Low Blood Flow Estimates in Lower-Leg Arteries Predict Cardiovascular Events in Japanese Patients With Type 2 Diabetes With Normal Ankle-Brachial Indexes Free

We enrolled 158 type 2 diabetic patients and 35 age-matched nondiabetic subjects ranging in age from 50 to 75 years who had been consecutively admitted to our hospital between September 1997 and March 2003. Patients were followed 4.8 ± 1.5 years (range 3.0–8.2) or until their first event of CVD. During the study, 23 patients dropped out. Six patients were excluded from the present study because they did not meet study criteria. Therefore, the remaining 129 patients were studied. All patients were admitted for strict glycemic control or assessment of diabetes complications, and no patients had history of cerebrovascular disease, coronary arterial disease, and/or PAOD. Patients were considered to have cerebrovascular disease if they had a history of sudden focal neurological deficit. Coronary arterial disease was diagnosed if the patients had a history of myocardial infarction or showed abnormal electrocardiographic findings. PAOD was diagnosed if the patient had an abnormal ABI (<0.9) (16); therefore, all patients had a normal ABI. Patients who had foot edema caused by heart failure, liver cirrhosis or nephropathy (serum creatinine >133 μmol/l), malignant neoplasm, alcohol abuse, or acute illness were excluded from the study. Hypertension was defined as either blood pressure >140/90 mmHg or initiation of antihypertensive treatment (17). Dyslipidemia was defined as level of total cholesterol >5.69 mmol/l, HDL cholesterol <1.03 mmol/l, or triglycerides >1.68 mmol/l as well as treatment with cholesterol-lowering agents (18). The ethics committee of our institution approved the study, and informed consent was obtained from all patients before the examinations.

Blood pressure was measured three times using automatic devices on the right arm after a 15-min rest in the sitting position, and the means of the second and third recordings were used as the baseline. Blood samples were taken from the cubital vein after overnight fasting. The ABI was examined using a handheld ultrasound Doppler device (ES-1000SP; Nihon Kohden, Tokyo, Japan) to assess PAOD (16). A trained ophthalmologist carried out fundus ophthalmoscopies and classified diabetic patients as either without retinopathy or having simple retinopathy corresponding to levels 21–53 or proliferative retinopathy corresponding to levels 60–80 of the modified Airlie House System (19). The diabetic patients were classified as having normoalbuminuria, microalbuminuria, or overt proteinuria when the urinary albumin excretion rate was <15, 15–200, or >200 μg/min, respectively, based on 24-h urine collection in our hospital. Patients with diabetes were screened for distal symmetric polyneuropathy using a 128-Hz tuning fork applied to the bony prominence at the dorsal surface of both great toes, just proximal to the nail bed (20). When the tuning fork was placed on the foot for 10 s, if the patient required >10 s to detect the vibration, vibration perception was regarded as compromised. These patients were also classified based on their smoking habits as being a current smoker or nonsmoker. Nonsmokers were defined as not having tobacco consumption for at least the previous 3 years.

An MRI scanner operating at 1.5-Tesla (Signa Horizon-LX; GE Medical Systems, Milwaukee, WI) was used for the following experimental protocols as previously described (21). Briefly, after at least 15 min of rest, all patients were evaluated in the supine position in a temperature-controlled 25°C room. To set up the individual flow analysis, the popliteal artery was depicted by 2D time-of-flight magnetic resonance angiography. A single slice with 5-mm thickness was oriented perpendicular to the flow direction, and flow data were obtained using 2D-cine-PC MRI with 80-cm/s velocity encoding triggered by peripheral gating. Heart rate was monitored by peripheral gating. Flow data were analyzed on an Advantage Windows version 3.1 workstation (GE Medical Systems) to determine the direction and velocity through the cardiac cycle. The instantaneous flow volume at 16 equally spaced time points through the cardiac cycle was calculated from the individual velocity images by integrating the velocity across the area of a vessel. The resultant 16 flow volumes allowed assessment of flow variations during the cardiac cycle. Total flow volume was calculated from the integration of waveform. In 2D-cine-PC MRI, pulsatility or resistive index was used to assess flow profiles at the popliteal artery (22). A resistive index, which associates with arterial resistance to blood flow, has been defined as (A − B)/A, where A is the systolic peak velocity and B is the end-diastolic velocity (23). Therefore, resistive index was calculated from the originally obtained 16 velocity images. A large 28 × 28-cm field of view was used to allow assessment of bilateral lower legs, and the average of those data was used.

The primary composite end point in the present study was any new CVD event, including cerebral infarction, myocardial infarction, coronary revascularization by percutaneous coronary intervention or coronary artery bypass grafting, coronary-related death, development of PAOD, or ischemic foot gangrene assessed by total flow volume at the popliteal artery of <50.8 ml/min (24).

