Methods

We first constructed a model of OM3s. There were 2 parts to the modeling: (1) the effect of OM3s on lipids and blood pressure and (2) the “pleiotropic effects,” which we defined to be the effects on CVD outcomes beyond those caused by biomarker changes as determined from the Archimedes model. Such pleiotropic effects may correspond to pathways not present in the model. For OM3s these pathways might involve explicit actions on cardiac and arterial functions, arrhythmia, and plaque stability.

The triglyceride decrease depends on the OM3 dose and a patient’s initial triglyceride value. The model reproduces triglyceride decreases of 18 studies. When triglycerides decrease through OM3 use, the Archimedes lipid model naturally generates changes in total cholesterol and high-density lipoprotein cholesterol of 20 studies. These independent validations provide confidence in the model for analyzing P-OM3s. Results and an analysis for blood pressure are provided in an on-line supplement ( http://www.AJConline.org ).

In building the model of OM3s, we made 4 main assumptions. (1) Because of a lack of outcome studies at high doses, we needed to assume that the effects of P-OM3 on CVD could be ascertained from low-dose studies. (2) Because of a lack of health outcome studies in SHTG populations, we assumed that the effects on CVD outcomes for a population with triglyceride levels ≥500 mg/dl could be determined from studies involving subjects with lower triglyceride levels. (3) We assumed that benefits of OM3s as determined from decreases in myocardial infarctions (MIs) and coronary heart disease (CHD) death extended to all heart-related outcomes including angina and heart failure. Angina is highly correlated with CHD, which is a main cause of MI. Also, previous MI is a risk factor for heart failure. Physiologic support for the assumption comes from evidence that OM3s directly improve cardiac function and arrhythmia. (4) The Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto Miocardico (GISSI) and Japan Eicosapentaenoic Acid Lipid Intervention Study (JELIS) studies have indicated some benefit from OM3 use for secondary prevention of cardiac events. These modest benefits are not inconsistent with the greater benefits seen for primary prevention because participants in the 2 studies were eating much fish. For further discussion, see the on-line supplement ( http://www.AJConline.org ). Therefore it is reasonable to assume that the pleiotropic effects of OM3s for secondary prevention are the same as for primary prevention.

Because there appears to be a dose saturation threshold beyond which there is little additional pleiotropic benefit from OM3 use, assumptions 1 and 2 have the following consequence for our simulations. If a subject is already consuming the equivalent of ≥3 fish meals per week, then that subject is not likely to achieve any pleiotropic CVD benefit from OM3s. If a subject is eating >1 fish meal but <3 fish meals per week, then that subject may achieve a pleiotropic cardiac benefit but not an ischemic stroke benefit. This leads to a fifth assumption, namely that the statistics for OM3 consumption for the general United States population apply to the SHTG population. These statistics indicate that approximately 25% of Americans eat ≥3 fish meals per week and/or take a fish oil supplement and that 8% of Americans eat >1 fish meal but <3 fish meals per week and do not take fish oil supplements.

Hence, in our main simulation 25% of the SHTG population do not achieve a pleiotropic cardiac benefit from P-OM3 use and 33% do not achieve an additional pleiotropic benefit for ischemic stroke and blood pressure improvement. These percentages and those in assumption 5 incorporate those who benefit in part from fish consumption because they consume near the thresholds.

Simulations with the Archimedes model mimicked what happens in an idealized clinical trial. The simulated trial “enrolled” subjects based on 2 entrance criteria: age 45 to 75 years and triglyceride levels 500 to 2,000 mg/dl. Subjects were generated based on adults in the National Health and Nutrition Examination Survey (NHANES) 1999 to 2006 to capture distributions and correlations of biomarkers, risk factors, and disease histories seen in real populations. The same population of 5,000 subjects was used in the 2 trial arms and followed for 20 years or until death. In the treatment arm P-OM3 was given to each subject at a dose of 4 g/day, which contained eicosapentaenoic acid plus docosahexaenoic acid 3.6 g. No subject in the control arm received P-OM3.

During the 20-year simulation subjects had regular doctor visits, went to the hospital emergency room when necessary, took medications, possibly developed adverse outcomes such as strokes and heart attacks, or developed diseases such as diabetes. Health care processes followed standard protocols except for statins, niacin, and fibrates; their use was prohibited. This allowed a clean comparison for use and nonuse of P-OM3s rather than obscuring their effect because of compensating use of statins (which is likely to be higher in the P-OM3 arm) and other triglyceride-lowering agents (which is likely to be higher in the control arm).

