The Effect of Aggressive Versus Standard Lipid Lowering by Atorvastatin on Diabetic Dyslipidemia: The DALI Study: a double-blind, randomized, placebo-controlled trial in patients with type 2 diabetes and diabetic dyslipidemia

The Diabetes Atorvastatin Lipid Intervention (DALI) Study is a randomized double-blind, placebo-controlled, multicenter study conducted in the Netherlands. Patients were recruited from outpatient clinics of the University Medical Centers of Leiden, Rotterdam, and Utrecht and surrounding community hospitals. The medical ethical committees of the three participating institutions approved the study, and written informed consent was obtained from all patients. The participants, aged 45–75 years, were male or female patients with type 2 diabetes with a duration of diabetes of at least 1 year and an HbA_1c level of 10% or lower. The main inclusion criteria were fasting total cholesterol level between 4.0 and 8.0 mmol/l and fasting triglyceride level between 1.5 and 6.0 mmol/l.

Patients were not included in the present study if they had a history of myocardial infarction, percutaneous transluminal coronary angioplasty, coronary artery bypass grafting, proven manifest coronary artery disease, severe or unstable angina pectoris (higher than grade II of the Canadian Cardiovascular Society), clinically manifest heart failure (higher than grade II New York Heart Association), or severe cardiac arrhythmia. Premenopausal women and patients with acute liver disease or hepatic dysfunction, impaired renal function (plasma creatinine >150 μmol/l), history of partial ileal bypass surgery, any surgical procedure or any systemic inflammatory disease within 3 months before randomization, malignancy, vasculitis, rheumatic arthritis, idiopathic lung fibrosis, ulcerative colitis, or Crohn’s disease were excluded. Patients who consumed more than four alcoholic drinks per day or who used systemic steroids, androgens, cyclosporin, other immunosuppressive drugs, erythromycin, or mibefradil were also excluded. When applicable, lipid-lowering drugs were withdrawn at least 8 weeks before the start of the run-in phase.

Patients who met the inclusion and exclusion criteria started with a placebo run-in period. If the lipid levels were still within the inclusion range after 2 weeks, patients were randomized to treatment with A10 A80 or placebo, administered once daily in the morning. Patients randomized to A80 started with 40 mg for 4 weeks, after which the dose was increased to 80 mg. The total treatment period was 30 weeks. Follow-up visits were scheduled at weeks 4, 10, 20, and 30; during these visits, adverse events were recorded, study medication was counted, blood pressure was measured, and fasting blood samples were collected for safety parameters and lipid profile. There were no changes in concurrent treatment, including hypoglycemic medication, during the study.

At baseline, a medical history was recorded and physical examination was performed. During the follow-up visits, patients were interviewed regarding possible adverse events.

Routine hematology and blood chemistry testing was performed after an overnight fast (12 h) at screening, at randomization, and at the end of the study. All laboratory measurements, except for the safety parameters, were performed at the Lipid Reference Laboratory, Rotterdam, the Netherlands. Standard plasma lipid variables (total cholesterol, HDL cholesterol, triglycerides, free fatty acids [FFAs], and apolipoprotein B [apoB]) were measured at each visit. Total cholesterol and triglyceride levels were measured by enzymatic colorimetric methods (CHOD-PAP and GPO-PAP, Boehringer Mannheim, Mannheim, Germany) on a Hitachi 911 analyzer (Boehringer Mannheim). HDL cholesterol was measured by a direct enzymatic HDL cholesterol method, based on PEG-modified enzymes method (Boehringer Mannheim) on a Hitachi 911 analyzer. LDL cholesterol was estimated by the Friedewald formula (21); apoB was determined on a Hitachi 917 analyzer, using immunoturbidimetric methods (Tina-quant apoB, cat. no.1551779; Boehringer Mannheim). Fasting FFAs were determined using an enzymatic colorimetric method (Wako, NEFA C, cat. no. 994-75409 D). The size of the LDL particles was measured at baseline and at the end of the study by polyacrylamide gradient gel electrophoresis (22). Standardization was achieved by inclusion of LDL samples with known size (donated by Dr. R.M. Krauss). Based on their size, LDL particles were divided into two classes: pattern A reflects the presence of predominantly large, buoyant LDL (>25.5 nm), and pattern B shows the presence of predominantly small, dense LDL particles (<25.5 nm). If both patterns were equally present, patients were classified as pattern AB (23). Lipoprotein lipase (LPL) activity was determined in plasma at baseline and at the end of the study after intravenous injection of heparin (50 IU/kg body wt) using an immunochemical technique. In a sample of 35 healthy subjects (mean age 55.5 years), the LPL activity measured with this technique was 147.4 ± 36.1 units/l (range 89–231).

