Cholesterol levels and development of cardiovascular disease in Koreans with type 2 diabetes mellitus and without pre-existing cardiovascular disease

Current evidence indicates that dyslipidemia is a strong predictor of cardiovascular disease (CVD). Lowering low-density lipoprotein cholesterol (LDL-C) level with statin therapy reduces incidence of major CVD [1‐5]. In the recent guidelines, the cholesterol targets are based on several primary and secondary prevention statin trials that have shown improved outcomes with more intensive LDL-C lowering [1‐3]. However, none of these trials used specific LDL-C targets to trigger adjustments in the medication dose, and LDL-C targets are usually extrapolated from these trials. The usefulness of the treat-to-target approach is controversial because of insufficient evidence. Recently, the standard versus intEnsive statin therapy for hypercholesteroleMic Patients with diAbetic retinopaTHY (EMPATHY) study demonstrated that, in patients with diabetic retinopathy, intensive statin therapy to achieve an LDL-C level < 70 mg/dL did not reduce incidence of the cardiovascular events compared with standard therapy to achieve levels of ≥ 100 and < 120 mg/dL [6].

Primary prevention research focuses on the thresholds for initiating treatment of modifiable risk factors and the optimal goals of risk factor modification. There is insufficient information available about the optimal cholesterol targets for preventing CVD development in people with type 2 diabetes mellitus (DM) [1‐4]. We analyzed data from the Korean National Health Insurance System database (NHIS DB) and investigated the relationship between high cholesterol level and the onset of CVD in people with type 2 DM. Because the NHIS DB represents the entire Korean population, it has been used in a number of population-based nationwide studies of type 2 DM in Korea [7‐12]. We classified the population into statin users and non-users, and we tried to identify a threshold for the cholesterol level at which the risk of CVD begins to increase in people with type 2 DM in each group.

In this large population-based prospective study of Koreans with type 2 DM, significant positive associations were observed between an increased risk of CVD and high LDL-C and non-HDL-C levels. We identified an increased risk of CVD with an LDL-C level ≥ 130 mg dL in patients with type 2 DM who were not taking statins. The risk of CVD was significantly increased in the statin-treated participants with an LDL-C level ≥ 70 mg/dL.

It is clear that people with low cholesterol levels gain less absolute benefit from cholesterol-lowering therapy than do people with high cholesterol levels [13]. For the present study population, the mean LDL-C and non-HDL cholesterol levels at baseline were 113 mg/dL and 146 mg/dL, respectively. Previous studies involving participants with higher mean LDL-C levels showed an association with the risk of coronary heart disease (CHD) for those in the higher LDL-C ranges. These studies include the Framingham study, in which the mean LDL-C levels at baseline were 139 mg/dL for men and 138 mg/dL for women [14]. In a cholesterol-lowering clinical trial of intermediate-risk patients (Heart Outcomes Prevention Evaluation-3 [HOPE-3]), the pretreatment LDL-C level was 129 mg/dL [15]. The goals and starting points of cholesterol treatment may be different in these low-cholesterol population. However, indication for starting statin therapy should also be related to the progression of atherosclerosis or clinical event regardless on a level of baseline cholesterol [16‐18]. For the prediction of MI or stroke, the area under the curve of LDL-C alone was low (data not shown), suggesting that including other potential risk factors such as smoking, BP, or family history of premature vascular disease to the risk model would be beneficial for a better risk prediction of CVD rather than LDL-C alone [18, 19]. Statins might be employed in patients without CVD but with elevated cholesterol and/or multiple atherosclerotic risk factors. In this study, we aimed to determine the level of “elevated cholesterol” in which statin therapy should be started in patients without CVD. This study suggested that statin therapy should be considered when LDL-C was above 130 mg/dL, regardless of CVD risk factors in diabetic patients without CVD. In this study, previous use of statin was dependent on the clinical judgement of the physicians. Statin users tended to have other cardiovascular risk factors, including hypertension and longer diabetes duration. According to the European Society of Cardiology guidelines [20], diabetic patients at high risk should be intensively treated with statin therapy with an LDL-C goal of < 70 mg/dl. Our results are consistent with the recommendation; The risk of MI and stroke increased among diabetic people with type 2 DM having an on-treatment LDL-C level ≥ 70 mg/dL.

