Summoning STRENGTH to Question the Placebo in REDUCE-IT

John A. Bostrom, MD; Joshua A. Beckman, MD, MS; Jeffrey S. Berger, MD


Circulation. 2021;144(6):407-409. 

The initial use of placebo medications in the late 18th century was not in the context of clinical trials. Placebos—or remedies given without curative intent—were often used by physicians aiming to satisfy a patient's demand for an evaluation and treatment. The use of a placebo as a comparator with an established treatment can be traced to John Haygarth. Skeptical of claims that widely popular "metallic tractors"—a pair of rods made of copper, zinc, and gold—could cure ailments ranging from gout to boils using electricity and magnetism, he designed what is regarded as one of the first placebo-controlled trials. Fashioning sham tractors out of wood, he treated 5 patients with either metallic or wood tractors. Equal results were obtained by both the true and false tractors; after the results were disseminated, metallic tractors soon fell out of favor.

Randomized, placebo-controlled trials are now considered the gold standard in clinical trial evidence. Ideally, placebos are indistinguishable from the active treatment and are inert substances. It is imperative that a placebo does not cause harm; interventions that worsen risk factors have a greater adverse effect than, conversely, the benefit achieved by improving risk factors by the same magnitude.

The data surrounding the utility of omega-3 fatty acid supplementation for the use of cardiovascular disease (CVD) are mixed, and aggregate results suggest no significant benefit. An outlier is REDUCE-IT (Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial), which randomized 8179 participants with diabetes (58%) or established CVD (71%) with elevated triglycerides (median baseline 216 mg/dL) despite background statin therapy to icosapent ethyl (IPE) versus placebo composed of mineral oil. The results were striking: Over a mean follow-up of 4.9 years, the primary end point composite of cardiovascular death, nonfatal myocardial infarction, nonfatal stroke, revascularization, or hospitalization for unstable angina occurred in 17.2% of the treatment group versus 22.0% of the placebo group.[1] Treatment with IPE led to a significant 20% reduction in cardiovascular mortality. On the basis of the results of REDUCE-IT, the Food and Drug Administration approved a label claim for CVD event reduction (for purposes of primary or secondary prevention) for IPE in high-risk patients with elevated triglycerides and diabetes.

After the significant reduction in cardiovascular events observed in REDUCE-IT, there was anticipation for the completion of the STRENGTH trial (Statin Residual Risk With Epanova in High Cardiovascular Risk Patients With Hypertriglyceridemia). STRENGTH was designed to evaluate another high-dose omega-3 fatty acid supplement, omega-3 carboxylic acid, versus placebo composed of corn oil. Among 13 708 participants at high risk for CVD (56% with established CVD), the trial was terminated early because of futility. The same primary end point used in REDUCE-IT occurred in 12.0% of the treatment group versus 12.2% of the placebo group.[2]

The difference in results between REDUCE-IT and STRENGTH could be a result of the different treatments (pure eicosapentaenoic acid [EPA] versus a combination of EPA and docosahexaenoic acid). However, a closer look offers a perplexing picture. The change in biomarkers of cardiovascular risk, including lipids and inflammation, was similar between the active treatment of both trials (Figure). Although a higher dose of pure EPA was used in REDUCE-IT compared with STRENGTH, the serum levels of EPA in STRENGTH participants rose dramatically in both STRENGTH (+269%) and REDUCE-IT (+394%). In a subgroup analysis of STRENGTH, participants with the highest increase in EPA levels (≥435% change—higher than what was achieved in REDUCE-IT) had no reduction in the primary composite outcome. In contrast with the consistent change in biomarkers among the treatment arms in both studies, significant differences were seen in the placebo arms. In REDUCE-IT, mineral oil was associated with a 10.2% increase in low-density lipoprotein cholesterol (LDL-C), 7.8% increase in apolipoprotein B, and 32% increase in high-sensitivity C-reactive protein. In contrast, the corn oil placebo used in STRENGTH did not increase any of these biomarkers (Figure). The mechanisms underlying the changes in lipid profiles in the mineral oil arm of REDUCE-IT are unclear, although indirect evidence suggests that mineral oil may block the proper absorption of background statin therapy.[3]


Percent change from baseline of select biomarkers between placebo arms and treatment arms of STRENGTH and REDUCE-IT trials.
Apo-B indicates apolipoprotein-B; CA, carboxylic acid; EPA, eicosapentaenoic acid; hsCRP, high-sensitivity C-reactive protein; LDL-C, low-density lipoprotein cholesterol; REDUCE-IT, Reduction of Cardiovascular Events With Icosapent Ethyl–Intervention Trial; and STRENGTH, Statin Residual Risk With Epanova in High Cardiovascular Risk Patients With Hypertriglyceridemia.

Discrepancies between these placebo arms remind us of the importance of an inert placebo. Although mineral oil has been used as a placebo in as many as 80 studies, a recent comprehensive review highlighted that its effects on lipid levels and inflammatory markers have been mixed.[4] The effect of mineral oil placebo has noted a wide range of effect on LDL-C (from –3.0% to +10.5%) and high-sensitivity C-reactive protein (from –15.9% to +33.3%). The majority of studies (10 of 16) reporting high-sensitivity C-reactive protein in mineral oil arms note an absolute increase in high-sensitivity C-reactive protein levels—including a 32% increase in REDUCE-IT and a 17% increase in ANCHOR.[3] It should be noted that the absolute difference in LDL-C between the placebo and IPE arms in REDUCE-IT (≈9 mg/dL) would only be estimated to account for a 3.0% relative risk increase in the primary outcome in the placebo group, suggesting that the observed differences in outcomes between arms were driven by an LDL-C–independent mechanism.[3] This estimate also supposes a mirror image of harm by raising LDL-C to the benefit of lowering it. This supposition is not supported in the literature: For example, a 5-mm Hg reduction in blood pressure in hypertension is associated with a 2% relative risk reduction in mortality. In contrast, a medication-mediated increase of 5 mm Hg was associated with a 58% relative risk increase in mortality and a significant increase in cardiovascular events.[5] Thus, we question the Food and Drug Administration estimate of harm in this case.

In light of the discrepancies between large, well-run trials with similar treatments in similar high-risk populations, it may be difficult for physicians and patients to confidently assign a benefit to omega-3 fatty acid supplementation. The differences between the placebo arms in REDUCE-IT and STRENGTH are so striking that a separate trial (ideally comparing IPE with a demonstrated bland placebo) would be helpful to provide a clear answer about the value of omega-3 fatty acid supplements for primary or secondary prevention of CVD. We believe a new trial is necessary for a definitive answer.