Author + information
- Received April 5, 2005
- Revision received October 6, 2005
- Accepted October 10, 2005
- Published online March 21, 2006.
- Lalit Kalra, PhD, FRCP⁎,⁎ (, )
- Curtis Rambaran, MRCP⁎,
- Elizabeth Iveson, MRCP†,
- Philip J. Chowienczyk, MD, FRCP⁎,
- Ian Hambleton, PhD†,
- James M. Ritter, MD, FRCP⁎,
- Ajay Shah, MD, FRCP⁎,
- Rainford Wilks, DM, FRCP† and
- Terrence Forrester, DM, PhD†
- ↵⁎Reprint requests and correspondence:
Dr. Lalit Kalra, Department of Medicine, King’s College London School of Medicine, Denmark Hill Campus, Bessemer Road, London SE5 9PJ, United Kingdom.
Objectives This study sought to compare vascular reactivity and carotid intima media thickness (CIMT) between Afro-Caribbean people in the United Kingdom (UK) and the West Indies and Afro-Caribbean and Caucasian people in the UK.
Background Attenuated vascular reactivity and increased CIMT in black patients is seen as evidence for predisposition to vascular disease, but no comparisons exist between Afro-Caribbean people in different settings, which can provide insight into non-inherited determinants of increased ethnic susceptibility.
Methods A representative community sample of 81 healthy Afro-Caribbean people and 101 Caucasian people in the UK was compared with 197 matched Afro-Caribbean people in Jamaica. Small vessel reactivity was assessed by measuring the absolute change from baseline in the reflection index (RI) of the digital volume pulse during intravenous infusion of albuterol (5 μg/min, ΔRIALB) and glyceryl trinitrate (5 μg/min, ΔRIGTN). The CIMT was measured ultrasonographically in the distal 1 cm of the common carotid artery.
Results Mean ΔRIALBwas 4.2 percentage points (95% confidence interval [CI], 2.3 to 6.1, p < 0.001) lower in UK Afro-Caribbean people compared with Jamaican Afro-Caribbean people and 2.6 percentage points (95% CI, 0.4 to 4.7, p = 0.02) lower compared with Caucasian people, after adjusting for vascular risk profile. Adjusted mean CIMT of UK Afro-Caribbean people was 0.13 mm (95% CI, 0.08 to 0.17, p < 0.001) greater compared with Jamaican Afro-Caribbean people and 0.05 mm (95% CI, 0.01 to 0.10, p = 0.02) greater compared with Caucasian people.
Conclusions Healthy UK Afro-Caribbean people have greater and Jamaican Afro-Caribbean people have less impairment of vascular reactivity and intima media thickness compared with UK Caucasian people, suggesting that potentially modifiable environmental interactions may contribute to excess vascular disease in Afro-Caribbean people.
Several studies have shown a higher incidence of vascular diseases in people of African descent compared with Caucasian people in Western environments (1–3). Although much of the excess risk is attributed to a higher prevalence of hypertension and other risk factors in this population, ethnicity itself has emerged as an important independent risk factor (4–6). Ethnicity represents a clustering of genetic and environmental factors. Phenotype expression can be represented schematically by:Hence, differences attributed to ethnicity may not be caused by genetic differences alone, but may also reflect differences in environmental influences or interactions between genetic heterogeneity and environmental pressures (7).
The concept of an inherited predisposition to vascular disease in people of African descent is supported by studies in the United Kingdom (UK) and the U.S., which show that healthy Afro-Caribbean patients have attenuated vascular responsiveness and increased artery intima thickness (8–15), widely regarded as functional and morphologic markers of early atherosclerotic processes (16,17). Other studies have shown differences in the frequency of gene polymorphisms for several proteins that modulate vascular function (18–20) between Afro-Caribbean and Caucasian patients, further supporting this concept. However, most observations of impaired vasomotor function in African-origin people have been based on direct comparisons between small numbers of selected Caucasian and Afro-Caribbean or African-American volunteers in Western settings, and are potentially vulnerable to sampling bias. Many studies lack the power to evaluate the effect of ethnicity independent of blood pressure, blood glucose, or serum cholesterol levels, all known to be higher in healthy Afro-Caribbean people. There are few studies comparing intermediate phenotypes of vascular disease between matched healthy Afro-Caribbean people living in the UK and in the West Indies, a design that could provide insight into the impact of non-inherited determinants of ethnic susceptibility to vascular disease.
