The risk of future coronary heart disease (CHD) in subjects with diabetes and “metabolic dyslipidemia” (high triglyceride [TGs] and low high-density cholesterol levels) remains a matter of concern. Little is known regarding the risk of CHD for this phenotype with low-density lipoprotein cholesterol (LDL-C) levels <100 mg/dl. We analyzed a diabetes cohort of 28,318 members (aged 30 to 90 years) of Kaiser Permanente Northern California during 2002 to 2011 (192,356 person-years [p-y] follow-up), with LDL-C levels <100 mg/dl and without known CHD. We compared the incidence and hazard ratios (HRs) for CHD events in groups using Cox models: normal high-density lipoprotein (HDL) and TG (reference; n = 7,278, 25.7%); normal HDL and high TG (≥150 mg/dl; n = 4,484,15.8%); low HDL (≤50 mg/dl for women and ≤40 mg/dl for men) and normal TG (n = 4,048, 14.3%); low HDL and high TG (metabolic dyslipidemia; n = 12,508, 44%). Patients with metabolic dyslipidemia had the highest age-adjusted CHD events/1,000 p-y (12.7/1,000 p-y and 19.0/1,000 p-y for women and men, respectively). After multivariate adjustment for age, gender, ethnicity, hypertension, smoking, statin use, duration of diabetes, and hemoglobin A1c, we observed an increased CHD risk in women (HR 1.35, 95% confidence interval 1.14 to 1.60) and men (HR 1.62, 95% confidence interval 1.43 to 1.83) with metabolic dyslipidemia compared to those with normal HDL and TG. Even in subjects with an LDL-C <100 mg/dl, presence of metabolic dyslipidemia in adults with diabetes is associated with an increased risk of CHD. In conclusion, effective CHD prevention strategies are needed for adults with diabetes and metabolic dyslipidemia.
The risk of coronary heart disease (CHD) associated with either low high-density lipoprotein cholesterol (HDL-C) levels or high triglyceride (TG) levels is well documented. The risk associated with the combination of low HDL-C and high TG levels, or “metabolic dyslipidemia,” was evaluated in a European cohort study of adults without diabetes and was found to be associated with a greater risk of CHD events. Burden and risk of CHD associated with metabolic dyslipidemia in a large contemporary diabetes population remains unexplored. Clinically, it is an important subset of patients at high risk with no conclusive therapy. In the Action to Control Cardiovascular Risk in Diabetes, in which adults with diabetes on statin therapy with low low-density lipoprotein cholesterol (LDL-C) levels were randomized to fibrate or placebo, there appeared to be a possible benefit for fibrate in a prespecified subgroup of patients (n = 941) in the highest tertile of TG (≥204 mg/dl) and lowest tertile of HDL-C (≤34 mg/dl). In the Atherothrombosis Intervention in Metabolic Syndrome With Low HDL/High Triglycerides and Impact on Global Health Outcomes (AIM-HIGH) trial, patients on statins were randomized to niacin or placebo; in a post hoc analysis in a subset of patients (n = 522) with the highest tertile of TG (≥198 mg/dl) and lowest tertile of HDL-C (<33 mg/dl), a nonsignificant trend toward reduction of cardiovascular risk was seen in the niacin group. The objective of our study was to evaluate the burden of metabolic dyslipidemia and associated risk of CHD events in a large cohort of patients with diabetes mellitus with low LDL-C levels.
Methods
Kaiser Permanente Northern California (KPNC) is a large, integrated health care delivery system, caring for more than 3 million people who are broadly representative of the service area. We constructed a diabetes cohort of 28,318 members with low LDL-C levels (<100 mg/dl) and without known CHD, aged 30 to 90 years, and followed them from 2002 to 2012. As noted, patients were only included if their LDL-C level was <100 mg/dl, and this level of control could have been either constitutive or a result of being treated with statin therapy as of baseline. We excluded subjects not continuously enrolled during the 12 months before baseline to avoid misclassifying new enrollees with preexisting CHD.
