A walking model to assess the onset of analgesia in osteoarthritis knee pain

Transcription

1 Osteoarthritis and Cartilage 18 (2010) 646e653 A walking model to assess the onset of analgesia in osteoarthritis knee pain E. Peeva y a, C.R. Beals y a, J.A. Bolognese z, A.J. Kivitz x, L. Taber k, A. Harman y, S.S. Smugar {, R.W. Moskowitz # y Clinical Research, Merck Research Laboratories, Rahway, NJ, United States z Clinical Biostatistics, Merck Research Laboratories, Rahway, NJ, United States x Altoona Center for Clinical Research, Duncansville, PA, United States k Pivotal Research Centers, Peoria, AZ, United States { Medical Communications, Merck Research Laboratories, North Wales, PA, United States # Department of Medicine, University Hospitals/Case Western Reserve University School of Medicine, Cleveland, OH, United States article info summary Article history: Received 28 July 2009 Accepted 21 December 2009 Keywords: Analgesia NSAID Onset Osteoarthritis Pain Tramadol Objective: To assess a walking model utilizing a set of standardized treadmill walks to measure acute analgesic response in osteoarthritis (OA) of the knee. Design: Randomized, double-blind, placebo-controlled, multiple dose, three-period crossover study. Patients 45 years of age (N ¼ 22) with symptomatic knee OA were randomized to naproxen 500 mg bid, tramadol/acetaminophen 37.5 mg/325 mg in forced titration, or placebo in each of three periods. Patients performed multiple 20-minute treadmill walks on Day 1 and Day 3 at a consistent self-selected pace predetermined at screening. Pain intensity (PI) during the walks was assessed on an 11-point numerical rating scale at 0, 3, 6, 9, 12, 15, 18, and 20 min. The primary endpoint was the time-weighted average (TWA) change from baseline PI on Day 3 for the two self-paced walks for the active treatments vs placebo. Time to moderate pain (TTMP) was a key secondary endpoint. Results: Compared with placebo, the TWA change from baseline PI on Day 3 was significantly better with tramadol/acetaminophen (P ¼ 0.043) but not with naproxen (P ¼ 0.089). TWA change from baseline on Day 1 was also significantly better with both tramadol/acetaminophen (P ¼ 0.001) and naproxen (P ¼ 0.048) compared with placebo. TTMP was significantly better for tramadol/acetaminophen and naproxen than placebo (P < to P ¼ 0.015) for walks on Day 1 after a single dose and on Day 3. Conclusions: This novel OA pain model was able to discriminate both tramadol/acetaminophen and naproxen from placebo after single and multiple doses. ClinicalTrials.gov identifier: NCT Ó 2010 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Introduction Osteoarthritis (OA) is a chronic disease 1. As such, most trials evaluating nonsteroidal anti-inflammatory drugs (NSAIDs) in OA have primary endpoints reflecting chronic therapy (e.g., 6 weeks) 2. Indeed, regulatory guidance recommends OA analgesia trials be 3 months in duration 3. However, despite the chronic nature of OA, the clinical course can be variable, with treatment needed only during exacerbations, often brought on by physical activity 1. This treatment pattern generally is confirmed by Address correspondence and reprint requests to: Elena Peeva, Merck Research Laboratories, 126 E Lincoln Ave, RY34B 272, Rahway, NJ 07065, United States. Tel: ; Fax: address: elena_peeva@merck.com (E. Peeva). a Drs. Peeva and Beals contributed equally to this paper. prescription patterns, showing that patients with OA receive only a few prescriptions each year, each for only approximately 1 month treatment 4e6. Thus, despite the chronic nature of OA, pain is often treated on a more acute, as-needed basis. Trials of NSAIDs in acute pain models 7e9 routinely evaluate acute onset of analgesia, but such measures of early onset are rarely assessed in OA NSAID trials 10e13, although it is not uncommon to evaluate the effect in the first days or week of treatment 14e16. Thus, there appear to be gaps among disease course, treatment patterns, and trial design in OA. Specifically, there appears to be a need to better assess acute onset of analgesia in OA. Moskowitz and colleagues demonstrated a proof-of-concept walking model to assess acute analgesia in OA, showing significant effect with single-dose rofecoxib or valdecoxib compared with placebo 11. We developed a walking model using multiple walks and a crossover design to evaluate single- and multiple-dose effects of /$ e see front matter Ó 2010 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. doi: /j.joca