Safety laboratory tests (creatine kinase, alanine transferase, and aspartate transferase) were performed at all follow-up visits at local laboratories. An increase of the alanine transferase or aspartate transferase levels to >3 times the upper limit of normal or an increase of creatine kinase >10 times the upper limit of normal, verified by repeat testing after 1 week, was considered clinically important and was reported as an adverse event.

All data were analyzed by intention to treat. Because patients were randomized, baseline values were not statistically tested between treatment groups. In subjects who did not complete the final visit, the “last observation carried forward” principle was applied. Mean differences between the study groups were analyzed using analysis of covariance, adjusted for baseline levels and study location. Intervention effects were also further adjusted for additional potential confounders, using analysis of covariance. If logarithmic transformation was applied to parameters with skewed distributions, the same results were obtained. Sample size of the study was based to prove a minimal reduction of 0.4 mmol/l in triglyceride levels (compared with placebo) or a difference of 0.4 mmol/l between the two treatment groups (SD 0.8, α = 0.05, power 85%). Analyses were performed using SPSS for Windows software (version 9.0; Chicago).

After a screening visit, 251 patients fulfilled the inclusion and exclusion criteria and entered the 2-week placebo run-in period. At baseline, 26 patients had normalized triglyceride levels (<1.5 mmol/l) and 8 patients refused to continue. Finally, 217 patients were randomized. The baseline characteristics of the study population are listed in Table 1. In all three groups, compliance with trial medication during the study was >95%. A total of 20 patients (9.2%) did not complete the study because of adverse events (n = 7), personal reasons (n = 10), protocol violation (n = 1), or loss to follow-up (n = 2). Personal reasons were defined as reasons not related to the study medication and were mostly due to inability to spend time for participation in the study.

Lipid and lipoprotein plasma values at baseline and after 30 weeks are shown in Table 2. In patients treated with A10, triglyceride levels were significantly lowered by 25% from 2.54 to 1.84 mmol/l (P < 0.001). Treatment with A80 resulted in a significant reduction of 35% from 2.85 to 1.78 mmol/l (P < 0.001). The difference in decrease of plasma triglyceride levels between the two intervention groups was not statistically significant (Fig. 1).

The effects on triglycerides were also analyzed in two strata of baseline plasma triglyceride levels to investigate whether the baseline levels of plasma triglycerides influenced the lipid-lowering effect of atorvastatin. The first group included all patients with baseline plasma triglyceride levels >2.3 mmol/l (n = 120), whereas the second group included 97 patients with baseline plasma triglyceride levels ≤2.3 mmol/l. In both groups, the same results were obtained. In patients with high baseline triglyceride levels, administration of A10 and A80 resulted in a triglyceride reduction of 29.6% and 39.4%, respectively (compared with placebo, both P < 0.001; A10 vs. A80, P > 0.5). The corresponding figures in patients with low baseline triglyceride levels were 23.4 and 27.9%, respectively (compared with placebo, both P < 0.001; A10 vs. A80, P > 0.4). Further adjustment for duration of diabetes and BMI did not change the results.

LDL cholesterol was dose-dependently improved to 2.2 mmol/l with administration of A10 (−40.8%, P < 0.001) and to 1.7 mmol/l with administration of A80 (−52.3%, P < 0.001). The difference between A10 and A80 was statistically significant (P < 0.001). Like LDL cholesterol, atorvastatin reduced apoB dose-dependently (Table 2). A10 and A80 increased HDL cholesterol significantly by 5–6%. Consequently, the total-cholesterol/HDL-cholesterol ratio improved, significantly more in patients treated with A80 (P < 0.005) (Table 2).

