Skip navigation

Coxibs in arthritis: update July 2002

 

Clinical bottom line

The evidence is overwhelming that at licensed dose the coxibs work as effectively as maximum daily doses of NSAIDs.

What we have is a huge amount of good evidence about efficacy. Trials are of high quality, are large, and of long duration. They eclipse all previous studies of NSAIDs in arthritis.


Coxibs are drug that selectively inhibit the cyclooxygenase-2 enzyme without inhibiting cyclooxygenase-1. This, of course, is at normal (licensed) doses, and as doses increase some inhibition of cyclooxygenase-1 is likely. The importance of the difference between coxibs and standard NSAIDs, is that NSAIDs inhibit both enzymes. There are many different ways of measuring cyclooxygenase activity and inhibition, and much is made about how different drugs inhibit different enzymes in different systems, little of which is relevant to the properties of the drugs in clinical practice.


The key issue is this. Cyclooxygenase-1 is largely constitutive and responsible for, among other functions, protecting the stomach and duodenum from acid attack. Inhibition leads to poorer protection, and the development of erosions and ulcers, bleeding, and even death from a bleeding ulcer. Cyclooxygenase-2 is largely inducible and responsible for actions that lead to pain. Inhibition should reduce pain. Cyclooxygenase-2 is constitutive in brain and kidney.


What we should expect from coxibs, therefore, is what we get from NSAIDs, but with better gastrointestinal safety. This supposes that there are no other issues that have yet to be revealed. Certainly, there should be no expectation for better efficacy, other than from longer duration of action.

Efficacy of coxibs

What we do have for the coxibs is a large body of evidence from large trials that are often also of long duration. The available evidence for rofecoxib is shown in Table 1, for celecoxib in Table 2 and etoricoxib in Table 3. They show the coxibs to be more effective than placebo, and as effective as maximum daily doses of standard NSAIDs (ibuprofen 2,400 mg, diclofenac 150 mg, naproxen 1000 mg daily).

