Ethnic Differences in the Relationship Between Albuminuria and Calcified Atherosclerotic Plaque: The African American-Diabetes Heart Study

Siblings concordant for type 2 diabetes were recruited from internal medicine clinics and community advertising in the Diabetes Heart Study (DHS) (8). Diabetes was diagnosed after the age of 34 years in the absence of historical evidence of ketoacidosis. In addition, 213 unrelated African American subjects were subsequently recruited in the AA-DHS using these same diagnostic criteria, except that type 2 diabetes was diagnosed after the age of 30 years. Subjects who underwent prior coronary artery bypass surgery or carotid endarterectomy were excluded from this analysis because it was felt that the CP mass score in relevant arteries would be affected by these procedures; those with prior myocardial infarction or stroke were included. The study was approved by the institutional review board at the Wake Forest University School of Medicine, and all participants provided written informed consent.

Participant examinations were conducted in the General Clinical Research Center of the Wake Forest University School of Medicine and included interviews for medical history, current medications and health behaviors, measurements of body size, resting blood pressure, 12-lead electrocardiogram, fasting blood sample, and morning spot urine collection. Laboratory assays included urine albumin and creatinine for ACR, total cholesterol, LDL, HDL, triglycerides, A1C, and fasting serum glucose. Renal function was assessed using serum creatinine concentration and Modification of Diet in Renal Disease estimation of GFR. History of CVD was provided by self-report and chart review. Hypertension was defined as self-report of a physician's diagnosis, blood pressure >140/90 mmHg, or use of antihypertensive medications. Antihypertensive medications were grouped into drug classes, particularly ACE inhibitors (ACEIs) and angiotensin receptor blockers (ARBs), medicines known to reduce urine ACR.

CP was measured in the coronary and carotid arteries and infrarenal abdominal aorta using single and multidetector CT systems incorporating a standardized scanning protocol based on those currently implemented in the National Heart, Lung, and Blood Institute's Multi-Ethnic Study of Atherosclerosis studies, and these methods have been reported previously (8,16). Traditionally, the Agatston score, also called the calcium score, has been used to report results. However, the nature of this scoring system adds noise to the CT measurement of CP compared with CP volume–based measures (17). In this report, we used the calcium mass score (SmartScore; GE Healthcare, Waukesha, WI), which is derived from the volume score but in addition accounts for the density of CP on a pixel-by-pixel basis. Additional scoring parameters included a 90-Hounsefield unit threshold and two adjacent pixels to define the maximum calcified lesion size, and the program accounted for slice thickness.

A series of generalized estimating equations assuming exchangeable correlation and using the empirical estimate of the variance to adjust for familial correlation was computed to test for associations between urine ACR and CP (18). The natural log of (urine ACR + 1) was calculated to minimize the influence of extremely large covariate values on parameter estimates in these models. The Box-Cox method was applied to identify the appropriate transformation of each outcome variable that would best approximate the distributional assumptions of conditional normality and homogeneity of variance of the residuals (19). The natural logs of (CorCP + 1), (CarCP + 1), and (AorCP + 1) were analyzed. We applied principal component analysis (PCA) to compute a linear combination of CorCP, CarCP, and AorCP that we used as the primary outcome variable. The principle component encompasses the burden of calcified atherosclerotic plaque in these three vascular beds. There is a strong correlation between the amount of CP in the coronary arteries, carotid arteries, and aorta of individuals with diabetes (20). PCA was conducted using the correlation matrix instead of the covariance matrix to account for the differences in measurement scales between the three variables. The first principal component explained ∼70% of the total variation observed in these three variables and is computed as

The variable PCA1_CP was computed using the standardized values of CarCP, CorCP, and AorCP. The Box-Cox transformation that best approximated the distributional assumptions of the model was the inverse PCA1_CP, i.e., (PCA1_CP)⁻¹. For PCA1_CP, models were adjusted for age, sex, BMI, and GFR. Subsequent analyses were further adjusted for use of ACEIs or ARBs, blood pressure, lipid levels, A1C, and smoking. Standard regression diagnostics for collinearity and influence were computed for each model reported.

A total of 1,427 individuals with type 2 diabetes had complete data for analysis. Of these, 1,008 were European Americans from 445 families and 419 were African Americans from 297 families. Coronary artery bypass grafting or carotid endarterectomy was performed in 173 European American and 26 African American participants, and these individuals were excluded. Tables 1 and 2 contain demographic and clinical information in study participants. Urine ACR was <30 mg/g in 60.3% of African American and 71.0% of European American participants, 30–299 mg/g in 29.0% of African American and 23.5% of European American participants, and ≥300 mg/g in 10.7% of African American and 5.5% of European American participants (Table 1). Median Modification of Diet in Renal Disease estimated GFR was 1.34 ml/s in African American and 1.09 ml/s in European American participants (P value for ethnic difference 6.3 × 10⁻²²), with graded reductions in GFR from normoalbuminuric to overt proteinuric participants in both ethnic groups (Table 2).