Clinical recognition of prevalent diabetes was based on a well-validated algorithm used by the KPNC Diabetes Registry, which includes any of the following: (1) diagnosis: inpatient diagnosis (principal diagnosis of International Statistical Classification of Diseases and Related Health Problems [ICD]-9: 250) or outpatient diagnosis (2 or more diagnoses with ICD-9: 250; excludes diagnoses collected in the emergency, optometry, or ophthalmology departments); (2) 2 abnormal outpatient laboratory results (fasting glucose ≥126 mg/dl; random or postchallenge [75 g] glucose ≥200 mg/dl; hemoglobin A1c [HbA1c] ≥6.5%), tested on separate days, within a 3-year period; or (3) pharmacy utilization for insulin or certain oral antihyperglycemic medications. Newly identified diabetes cases were excluded if they were based on a diagnosis for gestational diabetes (ICD-9: 648.8), identified because of the use of insulin sensitizers thiazolidinedione or metformin for conditions other than diabetes (e.g., prediabetes or polycystic ovary syndrome), or included because of a single criteria without subsequent diabetes related utilization within a 2-year period.
We collected information on prevalent CHD during the 5 years before cohort entry (January 1, 2002, baseline). New CHD events were identified during the 10 years of follow-up after baseline on the basis of inpatient primary diagnosis or procedure records: fatal and nonfatal myocardial infarction (ICD-9 diagnosis codes: 410 to 414, 429.2; ICD-10 diagnosis codes: I20 to I25), percutaneous coronary intervention (ICD-9 procedure codes: 36.01, 36.02, 36.05, 36.06, 36.07, 36.09, 00.66 or Current Procedural Terminology 4 codes: 92980, 92981, 92982, 92984, 92995, 92996), and coronary artery bypass surgery (ICD-9 procedure codes: 36.03, 36.1x or Current Procedural Terminology 4 codes: 33533, 33534, 33535, 33536, 33572, 33510, 33511, 33512, 33513, 33514, 33516, 33517, 33518, 33519, 33521, 33522, 33523, 33530). Fatal events were also identified through the California mortality file.
Age, gender, and race/ethnicity were ascertained from self-reported demographic data collected at clinic visits, during health plan enrollment, from member surveys, or on intake for hospitalization. Relevant ambulatory laboratory values were ascertained from KPNC databases: total cholesterol, LDL-C, TG, HDL-C, and HbA1c, along with blood pressure, smoking status, and medication use.
We calculated age-adjusted overall and gender-specific rates of CHD (number of events/1,000 person-years [p-y]) for each exposure category. We specified Cox proportional hazard regression models to estimate hazard ratios (HRs) as a measure of the difference in rates of CHD across the 4 groups: normal HDL and TG (reference); normal HDL and high TG (≥150 mg/dl); low HDL (≤50 mg/dl for women and ≤40 mg/dl for men) and normal TG; and low HDL and high TG (metabolic dyslipidemia). Multivariate-adjusted models included age, race/ethnicity, smoking, hypertension, statin use, duration of diabetes, and HbA1c. We carefully accounted for person-time during follow-up. Over the 10-year follow-up, a similar proportion (25% to 28%) left the health system across the 4 exposure groups and were censored in our time to event analysis. The similar censoring rates across the exposure groups suggest there was no substantive bias introduced.
All analyses were performed with SAS version 9.1 (SAS Institute, Cary, North Carolina), and associations were considered statistically significant at the 0.05 level. This study was approved by the Kaiser Foundation Research Institute Institutional Review Board.
Results
Of the 28,318 adults (aged 30 to 90 years) with diabetes and LDL-C <100 mg/dl at baseline, 7,278 subjects (25.7%) had normal HDL-C and TG levels, 4,484 subjects (15.8%) had normal HDL-C and high TG, 4,048 subjects (14.3%) had low HDL-C and normal TG, and 12,508 subjects (44.2%) had low HDL-C and high TG levels (metabolic dyslipidemia; Table 1 ). In the metabolic dyslipidemia group, the mean HDL-C was 35.3 (±6.3) mg/dl and mean TG was 340.5 (±176) mg/dl; history of smoking and hypertension was highest in this subset.
