Residual Hypertriglyceridemia and Estimated Atherosclerotic Cardiovascular Disease Risk by Statin Use in U.S. Adults With Diabetes: National Health and Nutrition Examination Survey 2007–2014

Elevated blood triglycerides (TGs) are a common lipid disorder with borderline hypertriglyceridemia (HTG) between 150 and 199 mg/dL and HTG if defined as TG of at least 200 mg/dL. Approximately 25% of the U.S. population has TG levels of 150 mg/dL or greater (1,2). Patients with diabetes are more likely to have elevated TGs because of both hepatic overproduction of VLDL cholesterol and reduced TG lipolysis due to decreased lipoprotein lipase activity (3). In the setting of insulin resistance, more free fatty acids reach the liver; one means of disposing of this fatty acid excess is to increase hepatic TG and VLDL production and secretion (4).

Epidemiological data on the prevalence of borderline or elevated TGs among patients with diabetes are limited, but earlier reports have indicated that such levels are prevalent in more than half of the U.S. population with diabetes (5). HTG is significantly associated with the risk of future cardiovascular disease (CVD) and is common in people with diabetes (6–8), even among those with well-controlled LDL cholesterol (LDL-C) levels on statin therapy (8).

However, while statins do reduce CVD risk in people with diabetes, substantial residual risk for CVD events remains (8,9), and few trials of TG-lowering agents have shown benefit. Previously, the VA HDL Intervention Trial (VA-HIT) showed people with diabetes receiving gemfibrozil had a 32% risk reduction in the composite end point of coronary heart disease (CHD) death, stroke, or myocardial infarction (MI) (10). Among patients with low LDL-C, elevations in plasma TGs increase CVD risk, and the greatest reduction in CVD risk occurs when both LDL and TGs are reduced (11). Although statins are currently the first-line treatment for patients with diabetes based on the most recent dyslipidemia guidelines (12), a substantial prevalence of elevated TG remains in patients with diabetes along with associated atherosclerotic CVD (ASCVD) risks despite statin treatment. Such information is important for understanding the role that newer therapies may play in reducing such risks.

The purpose of the current study is to estimate the proportion of people with borderline HTG and HTG among adults with diabetes in the U.S. with or without statin treatment, and further estimate their residual ASCVD risk in the following decade.

We included U.S. adults from the National Health and Nutrition Examination Survey (NHANES) 2007–2014. The methodology of NHANES data collection has been described elsewhere (13). Eligible participants for our analysis included adults aged 20 years and over, randomly assigned to a morning session that asked to be fasting for at least 8.5 hours, with available TG data. Participants were defined as having diabetes if one or more of the following criteria were met: 1) fasting glucose ≥126 mg/dL; 2) nonfasting glucose ≥200 mg/dL; 3) taking medication to lower blood glucose; 4) taking insulin; or 5) HbA_1c ≥6.5% (48 mmol/mol). A total of 1,448 adults aged 20 years and over with diabetes (projected to 24.4 million U.S. population) were included in this study. Of those, 1,390 adults (projected to 23.4 million) had available LDL-C data and 889 adults (projected to 14.7 million) had complete information to calculate ASCVD risk score (described below).

Examination and laboratory test results, medical history, and prescription medication information were used from the NHANES data. Obesity was defined as a BMI ≥30 kg/m². Hypertension was defined (based on the accepted definition at the time of the survey years) as a systolic blood pressure ≥140 mmHg and/or diastolic blood pressure ≥90 mmHg or currently taking prescription medication for hypertension. Total cholesterol and TGs were assayed using enzymatic reactions on a Roche/Hitachi Modular P Chemistry Analyzer, while HDL-C was analyzed through a modified traditional multistep precipitation reaction. LDL-C was computed using the Friedewald equation (defined as total cholesterol minus HDL-C and TG/5 in mg/dL) (13). Chronic kidney disease was determined by estimated glomerular filtration rate <60 mL/min/1.73 m² using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation (14). Information regarding smoking status, history of heart failure, CHD, stroke, and family history of heart attack were based on self-report. TG levels were classified as normal (<150 mg/dL), borderline high TG (TG 150–199 mg/dL), and HTG (TG ≥200 mg/dL), which is consistent with recent recommendations and statements (15–17).