Coronary CP was detectable in 81.7 and 93.3% of African American and European American participants, respectively (P value for ethnic difference 1 × 10⁻⁹). Because distributions of CP mass scores are highly skewed, mean scores should be interpreted cautiously and median values are more reflective of central tendency. In African American subjects, the median CorCP mass score was 53.5 (mean 697), median CarCP mass score was 3 (mean 170), median AorCP mass score was 761 (mean 4,934), and median ACR was 17.5 mg/g (mean ± SD 230 ± 758). In European American subjects, the median CorCP mass score was 291 (mean 1,277), median CarCP mass score was 35.5 (mean 275), median AorCP mass score was 3,237 (mean 9,803), and median ACR was 13.4 mg/g (110 ± 531); these ethnic differences were highly significant (P = 8.2 × 10⁻¹¹ to 2.39 × 10⁻¹⁴). Statistically significant and graded increases in median vascular CP mass score were observed with increasing levels of ACR in European Americans with P values = 1.6 × 10⁻⁴, 2.7 × 10⁻⁴, and 1.0 × 10⁻⁶, respectively, for AorCP, CorCP, and CarCP using traditional ACR cut points <30, 30–299, and >300 mg/g to stratify the sample. When stratification was done using quartiles of urine ACR, P values improved to 4.7 × 10⁻⁷ and 1.2 × 10⁻⁸ for AorCP and CorCP but decreased to 4.8 × 10⁻⁵ for CarCP. In African Americans, only CarCP had a significant positive relationship with urine ACR (P = 9.6 × 10⁻³ using ACR quartiles, P = 5.8 × 10⁻³ using traditional ACR cutoff values) (Table 2 and Supplementary Table 1, available in an online appendix at http://care.diabetesjournals.org/cgi/content/full/dc09-1253/DC1).

Table 3 focuses on the relationship between urine ACR and CP, separately in each vascular distribution and for the first principal component of CP. In age-, sex-, BMI- and GFR-adjusted analyses, significant associations were seen between urine ACR with each vascular bed as well as for the principal component in European Americans, whereas significant relationships were not present in African Americans. In addition, significant ethnic differences were observed in the relationship between urine ACR and the principal component for CP (P = 0.01) and a trend for urine ACR with CarCP (P = 0.06).

Table 4 reveals the parameter estimate for the relationship between log urine ACR and the first principal component for CP across the three vascular beds, adjusted for additional covariates including ACEIs or ARBs. The log urine ACR was significantly associated with the first principal component for CP in European Americans (parameter estimate ± SEM −0.0387 ± 0.008; P = 8.34 × 10⁻⁷), whereas a significant relationship was not observed in African Americans (parameter estimate −0.0068 ± 0.011; P = 0.54), and there was a significant ethnic interaction (P = 0.01). The Box-Cox transformation suggested using a decreasing function of the principal component; therefore, the direction indicated for the sign of the parameter estimate should be inverted when one reverts to the original scale. Ethnic differences in the relationship between sex and CP were also detected (P = 0.0001). Although there was no association between sex and the first principal component for CP in African Americans (P = 0.37), sex was strongly associated with CP in European Americans, with men having greater CP than women (P = 6.4 × 10⁻⁷).

This report characterized ethnic differences in the relationship between calcified atherosclerotic plaque and albuminuria in African Americans and European Americans with type 2 diabetes and without advanced renal dysfunction. Albuminuria and kidney disease are important risk factors for the development of atherosclerosis (21). After adjustment for the covariates of age, sex, BMI, and GFR, a significant and positive relationship was observed between albuminuria and CP in European Americans but not in African Americans. Consequently, the ethnic difference in this relationship suggests that lower amounts of CP in African Americans with diabetes, with attendant reduced risk for myocardial infarction relative to that in European Americans, could be related in part to a differential effect of albuminuria. Myocardial infarctions occur significantly less often in African American patients receiving renal replacement therapy compared with European American patients, and CVD is the leading cause of death in subjects with nephropathy (2,14). Our report intentionally examined European American and African American subjects with diabetes and relatively preserved kidney function, thereby minimizing the potential effects of exogenous vitamin D and vitamin D analogs, phosphate binders, and calcimimetic agents on CP. These medications and associated hyperphosphatemia may have an impact on risk for development of subclinical CVD in patients with advanced nephropathy (22).

