Ischemia Imaging and Plaque Imaging in Diabetes: Complementary tools to improve cardiovascular risk management

CAD is often silent in diabetic patients, and it is typically in advanced stages of development by the time it manifests (10). Hence, several investigators have utilized various forms of stress testing to detect silent obstructive CAD. Abnormal electrocardiographic stress tests have been reported in 12–31% of asymptomatic diabetic individuals (11–17). Among the different noninvasive imaging tests used to diagnose obstructive CAD, stress nuclear myocardial perfusion imaging (MPI) has established itself as one of the most reliable and informative tools. Obstructive CAD is demonstrated by detecting a reduced uptake (perfusion abnormality) of a radiotracer after stress in one or more areas of the myocardium perfused by a coronary artery with a critical (>50%) luminal stenosis. Perfusion defects can be elicited via treadmill exercise stress testing or after the injection of drugs with vasodilator (for example adenosine) or inotropic (for example dobutamine) effects. The most frequently employed radiopharmaceutical agents are thallium-201–or technetium-99m–based tracers (Tc-99m sestamibi or Tc-99m tetrofosmin), each with different pharmacodynamic and pharmacokinetic characteristics. Thallium-201 has a long half-life (∼73 h), a lower peak emission rate than technetium-99m, and recirculates in and out of viable cells utilizing the same membrane channels as potassium (18–20). Due to its rapid redistribution, imaging must proceed quickly after the injection of the tracer at peak exercise to detect stress-induced perfusion defects. Delayed imaging (usually 4 h from stress imaging) is then performed to verify whether a difference in perfusion (tracer uptake) can be detected between the stress and rest phase in any region of the left ventricular myocardium. Tc-99m has a higher peak emission rate and a shorter half-life (∼6.5 h) than thallium-201. It recirculates minimally in and out of viable cells and remains essentially trapped in the location of initial uptake, giving an instantaneous impression of the perfusion status at the time of injection. Due to its lack of redistribution, Tc-99m–based tracers must be injected at peak exercise and reinjected at rest to compare stress and rest images (18–20). The greater peak emission rate but shorter half-life of Tc-99m tracers allow the use of larger doses than thallium-201, with overall better image quality and performance of accurate gated segmental wall motion and left ventricular function. On the other hand, thallium-201 may be preferable to Tc-99m tracers for the performance of vasodilator stress (21) and myocardial viability (22) studies.

MPI has been employed in several series to test the prevalence of silent ischemia or to analyze the prognostic impact of perfusion abnormalities in diabetes. In the DIAD (Detection of Ischemia in Asymptomatic Diabetics) study (23), 1,124 type 2 diabetic patients were enrolled at 14 sites in the U.S. and Canada. Half of the patients were randomized to an adenosine Tc-99m sestamibi MPI and half were not. In the imaging cohort, 22% of the individuals showed abnormal MPI results, and 1 in every 18 subjects (5.5%) showed a moderate to severe perfusion defect indicative of poor prognosis. Since the entry criteria required an established diagnosis of type 2 diabetes, a normal electrocardiogram, and no known CAD, the authors concluded that obstructive CAD would remain undetected in as many as 41% of type 2 diabetic patients if the current American Diabetes Association recommendations (24) for CAD screening were strictly followed. Although this study helped establish the presence of silent ischemia in diabetes, it did not provide any information on the outcome of patients with abnormal scans, and a long-term follow-up has therefore been planned.

De Lorenzo et al. (25) submitted 180 asymptomatic adult-onset diabetic patients to exercise MPI with Tc-99m sestamibi to detect asymptomatic obstructive CAD. In this study, a short-term follow-up (36 ± 18 months) was conducted to correlate the imaging findings with patients’ outcome. A positive test result was reported in 26% of all subjects. During follow-up, 34 patients suffered cardiac events: 7 cardiac deaths, 6 nonfatal myocardial infarctions, 10 coronary artery bypass surgeries, and 11 percutaneous coronary angioplasties. Male sex and perfusion abnormalities were independent predictors of cardiac events. Inducible myocardial ischemia on MPI added incremental prognostic value to clinical variables and exercise stress test variables for the prediction of hard (χ² = 5.4, P < 0.001) as well as total (χ² = 7.4, P = 0.0001) events.

