Pathophysiology/Complications ORIGINAL ARTICLE

The Effect of Therapy on Endothelial Function in Type 2 Diabetes Without Manifest

1 4 EDITH D. BEISHUIZEN, MD J. CAREL M. VAN DER VIJVER, MD, PHD oxide overproduction and thus decreased 1 1 JOUKE T. TAMSMA, MD, PHD A. EDO MEINDERS, MD, PHD NO availability (3). Regarding insulin, its 2 1 J. WOUTER JUKEMA, MD, PHD MENNO V. HUISMAN, MD, PHD 3 vasodilatory capacity is at least in part NO MARCEL A. VAN DE REE, MD, PHD dependent (4,5), thus explaining how in- sulin resistance might be related to endo- thelial dysfunction. Flow-mediated dilation (FMD) of the brachial is a noninvasive technique OBJECTIVE — Cardiovascular disease (CVD) is the most important cause of mortality in patients with type 2 diabetes and is preceded by endothelial dysfunction. Flow-mediated dilation for measuring endothelial function. FMD (FMD) is a noninvasive technique for measuring endothelial dysfunction. We aimed to deter- of the brachial artery has been shown to mine the effect of long-term statin therapy versus placebo on FMD in patients with type 2 be the result of -derived NO diabetes without manifest CVD. release (6) and is related to coronary va- soreactivity (7). FMD has proven to be RESEARCH DESIGN AND METHODS — A randomized, placebo-controlled, double- predictive for the presence of coronary ar- blind trial was performed with 250 type 2 diabetic patients. Patients were given 0.4 mg ceriv- tery disease (8,9), for future cardiovascu- astatin or placebo daily. In August 2001, when cerivastatin was withdrawn from the market, the lar events (10–12), and for postoperative 0.4 mg cerivastatin was replaced by 20 mg simvastatin, without deblinding the study. The primary end point was the change in FMD, measured by B-mode ultrasound, after 2 years. cardiovascular events (13) in high-risk populations. Improvement in FMD pre- RESULTS — Determinants of baseline FMD were diabetes duration, common carotid intima- dicts a favorable cardiovascular outcome media thickness, and brachial artery diameter. FMD at baseline was 1.51% in the placebo group in postmenopausal hypertensive women and 1.66% in the statin group and did not change significantly after 2 years. (2). However, in patients at lower risk, FMD was not independently associated CONCLUSIONS — The 2-year statin therapy had no effect on FMD in type 2 diabetes. with outcome (14). FMD is impaired in Statin-induced improvement of cardiovascular risk in patients with type 2 diabetes may be patients with type 2 diabetes with re- mediated through mechanisms other than increased availability. ported FMD values of 4.47–12.3% in Diabetes Care 28:1668–1674, 2005 control subjects vs. 2.96–6.1% in type 2 diabetic patients in cross-sectional studies (15–22). Hydroxymethylglutaryl-CoA reduc- ardiovascular disease (CVD) is the ulation, and diminishes vascular smooth tase inhibitors () have been shown most important cause of mortality in muscle cell proliferation and migration. to reverse endothelial dysfunction in hy- patients with type 2 diabetes (1). Type 2 diabetes is associated with endo- C percholesterolemic nondiabetic patients, Endothelial dysfunction precedes the de- thelial dysfunction; the underlying mech- velopment of atherosclerotic plaques and anisms are complex and related to possibly through upregulation of endo- is believed to be reversible (2). Nitric ox- hyperglycemia (sorbitol, hexosamine, thelial NO synthase expression (23–25), ide (NO) is a key molecule in this process. protein kinase C, and advanced glycation resulting in increased NO production. It modulates blood flow and vascular per- end product pathways) and insulin resis- Statins also inhibit superoxide produc- meability, limits inflammation and coag- tance, resulting in mitochondrial super- tion (25), thereby reducing NO break- down. The net effect is an increase in NO ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● availability, theoretically within days after From the 1Department of General Internal Medicine, Leiden University Medical Center, Leiden, the Neth- starting statin therapy. This may explain 2 erlands; the Department of Cardiology, Leiden University Medical Center, Leiden, the Netherlands; the the rapid improvement in endothelial 3Department of Internal Medicine, Diakonessenhuis, Utrecht, the Netherlands; and the 4Department of Internal Medicine, Leyenburg Hospital, the Hague, the Netherlands. dysfunction observed in several studies in Address correspondence and reprint requests to E.D. Beishuizen, Department of General Internal Med- nondiabetic subjects (26). In patients icine, C1-R41, Leiden University Medical Center, P.O. Box 9600, 2300 RC Leiden, Netherlands. E-mail: with type 2 diabetes, the results of studies [email protected]. with short-term statin therapy are, how- Received for publication 2 December 2004 and accepted in revised form 10 April 2005. Abbreviations: CCA, common carotid artery; CVD, cardiovascular disease; FMD, flow-mediated dila- ever, contradictory with respect to FMD. tion; IMT, intima-media thickness; NMD, -mediated dilation. We therefore conducted a randomized, A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion placebo-controlled trial to evaluate the ef- factors for many substances. fect of 2 years of statin therapy on endo- © 2005 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby thelial function in patients with type 2 marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. diabetes without CVD.

