Lipid control and use of lipid-regulating drugs for prevention of cardiovascular events in Chinese type 2 diabetic patients: a prospective cohort study

The assessment methods, definitions and laboratory assays have been described [3]. In brief, patients attended the centre after an overnight fast of at least eight hours for clinical measurements and laboratory tests. Full lipid profile consisting of total cholesterol, triglycerides and high-density lipoprotein cholesterol (HDL-C) was measured by enzymatic methods on a Hitachi 911 automated analyzer (Boehringer Mannheim, Mannheim, Germany) using reagent kits supplied by the manufacturer. The low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald equation. All laboratory measurements were conducted by the Department of Chemical Pathology of the Prince of Wales Hospital, which was accredited by the Australian Quality Assurance Program.

The SAS (release 9.10; SAS Institute, Cary, NC) program was used for all analysis. Follow-up time in years was estimated from enrolment to the earliest date of a CVD event, death or censoring whichever came first. We used yes/no coding for use of statins, fibrates and other drugs during the follow-up period. Cox proportional hazard regression was used to obtain hazard ratio (HR) with 95% confidence intervals (95% CI) of lipids (LDL-C, HDL-C, triglyceride) and lipid regulating drugs (i.e. statins and fibrates) for risk of CVD.

Firstly, the linearity of lipids for CVD risk was examined using restricted cubic spline Cox model analysis [6]. Restricted cubic spline consists of piecewise cubic polynomials that are connected across different intervals of a continuous variable. It can fit sharply curving shapes [6]. As in our previous analyses [7, 8], we chose 4 knots at quartiles 0.05, 0.35, 0.65 and 0.95 which were suggested to offer adequate fit of the model with good compromise between flexibility and loss of precision due to overfitting of a small sample [6]. Initially, spline terms of LDL-C, HDL-C and triglyceride were entered the spline Cox model. We then performed a forward stepwise algorithm with P = 0.10 for inclusion and removal to identify possible confounding factors including clinical and biochemical covariates at enrolment, drug use during follow-up period (See table 1 for a list of candidate covariates) and years of enrolment. Based on the HR curves of lipids, we identified threshold value for CVD risk and categorized patients using these cutoff values. Otherwise, we treated the covariates as linearly associated with CVD. We performed Cox model analysis to obtain the HR of lipids for CVD with adjustments for other lipid parameters and covariates.

We then examined the HRs of use of statins and fibrates for CVD defined as drug use from enrolment to the first CVD event, death or censoring date whichever came first. We used logistic regression procedures and a stepwise algorithm with P = 0.30 for entry and stay to select baseline covariates including age, sex, year of enrolment, waist circumference, LDL-C, HDL-C, triglyceride, smoking status, body mass index, HbA_1c, systolic blood pressure, Ln [(urine albumin-creatinine ratio (ACR)) + 1], estimated glomerular filtration rate (eGFR), duration of disease, retinopathy and neuropathy to estimate propensity scores for adjustment of probabilities of use of statins and fibrates [9]. Age, waist circumference, LDL-C, HDL-C, triglyceride, HbA_1c, Ln(ACR + 1), eGFR, duration of disease, retinopathy and neuropathy were selected for the propensity score for statins (c-statistic = 0.79). Age, LDL-C, HDL-C, triglyceride, HbA_1c, systolic blood pressure, duration of disease, retinopathy and neuropathy were selected to predict use of fibrates during follow-up period (c-statistics = 0.79). Stratified models on deciles of both propensity scores were used to adjust for likelihood to use statins and fibrates during follow-up period. We further adjusted for covariates identified in the initial restricted cubic spline models to analyse the effects of statins and fibrates on CVD.

During a median follow up period of 4.9 years (interquartile range 2.8 - 7.0 years), 371 (8.2%) of 4521 subjects developed CVD. Patients who had new onset of CVD were older, more likely to be men and smokers, had longer disease duration, higher LDL-C, triglyceride, HbA_1c, ACR, lower HDL-C and eGFR than those without. These patients were also more likely to receive insulin and less likely to receive rosiglitazone and gliclazide (Table 1). In the whole cohort, only 23% and 6% of patients were prescribed statins or fibrates respectively during follow-up.

