Methods
Study Sample
Study participants were from the Oji-Cree community of Sandy Lake, Ontario, an isolated reserve located at the 55
th parallel of latitude, in the subarctic boreal forest of central Canada. Baseline demographic, clinical, and biochemical attributes were gathered during the Sandy Lake Health and Diabetes Project of 1993–1995, a prevalence study of type 2 diabetes [
3]. Seven hundred and twenty eight members of this community (72% of the total population) aged 10 years and above, participated in the original survey. In a follow-up study initiated in 2001 [
4], 278 adults free of coronary heart disease had US assessment of the carotid arteries. Of these, 161 had participated in the original prevalence study and had baseline measurements. For the current analysis, 49 subjects, 46 with type 2 diabetes and 3 with impaired glucose tolerance (IGT), were selected and matched for sex and age (± 3 years) with a normoglycemic control subject. Of the subjects with diabetes, 43.5% were receiving oral medication, and 4.4% were receiving insulin. For simplicity, from this point forward the 46 subjects with type 2 diabetes and 3 with IGT will be referred to as the "diabetic" group. Signed informed consent was obtained from all participants and study approval was granted by both the Sandy Lake First Nation Band Council and the institutional review boards of the University of Toronto and the University of Western Ontario.
Clinical and biochemical baseline analysis
Body weight, height, waist circumference, and blood pressure were measured by standardized procedures [
3]. Hypertensive individuals were defined as subjects having either a blood pressure reading >140/90 mmHg or taking anti-hypertensive medication. Measurements of fasting blood analytes, including triglycerides, insulin, lipoproteins, and total cholesterol were performed as described [
3].
Ultrasound examination
All subjects were examined using an HDI 5000 scanner equipped with Sono-CT compound imaging and a L12-5 transducer (Advanced Technology Laboratories, Bothell, Washington) that had been flown to the community and housed within the Diabetes Research Center. Common carotid US images for all participants were gathered over a 4-week period and from this data, IMT, total plaque area (TPA) and TPV measurements were determined. TPA was strongly correlated with TPV in these subjects (r = 0.921, P < 0.0001), and thus for simplicity, only TPV measurements were compared against IMT.
IMT measurement
IMT was determined as previously described [
5,
6]. Briefly, a single observer, blinded to subjects' vascular risk, measured combined thickness of intima and media of the far wall of both common carotid arteries. Images were recorded from an anterolateral longitudinal view. The still images were analyzed using computerized edge-detection software (Prowin™) [
7]. Using a step-wise algorithm, conditional sets of "edges" (consisting of lumen-intima and media-adventitia echoes) were located within the image and then tested for "edge strength", with the subsequent deletion of weak edge points. Once all acceptable edge points were identified, boundary gaps were filled by linear interpolation. The distance between lumen-intima and media-adventitia boundaries was then measured to calculate IMT. Mean IMT was computed from 120 measurements over a 10 mm span ending 5 mm proximal to the transition between the common carotid and bulb regions. Intra- and inter-operator coefficients of variation of 3.0 and 3.1%, respectively and intra- and inter-operator intraclass correlations were both 0.97 [n = 50] (both P < 0.01).
TPV measurement
TPV was determined as previously described [
5,
6]. Briefly, 3D ultrasound images were acquired with a mechanical linear scanning system and analyzed with L3Di visualization software [Life Imaging Systems Inc., London, Ontario]. Plaque volumes were measured using manual planimetry: each 3D image was 'sliced' transversely at an inter-slice distance of 1 mm, moving from one plaque edge to the other. Plaque boundaries were traced using a mouse driven cross-haired cursor. Slice areas were summed and multiplied by inter-slice distance to calculate plaque volume. For this analysis, TPV was defined as the sum of all plaque volumes on one side between the clavicle and angle of the jaw. Intra- and inter-observer reliability were 0.94 [n = 40] and 0.93 [n = 40], respectively (both P < 0.01).
Statistical analysis
SAS version 8.2 (SAS Institute, Cary, NC) was used for all statistical comparisons. Data are presented as means ± SE. The distribution of BMI, plasma total cholesterol, triglycerides, high density lipoprotein (HDL), and serum insulin, were non-normal in this data set, and thus were logarithmically transformed (natural log) and subjected to analysis of normality. IMT and TPV were also normalized using the inverse transformation of IMT and the square root transformation of TPV. The transformed variables were used for parametric statistical analyses, but the untransformed values are presented in Table
1. For continuous variables, differences between the groups were tested by the Student's
t test; categorical variables were tested by
γ
2 analysis. Statistical significance was taken at nominal
P < 0.05 for all comparisons. Correlation analysis between IMT and TPV was performed using Pearson correlation analysis.
