Author + information
- Received December 13, 2007
- Revision received March 7, 2008
- Accepted March 11, 2008
- Published online July 22, 2008.
- Stephen J. Nicholls, MBBS, PhD⁎,†,⁎ (, )
- E. Murat Tuzcu, MD⁎,
- Srinivasa Kalidindi, MD, MPH⁎,
- Kathy Wolski, MPH⁎,
- Keon-W. Moon, MD⁎,
- Ilke Sipahi, MD⁎,
- Paul Schoenhagen, MD⁎ and
- Steven E. Nissen, MD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Stephen J. Nicholls, Department of Cardiovascular Medicine, Mail Code JJ-65, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, Ohio 44195.
Objectives Our goal was to characterize coronary atherosclerosis progression and arterial remodeling in diabetic patients.
Background The mechanisms that underlie adverse cardiovascular outcomes in diabetic patients have not been well characterized.
Methods A systematic analysis was performed in 2,237 subjects in randomized controlled studies of atherosclerosis progression. The pattern of arterial remodeling, extent of coronary atherosclerosis, and disease progression was compared in subjects with and without diabetes.
Results In association with more risk factors, diabetic patients demonstrated a greater percent atheroma volume (PAV) (40.2 ± 0.9% vs. 37.5 ± 0.8%, p < 0.0001) and total atheroma volume (TAV) (199.4 ± 7.9 mm3 vs. 189.4 ± 7.1 mm3, p = 0.03) on multivariate analysis. A stronger correlation was observed between PAV and glycated hemoglobin (r = 0.22, p = 0.0003) than fasting glucose (r = 0.09, p < 0.0001), although the difference just failed to meet statistical significance after controlling for study. Diabetic patients exhibited a smaller lumen (291.1 ± 104.8 mm3 vs. 306.5 ± 108.2 mm3, p = 0.005) but no difference in external elastic membrane (494.9 ± 166.9 mm3 vs. 498.8 ± 167.2 mm3, p = 0.61) volumes. More rapid progression of PAV (0.6 ± 0.4% vs. 0.05 ± 0.3%, p = 0.0001) and TAV (−0.6 ± 2.5 mm3 vs. −2.7 ± 2.4 mm3, p = 0.03) was observed in diabetic patients on multivariate analysis. Smaller external elastic membrane (482.5 ± 160.7 mm3 vs. 519.9 ± 166.9 mm3, p = 0.03) and lumen (276.0 ± 100.3 mm3 vs. 310.1 ± 105.6 mm3, p = 0.001) volumes were observed in diabetic patients treated with insulin despite the presence of a similar TAV (206.5 ± 88.6 mm3 vs. 209.9 ± 90.2 mm3, p = 0.84). Intensive low-density lipoprotein cholesterol lowering in patients improved the rate of plaque progression, but only to the level observed in nondiabetic patients with suboptimal lipid control.
Conclusions Diabetes is accompanied by more extensive atherosclerosis and inadequate compensatory remodeling. Accelerated plaque progression, despite use of medical therapies, supports the need to develop new antiatherosclerotic strategies in diabetic patients.
The global spread of diabetes mellitus is a major factor contributing to the prediction that cardiovascular disease will become the leading cause of mortality worldwide by 2020 (1). Patients with diabetes have a markedly increased incidence of adverse cardiovascular events (2–6) and less favorable outcomes from myocardial infarction (7–10) or after coronary interventions (11,12). In addition to concomitant hypertension and dyslipidemia, increasing evidence suggests that impaired glycemic homeostasis has a direct influence on the formation and propagation of atherosclerotic plaque (13). This is likely to underscore the observation that even in the absence of ischemic symptoms the presence of diabetes confers a prospective risk of clinical events comparable to that observed in nondiabetic survivors of myocardial infarction (14). As a result, prevention and treatment of diabetes is a major component of strategies designed to reduce cardiovascular risk.
Elucidating the factors that promote cardiovascular disease in diabetes is critical for the development of new therapeutic approaches. The prevalence of hyperglycemia, hypertension, and dyslipidemia, in association with systemic inflammation and oxidative stress, accelerates the formation and propagation of atherosclerotic plaque (13). This underlies observations from small clinical (15–17) and necropsy (18,19) studies that diabetes is characterized by diffuse atherosclerosis, with a predilection for involvement of distal segments in relatively small vessels. However, no systematic assessment of the pattern of coronary atherosclerosis and associated arterial wall remodeling has been performed in a large cohort of diabetic patients by imaging of the entire thickness of the coronary artery wall.