The objective of this study was to investigate differences in vascular reactivity between Afro-Caribbean patients living in two different environments, as well as between Afro-Caribbean and Caucasian patients living in the same environment, after adjusting for differences in blood pressure, blood glucose, serum cholesterol, body mass index (BMI), and smoking status. Similar comparisons of differences in carotid intima media thickness (CIMT) were undertaken to assess early changes in arterial morphology.
A major threat to the validity of the study lies in potential differences in patient selection and measurement techniques between the UK and Jamaica. We validated and implemented rigorous standardized protocols for epidemiologic sampling, defining ethnicity and vascular risk factors, anthropometric measurement, and collection and storage of biological specimens. Pilot experiments were undertaken to standardize techniques for the measurements of vascular reactivity and intima media thickness between observers and to establish inter- and intra-rater reliability. The same make and model of digital venous plethysmography equipment was used to measure vascular reactivity in both settings and was calibrated for concordance before undertaking the study. The mean coefficient of variation for the change in reflection index with albuterol (ΔRIALB) was 4.5 ± 1.1 percentage points (within-patient) and 5.2 ± 3.2 percentage points (between patients). The CIMT was measured using the same ultrasound equipment for all patients in both settings (Accuson Sequoia in the UK and Logiq500GE in Jamaica). Because there was a possibility that a systematic inter-machine bias might confound comparisons between the two settings, concordance between the two machines was established by calibration using the same Gammex RMI precision small part phantom (RMI 404GSLE; Middleton, Wisconsin). The mean bias in CIMT measurements was 0.001 ± 0.08 mm (within patient) and 0.006 ± 0.11 mm (between sonographers). The intra-correlation coefficient for between manual and automated readings was 0.95 (95% confidence interval [CI], 0.93 to 0.97) and between the two machines was 0.92 (95% CI, 0.88 to 0.96).
The family practice register is the most accurate and comprehensive register of the general population in the UK (21), and individuals in the age range of 45 to 64 years registered at one large general practice were sampled randomly, stratified by age group, gender, and ethnicity. The practice was located in a relatively homogeneous and predominantly middle class area (social class II and III) (22) of South London and had a list size of 32,000 patients, of whom 28% were black people of Caribbean descent. Ethnicity was self-defined; only patients with both parents and all four grandparents belonging to the same ethnic group and of Caribbean descent were included (23). If a potential patient failed to meet inclusion criteria or declined, the next person on the list who matched age, gender, and ethnicity characteristics was selected.
A similar age- and gender-stratified sample was recruited from the Kingston Metropolitan area in Jamaica. Enumeration districts were selected according to techniques described previously (24), and five persons in each age/gender category were screened for eligibility from successive households in each enumeration district. No more than one individual in each age/gender grid was selected at each household or from a family group to avoid bias caused by large households or related families.
To minimize the confounding effects of conventional risk factors or pharmacological interventions, patients with a history of stroke, ischemic heart disease, peripheral vascular disease, hypertension, diabetes mellitus, or hypercholesterolemia or who were taking regular medication were excluded. Patients gave written informed consent, and the study was approved by the research ethics committees of King’s College Hospital and the University Hospital of the West Indies.
Details of age, gender, ethnicity, place of birth, duration of residence, and smoking habits were recorded in all patients (Table 1).Blood pressure was measured in seated patients using a mercury sphygmomanometer and appropriate-sized cuff. Three measurements were taken at five-minute intervals of quiet rest, and the average of second and third reading was used. The BMI and waist-to-hip ratio were calculated using standardized validated techniques. A fasting blood sample was taken for measuring blood glucose and lipid levels.
