Associations Between Polygenic Risk Score for Blood Pressure and Risk of Hypertension in Northeast Asian Individuals

Abstract

Background

Data on associations between genetic predisposition to high blood pressure (BP) and hypertension and its complications in non‐European populations are limited. The current study investigated associations between polygenic risk scores (PRSs) for BP and risks of hypertension, cardiovascular disease, and chronic kidney disease in Northeast Asian populations.

Methods

A genome‐wide association study of systolic BP (SBP) and diastolic BP (DBP) was conducted using data from the KoGES (Korean Genome and Epidemiology Study). Results were meta‐analyzed using summary statistics from Biobank Japan to construct PRSs.

Results

Compared with a PRS in the lowest 5 percentiles, a PRS in the highest 5 percentiles was associated with an increased risk of hypertension (hazard ratio [HR], 2.44 [95% CI, 1.67–3.56] for PRS for SBP; and HR, 1.77 [95% CI, 1.20–2.62] for PRS for DBP) and earlier onset of hypertension (by a median of 8.5 years for PRS for SBP and 8.0 years for PRS for DBP). These associations remained significant when continuous PRS was analyzed. The genetic risk of hypertension incidence was attenuated by moderate to vigorous physical activity. Adding the PRS for BP to the clinical risk factors improved the predictive value for hypertension (both area under the curve values, 0.787 [95% CI, 0.771–0.803]; P=0.063 for PRS for SBP and [95% CI, 0.771–0.804]; P=0.031 for PRS for DBP). However, neither PRS for SBP nor PRS for DBP was associated with the incidence of cardiovascular or chronic kidney disease.

Conclusions

The PRS for BP was associated with a higher risk of incident hypertension and earlier‐onset hypertension in a Northeast Asian population. PRS may facilitate early identification and targeted management of individuals at high risk of developing hypertension.

Nonstandard Abbreviations and Acronyms

CAVAS

Cardiovascular Disease Association Study

HEXA

Health Examinee

KoGES

Korean Genome and Epidemiology Study

MVPA

moderate to vigorous physical activity

PRS

polygenic risk score

PRS_DBP

polygenic risk score for diastolic blood pressure

PRS_SBP

polygenic risk score for systolic blood pressure

Clinical Perspective.

What Is New?

• A higher polygenic risk score for blood pressure was associated with elevated blood pressure levels, increased prevalence and incidence of hypertension, and earlier‐onset hypertension in Northeast Asian cohorts.

• Genetic risk for incident hypertension was attenuated by moderate to vigorous physical activity in Northeast Asian cohorts.

What Are the Clinical Implications?

• The polygenic risk score for blood pressure, whether used independently or alongside traditional clinical risk factors, can identify individuals at greater risk of developing hypertension and thereby enable early detection, prevention, and treatment to reduce the burden of hypertension and the incidence of associated complications.

Hypertension is a leading modifiable risk factor for cardiovascular disease (CVD) and chronic kidney disease (CKD). ¹ , ² Identifying high‐risk individuals for early detection and proper management of hypertension is crucial for reducing CVD and CKD risks. Blood pressure (BP) and hypertension are highly heritable, and genome‐wide association studies (GWAS) have identified single‐nucleotide polymorphisms (SNPs) associated with systolic BP (SBP), diastolic BP (DBP), and hypertension, providing insights into the underlying pathophysiology and potential therapeutic targets of hypertension, although these contributed only modestly to BP variation. ³ , ⁴ , ⁵ , ⁶

As BP is a highly polygenic trait, genetic risk scores, which incorporate the effects of a few significant SNPs into a single score, are used to evaluate the combined effects of risk alleles. ⁴ , ⁵ , ⁷ With increasing GWAS sample sizes and improved computational techniques, polygenic risk scores (PRSs) that capture cumulative SNP effects, regardless of statistical significance, have improved the prediction of multifactorial diseases. ⁸ , ⁹ , ¹⁰ A high PRS for BP is associated with elevated BP and increased risks of hypertension, early‐onset hypertension, and uncontrolled hypertension. ¹⁰ , ¹¹ , ¹² , ¹³ Genetic predisposition to elevated BP manifests early in life and is sustained throughout adulthood, causing prolonged high BP exposure and subsequent complications such as CVD and CKD. ¹⁴ The PRS for BP facilitates the identification of individuals at high risk for hypertension and its complications, who may benefit from earlier interventions. The provision of genetic risk information to individuals motivates those at higher risk to initiate lifestyle modifications and medical treatments. ¹⁵ , ¹⁶ Thus, the PRS for BP may reduce the burden of hypertension and the incidence of the associated vascular complications.

