East and South Asian-Specific Blood Pressure Trajectories and Cardiovascular Disease.
Feb 12, 2025DOI: 10.1161/hypertensionaha.124.23985Journal ArticleAHA
Written by: So Mi J ChoSarah UrbutYunfeng RuanAarushi BhatnagarShriienidhie GaneshWhitney E HornsbyRomit BhattacharyaMichael C HonigbergStephen P JuraschekEugene YangDaichi ShimboPradeep Natarajan
Despite several scientific and public health advances, high blood pressure (BP) remains the major modifiable risk factor for cardiovascular disease (CVD) globally [1,2]. Importantly, the epidemiology and contribution of elevated BP on cardiovascular outcomes differ across ethnic subgroups [3,4]. Notably, high BP at a young age contributes to CVD with the largest hazard increase for coronary death, heart failure, ischemic stroke, intracerebral haemorrhage, and peripheral arterial disease before age 60 years. [5].Furthermore, hypertension prevalence and management rates vary across populations with marked heterogeneity in both native and immigrant Asian populations. Importantly, the frequently used “Asian” category masks substantial gradient of hypertension prevalence across major subgroups, ranging from 34% among Chinese and Korean to 43% among South Asian adults living in the U.S [6]. Even across native Asian individuals, hypertension detection and management rates have been persistently low and stagnant over the past 30 years; strikingly, 68% of South Asian adults had either undiagnosed or untreated hypertension compared to 44% among East Asian adults [4].Cardiovascular disease and hypertension rates vary across demographics [7], including major Asian groups [8,9]. The extent to which longitudinal BP patterns relate to these differences may yield new insights into the determinants and prevention of hypertension-conferred CVD. Therefore, for each East and South Asian population living in the UK, we compared (1) life course BP trajectories, (2) hypertension prevalence and treatment characteristics, and (3) the age-specific association between BP and lifetime CVD risk.
Methods
Because of the sensitive nature of the data collected for this study, requests to access the dataset from affiliated and qualified researchers trained in human subject confidentiality protocols may be sent to Massachusetts General Hospital institutional review board.The UK Biobank study protocol was approved by the North West Multi-centre Research Ethics Committee (11/NW/0382) and the secondary data usage (UK Biobank application #7089) for the present analyses was approved by the Massachusetts General Hospital institutional review board (2021P002228). All participants provided electronically signed consent. Reporting followed the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.
The UK Biobank is a prospective cohort study of 0.5 million adults aged 40 to 69 years at recruitment between 2006 and 2010 [10]. To delineate longitudinal BP trajectory, we identified 148,872 participants with linkage to general practitioner data [11] provided by the National Health Service with at least two systolic and diastolic in-clinic BP measurements over two distinct healthcare encounters at age 18 years or older (Figure S1). The UK Biobank primary care data codes clinical events capturing diagnoses, history, symptoms, laboratory results, procedures as well as prescriptions, referrals, immunization, and mortality records. To mitigate the influence of potentially spurious BP readings, we excluded measurements with SBP ≥240 or <60 mmHg or diastolic BP (DBP) ≥160 or <30 mmHg [12].
We relied on two population classification approaches to determine East and South Asian background. Primary analysis categorized participants based on self-identified ethnicity collected from demographics survey at UK Biobank enrollment. Participants reporting multiple categories (e.g., White and Asian) or ambiguous background (e.g., Any other Asian) were excluded. We similarly delineated longitudinal BP patterns for self-identified White Europeans and African or Black populations.
Secondary analysis used genetically-inferred ancestry grouping based on genotype data. Central quality control and imputation of UK Biobank genotypic data were previously described [10]. Briefly, genotypes were obtained using either UK Biobank Axiom or UK BiLEVE Axiom arrays (Affymetrix Research Service Laboratory, Santa Clara, California, USA). The Haplotype Reference Consortium and the merged UK10K + 1000 Genomes were used as reference panels for imputation. Principal component analysis was performed using fastPCA based on a pruned set of 147,604 common independent single nucleotide polymorphisms among unrelated individuals to delineate population structure [13]. After removing outliers based on multidimensional centroid distances from average principal components, genetic ancestry was determined using random forest models (probability >0.8) that partitioned participants into five distinct ancestral groups [14] including East Asian (EAS), South Asian (SAS), African (AFR), European (EUR), and unassigned. Sensitivity analyses were conducted based on genetic ancestry assignment using random forest models probability >0.90 and >0.99.