Intravascular ultrasound (IVUS) permits evaluation of the influence of clinical characteristics on changes in coronary atheroma volume (20,21). More recently, IVUS has been employed to assess the impact of medical therapies on the natural history of plaque progression (22–26). The current study investigated the extent and progression of atherosclerosis and associated arterial wall remodeling in a large number of diabetic patients with coronary artery disease who underwent serial evaluation by IVUS.
Subjects participated in the REVERSAL (Reversal of Atherosclerosis with Aggressive Lipid Lowering) (22), CAMELOT (Comparison of Amlodipine versus Enalapril to Limit Occurrences of Thrombosis) (23), ACTIVATE (Acyl:Cholesterol Acyltransferase Intravascular Atherosclerosis Treatment Evaluation) (24), ASTEROID (A Study to Evaluate the Effect of Rosuvastatin on Intravascular Ultrasound-Derived Coronary Atheroma Burden) (25), and ILLUSTRATE (Investigation of Lipid Level Management Using Coronary Ultrasound to Assess Reduction of Atherosclerosis by Cholesteryl Ester Transfer Protein Inhibition and High-Density Lipoprotein Elevation) (26) studies. These were clinical trials that employed serial IVUS examination to assess the impact of intensive lipid lowering, antihypertensive therapy, experimental acyl:cholesterol acyltransferase inhibition, or cholesterol ester transfer protein inhibition on the progression of coronary atherosclerosis. All patients were required to have coronary artery disease, defined as having at least 1 lumen narrowing >20% in a major epicardial coronary artery on a diagnostic coronary angiogram performed for a clinical indication.
Acquisition and analysis of IVUS images
The acquisition and analysis of ultrasonic images have been described in detail previously (22–26). In brief, after anticoagulation and administration of intracoronary nitroglycerin, an imaging catheter containing a high-frequency ultrasound transducer (30 to 40 MHz) was inserted as far as distally possible within a coronary artery. The target vessel for imaging was required to have a segment of at least 30 mm in length that contained no lumen narrowing >50%, had not undergone previous revascularization, and was not considered to be the culprit vessel for a prior myocardial infarction. Continuous ultrasonic imaging was acquired during withdrawal of the catheter through the segment of artery at a constant rate of 0.5 mm/s. Images were stored on videotape and subsequently digitized for analysis in a single core laboratory by individuals who were blinded to the clinical characteristics and treatment status of the patients.
Matching arterial segments were defined from the images acquired at the baseline and follow-up studies on the basis of the anatomic location of proximal and distal side branches (fiduciary points). Images spaced precisely 1 mm apart in the segment of interest were selected for analysis. The leading edge of the lumen and the external elastic membrane (EEM) were defined by manual planimetry. The plaque area was defined as the difference in area occupied by the lumen and EEM borders. The total atheroma volume (TAV) was calculated by summation of the plaque area calculated for each measured image and subsequently normalized to account for differences in segment length between subjects: The percent atheroma volume (PAV) was also calculated by determination of the proportion of vessel wall volume occupied by atherosclerotic plaque: Volumes occupied by the lumen and EEM were similarly calculated by summation of their respective areas in each measured image and subsequently normalized to account for differences in segment length between subjects.
All analyses were performed using SAS version 8.2 (SAS Institute, Cary, North Carolina). Results are expressed as mean ± SD for continuous variables and percentage for categorical variables. Continuous variables that are not normally distributed (triglycerides, C-reactive protein [CRP]) are expressed as median (interquartile range). Comparisons between diabetic and nondiabetic groups were performed using a random effects model to account for any potential differences between each individual study that was pooled for the current analysis (Cochrane's Q statistic 8.9, p = 0.06 and I2 = 55%, indicating moderate heterogeneity). Changes in atheroma burden were expressed as the least square mean ± SEM after controlling for baseline values. The relationships between biochemical parameters and measures of atheroma burden and between changes in vessel dimensions were determined by calculation of the Pearson correlation coefficient. A p value <0.05 was considered statistically significant.