2 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e two analgesics with different mechanisms of action (naproxen or tramadol/acetaminophen). Methods This was a randomized, double-blind, placebo-controlled, threeperiod crossover study conducted at two sites in the United States from June 2008 to November The protocol (Merck protocol 105) and study procedures were conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice Guidelines, regarding the protection of the rights and welfare of human subjects participating in biomedical research, and were approved by appropriate institutional review boards. Written informed consent was obtained for all patients before study enrollment. Patients Patients were 45 years of age with knee OA >6 months based on clinical and radiographic criteria and had an American Rheumatological Association functional class of IeIII. Patients were required to be taking analgesics for knee pain for 15 days in the previous 30 days, with a self-reported history of positive therapeutic benefit with NSAIDs, cyclo-oxygenase-2 (COX-2) inhibitors, acetaminophen, codeine, or tramadol. Eligible patients had pain intensity (PI) while standing 5 on a 0e10 numeric rating scale (NRS). Patients met minimum criteria for worsening pain following analgesic washout: 1) 40 mm on patient-reported pain walking on a flat surface (Western Ontario and McMaster Universities [WOMAC] visual analog scale [VAS] 3.0), with an increase 15 mm from screening value on a 100-mm [VAS]; 2) a worsening investigator global assessment of disease status (IGADS) 1 point on a 5-point Likert scale; and 3) a PI score 1 unit higher on the post-washout treadmill walk than the screening treadmill walk. Patients also demonstrated PI increase 3 units during a screening treadmill walk above PI measured while standing (at rest). Exclusion criteria included a history of inflammatory arthritis, active gout, intraarticular fracture, or partial or complete arthrodesis or arthroplasy of the study joint; acute ligamentous or meniscal injury of the study joint within the previous 2 years or arthroscopy within 6 months; history of gastrointestinal bleeding or documented ulcer disease in the past 5 years; uncontrolled hypertension, impaired renal function, or significant hepatic disease; history of dizziness, ataxia, unsteadiness, or decreased proprioception; history of falling in the previous 30 days; routine use of a cane or walker for ambulation; intra-articular steroids within 3 months or hyaluronic acid injections within 6 months to the study knee; intravenous, intramuscular, or oral corticosteroids, intra-articular steroid or hyaluronic acid injections to any other joint within 1 month; current narcotic analgesic use other than codeine or tramadol; or any chronic medication used for less than 1 month. Patients taking a stable dose of glucosamine or chondroitin sulfate were permitted to continue treatment at the same dose. Patients were permitted to continue physiotherapy (e.g., physical therapy) throughout the study if begun at least 1 week prior to screening and did not change in session duration or frequency during the study. Acetaminophen up to 4000 mg daily was permitted for rescue analgesia during the study, but use was not permitted within 24 h of study treadmill walks. Study design At Visit 1 patients performed a treadmill walk at a pace that was to be as fast as they felt they could safely maintain for 20 min, with PI assessed on an 11-point NRS (0 ¼ No Pain and 10 ¼ Worst Pain You Can Imagine ). Patients discontinued analgesic treatments and returned (Visit 2) after an appropriate washout of 3e15 days based on the half-life specific medication(s) and demonstrated worsening pain as described above. Patients meeting all eligibility criteria established a self pace that would be used in all treatment period SP walks by increasing the treadmill speed from 0 miles per hour (MPH) to 1 MPH over 15 s, and then 0.4 MPH every 15 s thereafter until patients walked a brisk, sustainable pace 1 MPH that could be maintained for at least 3 min, figuring changes in pain at a self-selected pace would be more reflective of the patient's pain status when walking during daily living. A high-pace (HP) treadmill speed was also selected; patients were started at their SP speed, and the rate was increased until a speed of SP þ 30% was reached. The highest pace above SP that the patient could walk safely for at least 5 min was the HP. SP and HP walks were each 20 min in duration, with PI measured at 0, 3, 6, 9,12,15,18, and 20 min during the walks. Study medication was administered in double-dummy fashion with over-encapsulated pills. Patients received treatments as multiple oral doses for 3 days in each treatment period. Dosing occurred twice on Day 1 (morning and evening); three times on Day 2 (morning, afternoon and evening), and once in the morning on Day 3. Because tramadol/acetaminophen 37.5/325 mg/mg is a weak opioid, the dose was titrated over the 3-day study period, starting with one tablet bid on Day 1, increased to one tablet tid on Day 2 and reached full dosage of two tablets on the morning of Day 3. Thus, during the tramadol/acetaminophen period, patients received total daily doses of 75/650 mg/mg on Day 1 and 112.5/ 975 mg/mg on Day 2, and 75/650 mg/mg administered as a single dose on Day 3. Patients received 1000 mg total daily dose of naproxen on Days 1 and 2 and 500 mg on the morning of Day 3. Allocation to balanced