Most GWAS and PRS studies have focused on European populations ¹⁷ , ¹⁸ ; European population–derived PRS has shown decreased predictive performance in non‐European populations. ¹¹ , ¹⁸ , ¹⁹ Despite recent studies with larger, more diverse populations, Asian groups remain underrepresented. ²⁰ , ²¹ Thus, associations between PRS for BP and risks of hypertension, CVD, and CKD in European cohorts are poorly generalizable to cohorts of other ethnicities.

In the present study, we aimed to investigate whether Northeast Asian individuals with high PRSs for BP had a greater incidence of hypertension and its complications, such as CVD and CKD.

Methods

Data and Code Availability

GWAS summary statistics of Biobank Japan are publicly available at Biobank Japan PheWeb (https://pheweb.jp/). Genotype data from the KoGES (Korean Genome and Epidemiology Study) can be obtained upon reasonable request to the Korea National Institute of Health. Additional data mentioned in this manuscript are either included in the article and its Supplemental Information files or can be made available from the corresponding author upon request.

Study Population

The KoGES is a prospective cohort study including the Ansan and Ansung study, HEXA (Health Examinee) study, and CAVAS (Cardiovascular Disease Association Study). ²² The study design and methods have been previously described. ²² The Ansan and Ansung cohort is derived from a community‐based study of 10 030 participants aged 40 to 69 years who underwent medical examinations and completed self‐administered biennial questionnaires from 2001 to 2014. The HEXA cohort consists of 173 357 participants aged ≥40 years from an urban population–based study, recruited from the national health examinee registry. The CAVAS cohort is derived from a community‐based study of 28 338 participants aged ≥40 years from rural areas. Of these, 72 298 participants were genotyped (5493 from Ansan and Ansung, 58 700 from HEXA, and 8105 from CAVAS). GWAS was conducted using the 58 700 HEXA data (Figure S1).

The Ansan and Ansung data were used to assess the prevalence and incidence of hypertension, CVD, and CKD and the predictive value of PRS for BP. In the Ansan and Ansung study, those with missing baseline BP data (n=11) were excluded, leaving 5482 individuals for hypertension prevalence analysis. For hypertension incidence analysis, participants diagnosed with or treated for hypertension before baseline (n=825) or with baseline SBP ≥140 or DBP ≥90 mm Hg were excluded (n=1307). To account for the effects of underlying diseases, individuals diagnosed with or treated for CKD (n=167) or CVD (n=61) were excluded, leaving 3122 participants for analysis (Figure S1).

This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Ethics Committee of the KoGES at the Korea National Institute of Health, as well as the Institutional Review Boards of Soonchunhyang University Hospital (No. 2024‐03‐015) and Seoul University (No. E2101/003‐004). All participants provided written informed consent.

Data Collection

BP was measured by trained nurses using a mercury sphygmomanometer, with the patient in a sitting position after resting for at least 5 minutes. Measurements were taken at least twice, with intervals of >1 minute between measurements. If the difference between the last 2 measurements exceeded 5 mm Hg, additional measurements were taken until the difference was <5 mm Hg. The mean of the final 2 BP readings was recorded as the BP value for each visit. The arm with the higher BP reading was used in this study. Diabetes was defined as fasting glucose ≥126 mg/dL, postload glucose ≥200 mg/dL after a 75‐g oral glucose tolerance test, glycated hemoglobin ≥6.5%, diabetes history, or use of oral glucose‐lowering medication or insulin. Dyslipidemia was defined as a history of dyslipidemia or the use of lipid‐lowering medications. Moderate to vigorous physical activity (MVPA) was self‐reported and categorized as <150 or ≥150 min/wk.

