A randomized controlled trial to evaluate the slow-acting symptom-modifying effects of colchicine in osteoarthritis of the knee: A preliminary report

Abstract

Objective

To determine if colchicine added to nimesulide may have a beneficial effect on osteoarthritis (OA) of the knee.

Methods

Colchicine 0.5 mg twice daily or placebo was added to nimesulide (a nonsteroidal antiinflammatory drug) in 36 patients with OA of the knee in a randomized, double-blind, placebo-controlled trial over a 5-month period.

Results

The 30% improvement rate at 20 weeks was higher in the colchicine group than in the control group receiving placebo, as measured by total Western Ontario and McMaster University Osteoarthritis scores (57.9% versus 23.5%) and visual analog scale for index knee pain (52.6% versus 17.6%) (primary measures of response). The significance persisted on combined analysis by Mantel-Haenszel test (P = 0.062). Comparison of means also showed significant improvement in the colchicine group versus the control group in a multivariate analysis performed using T² test (P = 0.0115).

Conclusion

Among patients with OA of the knee, the group receiving colchicine plus nimesulide exhibited significantly greater symptomatic benefit at 20 weeks than did the control group receiving nimesulide plus placebo.

Osteoarthritis (OA) is traditionally considered to be an inherently noninflammatory disease, but acute flares are accepted as a component in the course of advanced OA (1). Inflammation in OA is frequently secondary to the presence of calcium-containing crystals, and leads to the production of interleukin-1 (IL-1), an important mediator of cartilage breakdown in OA (2). Because calcium-containing crystals are frequently seen in severe OA (3, 4), and colchicine has been shown to be beneficial in preventing calcium crystal–induced inflammation (pseudogout) (5), it is hypothesized that colchicine could have symptom-modifying or even disease-modifying effects in patients with OA. Even though the magnitude of benefit observed with colchicine in acute pyrophosphate arthropathy is not remarkable, a 12-week open clinical trial (6) showed that a regimen consisting of colchicine + piroxicam + intraarticular steroid was better than piroxicam + intraarticular steroid alone in patients with knee OA exhibiting inflammation and calcium pyrophosphate dihydrate (CPPD) crystals in joint fluid. A randomized double-blind placebo-controlled trial (Das SK: unpublished observations) showed significantly better symptom-modifying effects when colchicine was added to piroxicam plus intraarticular steroid over a 5-month period in patients with knee OA presenting with signs of inflammation irrespective of whether CPPD crystals were demonstrable.

Crystal deposition in OA is probably not an “on-off” phenomenon as previously thought (7), and may not only contribute to acute flares of inflammation in OA, but may also contribute to chronic low-grade, persistent, clinically nonapparent inflammation (1). This study was undertaken in patients with OA of the knee irrespective of the presence of clinical signs of inflammation. The purpose was to evaluate the benefit conferred by the addition of colchicine to nimesulide, a preferential cyclooxygenase inhibitor (8) with a half-life of approximately 3 hours, as compared with the addition of placebo to nimesulide in patients of OA of the knee, in a double-blind, randomized, placebo-controlled trial.

PATIENTS AND METHODS

Patients

Patients between 40 and 75 years of age, fulfilling the 1986 American College of Rheumatology (ACR) criteria for OA of the knee (9) without clinical or radiologic evidence of rheumatoid arthritis or other immunologic diseases, renal/hepatic diseases, allergies, or contraindications to the use of nimesulide, tramadol, and colchicine were included. All patients had a negative rheumatoid factor, serum uric acid <6 mg/dl, hemoglobin >10.0 gm/dl, total leukocyte count >4,000/mm³, serum creatinine <1.3 mg/dl, and transaminases <45 units/liter. All included patients had moderately severe symptomatic OA, determined by a minimum score of 25 on the Western Ontario and McMaster University Osteoarthritis total scale (WOMAC) (10). The WOMAC score was recorded after they had received at least 2 weeks of oral nimesulide, 100 mg twice a day. In the study, the WOMAC scale was recorded on a Likert scale of 0–4, where 0 = no pain/limitation; 1 = mild pain/limitation; 2 = moderate pain/limitation; 3 = severe pain/limitation; and 4 = very severe pain/limitation. The total score was the sum of all the responses.

Protocol

The research committee of our institution approved the trial protocol. On screening, the patients were assessed for eligibility criteria. All patients gave written informed consent. Clinical assessment was conducted at screening, at day 0, and at weeks 4, 8, 12, 16, and 20, and consisted of physical examination, and primary and secondary measures of disease assessment. A specially trained examiner (RS) performed all the clinical assessments. All investigators, including the examiner, had no knowledge of the patient's treatment assignments.