treatment sequences was determined according to a computer-generated schedule by the study statistician. Numbered containers were used to implement allocation. All study personnel, including investigators, study site personnel, patients, monitors, and central laboratory personnel, were blinded to treatment allocation throughout the study; the code was revealed to the researchers once recruitment, data collection, and laboratory analyses were complete. The study comprised three 3-day periods, with a 4- to 7-day washout between periods [Fig. 1(b)]. Treadmill walks were performed on Days 1 and 3 as shown in Fig. 1(a). Each walk was 20 min long with the patient resting for 40 min between each walk. Efficacy and safety assessments The primary efficacy endpoint was the change from baseline in time-weighted average (TWA) PI for both post-dose SP walks on Day 3, with TWA reflective of pain across the entire walk. Day 3 was chosen because a previous walking model in OA had not shown a significant difference between ibuprofen and placebo on TWA PI on Day Key secondary endpoints included TWA PI for all SP walks on Day 1 and for each individual walk on Days 1 and 3, the time to develop moderate pain (TTMP) using a 4-point Likert scale (none, slight, moderate, severe), and if applicable, the time to severe pain (TTSP). WOMAC questionnaire VAS 3.0 (100-mm VAS) was collected at the end of Days 1 and 3 after the completion of the last timed walks, with patients reporting pain, function, and stiffness results for the preceding 24 h. The incidence of overall adverse experiences (AEs), serious AEs, drug-related AEs, and discontinuation due to AEs was collected to evaluate tolerability and safety. Statistical methods The primary hypothesis was that naproxen and/or tramadol/ acetaminophen will have lower TWA PI for SP walks on Day 3 of

3 648 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e653 a Dose Dose Dose Dose Dose Dose SP SP SP HP SP SP HP SP -30min Day 1 Day 2 Day 3 b Treatment Period I Treatment Period II Treatment Period III acetaminophen Placebo Placebo acetaminophen acetaminophen Placebo D 1 D 2 D 3 D 1 D 2 D 3 D 1 D 2 D days 4-7 days Fig. 1. Study design and timing for dosing and walks in each treatment period. a. Timing for dosing and walks in each treatment period. Times shown represent hours post-initial dose. Observed doses administered at the clinic; all other doses administered at home. Treadmill walks: 20 min each followed by a 40-min rest period. HP: high-pace; SP: self pace. b. Study design. D ¼ Day. treatment compared with placebo. The primary endpoint was based on the two SP walks on Day 3 using all patients who received at least one dose of study medication and had baseline and at least one on-treatment observation. Only observed data were analyzed; no data were imputed. Baseline was defined as the TWA PI from the pre-treatment walk on Day 1 of each period. The same endpoint was assessed for the three SP walks on Day 1 and for each walk separately. The between-treatment differences were estimated and tested by the least squares (LS) mean for each endpoint compared between each active treatment group and placebo using an analysis of variance model for a 3-period crossover design. The model included terms for treatment, period, random subject effect, and baseline value included as covariate. Based on a measured standard deviation (SD) of 0.9 observed in a previous study (Merck protocol 057; ClinicalTrials.gov Identifier: NCT ) 17, a sample size of 18 patients provides 80% power to yield a statistically significant (a ¼ 0.05) between-treatment difference if the true underlying difference is 0.9 units on an 11-point NRS. Because each active treatment was compared with placebo independently, no adjustment for multiplicity was made for the primary endpoint. After results unblinding, we found that one outlier patient had exceptionally high PI scores; PI was >3 SD above the mean PI in the naproxen group for all walks in that treatment period. A review of the patient's medical history did not reveal trauma, injury, or any other explanation for his/her unusual response while taking naproxen. The previous walking model with 30 patients treated with ibuprofen showed no outliers 17. Additional post-hoc analyses of our primary and secondary efficacy measures were carried out excluding this patient. Most patients were female (63.6%), White (95.5%), and had a mean age of 59.6 years (Table I). Efficacy TWA change in PI For the primary endpoint, TWA change in PI from baseline on Day 3 SP walks, tramadol/acetaminophen (LS mean 1.7) had Results Patients Patient accounting is shown in Fig. 2. Nineteen patients (86.4%) completed the study, and all 19 were included in both the primary efficacy analysis and safety analysis. Two patients discontinued due to protocol violations, and one patient discontinued due to an AE (acute gouty attack), which was not considered to be drug-related. Fig. 2. Patient flow through the study.