We first calculated the weighted proportion and number of patients with diabetes within each TG category (<150, 150–199, and ≥200 mg/dL) with and without statin use in NHANES 2007–2014 and further stratified by sex and ethnicity. The χ² test was applied to determine whether the distributions were significantly different between groups. We also compared the distribution of demographic characteristics and proportion of cardiovascular risk factors in those with normal TG (<150 mg/dL), borderline HTG (TG 150–199 mg/dL), and HTG (TG ≥200 mg/dL), including age, sex, ethnicity, obesity, smoking status, hypertension, history of CHD, heart failure, stroke, chronic kidney disease, and family history of heart attack. In the overall sample of patients with diabetes, the proportion of adults with borderline HTG and HTG according to these cut points by LDL-C category (<70, 70–99, 100–129, 130–159, and ≥160 mg/dL) was also evaluated according to statin use. In addition, of interest is the prevalence of atherogenic dyslipidemia indicated by a TG ≥200 and HDL-C <35 mg/dL, which we had also evaluated according to statin use.

Next, the American College of Cardiology/American Heart Association pooled cohort 10-year ASCVD risk score (%) was determined for participants according to TG categories and was reported as the proportion of those with scores <5%, 5–7.4%, 7.5–19.9%, and ≥20.0%, respectively. The variables used for ASCVD risk score calculation included age, sex, race, systolic blood pressure, total cholesterol, HDL-C, diabetes status, smoking status, and antihypertensive treatment. The risk score was only calculated among those without prior ASCVD (CHD, myocardial infarction, stroke, and angina) and aged between 40 and 79 years old. The same procedure was then repeated among statin users.

Lastly, multiple logistic regression analyses with stepwise selection were performed to identify the independent predictors of TG ≥150 mg/dL among our statin-treated patients with diabetes. Initially, we included age, sex, ethnicity, BMI, HDL-C, LDL-C, cigarette smoking, hypertension, heart failure, CHD, stroke, and family history of heart attack as potential risk factors, and those who satisfied α-to-enter and α-to-stay levels of 0.15 (P values to allow entry and retaining of variables, respectively) were included in our final models through stepwise selection procedures (age, sex, and ethnicity were forced into final models).

All analyses applied an NHANES 8-year sample weighting procedure to project to the U.S. population in millions. SAS version 9.3 was used for data analysis.

In our sample of 1,448 participants with diabetes (weighted to 24.4 million) from NHANES 2007–2014, the overall prevalence of borderline HTG was 20.1% (projected to 4.9 million) and HTG was 22.5% (projected to 5.5 million). Approximately 19.5% and 25.3% in statin and nonstatin users had HTG, respectively. Specifically, the proportions of subjects with TG levels of <150, 150–199, and ≥200 mg/dL, respectively, were 60.5% (7.5 million), 20.0% (2.5 million), and 19.5% (2.4 million) in those with statin treatment and 54.6% (6.6 million), 20.1% (2.4 million), and 25.3% (3.0 million) in those without statin treatment (P < 0.0001). Compared with subjects with normal TG levels, those with borderline HTG and HTG were somewhat younger (mean age 59.2 and 57.8 vs. 60.8 years, P < 0.0001), less likely to be non-Hispanic black (8.5% and 8.0% vs. 19.2%, P < 0.0001), and more likely to have obesity (73.1% and 65.1% vs. 54.5%, P < 0.0001) and to be currently smoking (12.2% and 22.4% vs. 11.8%, P = 0.0011) (Table 1).