Standard erect anteroposterior and lateral projection radiographs were taken, and assessed independently by a rheumatologist (SKD) and a radiologist (R. Singh), using the Osteoarthritis Research Society (OARS) atlas (11). Biochemical evaluation and urinalysis were performed at baseline and at 20 weeks. Synovial fluid was aspirated from the more severely involved knee (index knee) under aseptic precautions in all patients showing at least 2 clinical signs of inflammation (n = 10), including palpable warmth, synovial effusion, soft tissue swelling, and joint margin tenderness. Adverse events were recorded on an open format for new problems and on a visual analog scale (VAS) for abdominal problems.

Treatment

Patients were randomized to receive either 0.5 mg of colchicine twice daily or a placebo. Both groups received compulsory nimesulide 100 mg twice daily for 4 weeks, which was continued if required by the patient. Concomitant therapy with other nonsteroidal antiinflammatory drugs (NSAIDs) and corticosteroids was not permitted. Tramadol 50 mg (an opioid analgesic with minimal addictive properties [12]) was used as rescue analgesic. Patients were advised not to use tramadol and nimesulide on the day of the visit.

The study drug was supplied in bottles and consisted of capsules containing starch and 0.5 mg of colchicine per capsule. The placebo was supplied in an identical capsule that contained starch alone. A person not involved in the trial performed randomization in blocks of 4, and the randomization code was sealed in individual envelopes until the completion of the study. It was not necessary to break the code for any patient.

Study end points

The primary measure of clinical response was 30% inprovement rates (responder rate) in total WOMAC scores and VAS for index knee pain at 20 weeks as compared with the respective scores at baseline. The other efficacy end points were: overall pain (i.e., “How severe is pain anywhere in your body”), patient assessment of severity of disease, and physician's assessment of disease severity. All end points were recorded on a VAS. Doyle's index (13), Modified Clinical Health Assessment Questionnaire (ModHAQ) (14), nimesulide and tramadol consumption were also recorded.

All VAS scores were recorded on a 15-cm scale. In patients who had equal involvement of both knees, the right knee was considered as the index knee. All HAQ scores were totaled.

The ModHAQ is a nonvalidated modification of a scale used at All India Institute of Medical Sciences, New Delhi, India and ClinHAQ (14) scale.

Statistical analysis

An intergroup univariate analysis by Student's t-test was performed between the assessment variables at 12 and 20 weeks in both groups. To take into account the correlation among the assessment variables, a multivariate analysis was also performed using T² statistics. Analysis was performed using BMDP statistical software version 7.1 (BMDP Statistical Software, Los Angeles, CA).

The proportions of patients in the colchicine and control groups showing ≥30% improvement at 12 and 20 weeks in the various assessment variables were compared using chi-square test. Mantel-Haenszel statistic was used to combine the odds ratio across the 2 × 2 contigency tables corresponding to the primary variables. An intent-to-treat analysis protocol was used. For patients who dropped out, their last assessment parameters were used for analysis (last observation carried forward).

A minimum sample size of 36 patients was calculated based on the assumption of a 30% improvement rate among 25% of the patients in the control group (nimesulide alone), and among 75% of the colchicine group (receiving nimesulide plus colchicine) in total WOMAC scores and VAS index knee pain at 20 weeks. A minimum power of 0.8 was seen, and the level of significance was 0.05.

RESULTS

The study population (n = 36) consisted predominantly of women (77.8%) with a mean age of 53.5 years, and mean duration of disease of 3.4 years (6 months to 20 years). On decoding, 19 patients had been assigned to the colchicine group and 17 patients to the placebo group. Baseline demographic, radiologic, biochemical, clinical, and evaluation measures were well matched in the 2 groups and showed no statistically significant difference. Twenty of the 36 patients had radiologic chondrocalcinosis. None had involvement of elbow, wrist, shoulder or metacarpophalangeal (MCP) joints suggestive of primary CPPD crystal deposition disease. Ten patients (27.8%; 6 colchicine and 4 control) had signs of inflammation. CPPD crystals were demonstrable in 8 patients (5 colchicine and 3 control). The differences between the colchicine and control groups were not statistically significant.