Biological sex was coded from fixed categories of female and male. Single-inverse normalized Townsend Deprivation Index was quantified based on occupation, car ownership, home ownership, and household overcrowding. Current smoking was defined as lifetime smoking of at least 100 cigarettes and without indication of cessation at enrollment. Dietary score was quantified based on self-recalled average intake of fruit, vegetable, whole grains, fish, dairy, vegetable oils, refined grains, meats, and sugar-sweetened beverages by the Eatwell Guide. Body mass index was measured using Tanita BC-418MA body composition analyser (Tanita, Tokyo, Japan). After 5 minutes of seated rest, BP was measured on two consecutive occasions with 1 minute interval using Omron 705 IT electronic BP monitor (OMRON Healthcare Europe, Hoofddorp, Netherlands). Blood biochemistry, including total, low- and high-density lipoprotein cholesterols, triglycerides, glycated A1c, and creatinine, were assayed within 24 hours of non-fasting sample collection.
We estimated hypertension prevalence by 5-year age increment. Hypertension was defined based on the 2023 European Society of Hypertension Guidelines for the management of arterial hypertension [15] (SBP ≥140 mmHg, DBP ≥90 mmHg, or antihypertensives intake) and 2017 American College of Cardiology/American Heart Association BP Guidelines [5] (SBP ≥130 mmHg, DBP ≥80 mmHg, or antihypertensives prescriptions).
We also quantified the age-specific association between BP and lifetime atherosclerotic CVD (ASCVD) risk across ethnic populations. To determine ASCVD incidence, we relied on both outpatient primary care records and the HESIN master table that entails information on inpatient episodes of care coded based on the International Classification of Diseases 9th and 10th revisions and Classification of Interventions and Procedures v4. The primary outcome was a first ASCVD event defined as fatal or non-fatal myocardial infarction, stroke, or peripheral artery disease (Table S1). In secondary analyses, we separately assessed the first event of each component of the primary outcome.
Baseline characteristics at UK Biobank enrollment, age-specific hypertension prevalence, and treatment status were compared with the χ2, analysis of variance, or Mann-Whitney U test. Mean BP level and BP trajectory over the life course were delineated for each age-year. BP trajectory is an integrative measure of empirically observed absolute BP levels and slope of change over lifetime [16]. Specifically, BP trajectory was fitted using polynomial regression as a function of age. To prevent clustering of serial measurements, for >1 BP readings recorded on the same day, we took the average of repeated measurements to ensure that a participant contributed no more than one BP reading for each date. Due to known sexual dimorphism in BP regulation [17], trajectories were fitted separately for sex by self-identified ethnicity or genetically inferred background. For each subgroup, we identified the most parsimonious polynomial model from a 1st (linear) up to a 4th order yielding the lowest Akaike information criterion to prevent overfitting [18]. Projected BPs and corresponding 95% confidence intervals (CIs) were computed via bootstrap repetitions of coefficients and nonparametric confidence limits that do not make distributional assumptions. To account for the effects of antihypertensive medication, 10 and 5 mmHg was added to measured SBP and DBP, respectively, since the time of initiation [19]. Two secondary analyses were performed. First, we illustrated underlying BP trajectory accounting for antihypertensives medication class-specific effects [20] for each unique prescription record (Figure S2). Prior external validation study [21] has confirmed this approach to yield the lowest mean difference between the measured untreated BP and the imputed underlying BP compared to truncated normal regression-based imputation method. Second, we delineated trajectory of untreated BP levels by censoring at the time of antihypertensives initiation.
Given that BP-associated genetic alleles predict incident hypertension and CVD, we also assessed the association between BP polygenic risk score (PRS) and age-specific measured BP levels across genetic ancestry based on linear regression adjusting for sex, genotyping batch, and the first 10 principal components. The BP genetic score was developed using Bayesian regression models with continuous shrinkage priors [22] leveraging ancestry-specific genome-wide association summary statistics from unrelated Million Veteran Program participants (92% male sex, 69% non-Hispanic White). The predictive performance of the BP PRS has been previously validated in UK Biobank cohort [23]. The BP PRS for the entire cohort was normalized to a standard normal distribution.
We estimated cumulative ASCVD incidence using the Kaplan–Meier method. We tested the associations of young adulthood (18–39 years), middle-age (40–64 years), and later-life (≥65 years) time-averaged BP with first ASCVD incidence using a generalized linear regression model adjusting for sex and antihypertensives prescription since its initiation. Time-averaged BP accounts for differences in measurement interval by averaging cumulative BP from adjacent consecutive primary care visits, indexed to duration in each stage of adulthood. In the context of primary prevention, time-averaged BP included BP measurements prior to ASCVD event. The odds ratio (OR) and 95% CI were expressed per 1 standard deviation (SD) increase in SBP or DBP. An interaction term was added to determine difference in association measure between time-averaged BP and ASCVD event by self-reported ethnicity or genetic ancestry.
