiropractic 2007;10:126–38. Information Service, U.S. Department of Commerce; 1995. [26] Hewitt PG, Poblete N, Wester RC, Maibach HI, Shainhouse JZ. In vitro cutaneous [52] Watkins PB, Kaplowitz N, Slattery JT. Aminotransferase elevations in healthy disposition of a topical diclofenac lotion in human skin: effect of a multi-dose adults receiving 4 grams of acetaminophen daily: a randomized controlled regimen. Pharmacol Res 1998;15:988–92. trial. JAMA 2006;296:87–93. [27] Heyneman CA, Lawless-Liday C, Wall GC. Oral versus topical NSAIDs in [53] Williams AC, Barry BW. Penetration enhancers. Adv Drug Deliv Rev rheumatic diseases. A comparison. Drugs 2000;60:555–74. 2004;56:603–18. L.S. Simon et al. / PAINÒ 143 (2009) 238–245 245

[54] Yocum D, Fleischmann R, Dalgin P, Caldwell J, Hall D, Roszko P. Safety and Kwoh K, Lohmander LS, Tugwell P. OARSI recommendations for the efﬁcacy of meloxicam in the treatment of osteoarthritis. Arch Int Med management of hip and knee osteoarthritis. Part I: critical appraisal of 2000;160:2947–54. existing treatment guidelines and systematic review of current research [55] Zacher J, Altman R, Bellamy N, Brühlmann P, Da Silva J, Huskisson E, Taylor evidence. Osteoarthritis Cartilage 2007;15:981–1000. RS. Topical diclofenac and its role in pain and inﬂammation: an evidence- [57] Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, Bierma- based review. Curr Med Res Opin 2008;24:925–50. doi:10.1185/ Zienstra S, Brandt KD, Croft P, Doherty M, Dougados M, Hochberg M, Hunter DJ, 030079908X273066. Kwoh K, Lohmander LS, Tugwell P. OARSI recommendations for the [56] Zhang W, Moskowitz RW, Nuki G, Abramson S, Altman RD, Arden N, Bierma- management of hip and knee osteoarthritis. Part II: OARSI evidence-based, Zienstra S, Brandt KD, Croft P, Doherty M, Dougados M, Hochberg M, Hunter DJ, expert consensus guidelines. Osteoarthritis Cartilage 2008;16:137–62.