Assessment of vascular reactivity
Vascular reactivity was measured non-invasively using digital volume pulse (DVP) plethysmography (25,26). All experiments were performed in the morning in temperature-controlled (24 ± 1°C) vascular laboratories with patients resting supine. An infrared photoplethysmograph (MicroMedical, Gillingham, United Kingdom) was placed on the index finger of the left hand of the patients. The DVP waveform consists of a systolic component (first peak) transmitted directly from the left ventricle and a diastolic component (second peak) formed by reflection from small muscular arteries (Fig. 1).The reflection index (RI) is the height of the diastolic peak expressed as a percentage of the systolic peak and is a measure of the amount of pulse wave reflection, determined predominantly by the tone of small muscular arteries. The beta2-adrenoceptor responses in these arteries are mediated through activation of the endothelial L-arginine/nitric oxide pathway (25–29), and the contour of the DVP is highly sensitive to vasodilation by small doses of beta-agonists and glyceryl trinitrate (GTN), which do not change heart rate or blood pressure.
After intravenous cannulation and infusion of 0.9% saline at 0.5 ml/min, patients rested supine for 30 min. Baseline measurements of heart rate, blood pressure (Omron 705CP, Kyoto, Japan) and DVP were obtained at 5-min intervals for 15 min; 5 μg/min of albuterol (Allen and Hanbury, Uxbridge, United Kingdom) was infused for 30 min, during which hemodynamic measurements were recorded at 3-min intervals. After a washout period of 60 min during which 0.9% saline was infused, all parameters returned to baseline. Next, 5 μg/min of GTN (Faulding, Warwickshire, United Kingdom) was then infused for 30 min and hemodynamic measurements were recorded at 3-min intervals. The means of the four readings immediately before starting the drug infusions were used as the baseline for each agent. The DVP response was measured as the mean absolute change in RI from baseline to between 12 and 21 min for albuterol (ΔRIALB) and from baseline to between 9 and 12 min for GTN (ΔRIGTN). A preliminary study had shown that the mean within-patient coefficient of variation of the response curve for ΔRIALBand ΔRIGTNwas least between these times.
Measurement of carotid artery intima thickness
An Accuson Sequoia 512 machine (Logiq500GE in Jamaica) with an 8-MHz transducer was used to image the carotid arteries. Each examination cycle included sequential longitudinal and transverse views of the common carotid artery, the carotid bifurcation, and the internal carotid artery bulb. Settings for depth-gain compensation, pre-processing, persistence, and post-processing were held constant. All ultrasonic examinations were stored digitally for subsequent offline processing. Images from the UK and Jamaica were analyzed together by an experienced ultrasonographer blinded to patient identity and ethnicity.
The CIMT was defined as the mean of differences between the blood/intima borderline and the media/adventitia borderline and measured on the far wall of the left and right common carotid artery. The distal 1 cm of the common carotid artery just proximal to the bulb was measured. Mean CIMT was measured by using a semi-automated computer analysis system over the 1 cm segment and confirmed by the mean of five manual readings at 20-mm intervals along the common carotid artery.
The primary outcome measure for the study on which sample size calculations were based was ΔRIALBbecause previous studies have consistently shown significant ethnic differences in endothelium-dependent vasodilatation in healthy patients (8,10–13). A preliminary study of digital venous plethysmography in 40 patients showed that the mean standard deviation for ΔRI was 11 percentage points and that a sample size of 77 patients in each group would have 80% power to detect a difference in ΔRI of 5 percentage points using a two-grouped ttest with a 0.05 two-sided significance level. Extending this calculation to three ethnic groups increased the required sample size to 80 per group using a one-way analysis of variance. This sample size also had 80% power at the 5% significance level to detect a difference of 0.07 mm in CIMT between the UK and the Jamaican Afro-Caribbean people.
Each continuous outcome measure, confounder, and potential risk factor was compared across the three ethnic groups using one-way analysis of variance. Statistically important results on this analysis were investigated further by three pairwise tests for each analysis of variance and adjusted for multiple testing using Bonferroni correction. The outcome measures of interest were ΔRIALBand CIMT. For each measure, vascular function was modeled in three stages. In stage one, the confounding effects of age, gender, height, baseline RI, and heart rate were assessed. In stage two, the univariate effect of each potential predictor (BMI, blood pressure, fasting glucose, total cholesterol, high-density lipoprotein, triglycerides, and smoking) was assessed after adjusting for confounders identified in stage one. In stage three, the independent contribution of ethnicity and all potential predictors variables that were biologically plausible or significant at the 0.2 level in the first two stages was assessed in a stepwise multiple regression model.