All statistical tests were two-sided, and statistical significance was set at P <0.05. All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).
No patients or members of the public were directly involved in this study owing to UK Biobank policy.
Results
Among UK Biobank participants with linkage to primary care utilization data, 376 (0.3%) self-identified as East Asian (64.9% women) and 3077 (2.0%) as South Asian (47.0% women) (Table 1). Based on genetic ancestry, 633 (0.4%) were inferred as East Asian and 3571 (2.4%) as South Asian (Table S2). Among 633 individuals of East Asian genetic ancestry, 371 (58.6%) also self-identified as East Asian; among 3571 individuals of South Asian genetic ancestry, 3017 (84.5%) self-identified as South Asian (Figure S3). Over lifetime, South Asian individuals (median [18–64] visits) had more general practitioner encounters than East Asian individuals (18 [10–38] visits) (Figure S4&S5).
At UK Biobank enrollment and relative to East Asian participants, South Asian individuals had significantly higher SBP (135.3 vs. 130.3 mmHg) and DBP (82.6 vs. 80.5 mmHg) and 3-fold higher antihypertensive medication prescription rates (21.6 vs. 6.7%) (P for all <0.001). Further, South Asian participants also had more adverse cardiometabolic profiles, including higher mean body mass index (27.6 vs. 24.2 kg/m2), high-sensitivity C-reactive protein (2.9 vs. 1.4 mg/L), and median triglycerides level (149.9 vs. 134.5 mg/dL) but lower low-density lipoprotein cholesterol (128.1 vs. 133.4 mg/dL) possibly owing to higher lipid-lowering medication prescription rate (27.5 vs. 10.6%) (P for all <0.001). Notwithstanding a more favourable overall dietary pattern, South Asian participants had a significantly higher proportion of individuals always adding salt to their diet compared to East Asian participants (11.3% vs. 8.3%; P <0.001) (Table S3).
The magnitude of associations between BP PRS and measured BP differed across life course and by ancestry. Each SD increase in SBP PRS was associated with 1.45 (1.22) mmHg time-averaged young adulthood SBP (standard error) among East Asians and 2.05 (0.48) mmHg among South Asians (Table S4). In later-life, whereas each SD increase in BP PRS associated with measured BP to a greater extent among East Asian participants (SBP: 3.14 mmHg and DBP: 1.28 mmHg), the associations attenuated among South Asian participants (SBP: 1.96 mmHg and DBP: 0.65 mmHg).
Further, trajectory of measured BPs over the life course distinguishably differed between East and South Asian participants. Over a median [IQR] follow-up of 58.6 [52.5 to 65.6] years beginning at age 18 years, participants contributed to a median [IQR] of 26 [14 to 58] primary care-based BP measurements. Age-specific mean BP level (Figure 1, Tables S5–S8, Figure S6) and polynomial regression-fitting of BP trajectory (Figure 2, Tables S9–S12) illustrate discernible life course BPs between East and South Asian individuals beginning in young adulthood. Irrespective of sex, South Asian participants generally underwent earlier and more rapid BP escalation at younger ages. At age 30 years, the predicted SBP for East versus South Asian participants was 120.7 mmHg vs. 124.9 mmHg in men and 105.7 mmHg vs. 107.4 mmHg in women, respectively. On average, the projected mean SBP of 140 mmHg in men was anticipated at age 50 years for South Asian participants and age 68 years for East Asian participants; for women, this was 55 years for South Asian and 65 years for East Asian individuals.
Consistent ethnicity- and sex-specific DBP patterns were observed. South Asian men generally underwent a rapid DBP increase until reaching its peak value of 86.2 mmHg at age 58 years, which, thereafter, declined with older age. Simultaneously, East Asian men maintained persistently lower DBP during young-to-middle adulthood with escalation beginning age 40 years and with a delayed decline beginning age 65 years (Figure 2). South Asian women had persistent higher DBP throughout the entire adulthood with a similar declining inflection at age 55 years as East Asian women. The Asian DBP gap was the most discernible between age 30–39 years in men and age 45–55 years in women. Similar age-varying BP changes were observed when stratified based on genetic ancestry (Figures S7–S9) or accounting for antihypertensives regimen and prescription changes (Figure S10). When inspecting untreated BP levels only, the ethnic discrepancy was less discernible, but South Asian participants continued to exhibit generally higher BP levels during young adulthood than East Asian participants (Figures S11&S12).
