Finally, I’m getting around to reviewing the two articles I’ve been threatening to review for the last week!

I’m referring to two articles published in the New England Journal of Medicine last week. Both are about reducing the risk of colon polyps by using celecoxib (Celebrex). The fact is that these articles are interesting not because of what they’re supposed to be studying – reducing the risk of colon polyps – but because they address the issue of the cardiovascular (CV) risk of this drug and this class of medication.

Part 1 – Introduction

Celecoxib is a relative of rofecoxib – otherwise known as Vioxx. I’m sure my readers have heard of Vioxx. It was pulled off the market in 2004 because it was found to be causing excess deaths from cardiovascular disease – in a study of its effect on colon polyps, incidentally. Hence our interest in celecoxib.

Celecoxib and rofecoxib are both COX2 inhibitors. COX2 inhibitors are a class of non-steroidal anti-inflammatory drugs (NSAIDs). NSAIDs are one of the most widely used types of drugs on the planet. They include pain killers such as aspirin, ibuprofen (Motrin, Advil), naproxen (Naprosyn, Alleve), indomethacin (Indocin), and many others. I’m pretty sure you’ve heard of them and have probably used them.

For a long time NSAIDs were the backbone of rheumatologists’ therapies because they not only reduce pain, but also reduce inflammation. Because so many of our diseases cause chronic inflammatory joint pain, they were used widely by us. In recent years, we have used them a little less because other drugs have been developed that work better and are safer for long term use.

However, NSAIDs are still very important. They are still the primary treatment for osteoarthritis (OA), one of the most common diseases around. OA is the common ‘wear and tear’ arthritis that you’ll see all over the place if you look around; in fact if you’re over 65, chances are you have OA. OA is not an inflammatory disease and hence does not respond to our new medications. NSAIDs are still the first line treatment for pain in OA.

Hence, we still prescribe a lot of NSAIDs. Rheumatologists were the doctors who primarily prescribed rofecoxib and who still primarily prescribe celecoxib. This is because this class of NSAID, the COX2 inhibitors, is supposed to be safer than traditional NSAIDs. You see, NSAIDs have a very well known side effect – they can (let me emphasize that this is not a certainty) cause gastrointestinal bleeding. This can range from a small ‘hole’ in the GI track which causes a little blood loss, to perforations causing massive bleeding and death.

We’ve known about this complication for many years. COX2 inhibitors are supposed to cause less GI bleeding. This is because this side effect is mediated through COX1. Traditional NSAIDs inhibit both COX1 and COX2. COX2 inhibitors affect only COX2, which is the major COX form that mediates inflammation, and therefore should not carry the same risk of GI bleeding; and indeed in clinical trials, this appears to be the case.

However, no one expected rofecoxib to cause more cardiovascular disease/events and its withdrawal for this reason was something of a shock. And it made us worry that this wasn’t just a risk of rofecoxib, but also a risk of other COX2 inhibitors like celecoxib. And in fact there is even some worry that traditional NSAIDs may carry an increased risk too.

(My own personal view is that this last statement is not well supported. The data that I’ve seen suggest that elevated cardiovascular risk is an effect of COX2 inhibitors only, not of NSAIDs in general. We have used NSAIDs for over 50 years, and there is a lot of data which, I believe, support their safety in this regard.)

This long introduction to COX2 inhibitors leads us back again to the second paragraph. These studies are the first official word on the cardiovascular safety of celecoxib. Because celecoxib is presently the only COX2 inhibitors approved for use in the US, it’s important to understand its safety profile. Hence, I and many like me are far more interested in this question than in these studies’ primary question: does celecoxib prevent colon polyps.

One last bit of explanation before we dig into the studies: colon polyps. Why does anyone care about them? The reason is that they are well established to be precursors for colon cancer. That doesn’t mean that all polyps will turn into cancer, but it appears that most if not all colon cancers arise from polyps. Hence, if you can catch and remove polyps early, you reduce the risk of colon cancer; and if you prevent their formation in the first place, you do the same.

Part 2 – Technical Stuff
Part 2.a – Methods

The first paper is by Arber, the second is by Bertagnolli, and that’s how I’ll refer to them. (If you want to see the actual articles you’ll have to have a subscription to the NEJM.)

Arber studied a once daily dose of 400 mg celecoxib compared to placebo. Bertagnolli studied 200 mg twice daily, 400 mg twice daily, and placebo. All subjects in both studies had a colonoscopy before enrollment and had had polyps removed. Hence, these are both secondary prevention trials, i.e. all patients already had ‘disease’, and the investigators were trying to prevent recurrence. After enrollment, patients were randomized to the placebo arm or one of the treatment arms.

All patients had regular examinations throughout the studies, and both studies gave the subjects colonoscopies at 12 and 36 months after enrollment to look for recurrence of polyps. Arber enrolled 1561 subjects, Bertagnolli 2035. All subjects had to abstain from other NSAID use throughout the length of the studies, though patients were allowed to use daily low dose aspirin; in fact, both studies stratified treatment by aspirin use (i.e. there were equal proportions of aspirin users in the treatment and placebo arms). The studies had similar inclusion and exclusion criteria which I won’t go into here.