1668 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Beishuizen and Associates

RESEARCH DESIGN AND Table 1—Baseline characteristics METHODS — The study design and baseline characteristics of the original pa- Placebo Statin tient population have been described n 79 103 elsewhere (27). Briefly, 250 patients aged Male sex 38 (48) 52 (51) 30–80 years, without CVD and with type Age (years) 59 Ϯ 10 59 Ϯ 11 2 diabetes for Ն1 year, were included in Race this randomized, double-blind clinical White 60 (76) 72 (70) trial. Patients were given 0.4 mg ceriva- Indo-Asian 10 (13) 21 (20) statin (Bayer, Mijdrecht, the Netherlands) Other 9 (11) 10 (10) or placebo daily for 2 years. After the BMI (kg/m2) 31.2 Ϯ 6.0 30.5 Ϯ 5.4 withdrawal of cerivastatin from the mar- Waist-to-hip ratio 1.00 Ϯ 0.09 0.98 Ϯ 0.08 ket, the 0.4-mg cerivastatin dose was re- Current smoker 19 (24) 27 (26) placed by 20 mg simvastatin (Merck 46 (58) 49 (48) Sharp & Dome, Haarlem, the Nether- Diabetes duration (years) 9 Ϯ 88Ϯ 7 lands) without deblinding the study. Insulin use 45 (57) 51 (50) Only patients who completed the study A1C (%) 7.68 Ϯ 1.31 7.50 Ϯ 0.98 were included in the present analysis. Microalbuminuria* 12 (15) 21 (20) There were no significant differences in CCA IMT (mm) 0.780 Ϯ 0.129 0.763 Ϯ 0.124 demographic or lipid parameters between Ϯ Ͼ Ͼ the full cohort (n ϭ 250) and the patients Data are means SD or n (%). *Defined for men as 2.5 g/mol creatine and for women as 3.5 g/mol creatine. in this study (n ϭ 182), except for race, since more whites than nonwhites com- pleted the study (data not shown). Eligi- During the study, all measurements were every beat during a 5-min period. After ble patients gave their written informed performed by the same two certified ul- a 10-min rest, another 15 R-wave– consent. The study was performed at the trasonographers. triggered beats were stored. Subse- Leyenburg Hospital, the Hague. The study Fasting subjects were examined in the quently, two puffs of nitroglycerin (0.8 was approved by the hospital’s medical supine position. Heart rate was continu- mg) spray were given sublingually, upon ethics committee. ously monitored by three-lead electrocar- which R-wave–frozen images were again The primary end point of the study diogram. Mean common carotid intima- recorded for every beat during a 5-min was the change in FMD between baseline media thickness (IMT) was measured as period. and 24 months. Secondary end points reported earlier (27). Briefly, the left and Lumen diameter (D) was defined as were the change in absolute diameter right distal 1.0 cm of the common carotid the distance between the media- Ϫ (Dmax D), the time to peak (tmax), the , near and far walls, were exam- adventitia interfaces of far and near wall. change in nitroglycerin-mediated dilation ined longitudinally in the angle resulting Using an automated contour detection (NMD), and the FMD-to-NMD ratio. in an optimal and maximal IMT (while system, D was measured semiautomati- Comparisons between standard measure- avoiding plaques). For each segment, cally by placing a cursor on the media- ments for FMD at 1 min after cuff defla- three R-wave triggered images were adventitia interfaces. FMD was defined as tion and for NMD at 3, 4, or 5 min after stored. Mean IMT was measured when the percentage increase in brachial artery nitroglycerin administration and real possible over the entire 1 cm of the vessel diameter within 30–120 s after ischemia Ϫ maximum values obtained by beat-to- segment. Mean common carotid artery ([Dmax D]/D). NMD is defined as the beat analysis were analyzed as exploratory (CCA) IMT was obtained by averaging the percentage increase within 5 min after end points. mean IMTs of the far and near wall, both nitroglycerin. Patients returned to the study site left and right. Earlier studies in our institute re- after a 12-h fast at 3, 6, 12, 18, and 24 For FMD, the right arm was placed in ported reliability coefficients of 99, 99, months when blinded lipid and safety extension in the elbow, hand in supina- and 67% for baseline diameter, peak di- measurements (creatine kinase and ala- tion, with wrist and elbow supported by ameter, and FMD, respectively (28). In a nine aminotransferase) were performed. foam cushions. An optimal longitudinal recent report on variability of FMD (using Ultrasound measurements were per- image of the brachial artery at, or just a continuous method like we did) in type formed at baseline and 24 months. A above, the elbow was established and 2 diabetes, coefficients of variation for 2-year follow-up for clinical events was kept stable using a specially designed fix- baseline diameter, peak diameter, and performed for all 250 patients. ative. To obtain clearer images, a water FMD were 2.7, 2.5, and 29.7%, respec- bag was placed between the transducer tively (29). Ultrasound measurements and the skin. At baseline, 15 consecutive Ultrasound imaging was performed with R-wave–triggered beats were stored. A Laboratory investigations an Acuson Aspen scanner with a linear cuff placed just distally from the elbow All laboratory measurements were per- array 7.5 MHz probe. All images were re- was inflated to 50 mmHg above systolic formed at the Department of Clinical corded digitally for offline, blinded anal- blood pressure (up to a maximum of 230 Chemistry and Hematology of the Leyen- ysis by an independent core laboratory mmHg) for 4 min. After deflation, R- burg Hospital, according to International (Heart Core, Leiden, the Netherlands). wave–frozen images were recorded for Standards Organization 15189 standard

