Study design and selection of patients
The British Study of Risedronate in Structure and Symptoms of Knee OA (BRISK) was a 1-year prospective, double-blind, placebo-controlled study conducted in 10 centres in the UK. Male and female subjects aged 40 to 80 years with mild to moderate medial-compartment knee OA, diagnosed according to the clinical and radiological criteria of the American College of Rheumatology , were recruited into the trial.
OA in at least one knee, designated the signal knee, was required to meet the following clinical and radiographic inclusion criteria. Clinical inclusion criteria were the presence of daily knee pain for at least 1 month out of the 3 months preceding the study, with at least one of the following: age >50 years, morning knee stiffness of <30 minutes, or knee crepitus. The radiographic criteria for inclusion were a joint-space width (JSW) of 2 to 4 mm in the medial tibiofemoral compartment in the semiflexed anterior–posterior (AP) view of the signal knee and a narrower width than in the lateral compartment of the same knee. Patients were also required to have at least one osteophyte in either the medial or the lateral compartment of the tibiofemoral joint. Major exclusion criteria were the presence of rheumatic diseases that could be responsible for secondary OA; use of intra-articular hyaluronic acid in the signal knee; knee injury or diagnostic arthroscopy of the signal knee in the 6 months preceding enrollment; a history of knee surgery (including arthroscopy requiring an incision of internal joint components) in the signal knee at any time; intra-articular corticosteroids in the 3 months preceding enrollment; the presence of non-OA causes of knee pain in the signal knee (e.g. anserine bursitis, fibromyalgia, or osteonecrosis); and the use of bisphosphonates within the 12 months preceding enrollment.
The subjects gave their written, informed consent before entering the study, which was conducted in accordance with the International Conference on Harmonization (ICH) guidelines for Good Clinical Practice (GCP) and was approved by the UK Multicentre Research Ethical Committee (MREC).
The subjects were randomly assigned in a 1:1:1 ratio to one of three arms to receive risedronate at 5 mg or 15 mg or placebo once daily for 1 year. Before randomization, patients were stratified according to their current use of oestrogen or a selective oestrogen receptor modulator.
The subjects were instructed to take their study medication with at least 120 mL of water, 30 minutes before breakfast, or, if the medication was taken later in the day, at least 2 hours before or after food intake and at least 30 minutes before bedtime. They were instructed to take their study medication while they were upright and not to lie down for at least 30 minutes afterwards.
Symptom outcome measures
The outcome instrument for evaluation of risedronate efficacy on symptoms of OA was the Western Ontario and McMaster Universities (WOMAC) OA index . The visual analogue scale (VAS) of the index was used, in which patients assessed each question using a 100-mm scale, with a higher score representing greater symptom severity. The total index score for the signal knee corresponded to the weighted composite of the 24 question scores standardized to a 100-point scale. Scores were also determined for the subscales of pain (5 questions), stiffness (2 questions), and physical function (17 questions). Other symptom outcome measures included the patient global assessment (PGA) of disease, consumption of pain medication, and the use of walking aids. For the PGA, patients answered the following question using a VAS: "Considering all the ways your OA affects you, how have you been in the last 48 hours?" Patients marked values on a scale from 0 to 100 mm.
A step-down reduction in the use of pain medication was effected 5 days before all symptom evaluations. Patients were provided with approximately 30 tablets each of paracetamol (500 mg) and diclofenac (50 mg) to be used as the only pain medication 3 to 5 days before the baseline assessment and at visits at months 3, 6, and 12 (the exit visit). No pain medications were to be used 2 days before the scheduled evaluation date or on the day itself, with the exception of low-dose acetylsalicylic acid (<350 mg/day) for cardiac protection. Rescue analgesia was permitted during the study except for the 2-day washout period before each visit.
Structure outcome measures
The outcome measure for assessment of joint structural changes was the mean change from baseline values in minimum JSW of the medial compartment of the knee. Radiographs of the knee were taken at baseline and at 1 year using a standardized radiographic method with fluoroscopic positioning of the joint in a semiflexed position [14, 15]. By the use of this technique, the anterior and posterior rims of the tibia were aligned (to within 1 mm) for reproducible positioning. Radiographs were subjected to extensive quality control at the radiographic facility before dispatch to the Central Analysis Center . Radiographs were read centrally and their quality control was rechecked before computer software was used to obtain the radiographic magnification. This was determined from measurement of a metal ball placed at the head of the fibula at the time of radiography and was used to adjust the computerized measurement obtained of the minimum medial compartment JSW . The test–retest standard deviation of the difference between radiographs taken 2 days apart for this technique was approximately 0.2 mm, based upon repeat measurements in 199 subjects . A retrospective analysis was performed taking into account the precision of the instrument. Retrospectively, clinically meaningful disease progression was defined as joint-space narrowing of ≥ 0.75 mm or a ≥ 25% loss from baseline values. The ≥ 0.75-mm value is almost four times the 0.2-mm standard deviation observed for the x-ray method.