Table 1: Clinical trials of rofecoxib in rheumatoid and osteoarthritis

Reference

Design

Patients

Treatments

Outcomes

Results

Discontinuations

Rheumatoid arthritis
Schnitzer et al. Clin Therap 1999 21: 1688-1702 Randomised, double blind, placebo controlled over 8 weeks RA with VAS pain score of >40 mm, >8 tender joints and >5 swollen joints Rofecoxib 5 mg (n=158)
Rofecoxib 25 mg (n=171)
Rofecoxib 50 mg (n=161)
Placebo (n=168)
(7,800 mg paracetamol allowed/2 weeks)
ACR 20 response, patient assessment of pain and disease activity. Adverse events Higher doses of rofecoxib significantly better than placebo for all efficacy outcomes. Maximum effect after 2-4 weeks Total discontinuations similar to placebo (15-20%). Smaller number through lack of efficacy.
Bombardier et al. NEJM 2000 343: 1520-1528. Randomised, double blind, comparison with naproxen for median of 9 months RA at least 50 years old, or 40 on steroids. Rofecoxib 50 mg (n=4047)
Naproxen 1000 mg (n=4029)
Global assessment of disease activity No difference between rofecoxib 50 mg and naproxen 1000 mg Total discontinuations similar in both groups (about 29%).
Significant reduction in perforations, ulcers, and bleeds with rofecoxib.
Osteoarthritis
Laine et al. Gastroenterol 1999 117: 776-783. Randomised comparison of rofecoxib with ibuprofen and placebo for 12 weeks Randomised comparison of rofecoxib with ibuprofen and placebo for 12 weeks Rofecoxib 25 mg (n=195)
Rofecoxib 50 mg (n=186)
Ibuprofen 2400 mg (n=184)
Placebo (n=177)
( paracetamol allowed daily)
No efficacy measures No efficacy results Total discontinuations were 30-34% for placebo and rofecoxib, 61% with ibuprofen, but these included endoscopic ulcers and erosions
Hawkey et al. Arth Rheum 2000 43: 370-377 Randomised comparison of rofecoxib with ibuprofen and placebo for 12 weeks Randomised comparison of rofecoxib with ibuprofen and placebo for 12 weeks Rofecoxib 25 mg (n=195)
Rofecoxib 50 mg (n=186)
Ibuprofen 2400 mg (n=184)
Placebo (n=177)
( paracetamol allowed daily)
No efficacy measures No efficacy results Total discontinuations were 29-34% for placebo and rofecoxib, 59% with ibuprofen, but these included endoscopic ulcers and erosions
Cannon et al. Arth Rheum 2000 43: 978-987. Randomised, double blind comparison of two doses of rofecoxib with diclofenac over 1 year Patients at least 40 years old with OA of hip or knee and at least 40 mm on VAS pain scale. Rofecoxib 12.5 mg (n=259)
Rofecoxib 25 mg (n=257)
Diclofenac 150 mg (n=268)
WOMAC pain on walking, patients assessment of response and physician assessment of disease status All three treatments were comparable on all three outcomes. Maximum effect after 2-4 weeks Total discontinuations similar in all groups (38-46%).
Day et al. Arch Intern Med 2000 160: 1781-1787. Randomised, double blind comparison of two doses of rofecoxib with ibuprofen and placebo over 6 weeks Patients at least 40 years old with OA of hip or knee and at least 40 mm on VAS pain scale. Rofecoxib 12.5 mg (n=244)
Rofecoxib 25 mg (n=242)
Ibuprofen 2400 mg (n=249)
Placebo (n=74)
(2,600 mg paracetamol allowed daily)
WOMAC pain on walking, patients assessment of response and physician assessment of disease status All three treatments were comparable on all three outcomes and better than placebo. Maximum effect after 2-4 weeks Total discontinuations were ranged from 6-14%, with more lack of efficacy discontinuations with placebo and more adverse event discontinuations with ibuprofen
Saag et al. Arch Fam Med 2000 9: 1124-1134.
Study 1
Randomised, double blind comparison of two doses of rofecoxib with ibuprofen and placebo over 6 weeks Patients at least 40 years old with OA of hip or knee and at least 40 mm on VAS pain scale. Rofecoxib 12.5 mg (n=219)
Rofecoxib 25 mg (n=227)
Ibuprofen 2400 mg (n=221)
Placebo (n=69)
( paracetamol allowed daily)
WOMAC pain on walking, physical functioning and morning stiffness All three treatments were comparable on all three outcomes and better than placebo. Maximum effect after 2-4 weeks Total discontinuations were ranged from 11-28%, with more lack of efficacy discontinuations with placebo
Saag et al. Arch Fam Med 2000 9: 1124-1134.
Study 2
Randomised, double blind comparison of two doses of rofecoxib with diclofenac over 1 year Patients at least 40 years old with OA of hip or knee and at least 40 mm on VAS pain scale. Rofecoxib 12.5 mg (n=231)
Rofecoxib 25 mg (n=232)
Diclofenac 150 mg (n=230)
WOMAC pain on walking, physical functioning and morning stiffness All three treatments were of comparable efficacy Total discontinuations similar in all groups (32-36%).
Geba et al. JAMA 2002 287: 64-71. Randomised double blind comparison of two doses of rofecoxib with celecoxib and paracetamol for 6 weeks Patients at least 40 years old with OA of hip or knee and at least 40 mm on VAS pain scale. Rofecoxib 12.5 mg (n=96)
Rofecoxib 25 mg (n=95)
Celecoxib 200 mg (n=97)
Paracetamol 2,400 mg (n=94)
WOMAC pain on walking, at night and at rest, and patient global evaluation Rofecoxib 25 mg was generally superior to other treatments Total discontinuations were 17-30%.
Lisse et al. (ADVANTAGE) 2002 Randomised, double blind, comparison with naproxen for 12 weeks Patients at least 40 years old with OA of hip or knee for at least 6 months Rofecoxib 25 mg (n=2785)
Naproxen 1000 mg (n=2772)
Patient global assessment, SF-36 and others No difference between treatments Lack of efficacy discontinuations 6.4% in both groups

 

Table 2: Clinical trials of celecoxib in rheumatoid and osteoarthritis

Reference

Design

Patients

Treatments

Outcomes

Results

Discontinuations

Rheumatoid arthritis
Simon et al. Arth Rheum 1998 41: 1591-1602. Randomised double blind comparison of celecoxib with placebo over 4 weeks Patients with RA in a flare state with at least six tender and three swollen joints Celecoxib 80 mg (n=81)
Celecoxib 400 mg (n=82)
Celecoxib 800 mg (n=82)
Placebo (n=85)
Physician and patient global assessment, pain, swelling, tenderness and others Higher doses of celecoxib better than placebo Lack of efficacy discontinuations were 4-18%
Emery et al. Lancet 1999 354: 2106-2011 Randomised double blind comparison of celecoxib and diclofenac over six months Patients with RA of at least six months duration Celecoxib 400 mg (n=326)
Diclofenac 150 mg (n=329)
Patient and physician assessments and ACR 20 plus pain plus others Treatments same efficacy for almost all outcomes Total discontinuations were 20% for celecoxib and 28% for diclofenac
Simon et al, JAMA 1999 282: 1921-1928 Randomised double blind comparison of celecoxib and naproxen and placebo over three months Patients with RA of at least three months duration and at least six tender and three swollen joints Celecoxib 200 mg (n=240)
Celecoxib 400 mg (n=235)
Celecoxib 800 mg (n=218)
Naproxen 1000 mg (n=225)
Placebo (n=231)
Number of tender and swollen joints, patients assessment, ACR-20 All active treatments similar and superior to placebo Total discontinuations were 33-56%, predominantly lack of efficacy
Osteoarthritis
Simon et al. Arth Rheum 1998 41: 1591-1602. Randomised double blind comparison of celecoxib with placebo over 2 weeks Patients with OA of the knee in a flare state Celecoxib 80 mg (n=73)
Celecoxib 200 mg (n=76)
Celecoxib 400 mg (n=73)
Placebo (n=71)
Physician and patient global assessment, pain, and others Higher doses of celecoxib better than placebo Lack of efficacy discontinuations were 1-14%
Bensen et al. Mayo Clin Proc 1999 74: 1095-1105. Randomised double blind comparison of celecoxib with naproxen and placebo over 6 months Patients with OA with worsening on treatment discontinuation Celecoxib 100 mg (n=203)
Celecoxib 200 mg (n=197)
Celecoxib 400 mg (n=202)
Naproxen 1000 mg (n=198)
Placebo (n=203)
Physician and patient global assessment, WOMAC and others Active treatments better than placebo, with maximum effect at 2-4 weeks. Overall discontinuation was 43%, mainly due to lack of effect, but numbers in tables do not add up.
Williams et al. J Clin Rheumatol 2000 6: 65-74. Randomised double blind comparison of celecoxib with placebo over 6 weeks Patients with OA of the knee with worsening on treatment discontinuation Celecoxib 200 mg (n=231)
Celecoxib 200 mg (n=223)
Placebo (n=232)
Physician and patient global assessment, functional capacity, and Lequesne index Active treatments similar and superior to placebo Overall discontinuation was 16-37%, with lack of effect predominating for placebo
Silverstein et al. JAMA 2000 284: 1247-1255. Randomised double blind comparison of celecoxib with ibuprofen and diclofenac for up to 13 months (six months reported) Patients over 18 years with RA or OA for at least 3 months Celecoxib 800 mg (n=3987)
Ibuprofen 2,400 mg or diclofenac 150 mg (n=3981)
Ne efficacy assessment No efficacy result Total withdrawals were 29% with celecoxib and 45% with NSAID (57% of patients had 6 months treatment)

