A reason for the initial dismissal of TGs as a potential marker for CVD was the finding that some subgroups of patients with familial genetic dyslipidemias (mainly Fredrickson-Levy classification type I and V), and severely elevated TGs (>1000 mg/dL), did not seem especially prone to developing atherosclerosis but rather were at high risk for chylomycronemia and pancreatitis [7]. Further research has shown that the genetic dyslipidemias with TGs in the 200–800 mg/dL range accompanied by the presence of TG rich lipoproteins (Fredrickson-Levy classification type IIb, III and IV) were inversely associated with premature atherosclerosis and CVD [8]. This paradox was accounted for by the thought that when TGs are >1000 mg/dL, remnants are too large to incorporate into the arterial intima and participate in the development of atherosclerosis.

In 2008, the INTERHEART study investigated apolipoproteins and conventional lipids as indices of risk for a first acute myocardial infarction (MI) [9]. This was an international, multi ethnic case control study with 12,461 cases and 14,637 age and sex matched controls in 52 countries. The investigators found that a high ApoB: ApoAI ratio had the highest population attributable risk for acute MI at 54 % which was statistically significant compared to conventional lipid ratios (p < 0.0001). Given the distribution of ApoB across all non-HDL lipoproteins, these results highlight the notion that lipid atherogenicity goes beyond LDL cholesterol, to include the variety of TRL’s and their downstream remnants: VLDL, IDL and chylomicron remnants.

The Emerging Risk Factors Collaboration released a 2009 meta-analysis (with data from 68 long term prospective studies analyzing 302,430 individuals) showing that raised TG levels regardless of fasting status were associated with an increased risk of Coronary Heart Disease (CHD) after adjustment for non-lipid risk factors [10]. However, this association was completely attenuated after adjustment for High Density Lipoprotein Cholesterol (HDL-C) and non-HDL-C (HR 0.99, 95 % CI, 0.94–1.05), suggesting that there was no added benefit of using TGs for CHD risk prediction beyond standard lipoprotein measurements. However, the latest guidelines recommend that risk prediction focuses on the combined endpoint of CHD and stroke, not just CHD. In addition, the usefulness by sex and in different ethnicities should also be considered.

Drawing on the previously mentioned role of small, dense LDL, a population based cohort study using the Multi-Ethnic Study of Atherosclerosis (MESA) database examined the discordance between LDL-C and LDL Particle concentration (LDL-P) [11]. This study examined 6814 subjects free of clinical CVD at baseline and followed up participants for 5.5 years during which time 319 incident CVD events occurred. The study showed that for patients with discordant LDL-C and LDL-P levels, only LDL-P was associated with incident CVD. Further, in patients with LDL-P greater than LDL-C, there was an increased incidence of hypertriglyceridemia and insulin resistance, with 54 % of those patients meeting criteria for metabolic syndrome. Finally, this study also showed that carotid intima-media thickness, a marker for subclinical atherosclerosis had a stronger correlation with LDL-P levels than LDL-C.

A subsequent Mendelian randomization study investigated the notion that if TG levels remained genetically low lifelong, there would be an expected decrease in all-cause mortality [12]. The study showed that genetically reduced nonfasting plasma TG levels due to allelic variants in LPL gene loci, resulted in lower levels of plasma TGs, inversely proportional to the number of alleles present (more alleles corresponded to lower TG levels and survival benefit). Further, the study did indeed find that genetically low concentrations of nonfasting TGs are associated with reduced all-cause mortality. The authors felt that TGs per se were not necessarily causal, but rather, RLPs were the more likely causal factor, based on other similar studies by the same group [13, 14].

The aforementioned select studies, with clear discordance in establishing a relationship between CVD and hypertriglyceridemia, highlight the extent of the controversy in the field. Despite this discordance, several conclusions can potentially be extrapolated: (i) there is some evidence for an association between hypertriglyceridemia and atherosclerosis in epidemiologic studies, (ii) Mendelian randomization studies do indeed provide relatively concrete evidence to establish causality, and (iii) elevated TG clearly alters the atherogenicity of lipoproteins, particularly remnant particles and small, dense LDL.

One of the earlier trials to examine TG lowering and CVD was the Helsinki Heart Study (HHS). This was a primary prevention trial examining 4081 dyslipidemic men without CVD who were randomized to gemfibrozil vs placebo [15]. While there was a 34 % reduction in CVD incidence in the gemfibrozil group, there was no difference in all cause mortality between the two groups. Further, despite reduction in TG levels in HHS patients, this was not associated with the observed reduction in CVD events.