At baseline, relatively few patients had small, dense LDL particles. Patterns A, AB, and B were present in 60.5, 15.3, and 24.2%, respectively, which did not differ between the three intervention groups. There was no overall effect of atorvastatin on the LDL particle size. As a result, the number of patients with dense LDL particles (pattern B) did not differ at the end of the study between the intervention groups: 21.3, 21.4, and 22.2% in the placebo, A10, and A80 groups, respectively. Atorvastatin had no effect on post-heparin LPL activity. Further adjustment for duration of diabetes and BMI did not change the results.

The effects of treatment were compared with target values defined by the American Diabetes Association (24). In both atorvastatin treatment groups, >75% reached the triglyceride treatment goals (79.5% in A10 and 76.4% in A80, NS). LDL cholesterol treatment goals were reached in 71.2% of the patients treated with A10 and 84.7% of the patients treated with A80, compared with 11.1% in the placebo group. This difference was statistically significant between A10 and A80. HDL cholesterol treatment goals were reached in 35.6% of the A10 group and 44.4% of the patients treated with A80 (NS).

Adverse events are summarized in Table 3. There was no difference in the number of patients reporting adverse events between the three treatment groups. Twelve serious adverse events were reported. One patient receiving placebo (nonfatal myocardial infarction), one patient treated with A10 (benign neoplasm of skin), and one patient treated with A80 (depressive episode) reported a serious adverse event that was considered “possibly related” to the study drug. The other serious adverse events, not considered related to the study drug, included self-limiting gastroenteritis (n = 1) and trauma (n = 2) in patients receiving placebo; self-limiting gastroenteritis (n = 1), diabetes dysregulation (n = 1), and adenocarcinoma of the lung (n = 1) in patients using A10; and admission to the hospital for hip replacement surgery (n = 1), depressive episode (n = 1), and benign neoplasm of the skin (n = 1) in patients using A80.

Atorvastatin 80 mg was associated with a slight increase in HbA_1c concentration from 8.4 to 8.6% after 30 weeks (P = 0.06). In both placebo and A10 groups, a slight decrease in HbA_1c was observed after 30 weeks. After 30 weeks, HbA_1c in the A80 group differed significantly from either placebo or A10 (both P < 0.05). Fasting glucose and blood pressure remained stable during the study in all treatment groups.

Treatment with atorvastatin 10 and 80 mg resulted in significant reductions in plasma triglyceride and LDL cholesterol levels and an increase in HDL cholesterol levels in patients with type 2 diabetes. A80 had a significantly greater effect on cholesterol-related variables than A10. The side effects of A10 and A80 did not differ from placebo.

The lower limit of plasma triglyceride levels for inclusion was 1.5 mmol/l. Levels >1.5 mmol/l were associated with an increased risk for cardiovascular disease in diabetic subjects in the Paris Prospective Study and the Bezafibrate Infarction Prevention Registry (14,25). Moreover, 98% of the participants had dyslipidemia, classified according to the American Diabetes Association guidelines (24). The results of the analyses stratified by baseline plasma triglyceride levels indicate that the effects of treatment are independent of baseline plasma triglyceride levels, as is not seen for reduction in plasma LDL cholesterol by statins. Our results show that in patients with type 2 diabetes, high-dose atorvastatin therapy (80 mg) does not have a significant additional effect on reduction of triglycerides compared with a standard dose of 10 mg. The triglyceride-lowering efficacy of atorvastatin in our study is in agreement with previous small studies of administration of 80 mg atorvastatin in nondiabetic patients with primary dyslipidemia (18,19,26). Of the few studies of high doses of other HMG-CoA reductase inhibitors (mainly simvastatin) in patients with diabetes, only one double-blind study was published; that study showed a 15% reduction in triglycerides after administration of simvastatin 40 mg for 24 weeks in 42 patients with type 2 diabetes (20).

The mechanism by which atorvastatin lowers plasma triglyceride levels is not known. Diabetic hypertriglyceridemia is often ascribed to overproduction of VLDL triglycerides as well as impairment of triglyceride clearing due to decreased LPL activity (12,27). In the present study, LPL was in the normal range and was not affected by atorvastatin treatment. Therefore, it is likely that the hepatic triglyceride secretion is affected. Plasma FFAs are the main precursors for hepatic triglyceride synthesis and secretion. Because atorvastatin reduced plasma FFAs, it may be that hepatic triglyceride synthesis and secretion are attenuated.