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1669 Effect of statins on FMD in type 2 diabetes procedures. Blood samples were collected cients. Comparison between beat-to-beat from the subjects after a 12-h fast. EDTA analysis and standard methods was per- value†

tubes were used for the determination of formed using the using Student’s paired t P HbA1c (A1C). Liver enzymes and lipids test and Bland Altman analysis (30). were measured in serum. A urine sample Analyses were performed using SPSS was collected for the determination of the 11.0 for Windows software. All analyses albumin-to-creatinine ratio. Serum or were two sided, with a level of signifi- plasma was isolated by centrifugation at cance of ␣ϭ0.05. 10 to 4) 0.37 0.04 to 0.11) 0.36 1.64 to 0.22) 0.13 0.08 to 0.08) 0.97 1,700g (2,900 rpm) for 5 min. 0.08 to 0.96) 0.10 Ϫ Ϫ Ϫ Ϫ Serum levels of total cholesterol and RESULTS — The characteristics of the Ϫ 3( triglycerides were measured by enzymatic study population are given in Table 1. No Ϫ methods on a Synchron LX20 analyzer statistical differences between the groups (Beckman Coulter, Brea, CA). LDL cho- were observed. Ϯ 0.26 0.04 ( 4.56 0.71 ( 0.69 0.00 ( 2.20 0.44 ( 0.10 0.02 (0.00–0.04) 0.10 lesterol was calculated according to the LDL cholesterol was 3.44 0.71 26 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Friedewald formula. If triglycerides were mmol/l at baseline and 2.58 0.95 Ϯ Ͼ4.5 mmol/l, LDL cholesterol was mea- mmol/l at 2 years (Ϫ25%, P Ͻ 0.001) in sured directly with the use of a reagent kit the statin group and 3.55 Ϯ 0.71 mmol/l (Genzyme Diagnostics). HDL cholesterol at baseline and 3.78 Ϯ 0.81 mmol/l at 2 levels were determined after dextran sul- years (8%, P ϭ 0.003) in the placebo 0.30 0.23 5.73 10.27 0.70 4.67 1.75 2.10 0.08 0.09 fate-magnesium precipitation of apoli- group (P Ͻ 0.001). HDL cholesterol was 28 61 Ϯ Ϯ Ϯ Ϯ Ϯ poprotein-B–containing lipoproteins. 1.23 Ϯ 0.39 mmol/l at baseline and Ϯ Creatine kinase and alanine aminotrans- 1.20 Ϯ 0.36 mmol/l at 2 years in the statin ferase were measured by an enzymatic group and 1.21 Ϯ 0.37 mmol/l at baseline rate method on a Synchron LX20 mul- and 1.22 Ϯ 0.38 mmol/l at 2 years in the tichannel chemistry analyzer, according placebo group. Triglycerides were 1.88 Ϯ to International Federation of Clinical 0.79 mmol/l at baseline and 1.72 Ϯ 1.22 value* Baseline 2 years Mean change

Chemistry methods. A1C was measured mmol/l at 2 years in the statin group and P by high-performance liquid chromatog- 1.82 Ϯ 0.97 mmol/l at baseline and raphy on a Variant II (BioRad). For the 1.60 Ϯ 1.38 mmol/l at 2 years in the pla- urine sample, a Jaffe´ rate method was cebo group. Changes in HDL cholesterol used for the measurement of creatinine on and triglycerides were not significantly a Synchron LX20 analyzer, whereas albu- different compared with baseline or the 11 to 8) 0.78 64 0.03 to 0.10) 0.31 0.19 0.87 to 0.94) 0.94 10.98 0.03 to 0.12) 0.22 4.67 0.50 to 0.66) 0.78 1.66 0.02 to 0.03) 0.61 0.08 Ϫ Ϫ Ϫ Ϫ Ϫ min was measured by rate nephelometry. placebo group. Average LDL cholesterol Ϫ

