Added: July 8, 2018
Updated: August 26, 2018

Cholesteryl ester transfer protein inhibitors (CETPi) entered clinical trials with a seemingly solid biological hypothesis and great clinical promise. This class of drugs raises high-density lipoprotein (HDL), the so-called “good cholesterol”. The hope was that raising HDL with a CETPi would lead to a reduction in cardiovascular risk, similar to how lowering LDL cholesterol with a statin is associated with a reduction in risk.

Unfortunately, one after the other, the CETPi’s failed to show benefit in clinical trials. One member of this class (torcetrapib) even increased the risk of serious cardiovascular events. However, in 2017, after more than a decade of CETPi trials and thousands of patients enrolled, Merck’s trial of anacetrapib found a small, but statistically significant benefit. Yet, shortly after that trial’s result was published, Merck announced that they were abandoning the drug anyway.

The dashboard below helps to shed some light on this development program. The "AERO graph" tab depicts the entire portfolio of CETPi trials, stratified by the individual drug and the phase of development. The "Histogram" tab shows the number of completed trials by phase each year. The "Outcome Matrix" tab summarizes the quality of evidence for each primary outcome across this portfolio of trials. The "Data Table" tab shows the data underlying these visualizations. The "Methods" tab describes the search strategy that produced the data and the original peer-reviewed publication.

Bokeh Plot


By: Spencer Phillips Hey, Ph.D., Brigham and Women's Hospital and Harvard Medical School
July 8, 2018

It is quite unusual for a company to not pursue a regulatory approval for a product that has been shown to be beneficial in a large randomized clinical trial. The decision to abandon anacetrapib is therefore puzzling, especially when one considers how much the pivotal clinical trial must have cost (it ran for about 6 years and enrolled more than 30,000 patients). It also seems like a potential loss for the patient population—some of whom do not tolerate statins or whose cholesterol is not well-controlled by other means.

But even though we do not have access to Merck’s reasoning, there are some clues as to why they might have chosen not go forward with trying to license anacetrapib. First, in 2016 the FDA withdrew indications for two other drugs that raised HDL (niacin and fenofibrate), and in that decision, the FDA stated that they no longer believed that increasing HDL was a sufficient biomarker for cardiovascular benefit. Thus, over the course of the CETPi program, the FDA’s position appears to have shifted to be more skeptical of the underlying biological hypothesis here.

Second, given the prior failures and safety risks with another members of the CETPi class, it is possible that the FDA was going to require another large clinical trial before they would license anacetrapib. And indeed, this would seem like a prudent regulatory decision in light of this total body of evidence. One positive result does not automatically outweigh the other largely negative findings in this class. If faced with having to conduct another long and expensive trial (and as always, the possibility that a subsequent trial would not confirm the positive finding), it could be that Merck decided it was not worth the additional investment.

This case also highlights some ethical tensions in research when there are multiple development programs conducted in parallel. To put the question provocatively: How much did we learn with each subsequent trial in this class? The negative torcetrapib result in 2016 is sufficient to show that CETPi’s are not going to be safe and effective as a class of agents. But, of course, this does not mean that there is no safe and effective CETPi. So one failed drug does not mean we will learn nothing (or little more) by continuing to test others.

However, as negative trials begin to add up, reversing the weight of this total body of evidence becomes increasingly unlikely. Again, this does not foreclose the scientific possibility that some member of the class is useful. But it does become an increasing problem for enrolling human subjects in more trials. The risks and burdens of experimenting on patients must be balanced by the benefits to future patients and society. While there may not be a time point in this research program where we can clearly say that we crossed a point of diminishing returns (in terms of the value of information gained and plausibility of future benefits for patients), it is nevertheless important to think about where that point might be—not just from the drug developers' perspective, but from patients' and society's perspective as well.

Read More About This Case