ARBs and Cancer -- Are We Looking at Junk Science?

June 28, 2010

This feature requires the newest version of Flash. You can download it here.

Hi. I'm Dr. Henry Black. I'm Immediate Past President of the American Society of Hypertension and a Clinical Professor of Internal Medicine at the New York University School of Medicine and a member of the Center for the Prevention of Cardiovascular Disease.

We're now faced with a very interesting issue in how you do science, how you interpret science, how science is covered, and what the public has to do with what is sometimes very confusing information. I'm referring to a recent paper in Lancet Oncology by Dr. Sipahi[1] concluding that angiotensin receptor blockers may be causing cancer. I say "may be" because that's what they said. But that's not necessarily what was picked up by our print and electronic media. What Sipahi and associates did was based on a finding from the Cardiovascular Health and Age Related Maculopathy (CHARM) study (which used candesartan), of an increase in cancer deaths; they began to look at the possibility that angiotensin receptor blockers might be responsible for cancer deaths.

Now when you do a study of this type, you have to start out with some biological reason for why a finding might make sense. The researchers refer back to some experimental studies showing that angiotensin receptors (maybe AT2 receptors, which are not blocked by angiotensin receptor blockers) might in some ways stimulate cell growth and maybe tumor growth. Yet in the studies that we need for US Food and Drug Administration approval, thousands and thousands of hours are spent, large numbers of experimental animals (mostly mice and rats and occasionally rabbits) are tested to see whether there's any increase in tumors. Nothing gets by, to even get to humans, until the data are past that barrier. Angiotensin receptor blockers, which are very widely used right now (in fact they have been available for 15 years) have never had any problem at that level.

So the researchers collected the available large, randomized clinical trials, most of which were less than 5 years in duration, and saw whether there was any increase in cancer occurrence or cancer death as well as the many other ways that they cut these data. There must be nearly 20 or 25 large clinical trials that they considered including, but they excluded trials that were less than 1 year in follow-up duration. They also excluded trials that had fewer than 100 volunteers, all of which was quite appropriate. Although, I'm not sure that less than 1 year is good enough for something like cancer because cancer is usually latent and we unmask it as we go along. We've learned that from the original accusation that calcium channel blockers cause cancer. In the Women's Health Initiative, where we used calcium channel blockers, we counted nothing unless it happened at least 2 years after treatment.

What Sipahi and associates found was about a 10% increase (maybe 15% increase when they pooled the large trials) in cancer. They were impressed with the number: 61,000 individuals. Almost all of them came, though, from 1 of 2 studies, ONgoing Telmisartan Alone and in combination with Ramipril Global Endpoint Trial (ONTARGET) and Telmisartan Randomized AssessmeNt Study in aCE iNtolerant Subjects with CV Disease (TRANSCEND), which used telmisartan. They also talked about Prevention Regimen for Effectively Avoiding Second Strokes (PRoFESS), which also used telmisartan, but somehow that got mixed up a little, because that trial used telmisartan plus dipyridamole and aspirin together and alone, and that was only about a 2.5 year study. So that trial shouldn't have been discussed at all. They included Trial of Preventing Hypertension (TROPHY), which did last 2 years, and used candesartan, but I'm not sure they read the protocol because after being on candesartan for 2 years everybody was switched to placebo. So that study shouldn't have been included at all. The curious thing, though, is that if you line the studies up as you do with the forest plot and look at what you see in the initial finding, no single study reached statistical significance. Only if you put them together, and meta-analyze them, do you get this modest increase in cancer.

Now this wouldn't matter too much -- but these are very popular drugs. We view them as the best tolerated of our cardiovascular drugs that block the immune and angiotensin systems, and that we give to our hypertensive patients, our heart failure patients, and maybe our patients with proteinuria and renal disease. So what happens when this information gets to the public is that someone who isn't really sure what to do is going to stop their medication, or they are going to call the doctor and say, "What should I do?" And you stop the medication if people are concerned about this.

Now what angiotensin receptor blockers replaced were angiotensin-converting enzyme (ACE) inhibitors, which have their own problems: angioedema, cough, and things that we may not know about just yet. They looked at the studies where ACE inhibitors were used and studies where they weren't used, and they tried to separate those out, and made the point in the paper that those investigators didn't allow patients to take ACE inhibitors. Well, that doesn't happen in trials. In trials there are certainly crossovers. So what you need are individual patient data that you analyze very carefully to see if there's something going on, and we don't have individual patient data. The other thing, there was no pattern about the cancers. There was an increase in lung cancer in 1 study, the Lifestyle Interventions and Independence for Elders (LIFE) study that used losartan, but not in the other studies. There was no increase in prostate or breast cancer.

If there is a common pathogenesis, the findings should be consistent. We would demand that of any study. Dr. Nissen (who commented on this) said "the more you look at a trial, and the more questions you ask, the more spurious findings you get." This happened with reserpine and breast cancer back in the 1970s when people blamed reserpine for causing breast cancer until we knew better. Calcium antagonists were blamed for cancer so we specifically look at the association with cancer in Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack (ALLHAT) and it wasn't there. We looked for it in Controlled Onset Verapamil Investigation of Cardiovascular End Points (CONVINCE), and it wasn't there. So I'm concerned that studies such as the one by Sipahi and colleagues need to be dissected very carefully and who you included and who you didn't has to clear, and we shouldn't be overwhelmed by these large numbers. I think right now this is irresponsible, it's dangerous, and it can cause patient harm. They also calculated what they call the "number needed to harm" and got a pretty robust number -- 100, 150 people -- but they didn't use comparative data. They picked people from 65 to 69 years old, which was the average age group of these trials, and they looked at population data. From where? The trials were done all over the world. The populations are going to be different. The incidence is going to be different. Very irresponsible, and it's junk science in my opinion. Thank you very much.