Kang et al. (26) also investigated the incremental prognostic value of MPI over clinical data for prediction of events in 1,271 diabetic and 5,862 nondiabetic patients. During an average follow-up period of 24 months, diabetic patients suffered almost twice the number of events (myocardial infarction, death, and coronary revascularizations) compared with the nondiabetic subjects. Findings on MPI added significant prognostic information (P < 0.001) to clinical and historical data in both subsets of patients.

In a single-center study, Berman et al. (27) reported on the incremental prognostic value of adenosine stress MPI in 1,222 consecutively tested diabetic and 4,111 nondiabetic patients. For the nondiabetic patients, a normal, mildly abnormal, and moderately to severely abnormal MPI study was associated with cardiac death rates ranging from 0.8 to 6.1% per year of follow-up. The event rates were similar among nondiabetic men and women. However, diabetic women with perfusion defects demonstrated significantly higher death rates than diabetic men (P < 0.0001). Furthermore, in the setting of a normal MPI the death rate increased from ∼0.6% for nondiabetic subjects to 1.8% for noninsulin-requiring to 2.5% for insulin-requiring diabetic patients, respectively. These observations were echoed by the report by Giri et al. (28). In this large multicenter study, 4,755 patients (20% had diabetes) were followed for an average of 2.5 years after a baseline MPI (28). The primary end points were the occurrence of cardiac death and nonfatal myocardial infarction or the performance of revascularization procedures. An abnormal MPI test was a significant predictor of cardiac death and myocardial infarction in both diabetic and nondiabetic subjects, though diabetic patients suffered more events in each category. A stepwise increase in cardiac death rate was observed with increasing numbers of ischemic segments of the left ventricle in both diabetic (χ² = 15.1, P = 0.002) and nondiabetic (χ² = 38.8, P < 0.0001) subjects. However, the stepwise increase was greater in diabetic patients where ischemia even in a single-vessel territory increased the risk of death significantly. As noted above, two more results from this large muticenter study are worth remarking on. Although a normal MPI was associated with a high event-free survival in the short range, diabetic patients with a normal perfusion stress test had a fourfold higher morbidity and mortality during follow-up than nondiabetic subjects with normal MPI tests. Furthermore, the “protective effect” of a normal MPI study in diabetic patients appeared to expire after ∼2 years from testing, when events started occurring very rapidly.

A retrospective analysis (29) of 1,427 diabetic patients referred for MPI to a single center was recently published. An abnormal MPI study was reported in 58% and a high-risk scan in 18% of the subjects. The best predictors of an abnormal MPI were the presence of Q waves on the resting electrocardiogram and the diagnosis of peripheral arterial disease. Not unexpectedly, the annual mortality rate was high (5.9%) in patients showing high-risk findings on MPI. However, the mortality rate was also very high in patients with low-risk scans (3.6% per year) (28). In this series, 46% of the patients from the community were asymptomatic and had been referred for CAD screening. Although this study was limited by substantial referral bias, it illustrates two important points in the care of diabetic patients: the need physicians feel to screen for CAD in asymptomatic diabetic subjects and the high event rate even in the presence of “seemingly reassuring” test results.

Hence, the actual cardiovascular risk of diabetic patients is greater than what can be assessed by simply seeking myocardial ischemic responses to stress to predict the presence of obstructive coronary disease. These observations lend support to the concept of refining risk stratification in diabetes using plaque imaging techniques. Indeed, these techniques identify the presence of arterial plaque burden not causing obstructive disease and may prove useful in improving risk prediction in diabetes.

