Results

Adverse events were reported by 22 patients receiving Omacor and 17 receiving placebo. These events were mostly minor except for one patient in the placebo group who died from acute myocardial infarction, for another patient in the placebo group who withdrew because of severe heartburn, and for one patient in the active treatment group who was withdrawn because of loose motions. Another patient receiving placebo withdrew because visits proved inconvenient. Fifty five patients thus completed the double blind, placebo controlled 24 week phase of the trial. Of these 46 agreed to take active Omacor for a further 24 weeks. The five patients who had earlier received placebo and the four who had received active treatment in the double blind phase, who declined to enter the open phase, did so because of the inconvenience of frequent hospital attendances. The results presented are for all evaluable patients at 0, 12, 24, and 48 weeks. Exclusion of all patients other than the 46 present at each of these time points does not alter the conclusions. Compliance, assessed by counting returned capsules, was > 80% in all but three patients. Patients were included in the statistical analysis regardless of compliance (intention to treat analysis). Serum phospholipid eicosapentaenoate and docosahexaenoate concentrations were similar in the placebo and Omacor groups at baseline, but in patients receiving Omacor the eicosapentaenoate concentrations rose by a mean of 260% (95% confidence interval (CI) 192% to 327%) at six months and by 361% (95% CI 276% to 447%) at 12 months, and the docosahexaenoate concentrations by 54% (95% CI 38% to 70%) and 52% (95% CI 35% to 69%) at the same time points. There were no significant changes in these parameters in patients on placebo at six months (eicosapentaenoate 12% (95% CI −33% to 57%) and docosahexaenoate –6.1% (95% CI −13% to 1%)).

The patients allocated to active and placebo treatment were similar in most respects at baseline (table 1). By chance, however, the mean (SD) serum triglyceride at randomisation was 4.6 (2.2) mmol/l in those allocated to the Omacor group and 3.8 (2.2) mmol/l in those allocated to placebo. The primary end points of the trial (percentage changes in serum triglycerides and VLDL cholesterol) in the patients on active treatment were, however, significantly greater by the per protocol analysis of covariance than in those on placebo (fig 1), even after adjusting for any effect of the baseline values (p < 0.005). The secondary end points (concentrations of serum cholesterol, triglycerides, VLDL, and LDL cholesterol) are summarised in table 2. Serum triglyceride and VLDL cholesterol concentrations decreased significantly compared to baseline on Omacor (p < 0.0005 and p < 0.005, respectively). At no stage was there any tendency for LDL cholesterol to rise or for HDL cholesterol to decrease.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Change in serum triglycerides and VLDL cholesterol concentrations as a percentage of baseline values during treatment with omega-3 polyunsaturated fatty acid concentrate at a dose of 2 g twice a day for 48 weeks (solid circles) or with placebo for 24 weeks followed by active omega-3 polyunsaturated fatty acid concentrate from 24 to 48 weeks (open circles) in patients with coronary heart disease and serum triglycerides exceeding 2.3 mmol/l despite treatment with simvastatin (n = 59 at 0 weeks, 57 at 12 weeks, 55 at 24 weeks, and 46 at 24 weeks). The trial was double blind from 0–24 weeks and open from 24–48 weeks. Values are mean (SEM). *Significant with baseline concentration as covariate (p < 0.005).

Changes in lipoprotein concentrations were unrelated to the dose of simvastatin, the mean decrease in serum triglycerides being 25% in those on 10–20 mg and 27% in those on 40 mg daily. The triglyceride, VLDL cholesterol, and serum cholesterol responses of diabetic patients to Omacor were also not significantly different from non-diabetics. Numerically they were greater, the decrease in triglyceride, for example, being 35% in the diabetic patients at the end of the double blind phase and 44% at the end of the trial. No significant change in fasting blood glucose or HbA_1coccurred in either the diabetic or non-diabetic patients receiving Omacor for 48 weeks. This was also true when the combined groups and the 15 diabetic patients receiving Omacor for 24 weeks were analysed. There were no apparent differences between placebo and Omacor in the frequency of laboratory values outside the reference range (full data available on request). No significant effects were observed on apo AI or B, Lp(a) lipoprotein, fibrinogen or blood pressure.