Ethnic differences were examined by comparing UK Afro-Caribbean people with Jamaican Afro-Caribbean people, UK Afro-Caribbean people with Caucasian people, and Jamaican Afro-Caribbean people with UK Caucasian people. The same regression model was used for these comparisons. The UK Afro-Caribbean people were set as the reference category, and comparisons were performed against UK Caucasian people and Jamaican Afro-Caribbean people. The UK Caucasian people were then reset as the reference category, and a third comparison was performed between UK Caucasian people and Jamaican Afro-Caribbean people. Each comparison was assessed for significance using a partial F test. All comparisons were made after adjustment for differences in heart rate, baseline RI, blood pressure, BMI, blood glucose, serum lipids, and smoking habits and weighted to adjust for the relative oversampling of the Jamaican population.
Of 379 patients studied, 81 were UK Afro-Caribbean people, 101 were UK Caucasian people, and 197 were Jamaican Afro-Caribbean people. Their mean age was 57.4 years (standard deviation, 13.8 years) and 50.7% were women. Although UK Afro-Caribbean people were younger and had higher mean BMI, waist-to-hip ratio, and diastolic blood pressure values compared with Jamaican Afro-Caribbean people (Table 1), only higher diastolic blood pressure in UK Afro-Caribbean people remained significant after adjusting for multiple testing (mean difference, 3.9 mm Hg, p = 0.02). Higher fasting total cholesterol and triglyceride levels and smoking were more common in Jamaican Afro-Caribbean people, but only cholesterol levels compared with UK Afro-Caribbean people remained significant after pairwise comparisons (mean difference, 0.52 mmol/l, p = 0.004).
Comparisons of vascular reactivity and structure
Statistically important differences in vascular reactivity were seen between the three groups (Table 1). Mean baseline RI was greater in Jamaican Afro-Caribbean people compared with UK Caucasian people (mean difference, 3.5%; p = 0.01). The mean ΔRIALBwas least in UK Afro-Caribbean people and greatest in Jamaican Afro-Caribbean people (difference, 5.9%; p < 0.001). The mean ΔRIALBin UK Afro-Caribbean people was also significantly less than that of UK Caucasian people (difference, 5.3%; p = 0.001). There were no significant differences in mean ΔRIGTNbetween the three groups. There also was a significant difference in CIMT measurements between the three groups (Table 1). The UK Afro-Caribbean people had the highest and the Jamaican Afro-Caribbean people had the lowest mean CIMT measurements (mean difference, 0.10 mm, p < 0.001). The mean CIMT of UK Afro-Caribbean people was also greater than that of UK Caucasian people (difference, 0.08 mm; p = 0.01).
Ethnic comparisons adjusted for differences in risk profile
Differences in ΔRIALBand CIMT between the three groups persisted after adjustment for the confounding effects of age, gender, baseline RI, baseline heart rate, BMI, blood pressure, blood glucose, cholesterol and triglyceride levels, and smoking profile in regression models (Fig. 2).The adjusted mean ΔRIALBwas 4.2 percentage points (95% CI, 2.3 to 6.1; p < 0.001) lower in UK Afro-Caribbean people compared with Jamaican Afro-Caribbean people and 2.6 percentage points (95% CI, 0.4 to 4.7; p = 0.02) lower compared with UK Caucasian people (Table 2).Differences in baseline RI did not remain significant after adjusting for confounding variables. The adjusted mean CIMT was 0.13 mm (95% CI, 0.08 to 0.17; p < 0.001) greater in UK Afro-Caribbean people compared with Jamaican Afro-Caribbean people and 0.06 mm (95% CI, 0.01 to 0.10; p < 0.02) greater compared with UK Caucasian people (Table 3).Interestingly, the adjusted ΔRIALBof Jamaican Afro-Caribbean people was greater than that of UK Caucasian people (mean difference, 1.7 percentage points; 95% CI, −0.08 to 3.43) but just failed to achieve statistical significance. Consistent with changes in ΔRIALB, the adjusted CIMT of Jamaican Afro-Caribbean people was significantly less than that of UK Caucasian people (mean difference, 0.07 mm; 95% CI, 0.04 to 0.10; p < 0.001).