The decrease in plasma triglycerides and the increase in HDL was not accompanied by an increase in LDL size. It should be noted that LDL size was relatively large. Moreover, based on epidemiological studies, it was found that the presence of small, dense LDL coincides with plasma triglyceride levels >1.6 mmol/l. To obtain a reversal of LDL size toward more buoyant particles, an even lower plasma triglyceride level might be necessary. Micronized fenofibrate has shown to improve LDL size in patients with type 2 diabetes who had higher fasting triglyceride and total cholesterol levels and a predominance (52%) of small, dense LDL at baseline compared with the patients in our study (28).

In the current study, total cholesterol, LDL cholesterol, and apoB levels were significantly reduced in a dose-dependent manner. This reduction is of the same magnitude as that found in studies of atorvastatin in nondiabetic hypercholesterolemic and hypertriglyceridemic patients (18,19). Compared with studies of simvastatin 40 mg in patients with type 2 diabetes, we found a larger reduction in total cholesterol level (39 vs. 30%) and LDL cholesterol level (52 vs. 24–42%) (20,29).

The significant increase from baseline in HDL cholesterol by atorvastatin was consistent with that reported in previous studies. Even small increases in HDL cholesterol may be clinically relevant because they contribute to the reduction of the total cholesterol:HDL cholesterol ratio. This ratio was one of the best predictors of future cardiovascular events in the Framingham study (30). Because treatment with A80 resulted in a significantly lower total cholesterol:HDL cholesterol ratio than A10, the high dose may have an additional protective effect on future cardiovascular events.

The number of serious adverse events and side effects were the same in the three intervention groups. The present study is the first study showing the safety of A80 for a longer treatment period in patients with type 2 diabetes who were using a variety of other medications concomitantly.

Possible limitations of the study are differences at baseline between the placebo and atorvastatin groups. These include BMI and duration of diabetes, which is not likely to influence the results because there is no relation between the duration of diabetes and the metabolism of triglycerides. Because visceral adipocytes are the main source for FFAs supplied to the liver as substrates for triglycerides, the waist-to-hip ratio and FFA levels are more important than BMI (31). Moreover, adjustment for duration of diabetes or BMI did not change the results.

Atorvastatin 80 mg induced a slight increase in HbA_1c levels, whereas in patients using A10 and placebo, the HbA_1c level decreased slightly. Other studies have shown inconsistent results of lipid-lowering therapy on glycemic control in patients with diabetes, and several clinical studies reported an increase in HbA_1c level with either fluvastatin or simvastatin (20,32,33,34).

In conclusion, administration of atorvastatin at doses of either 10 or 80 mg is effective in the treatment of diabetic dyslipidemia (elevated triglyceride, normal to elevated LDL cholesterol, low HDL cholesterol) in patients with type 2 diabetes. Both doses were well tolerated in our patient population.

Erasmus Medical Center Rotterdam, Department of Internal Medicine (I. Berk-Planken, N. Hoogerbrugge, H. Jansen); Erasmus University Rotterdam, Departments of Biochemistry and Clinical Chemistry (H. Jansen); Gaubius Laboratory TNO-PG, Leiden (H.M.G. Princen); Leiden University Medical Center (M.V. Huisman, M.A. van de Ree); University Medical Center Utrecht, Julius Center for General Practice and Patient Oriented Research (R.P. Stolk, F.V. van Venrooij); University Medical Center Utrecht, Division of Internal Medicine (J.D. Banga, G. Dallinga-Thie, F.V. van Venrooij).

The DALI Study was supported by a grant from Parke-Davis, the Netherlands.

We thank all of those who have contributed to the study as a patient, technician, or data manager. We also thank the referring physicians in the following hospitals: Albert Schweitzer Hospital (Zwijndrecht), Antonius Hospital (Nieuwegein), Diaconessenhuis (Leiden), Diakonessen Hospital (Utrecht), General Practitioner Center De Greev’ (Utrecht), Ikazia Hospital (Rotterdam), Lorenz Hospital (Zeist), Oudenrijn Hospital (Utrecht), Rijnland Hospital (Leiderdorp), and St. Franciscus Gasthuis (Rotterdam).