Discussion

Omacor 2 g twice a day for 48 weeks decreased serum triglycerides by 20–30% and VLDL cholesterol by 30–40% in patients with combined hyperlipidaemia and CHD already receiving simvastatin 10–40 mg daily. The patients were similar to those in the 4S trial in which the statin produced a 30% decrease in mortality over five years.1 In that trial, however, patients with serum triglycerides > 2.5 mmol/l before the start of treatment were excluded. In two later statin trials2 ,3 patients with higher triglyceride concentrations participated. In one of these, patients with higher than average triglyceride values showed no significant reduction in CHD incidents with statin treatment.3 No trial has yet attempted to discover whether there might be additional benefit from treatment to lower triglycerides in patients already receiving a statin. Fish oil in patients with hypertriglyceridaemia has been previously reported to raise LDL cholesterol.12 ,13 In the present trial, in which patients were receiving simvastatin, there was no adverse effect of Omacor on LDL cholesterol, which if anything decreased further. Fish oil treatment has also been associated with a deterioration in diabetic glycaemic control in some studies,14-18 but not in others19-22 and, as in non-diabetics, may increase LDL cholesterol.18 ,19In the present trial, glycaemic control and LDL cholesterol were not adversely affected by Omacor in eight diabetic patients during one year of treatment or in 15 patients treated with Omacor for 24 weeks.

Increased fish oil intake achieved by either increased consumption of omega-3 fats8 or direct supplementation10 can prolong survival following myocardial infarction. There has been speculation about whether the effect might be caused by diminished serum triglycerides, decreased platelet aggregation or suppression of ventricular dysrhythmias.13 Serum triglyceride concentrations do appear to be a major determinant of survival in patients with coronary heart disease even after revascularisation.23 In GISSI, however, the dose of omega-3 fatty acids was equivalent to Omacor 1 g daily,10which is one quarter of the dose used in the present study. Its effect on serum triglycerides, although significant, was small (−3.4% with respect to baseline and −4.8% compared to placebo). The patients in the GISSI trial were not selected for hypertriglyceridaemia and their average serum triglyceride value was 1.8 mmol/l. Almost half also received statin treatment. It was uncertain whether the very small decrease in serum triglycerides in GISSI was caused by the low dose of omega-3 fatty acids employed, an erosion of their effect on serum triglycerides due to the co-administration of statin treatment, or a lack of sustained effect. In the present study the dose of Omacor used (2 g twice daily) was chosen because it had previously been shown to lower serum triglyceride concentrations substantially in patients with hypertriglyceridaemia.11 The sustained decrease in serum triglycerides of 20–30% despite statin treatment, which we now report, strongly suggests that the smaller effect in the GISSI study was caused by the lower dose of omega-3 fatty acids used. In that trial the decrease in coronary mortality was attributed to the anti-dysrhythmic properties of omega-3 fatty acids. The question thus remains unanswered as to whether larger doses of omega-3 fatty acids with greater triglyceride lowering potential would be even more effective in reducing clinical end points, perhaps also affecting those related to atheroma progression.

Whatever the explanation for the decreased all cause mortality with fish oil, evidence for improved survival is lacking in the case of the more commonly prescribed class of triglyceride lowering agents, the fibrates,24 although in the most recent trial of gemfibrozil in myocardial infarction a 31% decrease in serum triglycerides was associated with a non-significant decrease in mortality of 11%.25 This result is, however, of limited relevance to the potential benefits of treating patients with hypertriglyceridaemia because the mean triglyceride concentration in the participants was only 1.6 mmol/l. Furthermore the combination of fibrate with statin treatment has the potential to cause myositis, and fibrates, unlike the statins and fish oil, are contraindicated in renal impairment.24 Statins themselves have a triglyceride lowering effect, which, although variable, tends to be related to the degree of LDL reduction achieved.26 Many patients with combined hyperlipidaemia with relatively greater increases in triglycerides than cholesterol thus have persistent hypertriglyceridaemia even after the introduction of statin treatment. The present study shows that Omacor can, under these circumstances and in the doses we used, safely decrease serum triglycerides by a further 20–30% and that it is well tolerated, which has not been our previous experience with whole fish oil in doses required to achieve similar triglyceride lowering. The decision was made in designing this trial not to study patients on a standardised dose of simvastatin, but on the dose to which the patient had been titrated in the clinic. This design has the advantage that the method of arriving at the dose of simvastatin mirrors clinical practice and also that it allows us to conclude that the triglyceride response to Omacor is not influenced by the dose of simvastatin.

Given the strength of the evidence for the benefit of statin treatment in myocardial infarction survivors,1-3 it is probably unethical to consider further trials of statin versus placebo in patients with established CHD. However, the present trial shows that, with minimal safety monitoring, it would be possible to conduct a trial in patients already receiving statin treatment to establish whether omega-3 PUFA in the form of Omacor in the doses used in this study will confer additional benefit over the lower doses used in GISSI.