I have to give the writers of this editorial credit for realizing that just because statins lower both heart attacks and LDL does not mean that lowering LDL would lower heart attack risk. They note that there may be other mechanisms at work “including amelioration of endothelial dysfunction, increased nitric oxide bioavailability, antioxidant properties, and inhibition of inflammation — that are unrelated to their lipid-lowering effect”.

So a lot of effort has gone into running trials that lower LDL without statins to see if those lower risk, but the authors note that these have not worked: “Several recent negative clinical trials have [shown] the addition of a nonstatin lipid-modifying agent to statin therapy conferred no significant increment in cardiovascular benefit over that seen with a statin alone”.

But all that has changed now, the editorial writers say, with the results of the IMPROVE-IT trial. They go on to say, “Perhaps the LDL hypothesis should now be considered the “LDL principle.””

In the IMPROVE-IT trial, researchers gave a control group a statin, and gave the experimental group the same statin plus a non-statin that reduces the absorption of cholesterol from the intestine.

Sure enough, the treatment group had 24% less cholesterol than the control group after just one year. And over the course of the 7 year trial, 34.7% of the control group hit a primary endpoint* compared to 32.7% of the treatment group. So there you go: proof that lowering LDL lowers risk.

But I’m not convinced, for several reasons.

First, at the one-year mark, patients in the treatment group had lower levels of triglycerides, apolipoprotein B, and high-sensitivity C-reactive protein when compared to the control group. These are known risk factors for heart disease! If treatment patients also had these factors in their favour, how do we know that it was reducing the LDL, and not these other things, that conferred this slight primary endpoint advantage? We don’t. But nowhere did the authors remark on this point in the discussion or the conclusion, as they attributed the difference in outcome solely on LDL.

Speaking of the outcome, that is a very small difference (34.7 vs 32.7), a relative risk difference of less than 6%. If you look further into the data, I’m not sure that “advantage” is worth much anyway. A higher percentage of patients who were in the treatment group died versus the control group (15.4% vs 15.3%)! This is worth emphasizing, because all of this effort is about prolonging life, and this treatment did not do that! A higher percentage of patients in the treatment group were also on aspirin, thienopyridine, and a beta-blocker, which also may have played a role in the results.

Finally, 42% of patients discontinued the medication prematurely. These patients were high-risk individuals who had been hospitalized for an acute coronary syndrome within the preceding 10 days, and yet they were rejecting what they were likely told were life-saving drugs. A high drop-out rate like this suggests to me there may be some major side effects to the drugs, though the trial authors did not cover this topic.

When people drop out, assumptions have to be made about what would have happened to them. When drop-out rates are high, the final results can be sensitive to these assumptions. Scientists are human; they want positive results. They are likely to make conscious or even unconscious assumptions that help them reach the conclusions they’re looking for. This study was funded by a major drug company.