levels were higher after the switch to sim- 1(

Statistical analysis vastatin (2.34 mmol/l before vs. 2.56 Ϫ The number needed to detect a difference mmol/l after the switch, P Ͻ 0.001). in FMD of 2% after 2 years (expected SD Placebo Statin 0.22 0.04 ( 4.32 0.04 ( 0.58 0.05 ( 1.84 0.08 ( 0.09 0.01 ( 4%) with a power of 80% (␣ϭ0.05) FMD measurements 29 Ϯ Ϯ Ϯ Ϯ Ϯ was 63 patients in each group. The pri- Baseline FMD was not significantly differ- Ϯ mary treatment comparison was between ent between the groups. Baseline FMD in placebo and statin therapy in patients the group of 182 patients who completed completing the study, as on-treatment the study was not significantly different analysis. Changes from baseline within from baseline FMD in the dropouts (data 0.19 0.18 4.40 10.28 0.55 4.82 1.73 1.59 0.08 0.08 each treatment group were analyzed us- not shown). For the 182 patients who 30 64 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ ing Student’s paired t test. Comparisons completed the study, FMD in the placebo 79 103 65 of the effects between the treatment group was 1.51% at baseline and 1.59% Baseline 2 years Mean change 0.07 groups were performed using Student’s at 2 years (P ϭ 0.78); in the statin group, independent samples t test. Stepwise re- it was 1.66% at baseline and 2.10% at 2 gression techniques were used to investi- years (P ϭ 0.10) (Table 2). There was no gate the effect on baseline FMD and on significant difference between the change ) (mm) 4.77 changes in FMD of baseline characteris- in FMD in the statin group and the pla- D tics, carotid IMT, and duration of cerivas- cebo group (mean difference 0.36% [95% tatin versus simvastatin use. To test the CI Ϫ0.42 to 1.13%]; P ϭ 0.37). We per- equivalence of 0.4 mg cerivastatin and 20 formed an intention-to-treat analysis for mg simvastatin, LDL levels before and af- the whole group of 250 patients by using SD or means (95% CI). *For change within the placebo group; †for change within the statin group. Ϯ ter the switch to simvastatin were com- the method of “last observation carried (mm) D pared using Student’s paired t test. forward” for missing values: FMD in the Parameters for endothelial function Ϫ Correlation between changes in FMD and placebo group was 1.69% at baseline and (s) max

ϭ max FMD-to-NMD ratio 0.14 NMD (%) 10.24 D Brachial artery diameter ( t changes in lipid levels were evaluated by 1.75% at 2 years (P 0.78); in the statin FMD (%) 1.51 Secondary end point Primary end point n calculating Pearson’s correlation coeffi- group it was 1.65% at baseline and 2.02% Table 2— Data are means

1670 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Beishuizen and Associates vial;NMD available; P 3— Table cnmdse l 3)1 admzdcontrolled, Randomized 19 (31) al. et Economides eroje l 2)4 admzdcontrolled, Randomized 46 (28) al. et Venrooij rsn td 0 admzdcontrolled, Randomized 103 study Present eiloe l 1)3 rs-vr randomized Cross-over, 30 (16) al. et Ceriello hue l 2)2 ornoie ra A20N ivsai 03 447 . . S1. 13.3 14.5 NS 7.7 6.1 4.71 24 36 20–40 Simvastatin 10 Simvastatin NA NA 200 200 NA NA trial Nonrandomized controlled, Randomized 6 trial Nonrandomized 14 21 (33) al. et Tsunekawa (32) al. et Sheu (21) al. et Sheu (ref.) Author

au oprscagsi M ewe ttnadpaeo Etmtdfo gr;†o oprsno M tflo-p(FMD follow-up at FMD of comparison †for figure; from *Estimated placebo. and statin between FMD in changes compares value at 2 years (P ϭ 0.10). There was no sig- nificant difference between the change in FMD in the statin group and the placebo

nevninsuiso h feto ttn nFDi ainswt ye2daee ihu CVD without diabetes 2 type with patients in FMD on statins of effect the on studies Intervention group (mean difference 0.32% [Ϫ0.89 to

bl 0.26%]; P ϭ 0.28). There was also no sig- M tbsln;NMD baseline; at NMD , nificant difference between the changes in absolute increase in diameter after isch- treated Statin- group emia, t , NMD, and the FMD-to-NMD

( max 30 3Aovsai 05 047 .83.10 3.18 4.77 30 51 80 Atorvastatin 43 n Design ) 8 6 ratio. Determinants for baseline FMD were ϭϪ ϭ obebidtrial double-blind obebidtrial double-blind obebidtrial double-blind ln trial blind double- controlled, pnlbltrial open-label age (r 0.145; P 0.055), systolic blood pressure (r ϭϪ0.192; P ϭ 0.011), f M tfollow-up. at NMD , diabetes duration (r ϭϪ0.160; P ϭ 0.034), and baseline brachial artery diam- eter (r ϭϪ0.582; P Ͻ 0.001). Baseline CCA IMT as a continuous variable was not a determinant of baseline FMD. How- A50 NA e 5s20 s 15 Per 010sbeat-to- s 30–120 59 beat-to- s 45–90 t6 5 oer eiattn01 . A40 8.5* 4.0* NA 0.5 2 0.15 Cerivastatin Forearm 250 s 60 At ever, when split into quartiles, FMD at etanalysis beat etanalysis beat baseline was significantly lower in the method

D highest CCA IMT quartile compared with max the three lower CCA IMT quartiles (0.94 vs. 1.77%; P ϭ 0.006). When in- cluded into a regression model, only the