Both studies had safety committees which monitored for adverse events. The Arber study formed the cardiovascular safety committee comparatively late, but when they did an analysis they showed no increased risk. However, by that point the Bertagnolli study had found an increased CV risk for celecoxib users, and stopped the study early. Even though the Arber group had not found an elevated risk, these finding prompted them to stop their study as well. (Interestingly, both studies had the same CV risk committee.)

Both studies were supported by Pfizer, the maker of celecoxib, though Bertagnolli also received funds from the National Cancer Institute.

Part 2.b – Results

Both studies found that celecoxib reduced the risk of colon adenomatous polyps (the precancerous kind). Bertagnolli found risk ratios (RR) of 0.43 (95% confidence interval 0.31-0.61) for the 200 mg twice a day group, and 0.34 (0.24-0.50) for the 400 mg twice a day group. Arber found an RR of 0.64 (0.58-0.75). Risks were similar in the aspirin and non-aspirin groups.

To put this in perspective, the authors are saying that, over 3 years, use of celecoxib roughly halves (or more!) the risk of recurrent colon polyps. That’s a big effect, and potentially an important one.

But let’s look at the safety data. I’ll concentrate on cardiovascular safety. Bertagnolli stopped the study early because of an elevated risk of CV disease (these included deaths ‘from cardiovascular causes, nonfatal myocardial infarction [‘heart attack’], stroke, or heart failure’). The RR for this was 2.6 (1.1-6.1) for the low dose group, and 3.4 (1.5-7.9) for the high dose group. In the Arber study, the RR for CV disease (same conditions) was 1.3 (0.7-2.6). The risks in Bertagnolli are ‘statistically significant’ (the 95% CI does not cross 1), while those in the Arber are not (the 95% CI does cross 1).

Once again, to put this in perspective, Bertagnolli is saying that the risk of a CV event in users of celecoxib is 2.5 to 3.5 times that in non-users.

In Bertagnolli, there was a larger risk for CV disease if subjects had a prior history of CV disease (not statistically significant), but this trend was not evident in Arber.

It’s a little hard to figure out exactly how many patients completed 3 years of study medication. It appears that about 87% of subjects had completed a whole 3 years of medication use in Bertagnolli, but I can’t find a comparable comparison in Arber.

Part 3 – Summing Up

So what does this mean? First, I think the Bertagnolli study strongly suggests that there is a significant risk of cardiovascular disease with celecoxib use and that the risk may be greater with higher daily doses. (Of note, 400 mg twice a day is a very hefty dose. We normally use on the order of 100mg-400mg a day.) This study also suggests that the risk is higher in patients who have a prior history of CV disease (not ‘statistically significant’ but the study was not designed to detect this kind of difference and probably had insufficient ‘power’ to show a difference; I suspect that the increased risk is really there).

Unfortunately, the Arber study doesn’t agree which makes things confusing. It’s hard to say why these studies gave different results, but my first question would be: what’s different in the Arber that didn’t allow them to find the effect that the Bertagnolli did?

My immediate suspicion is that this is the result of how many subjects completed a full 3 year course of treatment. The Bertagnolli is the earlier study (started in 1999), and therefore more patients probably completed a full 3 years of treatment. They say that 87% of subjects did. Arber began later (2001), and probably had fewer subjects who completed the full course; remember that they stopped the study early because of the findings in Bertagnolli. Unfortunately, Arber does not make clear how many patients completed a 3 year course – at least not that I found.

So my suspicion is that Arber simply didn’t have enough patients who received the drug long enough to show an effect.

So what do we do? Personally, I think that the risk of celecoxib has now been shown to outweigh the benefits in most patients. The fact that the same CV risk is evident in rofecoxib (Vioxx) supports the idea that this is a class effect of COX2 inhibitors. I do not think that I’ll be prescribing it routinely, especially as there are prescription-strength ‘antacids’ that reduce the risk of GI bleeding.

Having said that, I don’t think that celecoxib should be withdrawn from the market. In selected patients it might still be useful. For instance, in patients with no history of cardiac disease, but who do have a history of major GI bleeding and who you don’t want to use a traditional NSAID in , celecoxib in the lowest possible dose might be a good option. This is a decision to be made by patients and their doctors after a thorough discussion of the risks and benefits.

(Incidentally, I think the same is true of rofecoxib: I think it should probably be allowed back on the market. All drugs have potential side effects. That doesn’t mean we shouldn’t use them; it does mean that we should weigh the risks and benefits of treatment with them in each individual. Please note, though, that I’m making no comment here on whether Merck should be held accountable for failing to make the risks of rofecoxib known earlier. To weigh the risks of a drug, you have to know them!)

If you take celecoxib (once again, that’s Celebrex), don’t panic! It’s a risk, not a certainty, and the absolute risk of CV disease is still pretty low. But you might want to make an appointment with your physician to talk about the options in your case. Once again, I’m far from convinced that this increased risk hold true for all NSAIDs; I think it’s an effect of COX2 inhibitors. So, another NSAID may serve you just fine.