The early divergence of BP trajectory translated to different hypertension burden beginning in young adulthood. Based on the European Society of Hypertension definition, South Asian adults had nearly two-fold higher prevalence of hypertension (South Asian: 20.2% vs. East Asian: 10.9%) by age 40 years (Figure 3, Tables S13&S14, Figure S13). This gap expanded until the seventh decade in life but with marked sex differences between men (South Asian: 82.8% vs. East Asian: 64.4%) and women (South Asian: 74.8% vs. East Asian: 51.2%). Based on more stringent American threshold, 35.9% of South Asian individuals would attain either SBP/DBP over 130/80 mmHg or have already been prescribed antihypertensive medications by age 40. Nevertheless, treatment rates were ubiquitously suboptimal with only 15.8% of East Asian versus 45.4% of South Asian men and 36.4% East Asian versus 37.9% of South Asian women with hypertension receiving pharmacological treatment (Tables S15&S16).
Compared to East Asian individuals, South Asian individuals had younger mean (SD) age at first antihypertensive prescription (53.7 [8.5] years vs. 56.9 [8.6] years) and significantly higher proportion of triple therapy regimen (1.7 % vs. 0.0%) and beta-blockers prescription (24.5% vs. 21.0%) as first line of treatment (Table 2). Calcium channel blockers and thiazide-related diuretics were more frequently prescribed among East Asian participants (Table S17).
Throughout their lifetime, South Asian individuals had 4-fold higher cumulative ASCVD incidence than that of East Asian individuals (3.5 vs. 0.9 per 1000 person-year) (Figure S14). Across young adulthood to later-life, significant interactions by both self-reported ethnicity and genetic ancestry were observed between age-specific BP and CVD (Figure 4, Figure S15, Tables S18&S19). During young adulthood, each SD increase in SBP was associated with OR [95% CI] of 1.41 [1.12 to 1.75] for incident ASCVD among South Asian individuals, whereas young adulthood SBP did not significantly associate with ASCVD among East Asian individuals (P for interactionethnicity=0.014). Among East Asian participants, midlife SBP was associated with ASCVD (OR [1.37 to 3.50]) to a larger magnitude and most pronouncedly with ischemic stroke (3.84 [1.08 to 5.07]). SBP measured at age ≥65 years was significantly associated with all stroke types among East Asian participants.
Comparably, DBP exhibited ethnicity-specific age-varying patterns. In South Asian population, young adulthood DBP was significantly associated with lifetime ASCVD events and robustly with peripheral artery disease (OR [95% CI] per SD: 2.18 [1.09 to 3.10]). Among individuals with East Asian genetic ancestry, both midlife (OR [1.07 to 3.56]) and later-life (3.16 [1.26 to 5.40]) DBP were significantly associated with incident ischemic stroke.
Discussion
In this contemporary, UK population-based observational cohort, lifetime BP trajectories between East and South Asian populations diverged early in life. South Asian individuals generally presented higher, premature, and rapidly escalating BP compared to East Asian individuals. This BP differential was most evident during young adulthood in men and during midlife in women. Consequently, South Asian participants had comparatively higher hypertension burden and exhibited strong associations with lifetime ASCVD risk earlier in life. Although BP elevations were more delayed among East Asian individuals, midlife BP was associated with incident stroke risk. The distinct age-varying BP patterns permit several interpretations regarding BP management and associated CVD risk across Asian subgroups.
First, East and South Asian individuals had distinctive age-differential associations between BP PRS and time-averaged BP. These observations extend prior genetic analyses of UK biobank participants, in which each SD increase in SBP PRS among South Asian participants was associated with cross-sectional midlife SBP to a greater extent by a β [95% CI] of 2.58 [2.03 to 3.13] mmHg (versus East Asian: 1.85 [0.52 to 3.18] mmHg), whereas DBP PRS associated with measured DBP comparably to a lesser extent to a lesser extent by 1.31 [1.07 to 1.55] mmHg (versus East Asian: 1.58 [1.00 to 2.16] mmHg) [23]. Regardless, South Asian individuals were projected to have significantly higher SBP and DBP even beginning age 30 years, highlighting the accelerated BP elevation attributed to early cumulative exposures of causal atherogenic factors [24,25]. Whether this corresponds to a higher prevalence of deleterious dietary habits (e.g., use of reheated trans-fat-enriched oil) or other factors observed among South Asian immigrants requires further study [26]. Disparities in BP-related healthcare utilization may further amplify early disparities in BP as native and immigrant South Asian populations have persistently low hypertension treatment and antihypertensive therapy adherence rates—all of which contribute to the highest ischemic heart disease risk in Asia [3,4,27,28]. These observations underscore the criticality of 1) disaggregating the monolithic category of “Asian” exhibiting substantial heterogeneity of BP phenotype and 2) early recognition of suboptimal BP level and sustained management by targeting modifiable risk factors culturally or biologically specific to distinct Asian subpopulations.
