For years, non-specific anti-inflammatory drugs such as aspirin and paracetamol (acetaminophen) were the medications of choice for joint pain. But NSAIDs quickly became COX-2 inhibitors associated with adverse gastrointestinal effects such as peptic ulcers, leaving arthritis patients with a dilemma: manage the arthritis pain and risk serious stomach complications, or manage the ulcer and live with the pain.
Conventional NSAIDs work by decreasing the production of prostaglandins; they do this by inhibiting the enzyme cyclooxygenase (COX). Prostaglandins are responsible for inflammatory changes in the body.
But, despite being ‘demonised’ by modern medicine, prostaglandins also have important parts to play in everyday health. They are, for instance, involved in the normal functioning of blood platelets, the kidneys and lining of the gastrointestinal tract. Because of this, sweeping inhibition of prostaglandin synthesis can produce a variety of unwanted effects in addition to the desired helpful anti-inflammatory effect.
There are two distinct forms of COX enzymes in the body. The one known as COX-1 is released primarily in the gastrointestinal tract, kidneys and platelets. The other, COX-2, is released primarily in response to inflammation.
At least two COX-2 inhibitors, celecoxib (Celebrex, approved in December 1998) and rofecoxib (Vioxx, approved in May 1999) are available in the marketplace. In 2001, a third – valdecoxib (Bextra) – was added to the list. At their launches, the COX-2 inhibitors were considered a major advance in pain management.
But problems began to surface almost immediately. The first was calling these drugs ‘COX-2 inhibitors’, on the basis of a blood assay done in a lab, not in the body, despite the fact that the US Food and Drug Administration (FDA) has never approved the designation of any drug as ‘COX-2 specific’- that is, specifically designed to block only the actions of COX-2.
Currently available NSAIDs block both COX-1 and COX-2 to some degree (see box p 2), and it’s likely that inhibiting COX-1 causes the well-known toxic side-effects of NSAIDs – peptic ulcer, bleeding and perforation, and kidney damage. So, in theory, a drug that selectively blocks COX-2 will have anti-inflammatory effects, but without the toxic side-effects.
But, in reality, the actions of the two enzymes are not so well delineated. COX-1 and COX-2 serve identical functions in the body, and the involvement of the one more than the other depends on the other enzymes present – however specific they may act in the test-tube.
The FDA classes celecoxib, rofecoxib and valdecoxib as traditional NSAIDs. If the FDA approves a drug as a ‘COX-2 inhibitor’, it will probably have to demonstrate that it produces fewer serious gastrointestinal complications, such as bleeding, perforation and obstruction, than ordinary NSAIDs. But given the complex interactions of these and other enzymes in the body, none of the current so-called COX-2 inhibitors has yet been able to prove their specificity for COX-2 to the FDA’s satisfaction.
First, the good news
Human studies have found benefit with COX-2 inhibitors. When celecoxib and rofecoxib were compared with traditional NSAIDs (such as ibuprofen, naproxen and diclofenac) for osteoarthritis (OA) and/or rheumatoid arthritis (RA), results after six weeks showed pain relief that was significantly better than placebo in OA patients and comparable to that with traditional NSAIDs (J Rheumatol, 1999; 26: 2438-47). Similarly, patients with RA had significantly less pain, and joint swelling and tenderness with either rofecoxib or celecoxib than with a placebo (Lancet, 1999; 354: 2106-11; Clin Ther, 1999; 21: 1688-702).
Nevertheless, neither celecoxib nor rofecoxib has been taken for long enough periods of treatment in a large enough population to conclusively prove their safety.
Data from small trials do suggest a small reduction in GI events (JAMA, 1999; 282: 1919-33). Ulcers were far less frequent with celecoxib and rofecoxib in RA and OA patients compared with traditional NSAIDs. In OA, ulcers larger than 3 mm in diameter (which can develop into serious problems) were seen four to seven times less frequently with rofecoxib (depending on the dosage) than with ibuprofen (Gastroenterology, 1999; 117: 776-83).
In RA patients, the incidence of GI ulcers at 12 weeks with celecoxib was four to six times less frequent than with naproxen (Arthritis Rheum, 1998; 41: 1591-1602). However, with non-ulcer GI symptoms such as dyspepsia, rofecoxib and celecoxib have not convincingly proved able to reduce these symptoms compared with traditional NSAIDs.