Table 1
Clinical and biochemical attributes of Oji-Cree at baseline and carotid measurements after 7 years
number/females | 49/26 | 49/26 | |
attributes at screening | | | |
age (years) | 40.3 ± 1.8 | 40.4 ± 1.8 | NS (0.96) |
duration of diabetes (years) | 2.20 ± 0.62 | - | - |
current smokers (%) | 18.4 | 10.2 | NS (0.25) |
hypertensive (%) | 36.7 | 32.7 | NS (0.67) |
antihypertensive treatment (%) | 16.3 | 8.2 | NS (0.22) |
body mass index (kg/m2) | 29.6 ± 0.5 | 29.5 ± 0.6 | NS (0.75) |
waist circumference (cm) | 101 ± 1.4 | 99.8 ± 1.5 | NS (0.63) |
TC:HDL ratio | 4.95 ± 0.18 | 4.10 ± 0.17 | 0.0006 |
plasma triglycerides (mmol/L) | 2.25 ± 0.14 | 1.53 ± 0.11 | <0.0001 |
plasma glucose (mmol/L) | 10.79 ± 0.63 | 5.53 ± 0.07 | <0.0001 |
serum insulin (pmol/L) | 157 ± 12 | 133 ± 9 | NS (0.16) |
time elapsed since screening (years) | 7.34 ± 0.10 | 7.33 ± 0.10 | NS (0.96) |
mean IMT (μ m) | 795 ± 19 | 789 ± 26 | NS (0.49) |
mean TPV (mm3) | 109.9 ± 23.0 | 64.0 ± 17.0 | 0.037 |
Hypothetical sample sizes were calculated using the online calculator for normal power calculations (normal distribution 2-sample equal variances) found at the UCLA Department of Statistics website [
8]. This statistical tool calculates the sample size for two-sided tests of hypotheses on normal means, when the common population standard deviation is known, using the following formula:
where n
1 and n
2 are the sample sizes of the two groups, u
α/2and u
β
are the lower limits of the cumulative standard normal probability integrals,
σ is the known common standard deviation,
δ
0 is the least favourable non-negative difference consistent with the test hypothesis, and
δ
1 is difference in the population means [
9].
Using the normalized transformed means from the case-control study, the mean standard deviation as the common standard deviation (SD), and a significance level of 0.05, sample sizes were calculated. Transformed means and standard deviations were 1.29 (diabetic) vs 1.33 (non-diabetic), SD 0.237, for IMT (inverse transformation), and 7.81 (diabetic) vs 4.94 (non-diabetic), SD 6.71, for TPV (square root transformation). Power was tested at 0.70, 0.80 and 0.90.
Discussion
We report: 1) elevated TPV for diabetic subjects vs non-diabetic subjects following a 7 year period (P = 0.037); 2) increased sensitivity of the TPV measurement in comparison to IMT measurements for diabetic subjects.
A previous study by Hunt
et al., convincingly showed that early atherogenesis is present before the onset of diabetes, and thus is not solely dependent on the clinical manifestation of diabetes [
10], but rather, both conditions (diabetes and cardiovascular disease) originate from a "common soil" of pro-inflammatory and pro-atherogenic risk factors [
11]. Our study examined atherosclerosis burden after the diagnosis of diabetes had been made and found that diabetic subjects had TPV measurements that were 1.7-fold higher than non-diabetic subjects (
P = 0.037). Similar observations have been made previously, such as in the Insulin Resistance Atherosclerosis Study (n = ~1200), where subjects with diabetes had increased carotid wall thickness at baseline (~70
μ m increase in common carotid, ~130
μ m increase in internal carotid) [
12], and, over a five year time period, had IMT progression rates approximately twice as high as non-diabetic subjects (7.2 ± 1.9
vs 3.8 ± 1.3
μ m/year) [
13]. The Bruneck Study (n = 826) found that type 2 diabetes was a strong independent predictor (OR = 5.0,
P < 0.001) of US-determined, advanced stenotic atherosclerosis, defined by >40% lumenal narrowing [
14].
Significant differences were also noted in the lipid profile of diabetic subjects
vs controls, with a greater TC:HDL ratio and elevated triglycerides observed for those with diabetes. Glucose intolerance has been previously reported as an independent predictor of both triglycerides and HDL cholesterol [
15]. This worsening of lipids with glucose intolerance may potentially explain the differences between the two groups in terms of plaque volume progression.