Clinical characteristics are summarized in Table 1. Diabetic subjects were more likely to be women (32.7% vs. 26.3%, p = 0.008) and African American (7.9% vs. 4.0%, p = 0.001), had a greater body mass index (32.4 ± 6.3 kg/m2 vs. 29.7 ± 5.2 kg/m2, p < 0.001), and had a greater prevalence of hypertension (87.5% vs. 72.2%, p < 0.001) and hyperlipidemia (74.3% vs. 67.1%, p = 0.004). Subjects with diabetes were more likely to meet the criteria for the metabolic syndrome (77.1% vs. 41.2%, p < 0.001). Diabetic subjects were more likely to have undergone a previous percutaneous coronary intervention (52.7% vs. 43.2%, p = 0.001), were more likely to be treated with an angiotensin-converting enzyme inhibitor (65.1% vs. 47.6%, p < 0.001), and were less likely to receive aspirin (91.8% vs. 94.8%, p = 0.02). Among diabetic subjects, 37.0% were treated with insulin and 69.5% received an oral hypoglycemic agent.
Risk factor control at baseline and follow-up are summarized in Table 2. Predictably, diabetic subjects had higher levels of fasting plasma glucose (148.4 ± 60.3 mg/dl vs. 96.0 ± 16.9 mg/dl, p < 0.001) and glycated hemoglobin (7.4 ± 1.4% vs. 5.7 ± 0.6%, p < 0.001). Diabetic subjects had lower levels of low-density lipoprotein (LDL) cholesterol (102.7 ± 38.2 mg/dl vs. 110.9 ± 38.0 mg/dl, p < 0.001) and high-density lipoprotein (HDL) cholesterol (41.4 ± 10.3 mg/dl vs. 44.0 ± 11.8 mg/dl, p < 0.001) and higher levels of triglycerides (159.3 mg/dl vs. 139.0 mg/dl, p < 0.001), CRP (2.8 mg/l vs. 2.2 mg/l, p = 0.006), and systolic blood pressure (129.6 ± 17.1 mm Hg vs. 126.1 ± 16.0 mm Hg, p < 0.001) at baseline. During the course of the studies, the observed rise in HDL cholesterol was less in diabetic subjects (12.5 ± 24.5% vs. 17.0 ± 28.2%, p = 0.01).
Atheroma burden and vessel dimensions
Indexes of atheroma burden and vessel dimensions at baseline are summarized in Table 3. The PAV (40.7 ± 9.9% vs. 38.1 ± 9.3%, p < 0.001) and TAV (203.8 ± 90.4 mm3 vs. 192.3 ± 84.1 mm3, p = 0.03) were greater in diabetic subjects, consistent with the presence of more extensive atherosclerosis. Multivariate analysis controlling for differences in risk factors revealed that the presence of diabetes was an independent predictor of both PAV (p < 0.0001) and TAV (p = 0.03). A closer correlation was observed with glycated hemoglobin, compared with fasting glucose, and both PAV (r = 0.22, p = 0.0003 vs. r = 0.09, p < 0.0001) and TAV (r = 0.18, p = 0.002 vs. r = 0.05, p = 0.02), with the difference between these markers of glycemic control reaching statistical significance for both measures of plaque burden (p = 0.04). When controlling for study, differences in correlation between glycated hemoglobin (n = 280) and fasting glucose (n = 2,216) for both PAV (r = 0.19 vs. r = 0.09) and TAV (r = 0.15 vs. r = 0.04) just failed to meet statistical significance (p = 0.10). Of interest, diabetic subjects did not demonstrate more diffuse disease, as evidenced by a similar percentage of images containing plaque (76.0 ± 27.7% vs. 73.8 ± 27.5%, p = 0.15) as observed in nondiabetic subjects.
Differences in vascular dimensions were also observed between the groups. Despite the presence of more extensive atherosclerosis, lumen volume was smaller in diabetic subjects (291.1 ± 104.8 mm3 vs. 306.5 ± 108.2 mm3, p = 0.005), while there was no difference in EEM volume (494.9 ± 166.9 mm3 vs. 498.8 ± 167.2 mm3 in diabetic and nondiabetic subjects, respectively, p = 0.61).