Genotyping and Quality Control

Samples were genotyped using KoreanChip, a customized array based on the Korean genome structure, with high genomic coverage and abundant functional variants of low to rare frequency. ²³ We excluded samples with low call rates (<97%), high heterozygosity, singletons, cryptic first‐degree relatives, or sex discrepancies. ²⁴ SNPs with low genotype call rates (<95%) or Hardy–Weinberg equilibrium P values (<10⁻⁶) were excluded. Genotypes were phased using Eagle version 2.3 and imputed using IMPUTE4 with 1000 Genomes Project phase 3 data and the Korean reference genome as a reference panel. After excluding genetic variants with an imputation quality score <0.8 and minor allele frequency <1%, 8 056 211 variants remained for analysis.

GWAS and Meta‐Analysis

As information on antihypertensive medication use was available for only 1.45% of the participants in the HEXA cohort, blood pressure was not adjusted for antihypertensive medication use in the GWAS analysis. A linear mixed model implemented in SAIGE version 1.1.3 ²⁵ was used to conduct GWAS with covariates including the top 10 principal components, age, and sex. In the first step, 327 540 variants with an imputation quality score of 1 were used to obtain a genetic relationship matrix. A rank‐based inverse normal transformation method was applied to remove proximal contamination. Manhattan plots (Figure S2) and quantile–quantile plots with the λ genomic control values (Figure S3) were generated using LocusZoom. ²⁶

Biobank Japan is a multi‐institutional hospital‐based prospective biobank including genotype, laboratory, and clinical data from ≈200 000 participants. ²⁷ Information on genotyping, imputation, GWAS, and quality control has been published. ²⁷ The GWAS summary statistics for BP (145 505 and 145 515 individuals for SBP and DBP, respectively) were obtained from the Biobank Japan PheWeb (https://pheweb.jp/). We subsequently performed a Z score–based meta‐analysis for HEXA and Biobank Japan to obtain the combined P values. The effect sizes for risk prediction were obtained using the fixed‐effect inverse‐variance weighting method. A total of 6 907 487 variants were included in this study.

PRS Calculation

PRSs for SBP (PRS_SBP) and DBP (PRS_DBP) were calculated via PRS‐CS. ²⁸ The default “auto” version of PRS‐CS, which enables automatic learning of the global shrinkage parameter from GWAS summary statistics, was applied. Parameters used for PRS‐CS are described in Data S1. An East Asian linkage disequilibrium reference panel from the 1000 Genomes Project was used for this calculation. ²⁹ The parallel coordinates plot revealed overlap between the reference panel and our data (Figure S4). Individual‐level PRSs for BPs were generated using the sum modifier via PLINK version 1.9.

Definition of Outcomes

In accordance with the current Korean guideline, ³⁰ hypertension was defined as SBP ≥140 mm Hg, DBP ≥90 mm Hg, or the use of antihypertensive medications. In the cross‐sectional BP analyses, we adjusted baseline SBP and DBP by 15 and 10 mm Hg, respectively, for those on antihypertensive therapy. CKD was defined as an estimated glomerular filtration rate of <60 mL/min per 1.73 m² or the development of proteinuria during follow‐up. The estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation, and proteinuria was defined as ≥1+ on a dipstick test. CVD was defined as a new diagnosis or prescription of medication for myocardial infarction, coronary artery disease, congestive heart failure, cerebrovascular disease, or peripheral artery disease during follow‐up.