4 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e Table I Baseline patient demographics N ¼ 22, n (%) Age (years) Mean (SD) 59.6 (7.9) Range 47e73 Gender Male 8 (36.4) Female 14 (63.6) Race White 21 (95.5) Black 1 (4.5) Disease duration (years) Mean (SD) 8.5 (6.5) ARA class I 0 (0.0) II 20 (90.9) III 2 (9.1) Other OA joint involvement Hand 4 (18.2) Hip 1 (4.5) Shoulder 1 (4.5) Spine 2 (9.1) Prescreening analgesic use NSAID 16 (72.7) COX-2 selective inhibitor 3 (13.6) Acetaminophen 9 (40.9) codeine 1 (4.5) Prestudy WOMAC scores (mm), mean (SD) Pain subscale 34.7 (12.6) Stiffness subscale 42.8 (15.7) Physical function subscale 36.1 (12.4) statistically significantly lower PI than placebo ( 0.9; P ¼ 0.043) [Table II, Fig. 3(a)]. The comparison between naproxen ( 1.5) and placebo was not statistically significant (P ¼ 0.089). Post-hoc assessment excluding the outlier subject showed similar mean responses with reduced variability, such that the TWA PI was statistically significant for both naproxen ( 1.7; P ¼ 0.009) and tramadol/acetaminophen ( 2.0; P ¼ 0.001) compared with placebo ( 0.7) [Fig. 3(b)]. The TWA changes in PI from baseline for SP walks on Day 1 were 1.5 for tramadol/acetaminophen, 1.0 for naproxen, and 0.4 for placebo (P ¼ vs tramadol/acetaminophen; P ¼ vs naproxen) (Table II). For the HP walks on Day 1, the TWA change in PI was 1.0 for tramadol/acetaminophen, 0.5 for naproxen, and 0.2 for placebo (P ¼ vs tramadol/acetaminophen; P ¼ vs naproxen). For Day 3 HP walks, the results were 1.3 for tramadol/acetaminophen, 1.1 for naproxen, and 0.2 for placebo (P ¼ vs tramadol/ acetaminophen; P ¼ vs naproxen). Mean PI responses for each walk demonstrated increasing pain during each 20 min walk. The separation in mean PI values between placebo and active treatments is apparent early on Day 1, persisting through Day 3 [Fig. 3(c)]. PI differences between tramadol/acetaminophen and placebo were statistically significant for five of the seven individual walks and were statistically significant for one of seven walks when compared between naproxen and placebo. For the two walks that were non-significant for tramadol/acetaminophen, the P-values were 0.06 and For the naproxen vs placebo difference, one additional walk had P ¼ When the outlier was excluded, both active treatments were significantly different from placebo for all walks (P-value range 0.001e0.022). acetaminophen was significantly more effective than placebo at the earliest treadmill walk at 2 h on Day 1 (LS mean 1.4 vs 0.8; P ¼ 0.03). was significantly better than placebo beginning at the treadmill walk at 6 h on Day 1 ( 0.9 vs 0.2; P ¼ 0.049). TTMP TTMP was significantly improved (i.e., longer) for both tramadol/acetaminophen and naproxen than with placebo for all walks on both Days 1 and 3 (Fig. 4). For Day 1 SP walks, the LS mean TTMP was 17.8 min with tramadol/acetaminophen (P < vs placebo), 16.4 min with naproxen (P ¼ vs placebo), and 14.2 min with placebo. For Day 3 SP walks, these values were 18.0 min (P ¼ vs placebo), 18.4 min (P < vs placebo), and 14.6 min, respectively. WOMAC scores For the effect on Day 1, the only significant difference from placebo ( 5.2) was for tramadol/acetaminophen ( 8.3) for WOMAC physical function scale (P ¼ 0.038) (Fig. 5). The effect of naproxen ( 6.9) was not statistically different from placebo. For Day 1 WOMAC pain walking on a flat surface, scores were 10.0, 3.8, and 4.8 for tramadol/acetaminophen, naproxen, and placebo, respectively (P > 0.05), and for WOMAC pain subscale, 9.0, 5.8, and 5.6, respectively (P > 0.05). For Day 3, however, both naproxen and tramadol/acetaminophen were significantly better than placebo for all three measures; for WOMAC pain walking on a flat surface the scores were 23.4 for tramadol/ acetaminophen, 23.6 for naproxen, and 7.1 for placebo (P ¼ vs tramadol/acetaminophen; P ¼ vs naproxen). For WOMAC pain subscale, the values were 20.6, 23.8, and 6.8, respectively (P ¼ vs tramadol/acetaminophen; P < vs naproxen). For WOMAC physical function subscale, the values were 18.7, 22.1, and 6.2, respectively (P ¼ vs tramadol/acetaminophen; P < vs naproxen). The correlations between the TWA change in PI for all treatments and the three WOMAC measures were mild: r ¼ 0.4 (P < ) for WOMAC pain walking on a flat surface, r ¼ 0.2 Table II Mean difference from placebo in TWA change from baseline in PI Self-pace Day 3 Self-pace Day 1 High-pace Day 3 High-pace Day 1 Acetaminophen Acetaminophen Acetaminophen Acetaminophen Endpoint (all patients n ¼ 22) (0.089) (0.043) (0.048) 1.03 (0.001) (0.019) 1.07 (0.007) (0.052) 1.15 (0.003) Endpoint (excluding outlier n ¼ 21) (0.009) 1.28 (0.001) (<0.001) 1.32 (<0.001) 1.15 (0.002) 1.28 (<0.001) 1.12 (<0.001) 1.54 (<0.001) All values expressed as LS means. Self-pace walk results are pooled for each day. For each of these responses a mixed model (treatment, walk as fixed effects, subject as random effect and baseline PI as covariate) was used to compare both treatments with placebo.