The prevalences of HTG were 17.4% and 22.1% among male and female statin users, respectively, while in those without statin use, the prevalences were 25.6% and 24.9%, respectively (sex differences not significant). In both statin and nonstatin users, Hispanics had more than twice the prevalence of HTG than non-Hispanic blacks (28.1% vs. 9.9%; 32.2% vs. 14.4%; ethnic differences were significant among nonstatin users with P value = 0.0047, but not significant among statin users) (Fig. 1A and B). In those with LDL-C <70 mg/dL, 16.7% had HTG; these proportions were 15.4%, 22.0%, 15.9%, and 28.6% with higher LDL-C levels of 70–99, 100–129, 130–159, and ≥160 mg/dL. Among nonstatin users, HTG was present in 11.5%, 23.2%, 9.2%, 31.4%, and 39.0% among these LDL-C groups, respectively (P < 0.0001) (Fig. 2). Finally, we examined the prevalence of atherogenic dyslipidemia in our sample of diabetes patients indicated by a TG ≥200 mg/dL and HDL-C <35 mg/dL according to statin use. This prevalence was 3.9% (extrapolating to 1.0 million) and 3.3% (0.8 million) among nonstatin and statin users, respectively.

The majority of diabetes patients had 10-year estimated ASCVD risks ≥7.5%, with proportions of 68.7%, 62.7%, and 77.5% in the normal TG, borderline HTG, and HTG groups, respectively. The proportions were even greater among those on statin treatment, with over 85% of HTG individuals with diabetes at high risk of ASCVD over the next decade. One-third of patients with diabetes with HTG overall and ∼40% of statin users were at high ASCVD risk (≥20%) (Fig. 3). Mean 10-year ASCVD risks were 15.4, 17.0, and 17.1 among those not on statins, and 19.2, 19.0, and 20.7 among those on statins in the normal TG, borderline high TG, and HTG groups, respectively (P = 0.6078 across TG categories for both those on and not on statins).

Table 2 shows results for our logistic regression analysis. Variables remained after stepwise selection in predicting TG ≥150 mg/dL were age, sex, ethnicity, HDL-C, and LDL-C. Results from multiple logistic regression models showed that among statin-treated patients with diabetes, female sex was independently associated with an 88% higher risk of having TG ≥150 mg/dL (odds ratio [OR] 1.88, 95% CI 1.02–3.48, P < 0.01) and those of non-Hispanic black ancestry having nearly a two-thirds lower likelihood of having TG ≥150 mg/dL (OR 0.34, 95% CI 0.17–0.69, P < 0.01) compared with non-Hispanic whites. Higher HDL-C was significantly associated with lower risk of TG ≥150 mg/dL (OR 0.45 per SD, 95% CI 0.31–0.65, P < 0.0001). All the other covariates were not significantly associated with elevated TG.

To our knowledge, this study is unique in examining the prevalence of residual HTG despite statin therapy among a U.S. population–representative sample of adults with diabetes. We found that ∼40% of patients with diabetes have mild elevations in TG levels of ≥150 mg/dL, with half of these having levels of ≥200 mg/dL, regardless of statin use. No significant difference is observed between males and females with or without statin use. Of interest, among those on statin treatment despite having LDL-C <100 mg/dL, more than one-third of patients with diabetes have borderline HTG or HTG. Almost 40% of patients with diabetes and mild to moderate elevations of TG are at moderate or high risk for ASCVD in the following 10 years. In addition, about half of statin users with mild to moderate elevations of TG are at high risk for ASCVD. Finally, we note that despite statin treatment, among our U.S. sample of adults with diabetes, females have nearly double the likelihood of having HTG compared with males, and people of African American ancestry have nearly a two-thirds lower likelihood of having HTG compared with whites.

In part because of greater efforts at promoting lifestyle modification, the prevalence of borderline HTG or HTG (≥150 mg/dL) in people with diabetes has decreased from over 50% in NHANES 2009–2010 (4) to closer to 40% from more recent NHANES surveys as presented in the current study. However, the relatively large numbers of U.S. adults with diabetes who still have HTG is striking, especially among certain minority groups. Hispanics have the highest prevalence of HTG, while non-Hispanic blacks have the lowest. A study using electronic health records data from California, which has a comprehensive representation of multiple ethnic groups, also demonstrates the HTG prevalence has doubled among Hispanics compared with blacks in both males and females (18). Also, the mean TG level is almost 50% higher in Mexican Americans than in non-Hispanic blacks (19). The “TG paradox” of non-Hispanic blacks with diabetes with normal TG has been under investigation for years. Major arguments have focused on the role of lipoprotein lipase. A greater activity and less inhibition of lipoprotein lipase is observed in non-Hispanic blacks with diabetes, which may explain the paradox (20). Hispanics generally appear to be more likely to have HTG regardless of the presence of diabetes, and there is evidence suggesting a genetic predisposition to HTG (19). Nevertheless, more efforts are needed to investigate the etiology, prevention, and treatment of HTG in ethnic minority groups.