Serial changes in atheroma burden and vessel dimensions
Changes in atheroma burden and vascular dimensions during serial evaluation are summarized in Table 4. Greater progression of PAV was observed in diabetic subjects (0.8 ± 3.4% vs. 0.3 ± 3.3%, p = 0.003). Diabetic subjects were also more likely to undergo substantial atheroma progression (>5% relative increase in PAV, 30.3% vs. 23.2%, p = 0.002) and less likely to undergo substantial atheroma regression (>5% relative decrease in PAV, 15.9% vs. 20.3%, p = 0.04). Multivariate analysis revealed that the presence of diabetes was an independent predictor of progression of both PAV (p = 0.0001) and TAV (p = 0.03). A closer relationship was observed between baseline levels of glycated hemoglobin (n = 280), compared with fasting glucose (n = 2,216), with changes in TAV (r = 0.14, p = 0.02 vs. r = 0.02, p = 0.29), with the difference between these markers of glycemic control reaching statistical significance (p = 0.02). When controlling for study, the difference in correlation between glycated hemoglobin and fasting glucose for the change in TAV (r = 0.04 vs. r = 0.02) was not significant (p = 0.72). No significant differences were observed between groups with regard to changes in vascular dimensions. Changes in atheroma volume were accompanied by changes in EEM (r = 0.58, p < 0.0001 vs. r = 0.62, p < 0.0001) but not lumen (r = −0.02, p = 0.72 vs. r = 0.06, p = 0.008) in both diabetic and nondiabetic subjects, respectively.
Insulin therapy, atheroma burden, and arterial remodeling
Vascular dimensions and atheroma burden were further explored in diabetic subjects, classified according to use of insulin. Patients receiving insulin demonstrated smaller EEM (482.5 ± 160.7 mm3 vs. 519.9 ± 166.9 mm3, p = 0.03) and lumen (276.0 ± 100.3 mm3 vs. 310.1 ± 105.6 mm3, p = 0.001) volumes. This resulted in a larger PAV (42.4 ± 10.2% vs. 39.9 ± 9.6%, p = 0.02) despite the presence of a similar TAV (206.5 ± 88.6 mm3 vs. 209.9 ± 90.2 mm3, p = 0.84). This observation suggests that the remodeling pattern may differ in insulin-treated patients. On serial evaluation, no differences were observed in changes in PAV (0.9 ± 3.4% vs. 0.8 ± 3.4%, p = 0.54) and TAV (−1.6 ± 21.9 mm3 vs. −1.2 ± 25.7 mm3, p = 0.34) in diabetic patients treated with and without insulin, respectively. Diabetic patients treated with and without insulin also demonstrated similar changes in EEM (−13.4 ± 4.9 mm3 vs. −15.5 ± 4.8 mm3, p = 0.58) and lumen (−10.7 ± 3.0 mm3 vs. −12.6 mm3, p = 0.55) volumes.
Intensive lowering of LDL cholesterol and CRP and atheroma progression
The impact of intensive lowering of levels of LDL cholesterol below 80 mg/dl on progression of coronary atherosclerosis is summarized in Figure 1. Intensive lowering of LDL cholesterol had a beneficial impact on progression of PAV in both diabetic (0.4 ± 3.2% vs. 1.3 ± 3.6%, p = 0.008) and nondiabetic (−0.2 ± 3.0% vs. 0.6 ± 3.5%, p < 0.001) subjects and on progression of TAV in diabetic subjects (−5.0 ± 21.7 mm3 vs. 2.2 ± 24.8 mm3, p = 0.001) and nondiabetic subjects (−6.9 ± 19.5 mm3 vs. −1.5 ± 23.3 mm3, p < 0.001). Intensive lowering of LDL cholesterol also resulted in a greater proportion of subjects undergoing substantial regression (at least 5% relative reduction in PAV) in diabetic subjects (18.5% vs. 13.0%, p = 0.12) and nondiabetic subjects (23.4% vs. 17.5%, p = 0.002). The proportion of subjects undergoing substantial progression (at least 5% relative increase in PAV) was reduced with intensive lowering of LDL cholesterol in both diabetic subjects (26.9% vs. 35.2%, p = 0.07) and nondiabetic subjects (18.4% vs. 27.4%, p < 0.001). Of interest, intensive lowering of LDL cholesterol only improved progression rates of diabetic subjects to that observed in the nondiabetic subject without intensive control of LDL cholesterol. The impact of intensive lowering of CRP was investigated in subjects in whom CRP measurements were available (n = 1,727). While a similar trend was observed with regard to achieving a CRP level <2 mg/l, none of the comparisons achieved statistical significance (Fig. 2).