5 650 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e653 a LS mean difference with 90% CL c Pain Intensity Score (0-10) TWA PI LS mean difference from placebo SP Baseline 2 Hr Day 1 4 Hr HP Placebo acetaminophen 5 Hr (HP) Pace 6 Hr SP 52 Hr Walk (Each 0-20 min) LS mean difference with 90% CL TWA PI LS mean difference from placebo (excluding outlier) 0.0 Day 3 Day 1 Day Hr (HP) HP SP HP SP Pace acetaminophen 54 Hr b Fig. 3. Difference from placebo in PI. a. Difference from placebo in TWA change from baseline in PI on Days 1 and 3 (all patients). b. Difference from placebo in TWA change from baseline in PI on Days 1 and 3 (excluding outlier). c. Mean TWA PI (0e20 min) during all walks (all patients). All values are LS means. Bars represent LS 95% confidence limits (CL). HP: high-paced; SP: self-paced. <0.05, y 0.001, z <0.001, x <0.01, vs placebo. HP (P ¼ 0.01) for WOMAC pain subscale, and r ¼ 0.2 (P ¼ 0.08) for WOMAC physical function subscale. Safety A total of six patients had 10 AEs. There were four drug-related AEs: pruritis in a patient receiving naproxen, lightheadedness in a patient receiving tramadol/acetaminophen, drowsiness in a patient receiving tramadol/acetaminophen, and heartburn in a patient receiving naproxen. One patient discontinued due to an AE: acute gouty attack in a patient receiving tramadol/acetaminophen, which was not believed to be drug-related. There were no serious AEs. Discussion In this study we evaluated the acute analgesic effect of naproxen, tramadol/acetaminophen, and placebo in knee OA using a novel walking model modified from a previous model used in a truncated study 17, which failed to show a difference between a single dose of ibuprofen 800 mg and placebo during four 15-min walks over 4 h. Reasons hypothesized for the failed study include the lack of requirement for flare after analgesic washout, insufficient walk duration, insufficient overall observation time, and insufficient dosing duration. In the current study, we evaluated single- and multiple-dose response to two analgesics with different Time to Moderate Pain LS Mean difference from placebo WOMAC LS mean difference from placebo LS mean difference with 90% CL Day 1 Day 3 LS mean difference with 90% CL Flat Surface Pain Subscale Physical Function SP HP Pace SP HP Day 1 Day 3 Day 1 Day 3 Day 1 Day 3 Day acetaminophen acetaminophen Fig. 4. TTMP (all patients). HP: high-paced; SP: self-paced. <0.05, y <0.001, z <0.01 vs placebo. Fig. 5. Effect on WOMAC measures (all patients). WOMAC: Western Ontario and McMasters Universities OA Index. <0.01, y <0.001, z <0.05 vs placebo.

6 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e mechanisms of action in seven 20-min treadmill walks over 3 days. We found that tramadol/acetaminophen was significantly more effective than placebo for the primary endpoint of TWA PI change from baseline for SP walks on Day 3. acetaminophen was also significantly more effective than placebo at the earliest timepoint of 2 h. When excluding the outlier patient, both active treatments were significantly more effective than placebo for all evaluated endpoints. Both active treatments were also significantly better than placebo for TTMP for both SP and HP walks on Days 1 and 3. Significant differences between active treatment and placebo for WOMAC assessments were primarily noted only on Day 3. We observed single- and multiple-dose analgesic effects using treadmill-induced PI measures for treatments with established efficacy in treating OA pain 18,19. By including the WOMAC questionnaire, the endpoint most widely used for OA pain trials, we show that both treadmill-induced pain and WOMAC questionnaires pain subscale identify the treatment effects for both an NSAID and an opiate on Day 3. Individual patient data showed that change in treadmill-induced pain correlated with change in WOMAC pain subscale, but only to a small degree (r ¼ 0.2; P ¼ 0.01). Moreover, we utilized two different walk speeds to explore the relationship of walk pace and pain. It is known that patients with OA walk more slowly than healthy age-matched control subjects 20, and that selfselected pace is related to self-reported functional limitation 21.PI data for the two HP walks were compared with the SP walks revealing LS mean PI for HP walks was greater than for SP walks (LS mean difference w0.6 for each treatment, P < 0.001), indicating that increasing pain during treadmill walks is not merely an entrained response but reflects different levels of stimulus intensity. This performance-related walking model at self-selected pace has additional benefits in OA trial design and in identifying the attributes of analgesic medications. Our primary analysis was based on the active timed walks, which give a more objective measure of pain with function over time. As previously reported, there appears to be high and significant correlation between the WOMAC pain and physical function subscales 22, suggesting that they represent a single construct 23. Because of this, and because pain appears to drive physical function scores 24,25, patients may report improved WOMAC physical function scores despite worsening function on objective performance measures, such as timed walks 24,26,27. Thus, our walking model helps overcome these potential shortcomings by evaluating exertion-induced pain over the course of several timed walks. Moreover, given both the factorial importance of