Lessons to learn from the COX-2
The bad news for the COX-2 pain relievers started in September 2004.
Rofecoxib (Vioxx) was pulled from the market after a study testing
whether the pain reliever could prevent colon polyps instead showed
that it doubled heart attack and stroke risk.
In December 2004, the FDA put its strongest “black box” warning
on valdecoxib (Bextra), another COX-2 inhibitor. One study of celecoxib
(Celebrex), the best seller in the class, showed that it, too, increased
cardiovascular risk, although a smaller study gave it a clean bill of
health. Pfizer kept Celebrex on the market, but the FDA forced the company
to stop advertising it to the public.
To add to the confusion, the National Institute on Aging stopped an
Alzheimer’s disease study because of evidence that naproxen — the
active ingredient in over-the-counter Aleve — also increased cardiovascular
risk. Yet, many doctors literally couldn’t believe the naproxen
finding because there was no theoretical reason or experimental evidence
for the drug increasing heart disease risk. In fact, like its fellow
nonsteroidal anti-inflammatory drug (NSAID), aspirin, there was some
reason to think it could protect against heart attacks.
By the beginning of 2005, the COX-2 headache was wearing off a bit.
A consensus had developed: The evidence for cardiovascular side effects
is strongest for Vioxx and Bextra, intermediate for Celebrex, and weak
In the spring of 2005, the FDA asked Pfizer to voluntarily withdraw
Bextra from the market — and asked manufacturers of all prescription
NSAIDs to include a boxed warning that highlights the potential increased
risk for cardiovascular events and serious gastrointestinal bleeding
associated with these drugs. Manufacturers of over-the-counter NSAIDs
were also asked to change their labeling to provide more specific information
about the potential risks from these drugs and to remind consumers to
take these medications according to the package instructions.
Well before the COX-2 drugs were on the market, experts suspected that
they might raise the risk for narrowed arteries and blood clots, a combination
that would lead to heart attack and stroke. We are also reminded that
although these studies found a doubled or even tripled risk for heart
attack and stroke, the risk was less than 1% to begin with and developed
after the drugs were taken for a year and a half or more. A doubling
of risk is hardly trivial, but a small risk doubled remains a small risk.
Here are a few morals from the COX-2 story:
There’s no such thing as a free lunch — or
a medication free of side effects. One of the ironies
is that the COX-2 inhibitors were marketed as having fewer side effects
than older analgesics. The history of pharmacology could be written
as a search for medications with fewer side effects. But we’re
not at the end of this history. Any medication has some risk of side
The fuzziness in the risk-benefit math. Every
medication (and medical treatment) teeters between benefit and harm.
Randomized clinical trials are the best way to get objective information
about where the balancing point lies. But even such “gold standard” studies
leave plenty of room for debate. The results may be objective, but their
interpretation is not.
Moreover, the balance of benefit and risk differs from one person to
the next. While the COX-2 drugs are no more potent than older NSAIDs,
they were developed to provide stomach-sparing pain relief. So the risks
of the COX-2 drugs may outweigh the benefits for someone who has heart
disease and who hasn’t tried other pain relievers or nonmedical
But the benefit may outweigh the risk for a younger person without
heart disease who can’t tolerate the side effects of the older
pain relievers and who doesn’t get relief from acetaminophen (Tylenol,
other brands). That’s why many people were unhappy that some of
these drugs were pulled from the market.
The COX-2 inhibitors were showing real promise in reducing the development
of precancerous polyps and their growth into full-blown colon cancer.
In fact, the heart disease risks came to light in trials testing them
for that purpose. Certainly some would accept an increased risk of heart
disease as the price for lowering their risk of colon cancer.
“On average” needs a reality check. Randomized
clinical trials give us information about the risks and benefits for
the “average” person. But that average patient is a statistical
abstraction; individual experiences will differ considerably.
It’s clear that your genes can affect your response to a drug,
but in most cases, we don’t know yet which genes are important.
Nor do we have an easy, affordable way to test for them. Doctors hope
that this will change as the field of pharmacogenomics (drug therapy
tailored to individual genetic makeup) blossoms.
Your lifestyle can also dictate how you respond to a given medication.
People whose lifestyle makes them vulnerable to heart disease may be
at greater risk from a drug like Vioxx than those with a heart-healthy
Marketing changes the equation. The FDA issued
guidelines in 1997 that made it easier for drug companies to advertise
on television. As a result, the companies tripled their spending on direct-to-consumer
advertising in the late 1990s.
Vioxx and the other COX-2 drugs really only make sense for two relatively
small groups of people: those who don’t get pain relief from the
older NSAIDs, and those who do get relief but only at doses that cause
stomach irritation or gastrointestinal bleeding. Because of broader use
stimulated by marketing, people who stood to gain little from Vioxx were
exposed to unnecessary risks.
Why can’t all the studies reach the same conclusion? Size
matters hugely in research studies. Having fewer subjects means a quicker
answer but provides a less reliable estimate of both the risks and benefits.
Even a large study involving several hundred subjects is likely to miss
a side effect that occurs in one in every thousand. Similarly, longer
studies may pick up on problems that shorter ones will miss. And even
two studies of exactly the same size and length can differ if the subjects
differ in a material way, for example, if one study has more people with
heart disease risk factors. Some have suggested creating an agency for
monitoring side effects of drugs once they’re on the market.
Put more tools in the FDA toolkit. Right
now, the FDA has three basic choices: It can recall a drug, threaten
to recall it, or order the manufacturer to change the drug’s label.
Some reformers think the agency needs more options. For example, the
European Medicines Agency can suspend the sale of a drug for a year while
its safety record is examined.
May 2005 Update
Back to Previous Page