50 highest quartile CCA IMT, diabetes dura- mH)Cuff (mmHg) Inflation 0 oer ivsai 0305N . 7.3 4.8 NA 0.5 3 40 Simvastatin Forearm 300 Ͼ Ͼ Ͼ tion, and baseline brachial artery diame- BP BP BP ter remained significant determinants and sys sys sys together explained 11% of the variance in oer trattn2 11 . . . .7 2511.9 12.5 0.07† 5.6 4.2 3.7 14 3.20 3.41 4.89 41 30 0.4, Cerivastatin Forearm 20 Atorvastatin Forearm 46 10 Atorvastatin Forearm baseline FMD. Baseline FMD and changes in FMD were not correlated with LDL cholesterol or any other lipid parameter. Baseline ivsai 20 simvastatin

ttndose Statin FMD and changes in FMD were also not

(mg) related to sex, race, insulin use, antihy- pertensive medication, A1C, anthropo- metric parameters, and smoking habits. Changes in FMD were not related to base- Ͻ Ͼ line CCA IMT. Changes in FMD were neg- . mll1 A5613.6 5.6 NA 12 mmol/l 2.1 . mll1 A4482013 ANA NA 0.173† 8.2 4.4 NA 12 mmol/l 2.1 decrease atively correlated to changes in CCA IMT LDL (%) 81 A499.2 4.9 NA 14 28 11 A40 8.5* 4.0* NA 14 21 51446 .621 .71.810.27 10.98 0.37 2.10 1.66 4.67 104 25

f ϭϪ ϭ ihFDa aeie(FMD baseline at FMD with ) in the placebo group (r 0.259; P 0.029). Thus, an increase in CCA IMT in the placebo group during follow-up was Follow-up

(weeks) associated with a decrease in FMD. This could not be observed in the statin group. The effect of the two statins used was analyzed by correcting the change in FMD D for duration of cerivastatin treatment

FMD (range 6–23 months). This did not bl

.BP ). change the results. bl Dmax, FMD, and NMD as determined sys FMD

ytlcbodpesr;N,not NA, pressure; blood systolic , by beat-to-beat analysis were significantly higher compared with values obtained at f Ͻ Ͻ Ͼ Ͻ Ͻ Ͻ Ͼ fixed times. The extent of these differ- P .5N NA NA 0.05 .565 7.5* 6.5* 0.05 . .16.59 6.01 0.8 .0 ANA NA 0.001 .2†N NA NA 0.028† .570 7.4* 7.0* 0.05 . .06.87 6.80 0.8 au NMD value ences was not related to absolute values. However, standard deviations of the base- line values and 95% CIs of the changes after 2 years were not lower in the beat-

bl to-beat analysis (data not shown). When

NMD the analysis was repeated with fixed times values as an outcome measure, results did f not change.

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1671 Effect of statins on FMD in type 2 diabetes