Our results also suggest that only half of patients with diabetes in the U.S. are on statin treatment, and of those, 40% have borderline HTG or HTG in 2007–2014. It is not known whether levels have changed significantly since this period. According to the American Diabetes Association guidelines, moderate- or high-intensity statin therapy should be initiated in patients with diabetes with prior CVD or those aged 40 years and older with one or more CVD risk factors, and moderate-intensity statin therapy can be considered in younger people with other ASCVD risk factors (21). In addition, statin therapy remains the first-line therapy for those patients according to guidelines from the National Lipid Association and the American Heart Association (12). Our results also indicate that more than one-third of those on statins with well-controlled LDL-C still have borderline high TG or HTG. Conversely, what we show is a substantial prevalence of HTG and especially borderline high TG in those on statins whose LDL-C remains at 160 mg/dL or higher. While we do not have pretreatment levels of LDL-C and TG, we could speculate that at least in some people both the remaining elevated LDL-C and TG may be because of nonresponse to the statin or an inadequate dosage. This constitutes a significant opportunity to address this residual HTG in statin-treated adults with diabetes.

The high prevalence of CVD in individuals with diabetes and HTG has been investigated for several decades. One meta-analysis involving a total of 10,158 patients with incident coronary artery disease from 262,525 participants in 29 studies shows that a moderately strong association was consistently observed between TG concentration and risk of coronary artery disease (7). More recently, longitudinal, real-world administrative database analyses report increased CVD risk and direct health care costs associated with HTG despite statin therapy and controlled LDL-C compared with those with TG ≥150 mg/dL (8).

We also note the majority of those with diabetes who have HTG, even while on a statin, are at moderate or high risk of an ASCVD event in the next decade. Current guidelines (12) recommend at least a moderate-intensity statin in all those with diabetes, and high-intensity statin therapy for the many who have multiple risk factors. Ensuring recommended statin dosages is a first-line therapy for optimizing ASCVD risk reduction, including for people with diabetes regardless of TG levels. While the Action to Control Cardiovascular Risk in Diabetes (ACCORD) Lipid Trial showed that adding fenofibrate as an adjuvant therapy to simvastatin failed to meet its primary end point in reducing ASCVD events (22), those with HTG and low HDL-C did show a 31% risk reduction that was of borderline significance (P = 0.06), and in the Atherothrombosis Intervention in Metabolic Syndrome with Low HDL/High Triglycerides: Impact on Global Health Outcomes (AIM-HIGH) trial, this subgroup also showed a borderline significant (P = 0.07) 26% risk reduction (23). However, the potential benefit in the HTG and low HDL-C subgroups from ACCORD Lipid and AIM-HIGH can only be thought of as hypothesis generating, as those trials missed their primary cardiovascular end points. Recently, the Reduction of Cardiovascular Events with Icosapent Ethyl-Intervention Trial (REDUCE-IT) showed that patients with ASCVD or diabetes and additional CV risk factors treated with 4 g/day of icosapent ethyl (prescription eicosapentaenoic acid [EPA] omega-3 fatty acid) in addition to statins (compared with statins alone) are 25% less likely to have future ASCVD events over a median of 4.9 years of follow-up. This study included people mainly with TG of 200 mg/dL or greater, although initially a lower TG inclusion of 150 mg/dL with a 10% allowed variability around this resulted in patients with baseline TG as low as 135 mg/dL. Moreover, from subgroup analyses, those participants with diabetes had a similar 23% (hazard ratio 0.77, 95% CI 0.68–0.87) and 30% (hazard ratio 0.70, 95% CI 0.60–0.81) lower risk in the primary and key secondary end points (24), demonstrating similar benefits as in those without diabetes. Also, whether or not on-treatment TGs were ≥150 mg/dL or <150 mg/dL, there was similar benefit, suggesting that other non–TG-lowering effects of icosapent ethyl may have had an impact. Whether the risk reductions seen in REDUCE-IT are truly unique to icosapent ethyl pure EPA is still an open question. While recent trials of combined EPA/docosahexaenoic acid have not shown benefit over statin therapy, possibly because of inadequate dosages used, the currently ongoing Statin Residual Risk Reduction With EpaNova in High Cardiovascular Risk Patients With Hypertriglyceridemia (STRENGTH) trial (25) involving a 4 g per day EPA/docosahexaenoic acid combination product is due to report out in the near future. In addition, the ongoing Pemafibrate to Reduce Cardiovascular Outcomes by Reducing Triglycerides in Patients With Diabetes (PROMINENT) trial will examine the efficacy of pemafibrate in reducing cardiovascular outcomes in ∼10,000 patients with both diabetes and atherogenic dyslipidemia (HTG and low HDL-C), and investigations of apolipoprotein CIII inhibitors and angiopoietin-like 3 protein inhibitors will further examine the efficacy of modifying dyslipidemia in patients with diabetes in reducing residual ASCVD risk (26). While recent cholesterol management guidelines have not recommended nonstatin therapies for people with moderate HTG (12), the American Diabetes Association recently revised its 2019 Standards of Diabetes Care, providing a class A recommendation for consideration of icosapent ethyl treatment in patients with diabetes on a statin with controlled LDL-C who have either ASCVD or multiple risk factors and TGs of 135–499 mg/dL (27).