Of the 19 patients assigned to the colchicine group, 12 completed 5 months of the study; 1 communicated an inability to participate, 1 was not willing to continue because of associated cholelithiasis, 2 were lost to followup, and 1 withdrew due to drug-related diarrhea. Interestingly, 2 patients in the colchicine-treated group withdrew because they felt symptom-free after several months of colchicine treatment, (3 and 4 months, respectively). However, their best symptom reduction could not be recorded. They were categorized as dropouts due to poor compliance, and their last available responses were carried forward. Both patients reported recurrence of symptoms after 2 months of drug discontinuation. Of the 17 placebo-assigned patients, 12 completed 5 months; 1 was lost to followup, 1 patient was dropped because of noncompliance, and 3 withdrew because of lack of efficacy. No differences were noted in either the number of patients who withdrew, or the timing of withdrawal between the 2 groups.

Efficacy

The 30% responder rate at 20 weeks was significantly higher in the colchicine plus nimesulide group than in the nimesulide plus placebo group in total WOMAC score (57.9% versus 23.5%; P = 0.040) and VAS for index knee pain (52.6% versus 17.6%; P = 0.032) (Figure 1). Analysis performed to combine the odds ratio across the two 2 × 2 contigency tables showed a Mantel-Haenszel risk of 4.79, and a 95% confidence interval (95% CI) of 1.64–14.04. The test was highly significant (χ² = 7.49 at 1 degree of freedom; P = 0.0062, and the 95% CI did not include unity. Therefore, both statistics supported the same conclusion that there is a positive correlation between colchicine treatment and 30% responder rate in both total WOMAC score and VAS for index knee pain. Similarly, the proportion of patients who achieved at least 30% improvement at 20 weeks in other measures was significantly higher in the colchicine group in overall pain, patient's global assessment of disease, ModHAQ (Figure 1), physician's global assessment of disease, and nimesulide dose. The responder rate in Doyle's index did not achieve statistical significance.

Figure 1. Percentage of patients showing 30% improvement in different parameters at different visits.

On comparison of the mean between the 2 groups (Table 1), the significance levels using unadjusted total WOMAC scores at 20 weeks were marginal (F[1,34] = 4.3774, P = 0.044) using analysis of variance. However, when these scores were compared after adjusting for baseline values using analysis of covariance (ANCOVA) the difference was found to be highly significant (F[1,33] = 7.8, P = 0.0001). Similarly, the VAS score for index knee pain was significantly better in the colchicine-treated patients at 20 weeks (for unadjusted value F[1,34] = 12.74, P = 0.0011, and for baseline adjusted values F[1,33] = 11.2253, P = 0.002). Tramadol consumption was significantly less (P < 0.05) in the colchicine group at 4 and 20 weeks, and nimesulide consumption at 20 weeks (Table 1). The results were still significant when a multivariate analysis was performed using all the variables by T² test (T² = 36.4859, with F[9,26] = 3.1001, P = 0.0115).

Table 1. Outcome measures: comparison between the colchicine and control groups*

	Baseline			At 12 weeks			At 20 weeks
	Colchicine	Control	P	Colchicine	Control	P	Colchicine	Control	P
* Values are the mean ± SD. WOMAC = Western Ontario and McMaster University Osteoarthritis Index; VAS = visual analog scale; ModHAQ = Modified Clinical Health Assessment Questionnaire.
Primary
Total WOMAC scale	45.6 ± 12.2	41.3 ± 8.1	0.22	37 ± 14.5	39.8 ± 20.3	0.63	32 ± 15.9	44.9 ± 21.1	0.044
VAS—index knee pain	10.4 ± 2.5	11.3 ± 3	0.17	7.7 ± 2.8	9.1 ± 3.6	0.10	7.2 ± 3	10.6 ± 2.8	0.00054
Others
VAS—overall pain	8.6 ± 2.7	7.3 ± 3.3	0.11	5.7 ± 3.2	8.5 ± 3.4	0.007	5.7 ± 3.5	10.0 ± 3.1	0.00025
Patient's global assessment	9.0 ± 2.7	8.9 ± 3.1	0.49	6.4 ± 2.3	8.6 ± 3.3	0.012	6.4 ± 2.7	9.9 ± 2.6	0.00017
ModHAQ	7.4 ± 3.1	6.9 ± 2	0.63	4.4 ± 2.3	5.8 ± 2.8	0.098	3.6 ± 2.6	6.5 ± 2.4	0.0015
Doyle's index	4.7 ± 5.5	3.5 ± 3	0.42	3.1 ± 5	3.8 ± 6.9	0.70	2.7 ± 5.1	4.4 ± 6.8	0.41
Nimesulide intake, tablet/day	2 ± 0	2 ± 0	1.0	1.8 ± 0.4	2 ± 0	0.046	1.5 ± 0.6	1.9 ± 0.2	0.0049
Tramadol intake, capsule/month	0	0	—	2.5 ± 4.5	3.6 ± 4.1	0.44	0.8 ± 2.1	4.7 ± 4.6	0.0019
Physician's global assessment	8.2 ± 2	8.1 ± 2	0.459	6.5 ± 2.3	8.7 ± 2.8	0.006	6.1 ± 2.6	9.6 ± 2.4	0.00012