walking in the development of the WOMAC 28,29 as well as the importance patients place on walking 30,31 our findings of significantly prolonged TTMP further underscore the practical benefits of analgesic medications to allow additional functional performance before meeting a patient-defined threshold for discomfort. This walking model may be used to further study the interrelationship of pain and walking performance in OA and the impact of analgesics on both pain and performance domains. In a previous study using a walking model in acute OA pain, Moskowitz et al. 11 evaluated the effect of a single dose of rofecoxib 25 mg, valdecoxib 10 mg or placebo over a 6-h period comprising eight 10-min walks post-dose. They found valdecoxib was significantly better than placebo beginning at 4 h and rofecoxib better than placebo beginning at 5 h as measured by PI difference on a 100-mm VAS. However, our model appears considerably more efficient, as our Day 1 results were generally comparable with significant improvements with tramadol/acetaminophen at 2 h post-dose and with naproxen and at 6 h post-dose, but are considerably more sensitive, as we studied 22 patients rather than several hundred. The reasons for this greater sensitivity may be multiple, as we used a crossover design to reduce variability, controlled patient pace with a treadmill, required patients to demonstrate worsening on treadmill walk-induced pain prior to entry, employed longer walks, and measured single- and multipledose effects. The relative sensitivity of different efficacy endpoints can be quantified using the standardized effect size, which is the difference in treatment group means divided by the pooled SD, a reflection of treatment effect in relation to the underlying variability. Meta-analyses have estimated effect sizes for NSAIDs and COX-2 selective inhibitors in OA of the hip or knee of 0.23e0.69 using the WOMAC Questionnaire or Patient Global Assessment of Disease Status endpoint 32e34, indicating small-to-moderate changes compared with placebo. Our prespecified primary endpoint for treadmill walks on Day 3 yielded effect sizes of 0.5 and 0.6 for naproxen and tramadol/acetaminophen, respectively. When excluding the responses of one outlier patient, effect sizes increased substantially to 0.9 and 1.2, respectively. The improved power to observe treatment differences is illustrated by statistically significant effects of tramadol/acetaminophen on Day 1 with the treadmill walk pain endpoints but not with the WOMAC pain subscale. A more efficient clinical trial design such as this walking model may find utility in testing novel analgesics in patient populations and may be attractive for clinical researchers who have access to unique patient populations or wish to combine pain assessments with unique methodologies, such as imaging studies or neurophysiologic assessments. Acute pain relief in the context of an otherwise chronic disease is an important and possibly overlooked aspect of OA treatment. As such, this acute pain model may be relevant for the assessment of agents with rapid effects on OA pain. Despite the chronic nature of OA, clinical experience and prescription patterns show that many patients do not use analgesics on a daily basis. Rather, they tend to use them intermittently 4e6. Hawker and colleagues characterized both acute and constant features of the OA pain experience in focus groups, showing each component can be distinguished, and that the intermittent pain was more intense 35. Further, each of the mechanistic drug classes utilized in this study e acetaminophen, opiate, NSAID e is utilized chronically OA pain and also has activity in acute pain models such as third molar extraction, post-surgical pain, and acute ligamentous sprains. The particular tramadol/ acetaminophen combination used in this study is intended for short-term treatment of OA in some countries, so some regulatory bodies understand the need for acute, intermittent treatment of OA pain. Finally, the flare design of most OA pain trials e withdrawal of treatment to allow the patients' pain to worsen prior to randomized treatment assignment e demonstrates that this worsening of OA pain is transient, as most placebo-controlled studies show the resolution of the flare (PI) in many placebo-treated patients. Despite the importance of the acute pain component of the OA pain experience, the onset of analgesia in OA pain is not heavily studied. This clinical platform provides a tool with which to approach these questions. One limitation of this model is that we collected WOMAC scores at baseline and at the end of Days 1 and 3, but not before each treadmill walk. Such data would have allowed us to compare the TWA change in PI to the change in WOMAC scores over the same periods, which would provide a better measure of the relative sensitivities. Similarly, we did not collect any measure of ontreatment patient global assessment, which would have allowed us to further compare our model and WOMAC scores (e.g., as a receivereoperator characteristic curve). In addition, it should be noted that we compared naproxen, which is indicated for chronic symptomatic treatment of OA, with tramadol/acetaminophen, which is indicated for short-term (5 days) management of acute