Statistical evaluation was carried out on SPSS software version 11.0 for Windows (SPSS, Chicago, IL). A multiple comparison of significant differences among the three groups was carried out by one-way ANOVA followed by Scheffe’s F test. The Bonferroni’s test for two-by-three contingency tables was used to compare the frequencies among the three groups. Kaplan-Meier analysis was performed to estimate the probabilities of symptoms among the groups, and the level of significance was evaluated by the log-rank test. Multivariate Cox proportional hazards analysis was used to calculate hazard ratio (HR) and 95% CI of incident CVD events. Values are expressed as the means ± SD. We considered P values <0.05 to be statistically significant.

Clinical characteristics at baseline. On follow-up, 16 of 129 diabetic patients with normal ABI developed primary CVD events. These events consisted of cerebral infarction (n = 7), coronary revascularization by percutaneous coronary intervention (n = 6) or coronary artery bypass grafting (n = 1), coronary-related death (n = 1), and ischemic foot gangrene (n = 1). We investigated clinical differences at baseline among patients who had a primary CVD event (with CVD) (n = 16), those who remained without a CVD event throughout the follow-up period (without CVD) (n = 113), and age-matched nondiabetic subjects (n = 35). Clinical characteristics of all subjects are shown in Table 1. There were no significant differences among the groups for prevalence of male sex, age, BMI, total cholesterol, HDL cholesterol, triglycerides, prevalence of smoking, and ABI. However, compared with nondiabetic subjects, diabetic patients with CVD had higher fasting plasma glucose (P < 0.001), HbA_1c (A1C) (P < 0.001), and systolic blood pressure (P < 0.05) and lower diastolic blood pressure (P < 0.05). There were no significant differences between the groups with and without CVD for duration of diabetes, fasting plasma glucose, A1C, frequencies of patients taking cholesterol-lowering agents or antihypertensive treatment, or prevalence of retinopathy, nephropathy, or neuropathy.

Waveform analysis at the popliteal artery using gated 2D-cine-PC MRI is shown in Fig. 1. Nondiabetic subjects (Fig. 1A) and diabetic patients without CVD (Fig. 1B) had a typically triphasic waveform, which could be clearly separated into systolic, early diastolic, and late diastolic phases during the cardiac cycle. However, diabetic patients with CVD (Fig. 1C) showed abnormal flow reversal in late diastole, suggesting higher vascular resistance. Quantitative flow data are summarized in Table 1. There were no significant differences among the groups in heart rate and early diastolic flow reversal. Compared with diabetic patients without CVD, diabetic patients with CVD had lower total (P < 0.01), systolic (P < 0.01), and late diastolic (P < 0.05) flow volumes and a higher resistive index (P < 0.05), indicating that those parameters are characteristically altered in diabetes. Therefore, blood flow and vascular resistance were considered as possible predictors of CVD events. To clarify the association of blood flow and vascular resistance in the lower-leg arteries, simple linear regression analyses were performed. Total (r = −0.669, P < 0.001), systolic (r = −0.341, P < 0.001), and early (r = −0.339, P < 0.001) and late diastolic (r = −0.747, P < 0.001) flow volumes were each negatively correlated with resistive index, indicating that higher vascular resistance reduced blood supply in the lower-leg arteries among diabetic patients.

When diabetic patients were classified into three subgroups of equal numbers of patients based on their resistive indexes (0.877–0.993, 0.994–1.034, and 1.035–1.105), Kaplan-Meier analysis revealed no significant differences in the risk of developing CVD events among the groups (P = 0.1112, log-rank test) (Fig. 2A). However, when the patients were grouped into tertiles according to their total flow volumes (129.6–85.5, 85.3–63.3, and 62.7–23.8 ml/min), Kaplan-Meier analysis confirmed a significantly higher probability of developing CVD events in patients in the lowest tertile than in those in the highest tertile (P = 0.0199, log-rank test) (Fig. 2B). The clinical characteristics of those groups are summarized in Table 2. There were no significant differences among the groups for prevalence of male sex, age, fasting plasma glucose, A1C, total cholesterol, HDL cholesterol, triglycerides, frequency of patients taking cholesterol-lowering agents, diastolic blood pressure, prevalence of patients taking antihypertensive treatments, smoking status, retinopathy, nephropathy, neuropathy, or ABI. However, patients in the lowest tertile showed higher frequency of CVD events (P < 0.05), lower BMI (P < 0.01), longer duration of diabetes (P < 0.05), and higher systolic blood pressure (P < 0.05) than patients in the highest tertile. Compared with the highest tertile, the lowest tertile had greater early diastolic flow reversal (P < 0.05) and resistive index (P < 0.001) and lower systolic (P < 0.001) and late diastolic (P < 0.001) flow volumes, whereas there was no significant difference in heart rate between the groups.

We performed multivariate Cox hazards analysis to determine the association between incidence of CVD events and nine possible predictors of CVD events (Table 3). Low flow volume (<62.7 ml/min) (HR 8.60, 95% CI 1.61–45.97, P = 0.012), hypertension (3.99, 1.12–14.25, P = 0.033), and smoking status (12.01, 1.21–119.28, P = 0.034) were identified as significant independent predictors of CVD events.