CONCLUSIONS — Patients with (FMD 2.96%) (22) and to a cross- wall changes may not have an important type 2 diabetes have a high risk of cardio- sectional study (FMD 1.9%) in diabetic impact on FMD in this population. vascular events, and endothelial dysfunc- patients with microalbuminuria (35). In Therefore, we conclude that statin- tion can be viewed as an early sign of our study, with long-term statin therapy, induced cardiovascular risk reduction in . No long-term, blinded, more patients per treatment arm were in- type 2 diabetes is probably not mediated placebo-controlled trials on the effect of cluded than in any other study and we through improved NO availability. Other statin therapy on endothelial function in used the beat-to-beat analysis for optimal mechanisms, such as suppression of in- type 2 diabetes have been reported. The precision. Moreover, given the CI of the flammatory response, improvement of present study shows that in our patient mean difference in FMD change between plaque stability, and reduced thrombo- group endothelial dysfunction is not placebo and statin, there is a 95% cer- genic potential of the endothelial cell (42) reversible with medium-dose statin tainty that there is no treatment effect are possible alternative explanations for therapy. greater than an absolute difference in the beneficial effect of statin therapy in Several earlier studies have been FMD of 1.13%. diabetic subjects. Our results imply that performed to evaluate the effect of statin There is much debate whether statin- in patients with type 2 diabetes, FMD is therapy on FMD in patients with type 2 induced improvement of endothelial not a proper intermediate end point for diabetes. In Table 3, these studies are function is mediated through a change in statin studies. Until now, data on the summarized. In a randomized study, van lipid profile, so-called pleiotropic effects, prognostic value of FMD for future car- Venrooij et al. (28) did not find an effect or both. In the present study, we found no diovascular events in patients with type 2 of 30 weeks of atorvastatin (10 or 80 relation between (changes in) lipid profile diabetes are lacking. mg) versus placebo on FMD. Ceriello et and (changes in) FMD. There is also much We feel that the present study adds al. (16) reported an improvement in discussion about possible differing pleio- strongly to the evidence that medium- FMD after simvastatin 40 mg given for tropic effects between the different statins dose statin therapy has no effect on FMD only 3–6 days. Recently, Economides (36). In our study, because of unforeseen in type 2 diabetic subjects without mani- et al. (31) reported a nonsignificant im- circumstances, two different statins were fest CVD. FMD is impaired in diabetes of provement in FMD after 12 weeks of used and we found no difference in effect longer duration and with higher carotid atorvastatin 20 mg. The other studies are on FMD between the statins. IMT. Beat-to-beat analysis gives a more not randomized trials or open-label trials Until recently, the value of statin ther- precise estimate of Dmax but did not lead (21,32,33). apy in diabetic patients was not clear in to lower CIs in the present study. In pa- There are several explanations for the the setting of primary prevention. How- tients with type 2 diabetes, statin-induced discrepancy in the results of these studies. ever, a recent meta-analysis (37) and the improvement of cardiovascular risk may All studies have included patients without Collaborative Atorvastatin Diabetes Study be mediated through mechanisms other CVD, and age, diabetes duration, and trial (38), in which diabetic patients with than increased NO availability. A1C seem quite comparable. However, at least one additional cardiovascular risk FMD methodology was not always clearly factor were included, reported marked Acknowledgments— Bayer B.V. (Mijdrecht, defined. First, the way Dmax is determined cardiovascular risk reduction. We also is critical. Simply measuring once 1 min found a reduced cardiovascular event rate the Netherlands) supplied the study medica- tion and supported the study until cerivastatin after cuff deflation or measuring every in the statin-treated group in the present was withdrawn from the market (8 August 15 s can result in underestimation of study population as reported before (27). 2001). Merck Sharp & Dome (Haarlem, the FMD; however, with outliers, it can result Event reduction, on the one hand, and no Netherlands) supplied the simvastatin/ in overestimation of FMD. Beat-to-beat difference in FMD, on the other, imply placebo medication for the remaining study analysis results in a more precise estimate that statin-induced risk reduction in type period. J.W.J. is an established clinical inves- of Dmax but did not lead to lower CIs in 2 diabetes is either not mediated through tigator of the Netherlands Heart Foundation the present study. restoration of endothelial dependent dila- 2001 D032. Second, several authors do not men- tion or that FMD is not a proper test to We thank M. Spaans for his invaluable role tion their baseline lumen diameters, detect changes in endothelial dysfunction as research nurse, A. van Weert and analysts which is an established determinant of in type 2 diabetic patients. The latter pos- from Heart Core for their skilled ultrasound analysis, and all the patients for their partici- FMD (34). If lumen diameter is defined as sibility is less likely because forearm pation in the study. the distance between the intima-lumen blood flow measured by venous occlusion interfaces instead of media-adventitia in- plethysmography, another parameter for terfaces of the vessel wall, lumen diameter endothelial function, also showed no im- References decreases and FMD increases. provement after statin therapy in diabetic 1. Stamler J, Vaccaro O, Neaton JD, Went- Third, some authors do not mention patients (39,40). Other interventions in worth D: Diabetes, other risk factors, and whether the cuff is placed around the patients with recent-onset type 2 diabetes 12-yr cardiovascular mortality for men forearm or upper arm. This is a critical have resulted in an improvement in FMD screened in the Multiple Risk Factor In- issue because the latter location results in (41), indicating that FMD is not simply tervention Trial. Diabetes Care 16:434– 444, 1993 a higher FMD. Baseline FMD in our pa- irreversibly impaired in type 2 diabetes. 2. Modena MG, Bonetti L, Coppi F, Bursi F, tients was low in comparison to the dia- Moreover, diabetes duration, carotid Rossi R: Prognostic role of reversible betic populations in the intervention IMT, and vessel diameter together only endothelial dysfunction in hypertensive studies mentioned, but comparable to explain 11% of the variance in FMD, in- postmenopausal women. J Am Coll Car- another Dutch study, the Hoorn study dicating that irreversible diabetic vessel diol 40:505–510, 2002

1672 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 Beishuizen and Associates