A subgroup analysis that included only those without joint inflammation who did not have their joints aspirated (n = 26) showed a significant difference in the 30% improvement rate in VAS score for index knee pain at 20 weeks between the colchicine group (53.8%) and the control group (15.3%) (P = 0.043). The respective rates in total WOMAC scores were 53.8% and 23.1% (P = 0.114). However, when an ANCOVA was performed among this subgroup in total WOMAC for equality of adjusted means, a statistically significant difference was discernable at 20 weeks (F[1,23] = 5.0174, P = 0.035).

Safety and tolerability

The VAS score for abdominal problems was significantly worse at 8 weeks in the colchicine treated patients as compared with the control group, but not significantly worse at 20 weeks. The most common adverse events encountered with colchicine were loose bowel movements (n = 7 versus control n = 1) and pain in the abdomen (n = 4 versus control n = 3), and were usually mild. One patient withdrew because of drug-related diarrhea.

DISCUSSION

The addition of colchicine to nimesulide resulted in better symptom control in patients with moderately severe OA of the knee over a 5-month period. Significantly more patients in the colchicine group showed a 30% or greater response rate in the primary assessment variables. Similarly, the comparison of mean between both the colchine and the control group showed a statistically significant difference on univariate analysis. A multivariate analysis was performed to take into account the correlation among multiple outcome variables, and the results were still significant.

Ten patients had their knee joint aspirated. Knee joint aspiration may produce improvements by way of placebo effect or by removal of crystals from joint fluid. The degree of symptom modification produced after synovial aspiration is not known, although beneficial effects with joint lavage have been described (15). Because patients who had their synovial fluid aspirated were evenly distributed between the 2 groups and constituted only a minority, it is unlikely to influence the results significantly. A subgroup analysis performed excluding those who had their knee aspirated still showed a significant difference in VAS scores for index knee pain and total WOMAC scores. It should be noted, however, that the level of significance was less when compared with the significance with all patients included. This could be due to the smaller number of patients in the knee-aspirate group.

The dose of colchicine was well tolerated, and there was no significant increase in new problems compared with the control group. This tolerance of colchicine observed in our study correlates in general with the observed toxicity of colchicine on short- or long-term administration (16).

The colchicine-treated patients showed significant improvement on the overall pain scale (Table 1, Figure 1). Although this scale can be influenced by comorbid conditions like fibromyalgia, none of the patients had fibromyalgia at the inception of the study. Improvement in overall pain may suggest that colchicine was not only effective in knee joint pain but also was effective in osteoarthritic pain at other sites.

The pattern of joint involvement in primary CPPD crystal deposition disease is often clinically distinct from primary OA. Involvement of the elbow, wrist, shoulder, and MCP joints raises the suspicion of primary calcium crystal deposition disease (17). In this study, patients were selected from a population of patients with a diagnostic label of OA satisfying the ACR criteria for OA of the knee. Moreover, none of them had involvement of joints atypical for OA, and therefore can be labeled as primary OA. CPPD crystals found in 8 of the 10 synovial samples analyzed are likely to be related to OA and not to acute primary CPPD crystal deposition disease, for previously mentioned reasons. Calcium-containing crystals are quite common in primary OA, occurring in a majority of patients in 2 studies (3, 4). The role of crystals in worsening of symptoms of OA is underrated at present. Provided no other known effects of colchicine apart from prevention of crystal-induced inflammation are responsible for the symptom modification in OA, it can be said that crystal presence may have relevance in the symptom genesis in OA.

The acute effects of colchicine in pseudogout are not as impressive as those in acute gout. In chronic gout and pseudogout, the use of colchicine has been advocated for prevention of recurrent acute attacks. The hypothesis that colchicine will be effective in OA is based on the potential to prevent crystal-induced acute inflammatory flares. However, in the present study, patients without inflammation at baseline were benefited by colchicine. Hence, a different mechanism is possible. Interestingly, colchicine has been shown to inhibit elastase, a matrix metalloproteinase (MMP) in patients with chronic obstructive pulmonary disease, and thereby prevents the progression of emphysema in those who have stopped smoking (18). Given the fact that MMPs play a pivotal role in the degenerative process of OA (2), the action of colchicine on various MMPs may be the predominant mechanism that needs further evaluation. This may be responsible for the slowly progressive increase in the response observed with colchicine in the study.