There are several strengths and limitations in our study. A key strength of NHANES is its ability to use sample weighting to estimate the millions of U.S. adults with diabetes who have remaining HTG both among statin users and nonusers. NHANES also has uniform ascertainment of medical history and measurement of lipids and other risk factors across all study centers. Although NHANES relies on self-reported medical history information, previous investigations have shown a close correspondence with patients’ actual medical history (28). However, a limitation of NHANES is the unavailability of follow-up for ASCVD events because of its nature of having a cross-sectional study design. In addition, the event estimates only apply to those adults with diabetes aged 40–79 years without preexisting ASCVD and with complete data on necessary variables for the calculation of ASCVD risk; thus, potential underestimation of events may exist because of missing data on covariates and/or events occurring outside these parameters. Furthermore, no information regarding the duration, intensity, and adherence of statin use is available in NHANES; thus, it is impossible to evaluate how our results would be different based on a fixed dosage or intensity of a given statin. Finally, while we have defined borderline HTG and HTG at cut points of 150 mg/dL and 200 mg/dL, respectively, consistent with terminology from several societies and statements (15–17), not all societies use the same terminology or cut points. For example, the 2019 American Diabetes Association Standards of Care (21) notes elevated TGs to be 150 mg/dL or higher and moderate HTG as 175–499 mg/dL.

In conclusion, our study reveals that almost one in five adults with diabetes in the U.S. has residual HTG despite statin use. Moreover, many of these people are at moderate or increased risk of future ASCVD events in the next decade. This warrants greater attention toward improved efforts such as lifestyle management as the cornerstone for managing dyslipidemia, as well as consideration of newer evidence-based therapies to reduce residual ASCVD risk that remains despite statin-controlled LDL-C.

Duality of Interest. This project was supported by a contract from Amarin Pharma, Inc., to the University of California, Irvine, CA. N.D.W. receives research support from Amarin, Amgen, Boehringer-Ingelheim, and Novo Nordisk; is on the speaker’s bureau for Sanofi and Amarin; and is a consultant or on the advisory board for Amarin, Novartis, and Astra-Zeneca. S.P. and C.G. are employed at Amarin Pharma, Inc. P.P.T. is on the speaker’s bureau for Amarin, Amgen, Kowa, Nova Nordisk, Regeneron, and Sanofi and is a consultant or on the advisory board for Amarin, Amgen, Kowa, Novo Nordisk, Regeneron, and Sanofi.

Author Contributions. W.F. and N.D.W. wrote the manuscript and conducted the analysis. S.P., C.G., and P.P.T. provided critical review, comment, and editorial suggestions. W.F., S.P., C.G., P.P.T., and N.D.W. approved and determined the final manuscript to be accurate. N.D.W. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 78^th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018.