A recent report confirmed the lower risk of CVD in African Americans with advanced stages of CKD, relative to European Americans (23). There is mounting evidence that biological differences may contribute to ethnic differences in risk for CVD and development of CP with subsequent CVD (6,,,–10). Given equal access to health care, African Americans with type 2 diabetes and CKD face lower risks for CVD events than European Americans (4,5,14,23). In the current report, ACEIs, ARBs, and lipid-lowering medications were prescribed nearly equally in African American and European American participants. Previously, biological factors have been implicated in ethnic disparities in the development of kidney failure (24). Ethnic differences in MYH9 risk allele distribution reveal that inherited factors are capable of causing ethnic differences in the incidence of common diseases. It remains possible that other inherited factors contribute to the ethnic differences in the development of CP relating to albuminuria.

Markedly lower levels of CP are observed in African American subjects relative to those in European American subjects with and without diabetes (6,,,–10). Although ethnic differences exist in the presence of CP, the severity of CP is strongly associated with risk for coronary events among individuals of all ethnic groups. In the Multi-Ethnic Study of Atherosclerosis (MESA), Americans of European, African, Hispanic, and Chinese descent demonstrated equivalent risk for CVD events based on level of CP (12). Conventional CVD risk factors may have ethnic-specific effects on risk for CP (20). Therefore, it is important to assess whether ethnic differences in the relationship between albuminuria and CP exist because African Americans are known to develop type 2 diabetes-associated nephropathy with resultant albuminuria more often than European Americans (2). In addition, calcium metabolism clearly differs between African Americans and European Americans. Although African Americans typically ingest less dietary calcium than European Americans, they have denser bone with lower rates of osteoporosis and skeletal resistance to the effects of parathyroid hormone (25). Related phenomena may prove to be important in the development of CP because vascular endothelial cells assume osteoblastic phenotypes and deposit bone matrix in blood vessels.

Albuminuria markedly increases the risk for CVD in European Americans with type 2 diabetes (15). The current analyses emphasize the importance of ethnic differences in the effect of albuminuria on risk for CVD. We studied a population of subjects with median 8-year durations of diabetes, and diabetes would be expected to magnify atherosclerotic vascular disease. However, there is no a priori reason to suspect that the relationship between albuminuria and CP would be different in African Americans and European Americans without diabetes. Studies in these individuals remain to be performed. It is possible that the lower levels of CP in African Americans despite the presence of similar or more severe conventional CVD risk factors, as well as the differential effects of atherosclerotic risk factors on the amounts of CP, relate to inherited factors because CorCP is a heritable trait. The AA-DHS will soon perform mapping by admixture linkage disequilibrium in an attempt to detect genes underlying development of CP.

Potential weaknesses of this report are reliance on cross-sectional data, a single measurement of albuminuria, and inclusion of only subjects with diabetes. Equivalent diabetes durations and similar prescription of lipid-lowering and proteinuria-reducing medications in African American and European American participants, as well as consistency of CorCP scores when contrasted with other reports, suggest the lack of bias. Albuminuria is both a marker of renal impairment and a sign of systemic endothelial dysfunction. It seems likely that the balance between these relationships may differ between African Americans and European Americans. In the future, it will be important that longitudinal studies clarify the relationship between classic and novel CVD risk factors on ethnic differences in susceptibility to CP. These analyses need to be performed in subjects with and without diabetes.

In summary, albuminuria is strongly associated with CP in European Americans with type 2 diabetes but not in African Americans with type 2 diabetes. Ethnic differences in the relationship between CP and albuminuria may contribute to the lower levels of CP and reduced risk for myocardial infarction in African Americans with diabetes. It remains to be determined whether interventions that reduce albuminuria and preserve renal function will prevent the development of calcified atherosclerotic plaque and reduce CVD rates equally in African Americans and European Americans with type 2 diabetes.

This study was supported in part by the General Clinical Research Center of the Wake Forest University School of Medicine (Grant M01-RR-07122), National Institute of Diabetes and Digestive and Kidney Diseases (Grant R01-DK-071891) (B.I.F.), National Institute of Arthritis and Musculoskeletal and Skin Diseases (Grant R01-AR-048797) (J.J.C.), and National Heart, Lung, and Blood Institute (Grant R01-HL-67348) (D.W.B.).

No potential conflicts of interest relevant to this article were reported.

We acknowledge the cooperation of our participants and study recruiters Carrie Smith and Cassandra Bethea.