While a significant difference was found for TPV, no significant difference was found for IMT between diabetic and non-diabetic subjects, although IMT tended to be greater for diabetic subjects. The lack of a significant difference undoubtedly is related to the low number of subjects, but it is apparent that it may also be due to the relative insensitivity of carotid IMT as a surrogate marker for atherosclerosis in patients with type 2 diabetes. The potential increased sensitivity for TPV was reflected by our finding of a significant difference for a relatively small study sample. An important feature for enhanced sensitivity is found in the wider dynamic scale ranges for TPV compared to IMT: ~90% of the IMT measurements fall within a relatively narrow 0.55–1.0 mm range, whereas ~60% of TPV values fall within a range of 5–500 mm3. Thus, the dynamic range of measurements varied by ~100-fold for TPV compared to ~2-fold for the IMT. Furthermore, the quantity being measured (mm3 for TPV vs mm for IMT) is much larger for TPV, so that in relation to the resolution of the ultrasound method, TPV is much easier to quantify both accurately and reliably.
In designing studies it may be worthwhile to consider using TPV in addition to the traditional IMT measurement, as a primary endpoint, due to its potentially greater sensitivity and discrimination, which may have the benefit of greater statistical power, allowing for the use of a small sample size. For example, to observe a significant difference in IMT using values similar to those observed in this case-control study, (ie. 6
μ m difference in mean IMT) (Table
2), thousands of subjects are required to achieve a statistical power of even 0.70. In contrast, only a few hundred subjects are required to observe a statistical power of 0.90 for the TPV difference seen in our study (45.9 mm
3). Performing studies with over a thousand subjects [
10,
13] imposes limitations and difficulties, which could be minimized by using a more sensitive technique such as TPV measurement. Studies on the effects of therapy on atherosclerosis using TPV as the outcome have already been effectively carried out with much smaller sample sizes. Ainsworth
et al. [
16] have shown significant differences between active atorvastatin and placebo in 3 months, with a sample size of only 20 per group.
IMT and TPV, however, are not interchangeable. The correlation between IMT and TPV in this study, although statistically significant, was moderate (r<0.7) and, as has been noted previously, these different US-derived measures of carotid artery morphology likely represent distinct attributes of atherosclerosis [
5]. IMT may reflect wall hyperplasia or hypertrophy related to hypertension [
17] and TPV may reflect the later stages of plaque formation and the total carotid disease burden in a subject [
5]. This may be more relevant for the disease process of diabetes. It is also important to note that in our previous work [
5] IMT correlated better with hypertension and age, and as blood pressure and age were balanced between the groups, this might explain the lack of difference observed here for IMT. It may be the case that IMT would capture the atherosclerotic disease burden more effectively in hypertensive diabetics, and would be a more appropriate outcome measure for studies aimed at improving hypertension in diabetics. These implications must be taken into consideration when designing and analysing a study, and more research is needed to provide a complete understanding of how these US measures fit into the pathophysiology of atherosclerotic disease.
Considering that by nature atherosclerotic plaque is not evenly distributed along the arterial wall, it is logical to develop methods that will attempt to quantify the total plaque burden more accurately. With the relatively larger volumes being measured for TPV assessment
vs IMT or plaque area measurements, there are the accompanying benefits of more statistical power and less patients required per study, and also the potential benefit of less time required to observe significant differences between study groups. However, at present, using TPV as an assessment tool remains a labour-intensive task and has the additional disadvantage of not yet being widely used and standardized. Furthermore, in studies among children or other very young subjects, ultrasound evaluation of the arteries may be limited to IMT simply because plaque would not yet be developed in most cases. However, some plaque can be identified in most subjects above age 35 or 40 [
2].
Acknowledgements
The authors acknowledge Khalid Al-Shali for his work in measuring the plaque volume of the study subjects. The authors also gratefully acknowledge the chief, council and community members of Sandy Lake First Nation and the Sandy Lake community surveyors (Ken Goodwin, Edith Fiddler, Louisa Kakegamic, Tina Noon, Madeline Kakegamic, Elda Anishinabie, Annette Rae, Connie Kakegamic, and Mary Mamakeesic), whose partnership and co-operation was essential in the design and implementation of this project.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
RLP participated in the design of the study, analysis of the data, and writing of the manuscript. JDS, AAH, AF, AJGH, BZ, and SBH provided patients and data for the study, and assisted with manuscript revisions. RAH participated in the design of the study and writing of the manuscript. All authors read and approved the final manuscript.