Costs for medical processes (visits, hospitalizations, surgeries, etc.) were based on average 2006 national Medicare figures in most cases. Those for prescription drugs were obtained from public sources (e.g., http://www.Drugstore.com ). If a generic version of a drug existed, it and its cost were used. Cost of P-OM3 was based on the cost of the only medication currently approved by the Food and Drug Administration (Lovaza, GlaxoSmithKline, Research Triangle Park, North Carolina) and was $145.69 per month (Wholesale acquisition cost for Lovaza obtained using data from by Wolter Kluwer Health, Inc. October 2009). Quality-adjusted life-years were calculated based on the time subjects spent with different disorders. Estimated disutilities for these disorders originated from Sullivan and Ghushchyan. Details of quality-adjusted life-year calculations are available on request from the authors. Costs and quality-adjusted life-years were discounted at the standard rate of 3%.

We conducted a thorough analysis of the literature (up to September 2009) to determine (1) the action of OM3s on biomedical variables and (2) any additional pleiotropic effect on CVD outcomes. We found that P-OM3 substantially decreases triglycerides for subjects with SHTG. On average we found a modest decrease for total cholesterol, an increase in low-density lipoprotein cholesterol, and small improvements in high-density lipoprotein cholesterol and blood pressure.

OM3s may have effects on CVD beyond those that are generated by biomarker changes. This is evident for those with normal triglyceride levels, for which biomarker changes are small but up to 38% decreases in CVD have been observed. Our analysis of these pleiotropic CVD benefits involved a large number of studies including randomized controlled trials and prospective studies of fish oil and fish consumption. Most of these outcome studies involved doses of OM3s much lower than 4 g/day. We determined relative risk decreases at low doses and extrapolated these to higher doses. Several meta-analyses included plots of decreases versus dose, thus making this extrapolation straightforward.

Studies by Mozaffarian and Rimm, Hibbeln et al, and He et al have indicated a CHD risk that increases significantly as OM3 intake goes to 0, as represented by the dashed curve in Figure 1 , which is derived from several meta-analyses. At high intake the curve asymptotes, suggesting little additional CHD benefit beyond a threshold of about 2.5 fish servings a week or roughly OM3s 300 mg/day. To determine asymptotic relative risk for CHD, we combined results of 4 meta-analyses and found the relative risk factor to be 0.68. After taking into account biomarker changes, the pleiotropic CHD decrease of OM3s became 29%. Those who consumed enough OM3s through their diets to have above dose saturation thresholds did not achieve pleiotropic CVD decreases from OM3s and in our main simulation they did not benefit; however, their CVD rates were still lowered through improved biomarkers.

Schematic of predicted risk versus dose of ω-3 fatty acids shows quantitatively correct doses of >350 mg/day for coronary heart disease and >200 mg/day for ischemic stroke; other doses are qualitative. EPA = eicosapentaenoic acid; HDA = docosahexaenoic acid.

Analysis for ischemic stroke was similar to CHD. For the effect of OM3s on stroke we started with the meta-analysis by He et al and added results of some other studies. The meta-analysis by He et al involved 200,575 participants and 3,491 stroke events and the follow-up period was 4 to 30 years. Relative risks of ischemic stroke remained constant within error bars for 1 serving per week (0.68, 95% confidence interval 0.52 to 0.88), 2 to 4 servings per week (0.66, 95% confidence interval 0.51 to 0.87), and ≥5 servings per week (0.65, 95% confidence interval 0.46 to 0.93). He et al reported a relative risk for hemorrhagic stroke statistically consistent with 1.0.

Risk of stroke versus OM3 dose deduced from He et al was qualitatively similar to the risk for CHD ( Figure 1 ). However, the risk curve is steeper and the dose saturation threshold appears to be lower: there is little additional ischemic stroke benefit beyond an OM3 dose threshold of docosahexaenoic acid plus eicosapentaenoic acid ∼150 mg/dl, which corresponds to roughly 1 fish serving per week. A similar dose–benefit curve was seen in the studies by He et al and Hibbeln et al. We determined the asymptotic benefit by combining the results of He et al with the Cardiovascular Health Study and the European Prospective Investigation into Cancer (EPIC) study. The resulting asymptotic relative risk benefit for ischemic stroke was 0.68. After taking into account effects of other biomarker changes such as blood pressure in decreasing stroke risk, we determined the pleiotropic decrease for ischemic stroke to be 29%.

Some studies including the Finnish Mobile Clinic Health Examination Survey, the Japan Collaborative Cohort Study, GISSI, and JELIS did not observe such a dramatic decrease in CVD outcomes with OM3 use. However, most of these studies involved subjects who already were eating fish regularly. Given the low OM3 threshold of the dose–response curve for ischemic stroke presented in Figure 1 , it is not surprising that these studies did not observe stroke benefits. The pleiotropic CVD decrease of 29% for OM3 use compared to those with little OM3 intake is in the middle of a wide range of what we obtained from studies analyzed individually. However, given the totality of available data, the 29% decrease is accurate at the 95% confidence limit to within 6% for MIs and 8% for ischemic stroke.