Regression models showed that increasing BMI, blood pressure, blood glucose levels, and smoking also had a significant effect on ΔRIALBindependent of ethnicity or environment. The ΔRIALBdecreased by a mean of 0.2 percentage points for every unit increase in BMI (p = 0.04), 0.5 percentage points for every 5-mm increase in systolic blood pressure (p < 0.001), and 0.9 percentage points for every 1-mmol/l increase in fasting blood glucose (p = 0.001), and was 4.8 percentage points lower in smokers (p < 0.001) (Table 2). The CIMT increased by a mean of 0.04 mm for every 5-year increase in age (p < 0.001) and 0.01 mm for every 5-mm increase in diastolic blood pressure (p = 0.03) (Table 3). Although the mean CIMT of women was less than that of men (0.74 ± 0.17 mm vs. 0.76 ± 0.21 mm), it did not achieve statistical significance in this sample. There also was a correlation between ΔRIALBand CIMT within the whole group on univariate analysis (coefficient, 8.3; 95% CI, 3.2 to 13.4; p = 0.0001), but this association became non-significant when other variables such as blood pressure, blood glucose, lipid profile, and BMI were entered into the multivariate model.
This study, in age- and gender-matched samples of Afro-Caribbean people in two different environments and Afro-Caribbean and Caucasian patients in the same environment, showed that vascular reactivity to albuterol, but not to GTN, was impaired and carotid intima media thickness was increased in healthy Afro-Caribbean people in the UK compared with Jamaica, despite belonging to the same ethnic group. Consistent with previous studies (8–15), albuterol-dependent vasodilatation was attenuated in UK Afro-Caribbean people compared with Caucasian people, but in contrast to expectations, healthy Jamaican Afro-Caribbean people showed greater endothelium-mediated vasodilatation and lesser CIMT changes than Caucasian people, even after adjusting for differences in vascular risk factors. These observations are consistent with a recent epidemiologic study that showed that the incidence rate for stroke in Afro-Caribbean people in the West Indies was half that of UK Afro-Caribbean people and comparable to that of the Caucasian population (30). Therefore, a significant proportion of the excess of vascular disease observed in people of African descent in Western settings may be attributable to non-inherited influences or gene-environment interactions, which, in contrast to inherited factors, are potentially amenable to interventions.
Explanations for population differences in disease susceptibility must be sought in representative samples of the general population to ensure unbiased and generalizable comparisons (31). A limitation of previous studies, which seemed to support inherited differences in pharmacologically mediated vascular responsiveness, is that they were undertaken in relatively small and selected samples from the same environment (8–13). In contrast, this study used epidemiologic methods to obtain samples representative of local populations. Bias attributable to differences in measurement techniques between the UK and Jamaica was minimized using identical protocols, definitions, measurement techniques, equipment, and reagents. The comparability of ultrasound equipment was established using the same phantom, transported across sites. Proficiency and comparability of procedures and intra-observer and inter-observer agreement were established in preliminary studies before the main study. The dose of GTN and albuterol and the times of evaluation of ΔRIALBand ΔRIGTNwere selected to achieve maximum change in RI with only modest changes in heart rate and minimum within-patient variability. The CIMT measurements on recorded images were undertaken by a senior vascular physicist masked to the identity and location of patients.
Previous studies in migrant populations have shown that their risk of vascular diseases increases from low levels to match that of the local population because of acquisition of risk factors prevalent in native cultures (32–36). Although a similar phenomenon may be predicted for Afro-Caribbean people living in the UK, it does not fully explain why their vascular risk should exceed that of the local Caucasian population (1). It also does not explain our findings of decreased vascular reactivity and increased CIMT in healthy Afro-Caribbean people with no risk factors compared with Caucasian people in the UK or greater changes of arterial morphology and function in UK Afro-Caribbean people compared with Jamaicans, despite lipid and smoking levels being the higher in Jamaica (Fig. 2). This suggests that there may be environmental (e.g., salt, fish oils, micronutrient or anti-oxidant intake) or behavioral (e.g., physical activity, psychosocial stress) influences within the Jamaican setting that protect against early onset of vascular impairment (37). It is also possible that differences in the frequency of certain genetic polymorphisms between Afro-Caribbean people and Caucasian people, which previously conferred advantages in the Caribbean environment, contribute to a deleterious phenotype in the UK environment. This is exemplified by the well-known associations between salt sensitivity, urbanization, and hypertension in people of African descent (38).