The same study also found that neither of these COX-2 inhibitors interfered with blood clotting or bleeding time. However, both can slightly slow blood clotting in warfarin (Coumadin)-treated patients. Consequently, caution is now urged if patients taking anticoagulants are treated with selective COX-2 agents.
In the US, where drug companies are allowed to advertise directly to the public, ads on national television and in popular magazines regularly accentuate the positive side of the COX-2 inhibitors. Most reassure the public that these new anti-inflammatory drugs will safely reduce pain without the risks of adverse effects associated with older anti-inflammatories such as aspirin.
Unfortunately, the truth is that postmarketing medical research is rapidly proving that COX-2 inhibitors can be associated with severe and sometimes life-threatening toxic effects.
A price to pay
As with all ‘miracle cures’, the relief brought about by COX-2 inhibitors comes with a price. Much of the hoopla over these drugs has been based on the ‘fact’ that they cause fewer GI problems than conventional NSAIDs.
But, four years to the day after Celebrex was approved, a study suggests that the drug is no more effective in preventing a recurrent bleeding ulcer in arthritis patients than a traditional NSAID plus a proton-pump inhibitor such as Prilosec (N Engl J Med, 2002; 347: 2104-10). A popular treatment for heartburn that can help prevent stomach ulcers, Prilosec is often given to arthritis patients with a history of bleeding ulcer to counteract the high risk of NSAID aggravation of the problem.
In this study, nearly 5 per cent of patients taking Celebrex had recurrent bleeding – a potentially life-threatening condition – compared with 6 per cent of those taking a traditional NSAID plus Prilosec.
To medical news observers, this result was no surprise. In 1999, the findings of controlled clinical trials were already showing problems. In one of over 1000 patients with RA given celecoxib and compared with those given naproxen or a placebo, a similar rate of GI symptoms occurred with celecoxib and naproxen – 26 and 31 per cent, respectively.
The only real difference was with ulcers: naproxen resulted in five times more ulcers than celecoxib. Although the ulcers were usually less than 3 mm in diameter (and thus not clinically important), nevertheless, they can lead to serious ulcer complications such as perforation or bleeding (JAMA, 1999; 282: 1921-8).
In the same year, a review pooled the results of eight similar randomised controlled studies comparing rofecoxib, traditional NSAIDs and placebo in more than 5000 OA patients. The analysis focused on complications of GI ulcers (perforation, pain and bleeding). In a 12-month period, twice as many bleeding peptic ulcers were seen with NSAIDs than with rofecoxib. Rates for dyspepsia were similar in both groups – somewhere around 25 per cent (JAMA, 1999; 282: 1929-33).
An editorial accompanying the review suggested that to prevent one ulcer complication among low-risk patients in one year, some 500 patients would need to be treated with COX-2 inhibitors – at a cost of nearly $400,000 (JAMA, 1999; 282: 1961-3).
In the last couple of years, more disturbing adverse effects have come to light. An analysis by the World Health Organization of the side-effects of COX-2 inhibitors that have emerged since their release onto the marketplace suggests that the risks of kidney and cardiovascular problems (hypertension and cardiac failure) associated with the use of rofecoxib may be significantly greater than those associated with celecoxib or NSAIDs such as diclofenac and ibuprofen (Clin Ther, 2001; 23: 1478-91).
Other study results have only served to strengthen the WHO’s position. Indeed, the bad news has been uncovered almost by accident.
For instance, a recent review of smaller studies pooled the findings from 23,407 patients with data submitted to the US FDA by pharmaceutical companies (JAMA, 2001; 286: 954-9). Among these studies were two major randomised trials that originally investigated the gastrointestinal safety of rofecoxib and celecoxib: the Vioxx Gastrointestinal Outcomes Research Study (VIGOR), which involved 8076 patients, and the Celecoxib Long-term Arthritis Safety Study (CLASS), involving 7968 patients.
In the VIGOR, RA patients taking rofecoxib had a 238 per cent higher risk of having a cardiovascular event such as a heart attack, unstable angina, cardiac thrombus (blood clot), cardiac arrest, sudden or unexplained death, ischaemic stroke (due to narrowed or blocked blood vessels) and transient ischaemic attacks (mini-strokes in the brain) compared with those taking naproxen.