In the entire cohort and before adjustment, CVD risk was positively associated with LDL-C and triglyceride but negatively with HDL-C (figure 1, black line). After adjustment for age, smoking status, duration of diabetes, HbA_1c, Ln (ACR + 1), use of gliclazide or rosiglitazone during follow-up, years of enrolment (selected by the stepwise algorithm with P = 0.10 for entry and stay); the HR of CVD increased sharply when LDL-C was ≥ 3.0 mmol/L (Figure 1, blue line). The CVD risk declined linearly with increasing HDL-C with no definite threshold (Figure 2, blue line). The HR of CVD with triglyceride became insignificant after adjusting for confounders (Figure 3, blue line).

Using LDL ≥ 3.0 mmol/L as the cutoff point, the HR for CVD risk was 1.41 (1.13 - 1.75, p = 0.0024) as compared to those with LDL < 3.0 mmol/L. After adjusting for covariates, the HR was attenuated to 1.36 (1.08 - 1.71, p = 0.0095). Using the whole range of HDL-C values, with every 1 mmol/L increase, the HR for CVD was 0.61 (0.45 - 0.84, p = 0.0080), which decreased to 0.59 (0.42 - 0.82, p = 0.0020) after adjustment for covariates (Table 2).

In this new-user cohort, use of statins reduced CVD risk by 34% after controlling for baseline covariates and use of other medications during follow-up. Use of fibrates did not reduce CVD risk after adjustment for other covariates (Table 3). The trend of risk reduction was consistent in nearly all subgroups stratified by LDL-C and HDL-C levels, reaching significance for some subgroups. In patients with LDL-C ≥ 3 mmol/L, statins significantly reduced CVD risk by 40%. In patients with LDL-C < 3 mmol/L, fibrates reduced CVD risk by 66%. In patients with low HDL-C (< 1.0 mmol/L in men and < 1.3 mmol/L in women) and those with concomitant metabolic syndrome [10], statins reduced CVD risk by 51% and 42% respectively.

Epidemiological studies and interventional trials confirmed the predictive value of high LDL-C on risk of CVD. These risks were accentuated by 2-3 folds in type 2 diabetic patients for the same number of risk factors [11, 12]. Furthermore, type 2 diabetic patients are often insulin resistant with a typical dyslipidaemic pattern of low HDL-C and high triglyceride [13‐15]. Despite the similar LDL-C levels between type 2 diabetic and non-diabetic subjects, the high triglyceride contributes to increased concentration of small dense LDL-C particles which are more atherogenic. Type 2 diabetic patients are also more likely to have oxidized and glycated LDL-C which are readily taken up by macrophages to form atheromatous plaques. Given the consistency of interventional data, lowering of LDL-C using HMGcoA reductase inhibitors (statins) remains the primary target in preventing CVD in type 2 diabetes [2, 16, 17].

As early as in 1977, the Framingham study demonstrated an inverse association between HDL-C and incidence of coronary heart disease after adjustment for other lipid parameters and risk factors [18]. In the Asia Pacific Cohort Studies Collaboration consisting of more than 50,000 data, HDL-C, total cholesterol and LDL-C were independent risk factors for coronary heart disease in Asians [19]. By contrast, the prognostic significance of triglyceride in CVD remains controversial [19‐23]. While some researchers have reported triglyceride to be the most predictive lipid parameter for CVD, in a recent meta-analysis by the Emerging Risk Factors Collaboration, such risk association was not confirmed after controlling for other confounders [23]. In this analysis, triglyceride also did not predict incident CVD in our Chinese type 2 diabetic patients after adjusting for confounders.