The relation between knee OA symptoms and radiographic joint-space narrowing was assessed retrospectively; the mean change in symptom scores between baseline to month 12 of the total WOMAC score and pain and function subscales was compared with the magnitude of change in JSW over the study period.
Bone and cartilage markers
Early-morning fasting urine and serum samples were collected at baseline and at months 3, 6, and 12 for assessment of markers of bone and cartilage turnover. Bone resorption was assessed by measurement of urinary levels of the N-terminal crosslinking telopeptide of type I collagen (NTX-I, Osteomark; OrthoClinical Diagnostics, High Wycombe, Bucks, UK) . Bone formation was assessed by measurement of bone-specific serum alkaline phosphatase (Ostase, Beckman-Coulter, San Diego, CA, USA)  and cartilage degradation was assessed by measurement of urinary levels of C-terminal crosslinking telopeptide of type II collagen (CTX-II, Cartilaps, Nordic Bioscience, Herlev, Denmark) . The intra-assay and interassay coefficients of variation were lower than 10%.
Evaluation of safety
Patient-reported adverse events (AEs) were recorded throughout the study. AEs were categorized using the Coding Symbols for Thesaurus of Adverse Reaction Terms (COSTART®) coding dictionary. Clinical laboratory measurements for safety monitoring were made throughout the study. Serious AEs were defined as any that resulted in death; were life threatening; resulted in hospitalization; resulted in persistent or significant disability or incapacity; or were judged to be medically significant.
Upper-GI AEs included the following symptoms and conditions: substernal chest pain; duodenitis; dyspepsia; dysphagia; oesophagitis; gastritis; bleeding gastritis; gastro-oesophageal reflux; oesophageal bleeding; GI bleeding; haematemesis; melena; abdominal pain; ulcers (duodenal, oesophageal, peptic, gastric); bleeding ulcers (duodenal, peptic, gastric); perforated ulcers (duodenal, peptic, gastric); perforated and bleeding ulcers (duodenal, peptic, gastric); and reactivated ulcers (duodenal, peptic, gastric).
To ensure 80% power to detect a 20% effect of risedronate treatment versus placebo with respect to pain modification (quantified according to the WOMAC pain subscale, assuming a standard deviation of 70 mm on a 0- to 500-mm scale), a 1-year dropout rate of 20%, and a type I error rate of 5% without adjustment for two comparisons with placebo control, the sample size requirement was 100 patients per treatment group.
Analyses were undertaken on the intention-to-treat (ITT) population. This was defined as all randomized patients who received at least one dose of study medication. All statistical analyses were performed using a two-sided statistical test with a type-I-error rate of 0.05. Baseline characteristics were compared using Fisher's exact test for categorical variables and the Kruskal–Wallis test for continuous variables. Extended Mantel–Haenszel tests with pooled centres as strata were used for end points with categorical responses. Analysis of variance (ANOVA) methods were used. Symptom analyses were adjusted for baseline value (PGA < WOMAC total or subscale value, as appropriate), pooled study centres, baseline use of oestrogen or selective oestrogen receptor modulators, gender, age, body mass index, and baseline JSW. Mean JSW analyses were adjusted for pooled study centres, baseline use of oestrogen or selective oestrogen receptor modulators, gender, age, body mass index, and baseline JSW as covariates. Each risedronate group was compared with the placebo group. For walking aids, the percentages were compared with placebo using the Cochran–Mantel–Haenszel test after adjusting for pooled centres. Individual AEs and the proportion of clinically meaningful JSW progressors were analysed using Fisher's exact test.
The WOMAC scores were calculated in accordance with the WOMAC User's Guide . The total scores were composed of subscales weighted as follows: pain = 42%, stiffness = 21%, and function = 37% . For each subscale, the reported response was the patient's average. If at least two pain items, both stiffness items, or more than four physical function items were omitted, or if the patient's response was unclear, the items were regarded as invalid and the relevant subscale was not included.