7 652 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e653 pain only. Thus, the results of this 3-day study cannot be extrapolated to chronic OA in general. Lastly, although each primary hypothesis was considered separately, had we used a multiplicity adjustment for the primary hypotheses, neither naproxen nor tramadol/acetaminophen would have been declared effective at 3 days. However, without the statistical outlier patient, both agents would have been declared effective at 3 days even with multiplicity adjustment, because both P-values were <0.01 and the Bonferroni multiplicity adjustment would require both P-values to be < Likewise, we performed multiple analyses to assess the sensitivity and suitability of this walking model, but we did not adjust for multiplicity for the secondary, exploratory, or post-hoc analyses. These results should be interpreted accordingly and viewed as hypothesis-generating. The totality of the data, considering effects at Day 1 and across each of the individual walks (Table II, Fig. 3) suggests that the model may be capable of discriminating active analgesics from placebo after short exposure, which should facilitate the rapid development of novel agents. In conclusion, this novel treadmill walking model to evaluate acute analgesia in OA of the knee was able to discriminate active analgesic treatments from placebo within 2 h of the first dose and at multiple timepoints over 3 days of treatment, with effect sizes substantially larger than those observed with traditional OA trial endpoints. Conflict of interest Dr. Kivitz reports being an investigator in OA and rheumatoid arthritis trials for Greenwhich, Pfizer, Boston Life, Millennium, Searle, Schering-Plough, Boehringer Ingelheim, Immunex, Parke-Davis, AutoImmune, RW Johnson, Eli Lilly, Roche, Inflamatics, Corixa/Anergen, Amgen, Fujisawa, Centocor, Knoll, Alexion, Roche, Novartis, Abgenix, Bristol-Myers Squibb, Regeneron, MedImmune, Angiotech, Xoma, Scios, GenMab, Elan, Wyeth, Genentech, Abbott, Aventis, GlaxoSmithKline, Takeda, Purdue, Johnson & Johnson, and McNeil. Dr. Taber reports being an investigator in OA and rheumatoid arthritis trials for Abbott, Forest, Merck, Pfizer, Pozen, and ZARS. Dr. Moskowitz reports consultancies or advisory committee relationships with Pfizer, Sanofi-Aventis, Novartis, Nicox/SA, Endo Pharmaceuticals, Merck Research Laboratories, Nordic Bioscience, Adolor, Bioiberica, Rotta Pharmaceuticals, and Gelita Co. Drs. Peeva, Beals, and Smugar, Mr. Bolognese (formerly) and Ms. Harman are employees of Merck & Co., Inc., who may potentially own stock and/or hold stock options in the Company. All authors are responsible for the work described in this paper and were involved in at least one of the following: conception, design, acquisition, analysis, statistical analysis, interpretation of data; and drafting the manuscript and/or revising the manuscript for important intellectual content. All authors provided final approval of the version to be published. Acknowledgments The authors thank Chris Chung for his contribution to the conduct of the study, and Jennifer Pawlowski for her help in preparing the manuscript. No writing assistance was provided. Role of the Funding Source: This study was sponsored by Merck & Co., Inc. The sponsor's roles included study design, data collection, analysis and interpretation of data, writing of the manuscript, and the decision to submit the manuscript for publication. Supplemental material Supplementary information for this manuscript can be downloaded at doi: /j.joca References 1. Vignon E, Valat JP, Rossignol M, Avouac B, Rozenberg S, Thoumie P, et al. Osteoarthritis of the knee and hip and activity: a systematic international review and synthesis (OASIS). Joint Bone Spine 2006;73:442e Pham T, van der Heijde D, Altman RD, Anderson JJ, Bellamy N, Hochberg M, et al. OMERACT-OARSI initiative: Osteoarthritis Research Society International set of responder criteria for osteoarthritis clinical trials revisited. Osteoarthritis Cartilage 2004;12:389e Guidance for Industry: Clinical Development Programs for Drugs, Devices, and Biological Products Intended for the Treatment of Osteoarthritis (OA). Rockville, MD: United States Food and Drug Administration; Cullen DJ, Seager JM, Holmes S, Doherty M, Wilson JV, Garrud P, et al. Pharmacoepidemiology of non-steroidal anti-inflammatory drug use in Nottingham general practices. Aliment Pharmacol Ther 2000;14:177e Scarpa R, Sarzi-Puttini P, Cimmino MA, Caporali R, Parazzini F, Zaninelli A, et al. Analysis of pharmacologic and nonpharmacologic prescription patterns of general practitioners and specialists in the AMICA study. Semin Arthritis Rheum 2005;35:24e Rossignol M, Abouelfath A, Lassalle R, Merlière Y, Droz C, Bégaud B, et al. The CADEUS study: burden of nonsteroidal antiinflammatory drug (NSAID) utilization for musculoskeletal disorders in blue collar workers. Br J Clin Pharmacol 2009;67:118e Desjardins PJ, Mehlisch DR, Chang DJ, Krupa D, Polis AB, Petruschke RA, et al. The time to onset and overall analgesic efficacy of rofecoxib 50 mg: a meta-analysis of 13 randomized clinical trials. Clin J Pain 2005;21:241e Chelly JE, Nissen CW, Rodgers AJ, Smugar SS, Tershakovec AM. The efficacy of rofecoxib 50 mg and hydrocodone/acetaminophen 7.5/750 mg in patients with post-arthroscopic pain. Curr Med Res Opin 2007;23:195e Desjardins PJ, Black PM, Daniels S, Bird SR, Fitzgerald BJ, Petruschke RA, et al. A randomized controlled study comparing rofecoxib, diclofenac sodium, and placebo in post-bunionectomy pain. Curr Med Res Opin 