 

Table 3: Clinical trials of etoricoxib in rheumatoid arthritis

Reference

Design

Patients

Treatments

Outcomes

Results

Discontinuations

Rheumatoid arthritis
Collantes et al. BMC Family Practice 2002 3:10. Randomised double blind comparison of etoricoxib with placebo or naproxen over 12 weeks Patients with RA in a flare state with at least six tender and three swollen joints, and have RA for at least 6 months Etoricoxib 90 mg (n=353)
Naproxen 1000 mg (n=181)
Placebo (n=357)
Many different clinical and laboratory findings Etoricoxib and naproxen were equally efficacious, and both better than placebo Lack of efficacy discontinuations were 25% with placebo, 13% with etoricoxib and 11% with naproxen

 

Discontinuations

Much attention has been given to the gastrointestinal adverse effects of coxibs compared with NSAIDs. In clinical practice this is just one of the reasons why patients discontinue treatment. For instance, we know that over about six weeks, 5-10% of patients discontinue with NSAID or placebo because of adverse events. Knowing the rate of total discontinuation, adverse event discontinuations and discontinuations because of lack of effect is useful, and the results for placebo, paracetamol, NSAID and rofecoxib and celecoxib are shown in Table 4 and total discontinuations in Figure 1.

Table 4: Discontinuations in clinical trials of coxibs

2-6 weeks

8-12 weeks

24-26 weeks

36-52 weeks

Drug

Percent

Number

Percent

Number

Percent

Number

Percent

Number

Total discontinuation
Placebo

28

531

31

973

Paracetamol

31

94

NSAID

13

470

29

423

43

4640

30

4527

Rofecoxib

12

1123

16

332

24

769

31

5026

Celecoxib

14

864

27

1295

39

4269

Adverse event discontinuation
Placebo

6.0

531

4.6

973

Paracetamol

6.4

94

NSAID

6.4

470

4.7

423

20

4311

16

4527

Rofecoxib

5.3

1123

5.4

332

9.8

769

16

5026

Celecoxib

4.4

864

5.9

1295

19

3943

Lack of efficacy discontinuation
Placebo

19.0

531

20.0

973

Paracetamol

17.0

94

NSAID

6.0

470

8.3

3195

14

4311

7.1

4527

Rofecoxib

5.1

1123

6.4

3117

16

769

8.0

5026

Celecoxib

6.9

864

19.0

1295

13

3943

Figure 1: Total discontinuations in coxib trials


Total discontinuations with placebo are high early, mainly because of lack of effect (though paracetamol is usually given as escape analgesic). Paracetamol has much the same discontinuation rate. NSAIDs and coxibs have more discontinuations later. The message, though, is that even with the potentially greater safety of coxibs, 30-40% of patients will need their medicine changed for some reason within six months to one year. With paracetamol it may be higher.

Comment

Efficacy is not an issue, in the sense that the coxibs work as effectively as NSAIDs. What we do not have is any analysis to tell us which patients do well on what drugs, and why. We are still left with patients with arthritis whose average pain is moderate while on treatment. Even more relevant, but what we don’t have, is some sensible care pathway, based on good evidence.

What we do have is a huge amount of good evidence about efficacy. Trials are of high quality, are large, and of long duration. They eclipse all previous studies of NSAIDs in arthritis.