Assumptions have been made that Afro-Caribbean people in the UK and Jamaica share ethnicity and Afro-Caribbean people and Caucasian people in the UK share environment. Eligibility for participation was limited to African-origin people of Caribbean descent, and the study excluded patients who had emigrated from African or other countries. UK Afro-Caribbean patients were of Jamaican origin, and both parents and all four grandparents belonged to the same ethnic group. Although this minimized genetic or racial differences between the two Afro-Caribbean groups in the study, the possibility of a bias caused by self-selection of people who emigrate cannot be excluded. Afro-Caribbean and Caucasian patients in the UK were selected from a population with homogeneous socioeconomic status. Although differences in vascular function could not be attributed to measurable socioeconomic differences, a detailed analysis of lifestyles and diet was not undertaken and could have influenced observed ethnic differences in the UK. This becomes particularly important in the context of findings in the Jamaican cohort and supports the need for more research into the relationship between migration, diet, lifestyles, and vascular health.
Another limitation of the study is its cross-sectional design; a longitudinal study would be the optimal design to investigate the relationships between environment, intermediate phenotypes, and vascular events in people of African descent living in different settings. A longitudinal study would also provide better understanding of the relationships between functional (e.g., vascular reactivity) and morphologic (e.g., CIMT) impairments in vascular diseases. Despite these limitations, the study clearly shows that changes in vascular reactivity (ΔRIALB) were associated with parallel changes in arterial morphology (CIMT) and that the direction of these changes was biologically consistent across the three groups.
It is possible that the diminished responsiveness to albuterol seen in UK Afro-Caribbean people may be attributable to differences in the frequency of beta2-adrenoceptor polymorphisms between Afro-Caribbean people and Caucasian people (20,39). This is unlikely because the ΔRIALBin Jamaican Afro-Caribbean people was significantly greater than in UK Afro-Caribbean people and comparable to that of Caucasian people. Bias from this source should cause equal attenuation of ΔRIALBin UK and Jamaican Afro-Caribbean people compared with Caucasian people because all Afro-Caribbean patients in the UK had Jamaican origins and would be expected to have the same frequency of beta2-adrenoceptor polymorphisms.
The relationship between inherited traits, environmental pressure, and cardiovascular disease is complex, and this study suggests that non-inherited influences within the environment may play a significant, but potentially modifiable, role in ethnic predisposition to vascular disease. The results of this study support a focus on the investigation of multiple environmental factors and gene environment interactions, specific to ethnic groups in different environments, as an approach to eliminating the excess of vascular disease in such people.
This research was supported by a project grant from the Wellcome Trust (GR061574FR). Dr. Shah is supported by the British Heart Foundation. Dr. Chowienczyk advised Micromed on the development of the equipment used for digital volume photoplethysmography.
- Abbreviations and Acronyms
- body mass index
- confidence interval
- carotid intima media thickness
- digital volume pulse
- glyceryl trinitrate
- reflection index
- change in reflection index with albuterol
- change in reflection index with glyceryl trinitrate
- Received April 5, 2005.
- Revision received October 6, 2005.
- Accepted October 10, 2005.
- American College of Cardiology Foundation
- Stewart J.A.,
- Dundas R.,
- Howard R.S.,
- Rudd A.G.,
- Wolfe C.D.A.
- Kissela B.,
- Schneider A.,
- Kleindorfer D.,
- et al.
- McGruder H.F.,
- Malarcher A.M.,
- Antoine T.L.,
- Greenlund K.J.,
- Croft J.B.
- Kelsey R.M.,
- Alpert B.S.,
- Patterson S.M.,
- Barnard M.
- D’Agostino R.B. Jr.,
- Burke G.,
- O’Leary D.,
- et al.
- Markus H.,
- Kapozsta Z.,
- Ditrich R.,
- et al.
- Kuvin J.T.,
- Karas R.H.
- Bone M.
- Rose D.,
- Pevalen D.J.
- Balarajan R.,
- Soni-Raleigh V.
- Chowienczyk P.J.,
- Kelly R.,
- MacCallum H.,
- et al.
- Corbin D.O.,
- Poddar V.,
- Hennis A.,
- et al.