Results

Characteristics of the simulated population at baseline are presented in Table 1 . These baseline values reflect averages in NHANES 1999 to 2006 and indicate that those with SHTG who are 45 to 75 years old are more likely to be men, be overweight or obese, smoke, and have diabetes compared to the general United States population. On average they also have very high total cholesterol and low high-density lipoprotein cholesterol and low-density lipoprotein cholesterol. Within the Archimedes model subjects with this type of lipid profile are 1.8 times more likely to have an MI compared to subjects with normal lipid panels. The higher prevalence of diabetes for the SHTG population increases the risk further, approximately doubling it.

Table 1 presents changes after P-OM3 takes effect. After taking P-OM3, triglycerides decreased an average of 41%, and 55% of subjects obtained triglyceride levels <500 mg/dl. Total cholesterol decreased by 12%, whereas high-density lipoprotein cholesterol and low-density lipoprotein cholesterol increased by about 4% and 24%, respectively.

Results presented in Table 1 are consistent with those of another P-OM3 study of subjects with SHTG, although that study involved only 42 subjects, had a different age criterion (18 to 75 years), and employed additional exclusion criteria. With a dosage of P-OM3 4 g/day, triglycerides and total cholesterol decreased by 45% and 15%, respectively, in that study in agreement with our results within statistical uncertainties. The increase in low-density lipoprotein cholesterol was 31%, which is slightly higher than the 24% increase that we observed but decreased within error bars. The increase in low-density lipoprotein cholesterol as triglycerides decrease can be considered part of the normal physiologic process in correcting HTG in which very low-density lipoprotein overproduction is decreased. Because the main purpose of using P-OM3 is to lower triglycerides, we provide its 20-year evolution ( Figure 2 ). The downward drift in triglycerides for the 2 arms is probably caused by 2 effects: (1) triglycerides tend to decrease after 45 years of age in men and stop increasing after 65 years of age in women and (2) those with higher triglycerides are more likely to die, thus leaving subjects with lower values in the simulation at later years. Effect 2 is likely responsible for the steeper decrease in the control arm.

Evolution of triglycerides (milligrams per deciliter) over time.

As presented in Table 1 at baseline many of the SHTG population had diabetes that was undiagnosed in 1/3. The percentage of simulated subjects with diabetes increased in the 2 arms from 42% to 66% by year 12 and remained steady thereafter.

Results for health outcomes at the end of the trial for the 2 arms are presented in Table 2 . Outcomes reported as “per initial subject” provide the cumulative sum of events during the 20-year period divided by the number of subjects (n = 5,000) at baseline; a subject having >1 outcome contributes >1 to the sum. Major adverse cardiovascular event is the composite of nonfatal MI and stroke and CVD death. In the model CVD death is the sum of stroke death and CHD death. The latter is defined as death at the time of MI or subsequently if attributable to the event.

Table 2

Health outcomes at end of 20-year simulation

Outcome	No P-OM3	P-OM3	Relative Decrease
Myocardial infarction (%)
First myocardial infarction (Kaplan–Meier event rate)	33 ± 1.7	22 ± 1.5	31 ± 5
Total myocardial infarctions	42 ± 2.8	27 ± 2.3	36 ± 5
Stroke (%)
First stroke (Kaplan–Meier event rate)	15 ± 1.3	12 ± 1.2	19 ± 8
Total strokes	14 ± 2.0	12 ± 1.8	19 ± 13
Ischemic strokes	12 ± 1.2	9.0 ± 1.1	24 ± 9
Hemorrhagic strokes	2.5 ± 0.6	2.6 ± 0.6	−3.2 ± 24
Death (%)
Coronary heart disease deaths	16 ± 1.4	11 ± 1.2	29 ± 7
Stroke deaths	7.3 ± 1.0	6.4 ± 0.9	13 ± 12
Cardiovascular disease deaths	23 ± 1.6	18 ± 1.4	24 ± 6
Deaths	46 ± 1.8	42 ± 1.8	10 ± 4
Major adverse cardiovascular event (%)
Cumulative subjects with major adverse cardiovascular events	43 ± 1.8	32 ± 1.7	24 ± 4

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. ST-Segment Deviation in Lead aVR on Admission Is Not Associated With Left Ventricular Function at Predischarge in First Anterior Wall ST-Segment Elevation Acute Myocardial Infarction
2. Platelet Response to Aspirin 50 and 100 mg in Patients With Coronary Heart Disease Over a Five-Year Period
3. Eicosapentaenoic Acid Ethyl Ester (AMR101) Therapy in Patients With Very High Triglyceride Levels (from the Multi-center, plAcebo-controlled, Randomized, double-blINd, 12-week study with an open-label Extension [MARINE ] Trial)
4. Prediction of Response to Cardiac Resynchronization Therapy Combining Two Different Three-Dimensional Analyses of Left Ventricular Dyssynchrony

Stay updated, free articles. Join our Telegram channel

Join