A previous Japanese population-based cohort study showed a nonlinear association. The multivariable HR for CHD was 1.68-times higher for people with an LDL-C level 126–150 mg/dL compared with men and women whose level was ≤ 102 mg/dL, and the risk plateaued at < 125 mg/dL [21]. Another Japanese cohort study showed that the mean LDL-C levels were 110.5 mg/dL for men [13]. In that study, men with an LDL-C level ≥ 140 mg/dL had twofold higher age-adjusted risk of mortality for CHD than did those with an LDL-C level < 80 mg/dL [13]. In Chinese patients with type 2 DM without a history of CVD and no lipid-modifying drug use, the HR for CVD increased sharply when the LDL-C level was > 116 mg/dL [22]. We observed a graded positive trend for MI risk starting from an LDL-C level of 130 mg/dL, which increased for the higher LDL-C categories among people with type 2 DM who were not taking statins. Our data suggests that statin therapy might be considered for people whose LDL-C level remains ≥ 130 mg/dL after appropriate lifestyle modifications.

In the present study, an association between non-HDL-C level and MI risk was seen at a lower level of non-HDL-C than for the corresponding LDL-C level. In a recent study of relatively low-risk people without a history of atherosclerotic CVD (ASCVD) or DM at baseline, the association between LDL-C (or non-HDL-C) levels and ASCVD mortality was evaluated over a median follow-up of 27 years [23]. Compared with participants with an LDL-C level < 100 mg/dL, those with an LDL-C level ≥ 160 mg/dL had an increased risk for CVD mortality. In that study, a non-HDL-C level ≥ 160 mg/dL was significantly associated with CVD death compared with a non-HDL-C level < 130 mg/dL [23]; our findings are consistent with the results of that study. We also found that, among diabetic people not taking statins, LDL-C and non-HDL-C levels ≥ 130 mg/dL were independently associated with a significant increase in the incidence of MI. The non-HDL-C level captures the risk associated with both cholesterol-rich lipoproteins and TG-rich remnant lipoproteins [24]. We note that all recent outcomes trials of proprotein convertase subtilisin/kexin type 9 inhibitors used both LDL-C and non-HDL-C thresholds for deciding eligibility, and may therefore have included people with an elevated non-HDL-C level despite having an LDL-C level below the eligibility criterion [25].

In the present study, the risk of CVD was higher in participants whose LDL-C level was greater than the on-treatment level of > 70 mg/dL; however, no increase in risk was observed for those whose LDL-C level was in the range 55–69 mg/dL. It was recently reported that the intensive statin therapy target LDL-C level of < 70 mg/dL did not reduce the incidence of composite CV events more significantly than the standard therapy target LDL-C level of ≥ 100 and < 120 mg/dL [6]. However, in that study, targeted levels were achieved in < 50% of patients. When analyzed the data from 1909 patients who achieved the target LDL-C level, achieving an LDL-C level < 70 mg/dL was associated with a more effective reduction in CV events incidence than the target level of 100 to 120 mg/dL in patients with type 2 DM with retinopathy [26].

Application of the American Heart Association–American College of Cardiology–atherosclerotic cardiovascular disease (AHA-ACC-ASCVD) score overestimated cardiovascular events in a multiethnic cohort, including Chinese people, without baseline clinical ASCVD [27]. The Framingham risk score was developed based on data collected from the Caucasian population and overestimates the 10-year risk of CHD in the Asian population [28, 29]. A study using data from the Hong Kong Diabetes Registry, established in 1995, also reported that the United Kingdom Prospective Diabetes Study Risk Engine overestimates the risk of CVD with suboptimal discrimination [30]. There is significant heterogeneity in CVD prevalence between populations with diabetes. These discrepancies in CVD prevalence may be related to differences in disease profile and other determinants such as genetics, health care policy, and culture. A population-based CVD risk prediction model for Asian people with type 2 DM should be developed to allow for better management. The thresholds for initiating treatment of abnormal lipid levels and the optimal goals of lipid-altering therapy may also differ according to ethnicity or population [31‐33].