In recent years, IVUS has been increasingly used to study the progression of coronary atherosclerosis and the impact of pharmacological therapies on the arterial wall (22–26). Given that each of these clinical trials has included patients with diabetes, a pooled analysis of the studies enables a systematic assessment of the differences in the pattern of atherosclerosis in diabetic patients compared with that in nondiabetic patients. In the current pooled analysis of 5 prospective clinical trials, the presence of diabetes was associated with a greater atherosclerotic burden and impaired compensatory remodeling of the artery wall. Furthermore, atheroma progression, despite the high use of established medical therapies, was more rapid in patients with diabetes. This highlights the important mechanistic links that underscore the aggressive nature of atherosclerotic cardiovascular disease in patients with diabetes.
The presence of more extensive disease confirms through direct observations of the vessel wall the finding that diabetic patients have an accelerated form of atherosclerotic disease. A number of pathophysiologic abnormalities likely explain the more rapid progression of disease. The presence of diabetes is accompanied by a greater prevalence of established atherosclerotic risk factors including hypertension; low levels of HDL cholesterol; hypertriglyceridemia; the presence of small, dense LDL particles; and obesity. However, conventional risk factors alone represent only a portion of the excess disease burden. Hyperglycemia and the potential generation of advanced glycation end products also seem to play an important role (27). The chronic influence of hyperglycemia is supported by the observation of a correlation between measures of atheroma burden and glycated hemoglobin, a marker of long-term glycemic control. The presence of elevated systemic markers of inflammation and oxidative stress provide additional mechanisms that may contribute to the accelerated form of atherosclerosis.
The current observations also suggest that abnormalities of arterial remodeling may influence the clinical expression of disease in diabetes. Angiographic studies in diabetics typically reveal diffuse disease with a predilection for distal segments of relatively small vessels (28). The current findings suggest that the pattern of disease in diabetes is not more diffuse in the patient with angiographic abnormalities. Indeed, the presence of more extensive atherosclerosis involving a similar proportion of images is consistent with a pattern of disease that may be more focal than previously considered. The finding that lumen volumes are smaller in the diabetic subjects is consistent with the observation of smaller vessels on angiography. A similar EEM volume and smaller lumen, despite the presence of more extensive atherosclerosis, suggests that diabetes is accompanied by impaired compensatory remodeling of the arterial wall. The mechanisms that promote impaired arterial remodeling in diabetes remain to be defined. However, it is possible that increasing deposition of fibrous (29,30) and calcific (31) tissue in the arterial wall, in addition to impaired endothelial-dependent relaxation (32), may limit vessel wall expansion with plaque accumulation. Given its role in the promotion of obstructive disease, inadequate arterial remodeling represents an additional target for therapeutic modification in diabetic patients with coronary artery disease.
Impairment of compensatory remodeling appears to be prominent in diabetic patients treated with insulin and supports previous observations in studies of native atherosclerosis (33,34) and heart transplant recipients (35). These patients demonstrated smaller vascular dimensions and a greater PAV, despite the presence of a similar TAV. This suggests that inadequate compensatory remodeling is particularly evident in the insulin-treated patient. Proliferation of smooth muscle and fibrous tissue in response to insulin (36) might increase vascular stiffness and further impair the ability of the artery wall to expand in response to accumulation of plaque.
This study provides insight into the impact of established medical therapies on atheroma progression in the setting of diabetes. In previous studies of patients with established coronary artery disease, lipid-modifying therapies slowed progression of obstructive disease by angiography (37). This is consistent with the observation that the greatest clinical benefit of medical management of diabetic patients is derived from optimal control of blood pressure (38) and lipids (39–41). However, despite the high rate of use of blood pressure and lipid-modifying therapies, accelerated disease progression was observed in diabetic patients. Furthermore, while intensive lowering of LDL cholesterol had a favorable impact on plaque progression, the diabetic subjects continued to demonstrate greater increases in atheroma volume. This residual progression suggests that an incremental benefit might be derived from use of emerging medical therapies, which modify additional targets including hyperglycemia and inflammatory mediators of disease in the arterial wall. The more recent observation that improving control of glycemic, lipid, and inflammatory markers with the peroxisome proliferator-activated receptor-gamma agonist pioglitazone slows progression of carotid intimal-medial thickness (42) suggests that this strategy may also have a beneficial impact on established coronary atherosclerosis.