3. Creager MA, Luscher TF, Cosentino F, Watkins MT, Menzoian JO, Vita JA: Risk 23. Kaesemeyer WH, Caldwell RB, Huang J, Beckman JA: Diabetes and vascular dis- stratification for postoperative cardiovas- Caldwell RW: Pravastatin sodium acti- ease: pathophysiology, clinical conse- cular events via noninvasive assessment of vates endothelial quences, and medical therapy: Part I. endothelial function: a prospective study. independent of its cholesterol-lowering Circulation 108:1527–1532, 2003 Circulation 105:1567–1572, 2002 actions. J Am Coll Cardiol 33:234–241, 4. Steinberg HO, Brechtel G, Johnson A, 14. Fathi R, Haluska B, Isbel N, Short L, Mar- 1999 Fineberg N, Baron AD: Insulin-mediated wick TH: The relative importance of vas- 24. Laufs U, La F, V, Plutzky J, Liao JK: Up- skeletal muscle is nitric oxide cular structure and function in predicting regulation of endothelial nitric oxide syn- dependent: a novel action of insulin to cardiovascular events. J Am Coll Cardiol thase by HMG CoA reductase inhibitors. increase nitric oxide release. J Clin Invest 43:616–623, 2004 Circulation 97:1129–1135, 1998 94:1172–1179, 1994 15. Tan KC, Chow WS, Ai VH, Metz C, Bucala 25. Wagner AH, Kohler T, Ruckschloss U, 5. Baron AD: Vascular reactivity. Am J Car- R, Lam KS: Advanced glycation end prod- Just I, Hecker M: Improvement of nitric diol 84:25J–27J, 1999 ucts and endothelial dysfunction in type 2 oxide-dependent vasodilatation by HMG- 6. Joannides R, Haefeli WE, Linder L, Rich- diabetes. Diabetes Care 25:1055–1059, CoA reductase inhibitors through atten- ard V, Bakkali EH, Thuillez C, Luscher 2002 uation of endothelial superoxide anion TF: Nitric oxide is responsible for flow- 16. Ceriello A, Taboga C, Tonutti L, Qua- formation. Arterioscler Thromb Vasc Biol dependent dilatation of human peripheral gliaro L, Piconi L, Bais B, Da Ros R, Motz 20:61–69, 2000 conduit arteries in vivo. Circulation 91: E: Evidence for an independent and 26. Vogel RA, Corretti MC, Plotnick GD: 1314–1319, 1995 cumulative effect of postprandial hyper- Changes in flow-mediated brachial artery 7. Anderson TJ, Uehata A, Gerhard MD, triglyceridemia and hyperglycemia on en- vasoactivity with lowering of desirable Meredith IT, Knab S, Delagrange D, dothelial dysfunction and cholesterol levels in healthy middle-aged Lieberman EH, Ganz P, Creager MA, generation: effects of short- and long-term men. Am J Cardiol 77:37–40, 1996 Yeung AC: Close relation of endothelial simvastatin treatment. Circulation 106: 27. Beishuizen ED, van de Ree MA, Jukema function in the human coronary and pe- 1211–1218, 2002 JW, Tamsma JT, van der Vijver JC, ripheral circulations. J Am Coll Cardiol 26: 17. Kawano H, Motoyama T, Hirashima O, Meinders AE, Putter H, Huisman MV: 1235–1241, 1995 Hirai N, Miyao Y, Sakamoto T, Kugiyama Two-year statin therapy does not alter the 8. Schroeder S, Enderle MD, Ossen R, Meis- K, Ogawa H, Yasue H: Hyperglycemia progression of intima-media thickness in ner C, Baumbach A, Pfohl M, Herdeg C, rapidly suppresses flow-mediated endo- patients with type 2 diabetes without Oberhoff M, Haering HU, Karsch KR: thelium-dependent vasodilation of bra- manifest cardiovascular disease. Diabetes Noninvasive determination of endotheli- chial artery. J Am Coll Cardiol 34:146– Care 27:2887–2892, 2004 um-mediated vasodilation as a screening 154, 1999 28. van Venrooij FV, van de Ree MA, Bots ML, test for : pilot 18. Goodfellow J, Ramsey MW, Luddington Stolk RP, Huisman MV, Banga JD: Aggres- study to assess the predictive value in LA, Jones CJ, Coates PA, Dunstan F, Lewis sive lipid lowering does not improve en- comparison with angina pectoris, exercise MJ, Owens DR, Henderson AH: Endo- dothelial function in type 2 diabetes: the electrocardiography, and myocardial per- thelium and inelastic arteries: an early Diabetes Atorvastatin Lipid Interven- fusion imaging. Am Heart J 138:731–739, marker of vascular dysfunction in non- tion (DALI) Study: a randomized, double- 1999 insulin dependent diabetes. BMJ 312: blind, placebo-controlled trial. Diabetes 9. Teragawa H, Kato M, Kurokawa J, Yama- 744–745, 1996 Care 25:1211–1216, 2002 gata T, Matsuura H, Chayama K: Use- 19. Balletshofer BM, Rittig K, Enderle MD, 29. West SG, Wagner P, Schoemer SL, fulness of flow-mediated dilation of the Volk A, Maerker E, Jacob S, Matthaei S, Hecker KD, Hurston KL, Likos KA, Bo- brachial artery and/or the intima-media Rett K, Haring HU: Endothelial dysfunc- seska L, Ulbrecht J, Hinderliter AL: Bio- thickness of the carotid artery in predict- tion is detectable in young normotensive logical correlates of day-to-day variation ing coronary narrowing in patients sus- first-degree relatives of subjects with type in flow-mediated dilation in individuals pected of having coronary artery disease. 2 diabetes in association with insulin with type 2 diabetes: a study of test-retest Am J Cardiol 88:1147–1151, 2001 resistance. Circulation 101:1780–1784, reliability. Diabetologia 47:1625–1631, 10. Neunteufl T, Heher S, Katzenschlager R, 2000 2004 Wolfl G, Kostner K, Maurer G, Weidinger 20. Tan KC, Ai VH, Chow WS, Chau MT, Le- 30. Bland JM, Altman DG: Statistical methods F: Late prognostic value of flow-mediated ong L, Lam KS: Influence of low density for assessing agreement between two dilation in the brachial artery of patients lipoprotein (LDL) subfraction profile and methods of clinical measurement. Lancet with chest pain. Am J Cardiol 86:207–210, LDL oxidation on endothelium-depen- 327:307–310, 1986 2000 dent and independent vasodilation in 31. Economides PA, Caselli A, Tiani E, Kha- 11. Chan SY, Mancini GB, Kuramoto L, patients with type 2 diabetes. J Clin En- odhiar L, Horton ES, Veves A: The effects Schulzer M, Frohlich J, Ignaszewski A: docrinol Metab 84:3212–3216, 1999 of atorvastatin on endothelial function in The prognostic importance of endothelial 21. Sheu WH, Juang BL, Chen YT, Lee WJ: diabetic patients and subjects at risk for dysfunction and carotid atheroma burden Endothelial dysfunction is not reversed type 2 diabetes. J Clin Endocrinol Metab in patients with coronary artery disease. by simvastatin treatment in type 2 dia- 89:740–747, 2004 J Am Coll Cardiol 42:1037–1043, 2003 betic patients with . 32. Sheu WH, Chen YT, Lee WJ: Improve- 12. Gokce N, Keaney JF Jr, Hunter LM, Diabetes Care 22:1224–1225, 1999 ment in endothelial dysfunction with LDL Watkins MT, Nedeljkovic ZS, Menzoian 22. Henry RM, Ferreira I, Kostense PJ, Dekker cholesterol level Ͻ80 mg/dl in type 2 di- JO, Vita JA: Predictive value of noninva- JM, Nijpels G, Heine RJ, Kamp O, Bouter abetic patients. Diabetes Care 24:1499– sively determined endothelial dysfunc- LM, Stehouwer CD: Type 2 diabetes is as- 1501, 2001 tion for long-term cardiovascular events sociated with impaired endothelium- 33. Tsunekawa T, Hayashi T, Kano H, Sumi in patients with peripheral vascular dis- dependent, flow-mediated dilation, but D, Matsui-Hirai H, Thakur NK, Egashira ease. J Am Coll Cardiol 41:1769–1775, impaired glucose metabolism is not: the K, Iguchi A: Cerivastatin, a hydroxymeth- 2003 Hoorn Study. Atherosclerosis 174:49–56, ylglutaryl coenzyme a reductase inhibitor, 13. Gokce N, Keaney JF Jr, Hunter LM, 2004 improves endothelial function in elderly

DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005 1673 Effect of statins on FMD in type 2 diabetes

diabetic patients within 3 days. Circula- 294–299, 2004 Impaired endothelium-dependent vaso- tion 104:376–379, 2001 36. Sakabe K, Fukuda N, Wakayama K, Nada dilation in type 2 diabetes mellitus and 34. Corretti MC, Anderson TJ, Benjamin EJ, T, Shinohara H, Tamura Y: Time course the lack of effect of simvastatin. Cardio- Celermajer D, Charbonneau F, Creager differences for statin-induced pleiotropic vasc Res 52:299–305, 2001 MA, Deanfield J, Drexler H, Gerhard-Her- effects in hypercholesterolemic patients. 40. van Etten RW, de Koning EJ, Honing ML, man M, Herrington D, Vallance P, Vita J, Int J Cardiol 94:111–117, 2004 Stroes ES, Gaillard CA, Rabelink TJ: In- Vogel R: Guidelines for the ultrasound 37. Vijan S, Hayward RA: Pharmacologic lip- tensive lipid lowering by statin therapy assessment of endothelial-dependent id-lowering therapy in type 2 diabetes does not improve vasoreactivity in pa- flow-mediated vasodilation of the bra- mellitus: background paper for the Amer- tients with type 2 diabetes. Arterioscler ican College of Physicians. Ann Intern Med chial artery: a report of the International Thromb Vasc Biol 22:799–804, 2002 Brachial Artery Reactivity Task Force. 140:650–658, 2004 41. Caballero AE, Saouaf R, Lim SC, Hamdy J Am Coll Cardiol 39:257–265, 2002 38. Colhoun HM, Betteridge DJ, Durrington 35. Papaioannou GI, Seip RL, Grey NJ, Katten PN, Hitman GA, Neil HA, Livingstone SJ, O, Abou-Elenin K, O’Connor C, Logerfo D, Taylor A, Inzucchi SE, Young LH, Thomason MJ, Mackness MI, Charlton- FW, Horton ES, Veves A: The effects of Chyun DA, Davey JA, Wackers FJ, Iskan- Menys V, Fuller JH: Primary prevention of troglitazone, an insulin-sensitizing agent, drian AE, Ratner RE, Robinson EC, Caro- cardiovascular disease with atorvastatin on the endothelial function in early and lan S, Engel S, Heller GV: Brachial artery in type 2 diabetes in the Collaborative late type 2 diabetes: a placebo-controlled reactivity in asymptomatic patients with Atorvastatin Diabetes Study (CARDS): randomized clinical trial. Metabolism 52: type 2 diabetes mellitus and microalbu- multicentre randomised placebo-con- 173–180, 2003 minuria (from the Detection of Ischemia trolled trial. Lancet 364:685–696, 2004 42. Wolfrum S, Jensen KS, Liao JK: Endothe- in Asymptomatic Diabetics-brachial ar- 39. van de Ree MA, Huisman MV, de Man FH, lium-dependent effects of statins. Arterio- tery reactivity study). Am J Cardiol 94: van der Vijver JC, Meinders AE, Blauw GJ: scler Thromb Vasc Biol 23:729–736, 2003

1674 DIABETES CARE, VOLUME 28, NUMBER 7, JULY 2005