The optimal LDL-C threshold for starting statin therapy may also vary by age and sex [18, 34]. The association between LDL-C and MI risk in elderly patients was weak. Lipid levels measured in late life may decrease as a result of behavioral changes or because of the presence of comorbidities.

The association between LDL-C and risk of MI was stronger in men than in women, suggesting that there are more atherogenic characteristics of LDL-C in men. It was also reported that risk of CHD was associated with high TG and low HDL-C in women and high LDL-C in men [35]. There has been controversy over whether statins are effective for primary prevention in women [20]. The gender gap in risk of CHD persists throughout life but the relative risk declines with age [36]. What causes the gender contrast in risk is still not clear. An adverse effect of male hormones, gender heterogeneity in insulin resistance mechanisms, in LDL-C characteristics and/or in aging processes influencing arterial stiffness, are some of the theories suggested [36]. The epidemiological evidence for an association of high LDL-C with an increased risk of ischemic stroke is relatively weak [21]. We also found that a stroke was less related to the cholesterol level than MI. It may be due to the heterogenous etiology of stroke. Lacunar infarction and cardioembolic infarction seem to have a less potent relation with elevated LDL-C [21]. Although it was difficult to clearly define the subtypes of stroke, we excluded hemorrhagic stroke (ICD-10 codes I60, I61, I62). We provided cut-offs for the LDL-C levels (e.g. 70, 100, 130 mg/dL), which were different from the observed quintiles. In major guidelines on the treatment of blood cholesterol to reduce CVD [20, 37], LDL-C goals < 55, 70, 100, or 130 mg/dL are recommended depending to the CV risk.

There are several limitations to our study. First, day-to-day variability because of laboratory error or biological variability may have affected the results because we used a single measurement of cholesterol levels. Repeated measurement would have been more accurate for classifying the participants. Second, several studies [38, 39] have shown that even when the TG level is < 400 mg/dL, a higher TG level can increase the degree to which the Friedewald formula underestimates LDL-C level. We excluded participants with a TG level > 400 mg/dL, but we also analyzed non-HDL cholesterol levels. Third, because the analysis was performed using an NHIS health examinee cohort, diabetic patients with severe disabilities would likely have been excluded from undergoing regular health examinations. Because patients with underlying CVD at baseline were excluded, the study population likely included mostly relatively low-risk diabetic participants. Fourth, cholesterol levels may increase over time and trigger the use of statins among non-users, such that the observed risk of CVD would be mitigated. Our study was limited in that time-varying Cox regression was not performed. Fifth, we were unable to obtain clinical information on hemoglobin A1c levels, high sensitivity C-reactive protein [40], and other medications that an affect the development of CVD. Lastly, because only the Korean population was included, our findings cannot be extrapolated to people with different ethnicities. Despite these limitations, this study evaluated longitudinal data from the entire Korean adult population. Therefore, our findings reflect ‘real-world’ data, on a national scale, regarding the effects of LDL-C and non-HDL-C levels on CVD risk in people with type 2 DM.

Although the existence of a threshold does not always mean that the optimal cutoff level should be the same value, the absence of an increase in risk below this threshold suggests that the LDL-C cutoff point for Korean people with DM that increases the risk of CVD may be around 130 mg/dL. Because we observed a graded positive trend in MI risk starting from a LDL-C level of 130 mg/dL, and this risk increased for the higher LDL-C categories, we recommend initiating statin treatment for primary prevention of CVD in Korean patients with type 2 DM when the LDL-C level is ≥ 130 mg/dL. We found a significantly increased risk of CVD in statin-treated participants with an LDL-C level ≥ 70 mg/dL. For patients with type 2 DM without preexisting CVD, our recommendations for starting statin therapy for primary prevention of CVD are: 1) an LDL cholesterol level ≥ 130 mg/dL, and 2) a goal LDL-C level of < 70 mg/dL.