Stress echocardiography (SE) was introduced as an alternative to nuclear perfusion imaging and has demonstrated very high diagnostic accuracy. With SE, myocardial ischemia is demonstrated when segmental wall motion abnormalities of the left ventricle, such as new or worsening hypokinesia, new akinesia, or dyskinesia, are induced by either exercise or pharmacological stress testing. Due to poor acoustic windows, patient habitus, excessive respiratory motion artifacts, and limited technical experience of the operator, ∼5–10% of all SE tests are nondiagnostic. In comparison, the nondiagnostic rate with stress nuclear testing varies between 1 and 2%. However, SE offers the advantage of easy portability, lower cost, and no radiation exposure compared with nuclear stress testing, and it has therefore gained substantial popularity.

The diagnostic accuracy for obstructive CAD of SE in diabetic patients has been reported to be similar to that noted in nondiabetic subjects. Penfornis et al. (30) performed dobutamine stress echocardiography (DSE) in 56 asymptomatic diabetic subjects who were also submitted to electrocardiographic and MPI stress testing. They reported a positive predictive value of 60, 69, and 75% for electrocardiographic stress testing, DSE, and MPI, respectively, for the detection of obstructive CAD. The authors concluded that DSE and MPI are equivalent, though in this particular study both showed a moderate accuracy.

SE and DSE were utilized in a multicenter study (31) involving 937 predominantly type 2 diabetic patients with known or suspected CAD. The follow-up was as long as 9 years from the time of SE testing, and the primary end point was all-cause mortality. A very high mortality was recorded both for patients with resting as well as for stress-induced wall motion abnormalities (123 of 275 subjects, 45% mortality and 115 of 232 subjects, 50% mortality). In multivariable models, the strongest predictor of mortality was a referral for pharmacological stress testing (emphasizing the importance of limited exercise tolerance as an indicator of risk), followed by the visualization of ischemic changes on echocardiographic imaging, presence of heart failure, and age. Similarly, in a smaller study of 259 diabetic patients followed for an average of 24 ± 22 months, the presence of ischemic myocardial changes on SE was an independent predictor of cardiac death and nonfatal myocardial infarction (32). However, in the study by Marwick et al. (31), the yearly event rate for patients with a normal SE varied between 2 and 3%. Hence, as shown using nuclear techniques (27,28), the event rate for diabetic patients with nonischemic myocardial stress tests was much higher than that of nondiabetic patients with negative tests. These findings suggest once again that tests for atherosclerosis may aid refining risk stratification in diabetic patients beyond myocardial ischemia imaging.

CAC is intimately related to atherosclerosis, and the area of coronary artery calcification measured on cardiac computed tomography (CT) has been shown to closely correlate with the extent of atherosclerotic plaque (33). Hence, coronary calcium is considered an accurate marker of atheroslerosis. The first CT developed for this purpose was the electron-beam CT scanner, followed more recently by multidetector spiral CTs. The extent of calcification is measured by means of a calcium score calculated by the computer software on the basis of plaque size and density or as volume of calcified plaque (34). Though often misrepresented, the main purpose of calcium screening is not to identify patients with obstructive CAD but to detect vessel wall atherosclerosis. Identifying nonobstructive plaque, in fact, may be as important as assessing stenosis severity since many acute coronary events occur on the basis of nonobstructive disease (35,36).