Another study examining fibrate therapy was the Fenofibrate Intervention and Event Lowering in Diabetes trial (FIELD). Patients with Type 2 Diabetes, regardless of their lipid profile, were randomized to fenofibrate vs placebo [16]. While there were significant reductions in non-fatal MI, microvascular disease, total CVD events, coronary and non-coronary revascularization in the fibrate group, unfortunately – likely secondary to higher crossover to statins in the placebo treated patients – FIELD failed to show a difference in all-cause mortality. In fact, there was a trend toward increased mortality, pulmonary embolus and pancreatitis in the treated patients. Notably, post-hoc analysis of FIELD did show that fenofibrate was most beneficial in patients with metabolic syndrome and marked dyslipidemia (high TG, low HDL).

The most recent fibrate based trial was the Action to Control Cardiovascular Risk in Diabetes (ACCORD). This study was designed to investigate whether fibrate therapy provided mortality benefit in addition to statin therapy in patients with type 2 diabetes [17]. In this study, 10,251 patients were randomized to fenofibrate plus simvastatin vs placebo plus simvastatin. While there was significant, albeit modest, lowering in TG levels (186–170 mg/dL), there was no primary outcome reduction in the study group (fatal CVD, non fatal MI or non fatal stroke). However, further subgroup analysis showed that in patients with marked dyslipidemia (TG > 204 mg/dL and HDL < 34 mg/dL), there was a trend towards benefit.

Another trial investigating combination therapy was the Japan Eicosapentaenoic Acid Lipid Intervention Study (JELIS). Here, patients were randomized to pravastatin or simvastatin in addition to the fish oil eicosapentaenoic acid vs. a statin alone [18]. In follow-up, patients in the study group had about a 20 % reduction in primary and recurrent CVD events but no difference was seen in sudden cardiac death or overall mortality. Notably, here as well, the improved effects were likely not related to TG lowering over and above statins.

A landmark secondary prevention trial was the Veterans Affairs High-Density Lipoprotein Intervention Trial (VA-HIT). In this study, men with known CVD and HDL-C <40 mg/dL, LDL-C <140 mg/dL and TG <300 mg/dL were randomized to gemfibrozil vs placebo [19]. With mean follow up of about 5 years, patients who were treated with fibrate therapy had a significant 22 % reduction in the primary endpoint of nonfatal MI or death due to CVD. While there was a 31 % reduction in mean TG levels in the treated group, this was not associated with the reduction in primary endpoints. Initial statistical analysis pointed towards increased HDL levels as what may have accounted for the positive result given that LDL-C levels in both treatment groups were equivalent and not changed with gemfibrozil. However, further analysis with Nuclear Magnetic Resonance spectroscopy did show that treatment with gemfibrozil increased LDL size and decreased LDL-P, likely explaining the observed reduction in endpoints with therapy concomitant with TG lowering [20]. As with the primary prevention trials noted above, there was a further significant reduction in overall death and CVD-related death in patients with diabetes at baseline.

Another secondary prevention trial of fibrate therapy was Bezafibrate Infarction Prevention (BIP). Here 3090 patients with known prior MI and dyslipidemia were randomized to bezafibrate vs placebo. While there was no difference in the primary endpoint of fatal, nonfatal MI and sudden death in either group, patients with TG >200 mg/dL at baseline had a significant reduction in the primary endpoint [21].

While the PROVE-IT TIMI 22 trial successfully established that reduction in LDL-C to <70 mg/dL was superior to previous more conservative cutoffs in patients with Acute Coronary Syndromes (ACS), a subgroup analysis explored the impact of on-treatment TG levels on CHD. There was a reduced risk of the incidence of death, MI and recurrent ACS associated with low on-treatment TG levels (<150 mg/dL) even after adjustment for other covariates [22]. While only a post-hoc analysis, these results are consistent with evidence that hypertriglyceridemia may be associated with hypercoagulability, worsened in post-MI states.

Lastly, the most recent trial to examine the effects of niacin in secondary prevention was the Atherothrombosis Intervention in Metabolic Syndrome With Low HDL/High Triglycerides and Impact on Global Health Outcomes (AIM-HIGH) study [23]. Here, 3414 patients with known CVD and low HDL-C were randomized to high dose niacin versus placebo, with both arms receiving varying doses of simvastatin and ezetimibe to maintain LDL-C of 40–80 mg/dL. The study was halted early given no significant differences in primary outcomes. Further post-hoc analysis of a pre-specified subgroup with HDL-C <32 mg/dL and TG of >200 mg/dL showed that niacin decreased the primary endpoint by 37 % [24].