In Caucasians, meta-analyses of large scale randomized studies [15, 24, 25] have confirmed the benefits of statins on CVD in primary and secondary prevention studies which included diabetic and non-diabetic subjects. These beneficial effects have recently been extended to patients with LDL-C level less than the recommended treatment level of 2.6 mmol/L although these patients had high sensitivity C reactive protein [26]. In the post-hoc analysis of the FIELD study, treatment with fenofibrate reduced the risk of CVD by 27% in a subgroup of type 2 diabetic patients with high triglyceride and low HDL-C [27]. Echoing this finding, the recently published ACCORD lipid study showed benefit of fenofibrate in a subgroup of type 2 diabetic patients who had both high triglyceride ≥ 2.30 mmol/L and low HDL-C ≤ 0.88 mmol/L [28]. We found use of statins was associated with reduced risk of CVD in the whole group as well as in those with high LDL-C, low HDL-C and metabolic syndrome. The benefit of fibrates on CVD in the whole group was rendered insignificant after adjusting for confounders, probably due to the small sample size.

Recent genetic studies suggested possible inter-ethnic differences in patterns of risk variants associated with dyslipidaemia. For example, two single nucleotide polymorphisms of the KCNQ1 gene, a well-known type 2 diabetes gene, were associated with higher level of triglycerides and lower levels of HDL-C in Chinese population [29]. On the other hand, in Chinese type 2 diabetic patients with concomitant coronary heart disease, the PPARγC161 〿 T genotype was associated with reduced severity of atherosclerosis, in part medicated by improved lipid metabolism of decreased the triglycerides and apoB levels [30]. While these findings suggest that triglyceride may be implicated in CVD in Chinese subjects, after adjusting for cofounders, our overall results concord with most studies regarding the importance of LDL-C and statins in modifying CVD risk in Chinese.

Despite the growing epidemic of diabetes and expected increase in CVD in Asia, there have not been any large scale clinical trials to examine the efficacy of use of statins or fibrates in preventing CVD in Asian populations. In the absence of such evidence, treatment recommendations were largely empirical and derived from Western guidelines. To our knowledge, our present analysis is by far the largest prospective cohort of Asian type 2 diabetic patients with detailed documentation of risk factors, drug usage and clinical outcomes. Our diabetes registry provided useful information for guiding clinical practice and for future large randomized clinical trials in Chinese.

To circumvent potential bias inherent in an observational survey, we have used a new-user design and excluded existing statins or fibrates users. Such design is superior to traditional cohort studies as prevalent users cause indication and adherence bias [31]. Besides, all patients have undergone comprehensive assessment with detailed documentation of other confounders which were adjusted in our analysis. We further used propensity score to adjust for bias due to prescribers' preference although we were unable to fully adjust for factors such as volunteer bias which is often seen in patients more compliant to treatment or follow up. After careful analysis of the bias, we concluded that the drug use definition was robust and did not introduce major bias neither towards an increased nor decreased risk of the endpoint [32].

Our results need to be interpreted with the following caveats. We only measured a single fasting plasma lipid profile when plasma triglyceride level is known to have marked day-to-day and intra-individual variability. This may have attenuated the risk association between triglyceride level and CVD outcomes. Our cohort was predominantly clinic-based although given the heavily-subsidized health care system in Hong Kong, our public hospital clinics provide care to more than 90% patients with chronic diseases such as diabetes [7]. Despite our careful adjustment for potential confounders, a new-user cohort design cannot replace a randomized controlled trial which aims to remove bias due to unbalanced distribution of measured and unmeasured confounders.

In our survey, lipid levels were not systematically collected during clinic visits after the initial enrolment. To avoid bias due to unknown lipid levels during start of therapy, we opted to use yes/no coding rather than time-dependent use in our analysis. In these public hospital clinics, atorvastatin, fluvastatin, pravastatin, rosuvastatin and simvastatin were available with atorvastatin (10-20 mg daily) and simvastatin (10-40 mg daily) being the two most popular statins. For fibrates, bezafibrate and gemfibrozil were both available with gemfibrozil being the more popular choice. However, with the relatively small sample size of patients on statins and fibrates, we were not able to examine the effects of individual drug on CVD. We also did not document the drug compliance of the patients or check whether other private drugs were used. Although non compliance or occasional drug omission is not uncommon in patients on long term medications, this would tend to attenuate effect size and level of significance. Besides, surveys by our other colleagues have reported more than 90% compliance amongst patients taking lipid regulating drugs [33, 34].