IMPROVE-IT Substudy: Ezetimibe Benefit Restricted To Diabetics

The beneficial effects of ezetimibe are found almost exclusively in  patients with diabetes, according to an update of the influential IMPROVE-IT trial presented on Sunday at the European Society of Cardiology meeting in London.

The new finding may lead to questions about the widely accepted interpretation of the main finding of the trial, which is that it provided strong support for the “lower is better” LDL hypothesis. The presenter of the update, the TIMI group’s Robert Giugliano, noted the “caveats of subgroups” but pointed out that this was a “large prespecified subgroup” that was larger than the entire population of the PROVE IT and 4S studies.

The main IMPROVE-IT trial randomized  18,144 patients with acute coronary syndrome to the combination of ezetimibe and simvastatin or simvastatin alone. The primary endpoint of the trial– CV death, MI, documented unstable angina requiring rehospitalization, coronary revascularization after one month of treatment, or stroke– was significantly reduced in the ezetimibe group (32.7% vs. 34.7%, p = 0.016).

The ESC presentation focused on the 27% of patients in the trial (n= 4,933) who had diabetes. “We did see a quantitative, statistically significant interaction between diabetics and non diabetics,” said Giugliano.

Primary Endpoint (P value for interaction = 0.023)

  • ezetimibe vs. placebo in diabetics: 40.0% vs. 45.5%; HR 0.86,  CI 0.78-0.94
  • ezetimibe vs. placebo in non-diabetics: 30.2% vs. 30.8%, HR 0.98, CI 0.91-1.04

Myocardial Infarction (P value for interaction = 0.028)

  • diabetes: 16.4% vs. 20.8%
  • non-diabetes: 12.0% vs. 12.7%

Stroke (P value for interaction = 0.031)

  • diabetes: 3.8% vs. 6.5%
  • non-diabetes: 3.2% vs. 3.4%

There were no significant differences in the safety profile based on diabetes status, Giugliano reported.

As expected the diabetes patients differed from the non-diabetics in some important respects. Diabetes patients were older, more likely to be female, had a higher BMI, had more history of CV disease, and were more likely to have been been previously treated with statins. But they were less likely to be a smoker or to have had a STEMI. At baseline diabetes patients had slightly lower LDL and HDL levels but higher triglyceride and hsCRP levels. In response to treatment diabetes patients also had a significantly greater reduction in LDL and hsCRP levels.

Giugliano concluded that diabetic patients were at higher risk than non-diabetic patients and that they had a greater relative and absolute benefit from the addition of ezetimibe. The greater benefit in diabetics was driven by reductions of MI and ischemic stroke.

A number of experts outside the trial offered comments.

Sanjay Kaul wrote:

“The clinically relevant and statistically significant findings in the diabetic subset help optimize the use of ezetimibe in clinical practice. Given these results, it would be difficult to support its use in non-diabetics, nearly 3/4ths of the overall cohort in IMPROVE-IT.”

Harlan Krumholz wrote:

“The finding raises the possibility that the drug’s benefit is concentrated among diabetics, which prompts questions about its utility for non-diabetics. This could have occurred by chance, but it adds some uncertainty to who benefits from adding this drug to statin therapy after ACS. Unfortunately this is the only large outcomes trial of ezetimibe therapy and there are so many important questions that still abound about the role of this drug in our armamentarium – and we will have to live with that uncertainty since there is unlikely to be another trial anytime soon.”

Mary Walsh cautioned that “we should not extrapolate these findings to every stable diabetic patient with coronary disease because the patients studied in IMPROVE-IT had had a recent coronary event.”

Walsh and another expert, Robert Eckel, rejected the interpretation of the trial suggesting that ezetimibe might not be beneficial in the non-diabetes population. Eckel said that “the takeaway message is that lower LDL with a nonstatin agent has proven to be beneficial, the subgroup analysis simply raises questions about why there is a preferential benefit in diabetes.” But, he acknowledged, “these data make you pause a little bit about people without diabetes.”

The results of the study may also have implications beyond the use of ezetimibe in clinical practice. It is entirely possible that the data might be used to lend additional support to the use of the new PCSK9 inhibitors in diabetic patients, but it may also be used to support more restrictive use in nondiabetics.


Richard Becker raised an important point about the interpretation of these findings:

The reduction in the composite primary endpoint as well as its independent components is not only interesting but potentially relevant. If the LDL reduction in the diabetes patients is consistent with the findings from decades of clinical trials in support of the LDL cholesterol hypothesis then we may have our answer. If not it may be that ezetimibe has a unique property in individuals with diabetes that requires much deeper investigation.”

Related story: No Increase In Diabetes Found With Ezetimibe In IMPROVE-IT




  1. James Stein says

    I am underwhelmed by thus subgroup analysis. IMHO, the benefit with ezetimibe has nothing to do with diabetes, per se, just that diabetic patients are at higher risk. The intervention is weak so you see the benefit primarily in those at higher risk – the diabetic patients. All point estimates are in the left side of unity but the confidence intervals achieve significance only for diabetic patients. This is just like the effects of high dose statin therapy in TNT, with large effect sizes and smaller NNTs for those with diabetes and/or Metabolic Syndrome and much larger ones for those without. By analogy, I am sure we could find other high risk subgroups that show the same thing too, for example older smokers previously in statins with higher LDL and perhaps impaired fasting glucose. In the end it is about the risk patients start with and the amount of lipid lowering achieved with the drug (balanced by safety of course).

Speak Your Mind