2004;20:1523e Zacher J, Feldman D, Gerli R, Scott D, Hou SM, Uebelhart D, et al. A comparison of the therapeutic efficacy and tolerability of etoricoxib and diclofenac in patients with osteoarthritis. Curr Med Res Opin 2003;19:725e Moskowitz RW, Sunshine A, Hooper M, Olson NZ, Cawkwell GD. An analgesic model for assessment of acute pain response in osteoarthritis of the knee. Osteoarthritis Cartilage 2006;14:1111e Weaver AL, Messner RP, Storms WW, Polis AB, Najarian DK, Petruschke RA, et al. Treatment of patients with osteoarthritis with rofecoxib compared with nabumetone. J Clin Rheumatol 2006;12:17e Wittenberg RH, Schell E, Krehan G, Maeumbaed R, Runge H, Schlüter P, et al. First-dose analgesic effect of the cyclo-oxygenase- 2 selective inhibitor lumiracoxib in osteoarthritis of the knee: a randomized, double-blind, placebo-controlled comparison with celecoxib [NCT ]. Arthritis Res Ther 2006;8:R Bingham III CO, Sebba AI, Rubin BR, Ruoff GE, Kremer J, Bird S, et al. Efficacy and safety of etoricoxib 30 mg and celecoxib 200 mg in the treatment of osteoarthritis in two identically designed, randomized, placebo-controlled, non-inferiority studies. Rheumatology (Oxford) 2007;46:496e Kivitz AJ, Greenwald MW, Cohen SB, Polis AB, Najarian DK, Dixon ME, et al. Efficacy and safety of rofecoxib 12.5 mg versus

8 E. Peeva et al. / Osteoarthritis and Cartilage 18 (2010) 646e nabumetone 1000 mg in patients with osteoarthritis of the knee: a randomized controlled trial. J Am Geriatr Soc 2004;52:666e Battisti WP, Katz NP, Weaver AL, Matsumoto AK, Kivitz AJ, Polis AB, et al. Pain management in osteoarthritis: a focus on onset of efficacy e a comparison of rofecoxib, celecoxib, acetaminophen, and nabumetone across four clinical trials. J Pain 2004;5:511e A randomized, placebo-controlled single-dose 3-period crossover study to assess the tolerability and efficacy of ibuprofen 800 mg in a walking model of osteoarthritis knee pain. <http: // indication_id¼123&sort¼c.company_name&page¼1&drug_id ¼8335> [accessed ]. 18. Towheed TE, Hochberg MC. A systematic review of randomized controlled trials of pharmacological therapy in osteoarthritis of the knee, with an emphasis on trial methodology. Semin Arthritis Rheum 1997;26:755e Cepeda MS, Camargo F, Zea C, Valencia L. Tramadol for osteoarthritis. Cochrane Database Syst Rev 2006;3:CD Al Zahrani KS, Bakheit AM. A study of the gait characteristics of patients with chronic osteoarthritis of the knee. Disabil Rehabil 2002;24:275e Marks R. Reliability and validity of self-paced walking time measures for knee osteoarthritis. Arthritis Care Res 1994;7:50e Bingham III CO, Bird SR, Smugar SS, Xu X, Tershakovec AM. Responder analysis and correlation of outcome measures: pooled results from two identical studies comparing etoricoxib, celecoxib, and placebo in osteoarthritis. Osteoarthritis Cartilage 2008;16:1289e Ryser L, Wright BD, Aeschlimann A, Mariacher-Gehler S, Stucki G. A new look at the Western Ontario and McMaster Universities Osteoarthritis Index using Rasch analysis. Arthritis Care Res 1999;12:331e Witvrouw E, Victor J, Bellemans J, Rock B, Van Lummel R, Van Der Slikke R, et al. A correlation study of objective functionality and WOMAC in total knee arthroplasty. Knee Surg Sports Traumatol Arthrosc 2002;10:347e Faucher M, Poiraudeau S, Lefevre-Colau MM, Rannou F, Fermanian J, Revel M. Algo-functional assessment of knee osteoarthritis: comparison of the testeretest reliability and construct validity of the WOMAC and Lequesne indexes. Osteoarthritis Cartilage 2002;10:602e Stratford PW, Kennedy DM. Performance measures were necessary to obtain a complete picture of osteoarthritic patients. J Clin Epidemiol 2006;59:160e Stratford PW, Kennedy DM. Does parallel item content on WOMAC's pain and function subscales limit its ability to detect change in functional status? BMC Musculoskelet Disord 2004;5: Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833e Stratford PW, Kennedy DM, Woodhouse LJ, Spadoni GF. Measurement properties of the WOMAC LK 3.1 pain scale. Osteoarthritis Cartilage 2007;15:266e Bellamy N, Wells G, Campbell J. Relationship between severity and clinical importance of symptoms in osteoarthritis. Clin Rheumatol 1991;10:138e Bellamy N, Buchanan WW. A preliminary evaluation of the dimensionality and clinical importance of pain and disability in osteoarthritis of the hip and knee. Clin Rheumatol 1986;5:231e Bjordal JM, Ljunggren AE, Klovning A, Slordal L. Non-steroidal anti-inflammatory drugs, including cyclo-oxygenase-2 inhibitors, in osteoarthritic knee pain: meta-analysis of randomised placebo controlled trials. BMJ 2004;329: Lee C, Hunsche E, Balshaw R, Kong SX, Schnitzer TJ. Need for common internal controls when assessing the relative efficacy of pharmacologic agents using a meta-analytic approach: case study of cyclooxygenase 2-selective inhibitors for the treatment of osteoarthritis. Arthritis Rheum 2005;53:510e Zhang W, Doherty M, Arden N, Bannwarth B, Bijlsma J, Gunther KP, et al. EULAR evidence based recommendations for the management of hip osteoarthritis: report of a task force of the EULAR Standing Committee for International Clinical Studies Including Therapeutics (ESCISIT). Ann Rheum Dis 2005;64:669e Hawker GA, Davis AM, French MR, Cibere J, Jordan JM, March L, et al. Development and preliminary psychometric testing of a new OA pain measuredan OARSI/OMERACT initiative. Osteoarthritis Cartilage 2008;16:409e14.