| Full Cohort | HDL-C Normal, TG Normal | HDL-C Normal, TG High | HDL-C Low, TG Normal | HDL-Low, TG High | p-value | |
|---|---|---|---|---|---|---|
| HDL-C (mg/dl) | >50 (women) | >50 (women) | ≤50 (women) | ≤50 (women) | ||
| >40 (men) | >40 (men) | ≤40 (men) | ≤40 (men) | |||
| TG (mg/dl) | <150 | ≥150 | <150 | ≥150 | ||
| Number | 28,318 | 7,278 (25.7%) | 4,484 (15.8%) | 4,048 (14.3%) | 12,508 (44.2%) | |
| Men | 15,638 (55.2%) | 4,234 (58.2%) | 2,385 (53.2%) | 2,101 (51.9%) | 6,918 (55.3%) | <0.001 |
| Age (years), mean (SD) | 60.9 (12.2) | 62.1 (12.7) | 62.3 (11.1) | 60.8 (12.9) | 59.7 (12.0) | <0.001 |
| Ethnic group | <0.001 | |||||
| White | 51.7% | 47.3% | 49.6% | 50% | 55.4% | |
| Black | 9.2% | 15.5% | 7.2% | 12.4% | 5.2% | |
| Hispanic | 14.7% | 12.8% | 14.1% | 15.9% | 15.7% | |
| Asian | 14.8% | 15.9% | 19.2% | 12.4% | 13.4% | |
| Other | 9.6% | 8.5% | 9.9% | 9.2% | 10.2% | |
| Lipids (mg/dL), mean (SD) | ||||||
| Total cholesterol | 170.9 (32.6) | 160.8 (20.2) | 186.6 (31.9) | 143.2 (18.3) | 180.2 (34.8) | <0.001 |
| LDL cholesterol | 81.6 (14.9) | 83.7 (13.2) | 84.06 (12.9) | 82.8 (13.5) | 79.05 (16.5) | <0.001 |
| HDL cholesterol | 43.9 (13.4) | 57.3 (13.7) | 51.7 (9.9) | 37.8 (6.1) | 35.3 (6.3) | <0.001 |
| Triglycerides | 229.8 (168) | 94.4 (29.5) | 250.8 (115) | 108.2 (26.2) | 340.5 (176) | <0.001 |
| Statin use | 36.1% | 35.1% | 45.2% | 33.9% | 34.2% | <0.001 |
| Current Smoker | 19.5% | 17.8% | 17.6% | 19.4% | 21.2% | <0.001 |
| Ever Smoker | 44.7% | 42.7% | 44.5% | 43.9% | 46.2% | <0.001 |
| Hypertension history | 77% | 73% | 80.6% | 75.2% | 78.6% | <0.001 |
| Duration of diabetes (years), mean (SD) | 5.5 (4.0) | 6.3 (4.3) | 5.5 (3.9) | 5.7 (4.0) | 5.04 (3.7) | <0.001 |
| Duration of diabetes groups (years) | <0.001 | |||||
| 0-4 | 51.8% | 44.2% | 51.7% | 50.6% | 56.2% | |
| 5-9 | 33.5% | 35.2% | 34.0% | 33.8% | 32.2% | |
| >10 | 14.7% | 20.6% | 14.3% | 15.5% | 11.2% | |
| Hemoglobin A1C (%), mean (SD) | 7.8 (1.7) | 7.7 (1.6) | 7.7 (1.5) | 7.6 (1.7) | 7.9 (1.7) | <0.001 |
| Insulin therapy | 20.6% | 27.9% | 16.1% | 21.8% | 17.5% | <0.001 |
| Oral antihypoglycemic therapy | 70.5% | 61.8% | 71.4% | 70.7% | 75.2% | <0.001 |
We observed 3,467 new CHD events during median follow-up time of 10-year follow-up (192,356 p-y). The age-adjusted CHD rate ( Table 2 ) was lowest in subjects with normal HDL-C and normal TG (11.4, 95% confidence interval [CI] 10.4 to 12.5 per 1,000 p-y) and highest in those with metabolic dyslipidemia (16.0, 95% CI 14.8 to 17.2, per 1,000 p-y). Of note, the age-adjusted CHD rate per 1,000 p-y for subjects with metabolic dyslipidemia was much higher for men (19.0, 95% CI 17.1 to 20.9 per 1,000 p-y) compared to women (12.7, 95% CI 11.3 to 14.3). The final model adjusted for age, ethnicity, smoking, history of hypertension, statin use, duration of diabetes, and HbA1c showed that men with isolated high TG (HR 1.23, 95% CI 1.06 to 1.43) or low HDL-C (HR 1.39, 95% CI 1.19 to 1.63) had a higher risk of CHD events; both women (HR 1.35, 95% CI 1.14 to 1.60) and men (HR 1.62, 95% CI 1.43 to 1.83) with metabolic dyslipidemia had a greater risk of CHD events ( Figure 1 ).