Second, the BP trajectory for each East and South Asian population, whether defined by self-identified ethnicity or genetic ancestry, was sustained from early to later-life, thereby reinforcing the multifactorial etiologic nature of BP [29,30]. Our genomic and phenotypic delineation of BP suggests distinct interplay of allele frequencies and latently acquired CVD risk factors in shaping lifetime BP. Compared to East Asian population, we observed greater polygenic effect on young adulthood BP in South Asian population suggesting population-differential contribution of genomics to earlier onset BP elevation and acceleration. Given that modest abnormalities in early-life BP carry additive risks beyond risk factors ascertained in middle age [31], identifying ancestry-specific BP variants may advance the understanding of population-specific BP-CVD pathobiology and facilitate earlier preventive actionability.
Third, young adulthood BP associated with lifetime myocardial infarction and peripheral artery disease risk among South Asian individuals, whereas midlife BP robustly predicted incident stroke among East Asian individuals. A global age-specific meta-analysis on BP and CVD subtype reported that SBP between age 30 to 59 years conferred the greatest hazard ratio [95% CI] for myocardial infarction (1.39 [1.32 to 1.48]) and ischemic stroke (1.57 [1.39 to 1.77]); however, the associations diminished with older age [32]. The reasons for dynamic BP effects on different ASCVD outcomes may owe to unequal multimorbidity, multilevel socioeconomic deprivation, healthcare utilization, and cultural or religious organizations shaping behavioural risk factors [28, 33]. Whether BP lowering at different ages will yield cardiovascular benefits and adverse effects differently across ethnic groups merits further investigation. A previous UK Biobank study demonstrated that for an equivalent risk of cardiorenal complications observed in the White European population at SBP of 140 mmHg, the SBP threshold was substantially lower for South Asian individuals (123 mmHg) with a larger population attributable fraction (21.5 % vs. 11.3%) [34]. Given varying predictive utility of BP over life course and in align with region- and age-specific threshold of 10-year ASCVD risk [35], comparative effectiveness of demographic-specific preventive and therapeutic thresholds should be pursued across multiethnic and interregional cohorts.
Our study had notable strengths. Based on a contemporary nationwide cohort data, we leveraged longitudinal electronic health records to track BP levels, hypertension burden, and CVD incidence over life course for each ethnic and genetic ancestral group. Such delineation of demographic-specific BP trajectory unmasked substantial heterogeneity of BP epidemiology previously obscured in an aggregated Asian category.
Despite extensive BP characterization, our results should be interpreted in the context of potential limitations. First, the findings may have limited generalizable to native Asian populations or non-UK-based immigrants with different clinical practice, environment, and extent of acculturation. The variations in culture-specific lifestyle, barriers to health management, and extent of acculturation may differently affect BP. Nevertheless, our results align with U.S.-based study [9] reporting a divergence of hypertension prevalence between South Asian and East Asian immigrant populations beginning age 30 years and larger ethnic discrepancy observed during middle-age in women. Emerging efforts are dedicated to study health disparities across disaggregated Asian populations and assess efficacy of culturally adapted healthcare provisions. Future studies warrant identification of physical and structural environment, sociopolitical inequities, acculturation patterns, and traditionally unmeasured risk enhancers (e.g., language, legal, political restrictions) and quantification of their population-specific attributable risk on BP across geographical contexts of the Asian diaspora. Second, we used a sparse regression model as the availability of established CVD risk factors (e.g., lipids, glycated haemoglobin, smoking status) were not uniformly complete as BP measurements were based on general practitioner data without reassessment of each risk factor simultaneously although catchment is expected to be high in this nationalized health system. Third, the on-treatment BP values may have been inaccurately weighted if the antihypertensive medication adherence varied across ethnic groups. Fourth, survival bias may be possible, since the UK Biobank experimental design required the ability to enrol in middle-age. Finally, it remains uncertain how best to integrate ethnicity or genetic ancestry into risk stratification and clinical practice in an unbiased manner.
South Asian individuals had premature and rapid BP elevation with BP in young adulthood being significantly associated with lifetime ASCVD. East Asian individuals exhibited lower and delayed age-associated BP increase, in which midlife BP associated with incident stroke risk. The distinct age-related BP patterns associated with different CVD risks may elucidate heterogeneity in expression of overt ASCVD across Asian subgroups to inform tailored BP management.
PN
Dr. Pradeep Natarajan
Cardiology · Preventive CardiologyMassachusetts General Hospitalpradeep