Mielke, Shields, and Broemeling (37) studied a cohort of 3,389 patients suffering from type 2 diabetes and showed that diabetic patients tend to harbor larger amounts of CAC than nondiabetic patients of similar age and with a similar risk-factor profile. Moreover, the amount of CAC was similar to that of patients with established CAD but without diabetes. Similarly, Khaleeli et al. (38) showed that asymptomatic diabetic patients present the same prevalence of CAC as nondiabetic individuals symptomatic for CAD (89 vs. 73%, P = NS). Furthermore, both studies showed that diabetic women harbor as much CAC as diabetic men, confirming the clinical evidence that diabetes negates the well-known advantage of women over men in prevalence and extent of atherosclerosis (37,38). Schurgin, Rich, and Mazzone (39) screened a cohort of 139 asymptomatic diabetic patients and matched control subjects. They concluded that extensive CAC (calcium score >400) is more prevalent (25.9%) in diabetic patients than nondiabetic control subjects both without (7.2%) and with (14%) traditional risk factors. These findings are of interest since a calcium score >400 is associated with a high risk of myocardial perfusion impairment (40) and a high risk of any cardiovascular event in the short term (41,42). Furthermore, the calcium imaging data are supportive of the well-known clinical data showing that an asymptomatic diabetic patient presents the same cardiovascular risk as a patient with established CAD but without diabetes (43).

Recently Raggi et al. (44) published a report of 10,377 asymptomatic individuals (903 type 2 diabetic patients) followed for an average of 5 years after having been referred by a primary care physician for CAC screening. The primary end point of the study was all-cause mortality. In that study, the risk of all-cause mortality was higher in diabetic patients than nondiabetic subjects for any degree of calcification, and the risk increased as the calcium score increased. In diabetic patients, there was a 44% (95% CI 20–80) increased risk of death for every increase in coronary calcium score (CCS) grouping from 11–100 to 101–400, 401 to 1,000, and >1,000 (P = 0.0001). Finally, a risk-adjusted model showed that there was a significant interaction of CCS with diabetes (P < 0.00001), indicating that for every increase in CCS, there was a greater increase in mortality for diabetic individuals than nondiabetic subjects. Interestingly, the absence of CAC predicted a low short-term risk of death (∼1% at 5 years) for diabetic patients as well as nondiabetic subjects. Hence, both the presence and absence of CAC were important modifiers of risk in diabetes. Discordant with these results were the findings reported by Qu et al. (45). In that study, the investigators followed 269 type 2 diabetic patients for an average of 6.3 years. They noted an increased prevalence of CAC in diabetic patients compared with control subjects and observed an increased risk of cardiovascular events in the patients with CAC compared with those without CAC. However, they were unable to prove that CAC adds incremental prognostic value to diabetes for the prediction of future events. This apparent discrepancy may have been due to the smaller cohort used in the study by Qu et al. (45) and the utilization of high-risk subjects (mostly men, on average 10 years older than the patients included in the study by Raggi et al. [44]) .

Sequential CAC imaging has been proposed as a method to monitor effectiveness of medical therapy in the general population (46), and CAC progression appears to be linked to the occurrence of adverse outcomes (47). In a retrospective analysis, the progression of CAC and the occurrence of myocardial infarction were compared in 157 type 2 diabetic patients and 1,153 nondiabetic control subjects during a follow-up of 1–3 years (48). Both event-free diabetic subjects and diabetic patients who suffered a myocardial infarction demonstrated a significantly greater progression of CAC than did nondiabetic patients. Though statin therapy slowed the progression of CAC in all groups, such treatment was significantly less effective in diabetic subjects.

In the CACTI (Coronary Artery Calcification in Type 1 Diabetes) study (49), 656 diabetic patients showed a higher prevalence and extent of CAC than 764 age- and sex-matched control subjects. Though there tended to be more CAC in men than women, after adjustment for waist-to-hip ratio, waist circumference, or visceral fat, the sex difference in CAC was not significant in diabetic subjects. The authors concluded that insulin resistance, associated with an android deposition of fat, might be one of the mechanisms responsible for the increased prevalence of CAC in women with type 1 diabetes. Of interest, Colhoun et al. (50) reported that the increased prevalence of CAC in women affected by type 1 diabetes is not associated with traditional risk factors for atherosclerosis or the size and concentration of various lipoproteins, while it is associated with the extent of systemic inflammation.