It has been reported that vessel wall properties are not uniform along the arterial tree. Postmortem study demonstrated that leg arteries are more elastic in older adults than in younger subjects, in contrast to the carotid system (25). Our findings were obtained in the elderly. Therefore, we may have underestimated the association of abnormal peripheral circulation and arterial stiffness in the lower legs. Among diabetic patients, arterial stiffness is reported to be increased in the aorta (4), carotid, brachial, and femoral arteries (26) and lower-limb arteries (27) compared with those in healthy subjects. Abnormal peripheral circulation in type 2 diabetic patients with normal ABI associates with coronary atherosclerosis, large artery stiffness, and peripheral vascular resistance (28). Those parameters were positively correlated with each other, suggesting the parallel development of different aspects of vascular abnormalities in diabetes. Although vascular rigidity was not evaluated in the present study, we have previously reported that diabetic patients with normal ABI had a higher brachial-ankle pulse wave velocity, which is used as an indirect index of arterial stiffness, than that in age-matched nondiabetic subjects (28). Furthermore, total flow volume, brachial-ankle pulse wave velocity, and resistive index were associated with each other, indicating that higher vascular resistance caused by greater arterial stiffness reduced blood supply in the lower-leg arteries among diabetic patients (28). The peripheral circulation depends on not only vascular conditions in the periphery but also cardiac output. Diabetic patients with greater resistance to blood flow had less systolic flow volume, and therefore perfusion pressure in the periphery did not rise.

Waveform analysis at the popliteal artery using the new technique of 2D-cine-PC MRI is beneficial to assess peripheral circulation in both normal and diseased arteries (21). The normal subjects had a typically triphasic waveform, which could be clearly separated into systolic and early and late diastolic phases of the cardiac cycle. In late diastole, a positive waveform smaller than during systole occurs as the distended arterial reservoirs force blood antegrade through the arterioles into the venous circulation (22). Diabetic patients are reported to have two types of insufficient arterial blood flow to the lower limbs associated with the vessel wall properties. First, a major suggested cause of peripheral artery disease is atherosclerotic occlusion of the lower-leg arteries distal to the aortic bifurcation, resulting in a low ABI. Those patients showed an abnormal monophasic waveform, which is dampened in the diseased or collateral arteries, resulting in amplitude reduction and broadening of the systolic segment with absence of diastolic segments (21). Second, diabetic patients are known to have stiffer arteries even though they have a normal ABI. We have reported that diabetic patients with stiffer arteries characteristically show abnormal flow reversal in late diastole, suggesting higher vascular resistance (29).

Our data revealed that diabetic patients with CVD events frequently show not only abnormal flow reversal but also reduced blood flow in late diastole. There are important differences between elastic and muscular arteries, and arteries of both types are impaired in diabetes (26). Large arteries, including the aorta and its major branches, having elastic properties of the vessel wall, act as carrying vessels and blood supply reservoirs (30). When there is a decrease in arterial elasticity, less blood can be stored in these arteries, resulting in a decrease in diastolic forward flow. Nonenzymatic glycosylation of matrix proteins (6), increased intima-media thickness (7, 8), and medial arterial calcification (9,10) can all be involved in the pathogenesis of vascular rigidity. The medium- and small-caliber arteries and arterioles, which have functional smooth muscles in the vessel wall, act as resistance vessels regulating blood flow to the capillaries (30). Based on Poiseuille’s law, a decrease in vessel radius increases the resistance to blood flow (31). Endothelial dysfunction (32) and reduced lumen diameter in small vessels (33) are major determinants of peripheral vascular resistance. Increased formation of circulating advanced glycosylation end products (34) and carotid intima-media thickness (35) are known to be associated with endothelial dysfunction, suggesting that patients with stiffer arteries also have endothelial dysfunction.

Our data showed that patients with CVD had higher vascular resistance and lower blood flow in the lower-leg arteries than patients without CVD, indicating that those parameters may be possible predictors of CVD events among diabetic patients. When the patients were classified into three subgroups based on their levels of those parameters, Kaplan-Meier analysis confirmed a significantly higher probability of development of CVD events for blood flow but not for vascular resistance. Peripheral circulation depends on both cardiac output and vascular resistance. These results suggest that quantitatively assessed blood flow is a better predictor of CVD events than is vascular resistance. Multivariate Cox proportional hazards analysis showed that low flow volume, hypertension, and smoking status were significant independent predictors of CVD events. Hypertension (36) and smoking habit (37) are well-known causes of CVD events among individuals with diabetes. In the present study, patients with total flow volume <62.7 ml/min were at an 8.6-fold increased risk of CVD events compared with patients with total flow volume >85.5 ml/min. We have reported that reduction of blood flow to the lower limb associates with severity of coronary atherosclerosis, large artery stiffness, and peripheral vascular resistance among diabetic patients with normal ABI (28). Therefore, those patients may have increased risk of CVD events.

Our data were obtained in a Japanese population, and therefore it remains to be established whether these results can be generalized to other ethnicities. Therefore, we have concluded that low blood flow estimates in the lower-leg arteries may be predictive for CVD events among Japanese patients with type 2 diabetes, even though they have normal ABI.