Statistical Analysis

PRS was standardized for all analyses and categorized on the basis of PRS percentiles (≤5, >5–50, >50–95, and >95). Continuous variables were expressed as means±SDs or medians and interquartile ranges, as appropriate. Categorical variables were expressed as frequencies and proportions. Differences between PRS groups were assessed via ANOVA or the Kruskal–Wallis and Cochran–Armitage trend tests for continuous and categorical variables, respectively. Associations between PRS and baseline SBP and DBP were tested using multivariate linear regression models. Multivariate logistic regression analysis was used to investigate associations between PRS and hypertension prevalence. In prospective analyses, Cox proportional hazard regression analyses were used to investigate associations between PRS and incident hypertension, CVD, and CKD. All the regression models were fitted with robust SE estimates to account for heteroscedasticity. Three models were evaluated. Model 1 included age, sex, and the first 10 principal components. Model 2 also included body mass index, smoking status, MVPA, and sodium intake. Model 3 included all variables in Model 2 plus serum glucose and total cholesterol levels.

Receiver operating characteristic curve analysis was used to calculate the area under the curve (AUC) for predicting hypertension incidence. P<0.05 was considered statistically significant. Statistical analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria).

Sensitivity Analysis

Whether different PRS calculation methods influenced the results was assessed by calculating PRS using Lassosum. ³¹ Genotyped CAVAS data (n=8105) served as validation data for tuning hyperparameters in Lassosum (Figure S1).

Results

Baseline Characteristics

PRS_SBP and PRS_DBP were strongly correlated (r=0.758, P<0.001). Table 1 presents the baseline characteristics of participants according to PRS_SBP derived from PRS‐CS after excluding those with prevalent hypertension, CVD, or CKD. Participants with higher PRS_SBP percentiles had higher SBP and DBP (both P<0.001) and were more likely to have a family history of hypertension (P=0.002). Similarly, those in higher PRS_DBP percentiles had higher SBP (P=0.002) and DBP (P<0.001) and were more likely to have a family history of hypertension (P=0.035; Table S1).

Table 1.

Baseline Characteristics of Participants Stratified According to the PRS_SBP

PRS percentile	≤5 (n=193)	>5–50 (n=1522)	>50–95 (n=1290)	>95 (n=117)
Age, y	47.00 (43.00–58.00)	47.00 (43.00–55.00)	47.00 (43.00–55.00)	45.00 (42.00–53.00)
Women	111 (57.51)	784 (51.51)	706 (54.73)	70 (59.83)
SBP, mm Hg	110.00 (100.00–118.00)	114.00 (106.00–120.00)	114.00 (108.00–122.00)	116.00 (110.00–124.00)
DBP, mm Hg	74.00 (68.00–80.00)	76.00 (70.00–80.00)	78.00 (70.00–82.00)	78.00 (72.00–82.00)
Body mass index, kg/m2	23.56 (21.93–25.35)	23.92 (22.01–25.84)	24.02 (22.31–25.92)	24.24 (22.56–25.87)
Family history of hypertension	32 (16.58)	319 (20.96)	294 (22.79)	38 (32.48)
Smoking
Never	122 (64.55)	884 (58.62)	780 (61.13)	77 (66.38)
Former	28 (14.81)	245 (16.25)	174 (13.64)	17 (14.66)
Current	39 (20.63)	379 (25.13)	322 (25.24)	22 (18.97)
MVPA ≥150 min/wk	113 (60.75)	870 (59.06)	717 (57.13)	64 (57.14)
Sodium intake, g/d	2.93 (2.03–3.88)	2.94 (2.15–3.89)	2.85 (2.05–3.85)	2.71 (1.93–3.84)
Diabetes	18 (9.33)	152 (9.99)	124 (9.61)	5 (4.27)
Dyslipidemia	2 (1.04)	34 (2.23)	34 (2.64)	3 (2.56)
Fasting glucose, mg/dL	81.00 (76.00–87.00)	82.00 (77.00–89.00)	82.00 (77.00–88.00)	82.00 (77.00–87.00)
eGFR, mL/min per 1.73 m2	98.38 (89.65–105.50)	95.50 (83.53–104.64)	96.29 (83.53–105.37)	99.62 (89.23–106.12)
Total cholesterol, mg/dL	184.00 (164.00–209.00)	188.00 (166.00–209.00)	185.00 (165.00–211.00)	183.00 (163.00–212.00)

Values are expressed as the frequency and proportion for categorical variables and as the median and interquartile range for continuous variables.