Extensive vascular calcifications were detectable even in young (17–28 years of age) adults with type 1 diabetes (51) and have been associated with factors such as genetic polymorphism for hepatic lipoxygenase (LIPC-480 T) (52) or more conventional ones like smoking, elevated serum lipoprotein(a) (51), or suboptimal glycemic control (52,53).

No prospective outcome data exist in type 1 diabetic patients related to the presence of CAC, and only cross-sectional observational data are available. Olson et al. (54) investigated 302 type 1 diabetic patients (146 men and 156 women) with a history of myocardial infarction, angina, or evidence of ischemia on stress testing or surface electrocardiograms. Patients were participants in the Pittsburgh Epidemiology of Diabetes Complications study, a 10-year prospective follow-up study of risk factors for complications of type 1 diabetes diagnosed before the age of 17 years. Electron-beam CT imaging showed that the prevalence of CAC clearly increased with age (from 11% before age 30 years to 88% in individuals aged 50–55 years or older). Among subjects without clinical manifestations of CAD, 5% had a CAC score ≥400 (indicative of a large atherosclerosis burden), as opposed to 25% of the subjects with angina or ischemic electrocardiogram changes and 80% of the patients with myocardial infarction or luminal stenosis on coronary angiography. CAC showed a sensitivity of 84 and 71% for clinical CAD in men and women, respectively, and 100% sensitivity for myocardial infarction or obstructive CAD. In multivariable regression analyses, CAC was independently correlated with myocardial infarction or obstructive CAD in both sexes and was the strongest independent correlate for men. The authors suggested that CAC in type 1 diabetes may be a marker of risk, independently of other risk factors. Obviously, prospective investigations will be necessary to support and expand these observations.

The thickness of the wall of the carotid arteries measured ultrasonographically was shown almost 2 decades ago to bear a good correlation with the presence and extent of atherosclerosis of the aorta (55–57). Subsequently, several randomized and epidemiological studies demonstrated the value of the CIMT as a marker of increased risk for cardiovascular events. Increased CIMT, even in the absence of obstructive luminal disease of the carotid arteries, has been associated with risk of myocardial infarction and stroke in the elderly (>65 years) (58) as well as younger age-groups (59–61). Furthermore, an increase of merely 0.03 mm per year in CIMT has been associated with a twofold increase in relative risk of myocardial infarction and cardiac death during follow-up (60). CIMT bears a modest relationship with obstructive CAD in nondiabetic (62,63) as well as diabetic (64) subjects, indicating that this surrogate is a marker of atherosclerosis burden rather than an index of the severity of CAD.

Increased CIMT in diabetes has been associated with risk factors for atherosclerosis such as high serum triglyceride levels and low total-to-HDL cholesterol ratio (65,66), age and BMI (67), lipoprotein(a) (68), microalbuminuria (69,70), and endothelial dysfunction and low-grade inflammation (71,72). Insulin resistance alone, even in the absence of clinical diabetes, has been associated with an increased CIMT (73).

Diabetes is an independent predictor of CIMT progression (74,75), and CIMT appears to progress faster in diabetes than in other conditions (59,76,77). Such accelerated progression appears at least in part related to glycemic control expressed both as fasting serum glucose levels (78) as well as HbA_1c levels (79). Furthermore, baseline age and carotid wall thickness (79), as well as systolic hypertension and development of nephropathy (80), have been identified as predictors of CIMT progression.