In the CLASS, there were no significant differences in cardiovascular event rates (myocardial infarction, stroke and death) between celecoxib and conventional NSAIDs (ibuprofen and diclofenac).
However, participants in the CLASS were allowed to take aspirin while those in the VIGOR were not – a twist that was significant for the gastrointestinal safety of celecoxib.
Although both studies were able to show an overall twofold reduction in serious GI complications, in the CLASS, this benefit with celecoxib was cancelled out if the patient also took low-dose aspirin. This finding is particularly relevant given the large number of elderly adults who regularly take low-dose aspirin to protect against heart attack.
Taken altogether, the yearly heart attack rates for COX-2 inhibitors, according to the findings in both the VIGOR and CLASS, were significantly higher than usual: 0.74 per cent with rofecoxib and 0.80 per cent with celecoxib compared with 0.52 per cent among those taking a placebo.
Not surprisingly, the results of this meta-analysis caused some outrage among COX-2 enthusiasts. They noted – and correctly – that neither CLASS nor VIGOR had been originally designed to look at cardiovascular events. Furthermore, they claim that the analysis of these data was done on a post-hoc basis – concluding a causal relationship between events solely because they took place one after the other in time.
Similarly, others have questioned the tactic of using ‘historical’ controls – comparing the findings with results from earlier studies in the medical literature.
Nevertheless, these disturbing findings raise a cautionary flag that should not be ignored.
Smaller studies have also thrown up problems with cardiovascular risk. In one study, blood pressure was increased significantly in 17 per cent of patients taking rofecoxib and in 11 per cent of those taking celecoxib (Am J Ther, 2001; 8: 85- 95). Indeed, no study so far has proven that taking COX-2 inhibitors is safe for the heart.
A kidney punch
High blood pressure increases the risk for heart attack and stroke, and also increases the risk of kidney failure. And case reports are now emerging that describe kidney problems in association with taking COX-2 inhibitors. What’s more, these kinds of side-effects are often reported as mere footnotes to studies – as if they are of no major importance at all.
In one report, it was concluded that Celebrex was effective for pain relief. Yet, the data also showed that celecoxib resulted in kidney problems such as hypertension, peripheral oedema and kidney failure in one-quarter of all patients receiving the COX-2 drug (N Engl J Med, 2002; 347: 2104-10).
In a recent case study reported in The Lancet, doctors describe a patient whose kidneys started to fail after taking Vioxx. The patient’s condition improved following discontinuation of the drug, an indication that the drug was the cause of the kidney dysfunction (Lancet, 2001; 357: 1946-7).
In another study comparing the effects of rofecoxib with the conventional NSAID indomethacin during regular long-term use, both drugs hampered the kidney’s ability to filter waste. However, the authors also suggest that the problem is likely to apply to the entire class of medications. This means that people on sodium-restricted diets and diuretic therapy, and those with impaired kidney function, hypertension, congestive heart failure or liver disease should avoid NSAID therapy, including COX-2 inhibitors (Ann Intern Med, 2000; 133: 1-9).
As sodium excretion by the kidneys is at least partly assisted by the COX-2 enzymes, the use of COX-2 inhibitors can slow the elimination of sodium from the body, thereby leading to fluid retention and raising high blood pressure even further (Curr Opin Rheumatol, 1997; 9: 178-82).
The decision to prescribe COX-2 inhibitors should take into account more than just the degree of pain.
According to Dr Joan M. Bathon in a discussion of COX-2 inhibitors in OA (www.hopkins-arthritis.som.jhmi. edu), there have been clinical trials showing that both rofecoxib and celecoxib can cause serious swelling of the feet even in patients who have normal kidney function.
Disturbingly, though, none of the current COX-2 inhibitors has yet been tested for safety in patients with compromised kidney function
Pain as a result of arthritis can be both debilitating and discouraging. Given that medical science has so little to offer sufferers of either rheumatoid arthritis or osteoarthritis in the way of a cure, simply relieving the pain of arthritis has become the major focus of treatment.
There are many alternative approaches that can help not only to reduce pain, but also to improve mobility and quality of life. A look at the ‘bigger picture’ of the problems of arthritic pain needs to be part of the conventional medical thinking if doctors are ever to offer sufferers more than a trade-off of debilitating symptoms.