A number of caveats should be noted with regard to the current analysis. The results of this analysis were obtained by pooling data from a number of clinical trials. The diagnosis of diabetes was recorded on the basis of the clinical report form of each study. As duration of diabetes was not recorded, it is unknown whether the extent of atherosclerosis and inadequate remodeling was greater in those subjects with a longer history of impaired glycemic control. As all patients had a diagnosis of coronary artery disease at the time of a clinically indicated angiogram, it is unknown if the current findings can be translated to the setting of primary prevention. However, the observation that asymptomatic diabetic patients have a prospective cardiovascular risk comparable to that of nondiabetic survivors of myocardial infarction (14) suggests that the underlying disease is aggressive even in the patient who has yet to come to clinical attention. As the direct relationship between the extent of atherosclerosis on IVUS and clinical outcome continues to be defined, it remains to be determined to what extent the differences in plaque burden contribute to the adverse cardiovascular outcome observed in diabetic patients with established coronary artery disease.
The current analysis of a large number of diabetic subjects reveals the presence of more extensive atherosclerosis in association with impaired compensatory remodeling of the arterial wall. While a benefit on disease progression is observed with the use of medical therapies, an accelerated increase in atheroma volume is observed in comparison with that seen in patients without diabetes. This finding suggests that there is an ongoing need to develop new therapies that complement modification of established risk factors to achieve a greater benefit in halting disease progression in diabetic patients resulting in optimal reduction of cardiovascular risk.
The authors thank the Intravascular Ultrasound Core Laboratory of the Cleveland Clinic for its technical expertise.
The ACTIVATE study was sponsored by Sankyo Pharmaceuticals. The ASTEROID study was sponsored by AstraZeneca Pharmaceuticals. The CAMELOT, ILLUSTRATE, and REVERSAL studies were sponsored by Pfizer Pharmaceuticals.
Dr. Nicholls has received speaking honoraria from Pfizer, Takeda, Merck, Schering-Plough, and AstraZeneca and consulting fees from Pfizer, Roche, and AstraZeneca. Dr. Tuzcu has received speaking honoraria and research support from Pfizer and AstraZeneca. Dr. Sipahi has received an educational grant from Pfizer. Dr. Nissen has received research support from AstraZeneca, Eli Lilly, Pfizer, Takeda, Sankyo, and Sanofi-Aventis; he has consulted for a number of pharmaceutical companies without financial compensation. All honoraria, consulting fees, or other payments from any for-profit entity are paid directly to charity, so that neither income nor any tax deduction is received.
- Abbreviations and Acronyms
- C-reactive protein
- external elastic membrane
- high-density lipoprotein
- intravascular ultrasound
- low-density lipoprotein
- percent atheroma volume
- total atheroma volume
- Received December 13, 2007.
- Revision received March 7, 2008.
- Accepted March 11, 2008.
- American College of Cardiology Foundation
- Huxley R.,
- Barzi F.,
- Woodward M.
- Ho J.E.,
- Paultre F.,
- Mosca L.
- Kannel W.B.,
- McGee D.L.
- Stamler J.,
- Vaccaro O.,
- Neaton J.D.,
- Wentworth D.
- Chun B.Y.,
- Dobson A.J.,
- Heller R.F.
- Miettinen H.,
- Lehto S.,
- Salomaa V.,
- et al.
- Wannamethee G.,
- Whincup P.H.,
- Shaper A.G.,
- Walker M.,
- MacFarlane P.W.
- Larsen J.R.,
- Tsunoda T.,
- Tuzcu E.M.,
- et al.
- Nicholls S.J.,
- Tuzcu E.M.,
- Crowe T.,
- et al.
- Burchfiel C.M.,
- Reed D.M.,
- Marcus E.B.,
- Strong J.P.,
- Hayashi T.
- von Birgelen C.,
- Hartmann M.,
- Mintz G.S.,
- et al.
- Nicholls S.J.,
- Tuzcu E.M.,
- Sipahi I.,
- Schoenhagen P.,
- Nissen S.E.
- Stone P.H.,
- Muller J.E.,
- Hartwell T.,
- et al.,
- The MILIS study group
- Portik-Dobos V.,
- Anstadt M.P.,
- Hutchinson J.,
- Bannan M.,
- Ergul A.
- Anand D.V.,
- Lim E.,
- Darko D.,
- et al.
- Turnbull F.,
- Neal B.,
- Algert C.,
- et al.