Sequential CIMT measurements have been utilized to assess efficacy of medical therapy in several studies (81–85) in the general population but only in a limited number of studies in diabetic patients. Nathan et al. (86) conducted first a short- then a long-term follow-up study of 1,229 type 1 diabetic patients randomized to either standard hypoglycemic therapy or intensive therapy and age- and sex-matched individuals. Progression of CIMT was not different at the end of the 1st year of follow-up between control subjects and all diabetic patients, but CIMT progressed significantly more in diabetic than control subjects by the end of 6.5 years of follow-up. Furthermore, at the end of the long-term follow-up, CIMT was shown to have progressed significantly more in diabetic patients who received standard therapy than in those randomized to intensive therapy. A small study of 34 patients affected by type 1 diabetes demonstrated that successful transplantation of pancreatic islet cells was associated with a slower CIMT progression and fewer cardiovascular events at 3 years (87). In a recent study (88), sequential carotid artery IMT measurements were utilized to assess the effect of statin therapy compared with placebo in 250 type 1 diabetic patients. At the end of 2 years of follow-up, there was no difference in the progression of CIMT, though the number of cardiovascular events was significantly smaller (P = 0.006) in patients treated with statins. The effectiveness of statins on reduction of cardiovascular events in diabetes was clearly consonant with the results of the Heart Protection Study (89) and the Collaborative Atorvastatin Diabetes Study (90). However, to demonstrate an effect on CIMT progression, a longer follow-up time may have been required. Finally, in two small randomized trials of type 2 diabetic patients, antiplatelet therapy based either on aspirin, ticlopidine (91), or cilostazol (92) was shown to slow progression of CIMT after a follow-up of 3 years (91) and 1 year (92), respectively.

The evidence presented, though clearly indicative of a greater atherosclerosis burden and attendant risk in diabetic patients with increased CIMT, identifies important limitations of the current knowledge in this field. First, while CIMT is an independent predictor of cardiovascular events in the general population, such evidence is of modest strength in diabetes. An isolated example in diabetes is represented by the study performed by Yamasaki et al. (79) on 287 type 2 diabetic Japanese patients, where baseline CIMT was identified as an independent predictor of incident nonfatal CAD (angina and myocardial infarction) during a follow-up period of 3 years. Second, although IMT progression can be slowed both in diabetic and nondiabetic patients with various interventions, little is known of the prognostic significance of CIMT progression in either patient group. Finally, no study has yet demonstrated the incremental prognostic value of CIMT over other conventional risk factors in either diabetic or nondiabetic individuals.

This analysis does not constitute a systematic review of the current medical literature (utilizing metanalytic methods) and does not offer an assessment of the cost-effectiveness of integrating various imaging modalities for the detection and care of CAD in diabetic patients. Furthermore, the current evidence on use of nontraditional imaging modalities to improve risk prediction is limited and often based on preselected populations. It represents, however, an attempt at summarizing a vast amount of information related to a subject that has received a large amount of interest in the most recent literature.

Figure 1 is an algorithm with a proposed approach to the diagnosis of CAD in diabetes in an attempt to integrate ischemia and atherosclerosis imaging. The thresholds utilized to define risk were chosen based on personal opinion and the data published in the current medical literature. Whether all asymptomatic diabetic patients should be tested remains debatable and unlikely to be financially affordable for society. To make asymptomatic screening more affordable, at least one of the following conditions should probably be present: one additional cardiovascular risk factor, an abnormal resting electrocardiogram, microalbuminuria, or autonomic neuropathy.

Cardiovascular disease remains the main health concern in diabetes. Conventional imaging tools based on ischemia imaging perform well at detecting obstructive coronary luminal disease. Nonetheless, the high event rate seen in subjects with normal functional and perfusion studies, as well as the frequent occurrence of silent events in diabetic patients, provide support for disease detection in its preclinical stages. Atherosclerosis imaging modalities, such as CT for CAC and ultrasonography for CIMT, offer an opportunity to improve our ability to acquire prognostically important information. Indeed, the integration of myocardial ischemia imaging tools and tools to image atherosclerosis (93) may improve outcome in high-risk patients through the early implementation of aggressive medical and interventional therapies. Furthermore, sequential quantification of progression of atherosclerosis may allow monitoring of treatment efficacy and hopefully facilitate adherence to recommended treatment regimens. Continued research will be needed to confirm that the integration of several imaging modalities improves clinical outcome in a cost-effective manner.