DBP indicates diastolic blood pressure; eGFR, estimated glomerular filtration rate; MVPA, moderate to vigorous physical activity; PRS, polygenic risk score; PRS_SBP, polygenic risk score for systolic blood pressure; and SBP, systolic blood pressure.

PRS for BP and Baseline BP

A 1‐SD increase in the PRS for BP corresponded to an increase of 3.80 mm Hg in SBP (95% CI, 3.31–4.30; P<0.001) and 2.31 mm Hg in DBP (95% CI, 1.99–2.63; P<0.001) after adjusting for clinical risk factors in model 3. After excluding individuals with prevalent hypertension, CVD, or CKD, a 1‐SD increase in the PRS for BP corresponded to an increase of 1.49 mm Hg in SBP (95% CI, 1.11–1.88; P<0.001) and 0.94 mm Hg in DBP (95% CI, 0.69–1.20; P<0.001). Differences between the highest 5 and lowest 5 PRS percentiles were 6 mm Hg for SBP and 4 mm Hg for DBP (Table 1 and Table S1).

PRS for BP and Hypertension Prevalence

Higher BP PRS percentiles were associated with higher hypertension prevalence (Table 2). Compared with individuals in the lowest 5 PRS_SBP percentiles, those in higher PRS_SBP percentiles had significantly higher hypertension prevalence (all P<0.001). Similarly, all higher PRS_DBP percentiles were associated with higher hypertension prevalence than the lowest 5 PRS_DBP percentiles (P=0.002 for >5–50 percentiles; P<0.001 for the remaining percentiles). Moreover, a 1‐SD increase in the PRS for BPs was significantly associated with hypertension prevalence (P<0.001 for both SBP and DBP; Table 2).

Table 2.

Associations Between the PRS for BP and Hypertension Prevalence

	Unadjusted	Model 1	Model 2	Model 3
SBP
PRS categories
≤5	Reference	Reference	Reference	Reference
>5–50	1.87 (1.39–2.52)	2.01 (1.47–2.75)	2.08 (1.47–2.93)	2.00 (1.41–2.85)
>50–95	2.78 (2.06–3.74)	3.25 (2.38–4.44)	3.17 (2.24–4.48)	3.08 (2.16–4.39)
>95	4.30 (2.96–6.24)	5.40 (3.65–7.99)	5.07 (3.29–7.82)	4.87 (3.13–7.57)
PRS*	1.40 (1.32–1.48)	1.50 (1.41–1.59)	1.44 (1.35–1.54)	1.44 (1.34–1.54)
DBP
PRS categories
≤5	Reference	Reference	Reference	Reference
>5–50	1.78 (1.33–2.38)	1.87 (1.38–2.53)	1.67 (1.21–2.30)	1.66 (1.20–2.31)
>50–95	2.44 (1.82–3.27)	2.77 (2.05–3.75)	2.46 (1.78–3.40)	2.49 (1.79–3.46)
>95	4.26 (2.95–6.16)	5.17 (3.51–7.61)	4.14 (2.73–6.27)	4.03 (2.64–6.16)
PRS*	1.38 (1.30–1.46)	1.47 (1.38–1.56)	1.42 (1.33–1.52)	1.42 (1.33–1.52)

Multivariate logistic regression analysis was performed; the data are shown as odds ratios (95% CIs). Model 1: adjusted for age, sex, and top 10 principal components. Model 2: adjusted for the variables in model 1+body mass index, smoking status, moderate to vigorous physical activity, and sodium intake. Model 3: adjusted for the variables in model 2+serum glucose and total cholesterol levels. BP indicates blood pressure; DBP, diastolic blood pressure; PRS, polygenic risk score; and SBP, systolic blood pressure.

^*PRS is expressed as a continuous variable for SD increment.

PRS for BP and Hypertension Incidence

During a median follow‐up of 14.0 (interquartile range, 7.0–14.0) years, 1425 incident hypertension events were recorded. The hypertension risk increased in participants with higher PRS_SBP or PRS_DBP percentiles than in those with PRS_SBP or PRS_DBP in the lowest 5 percentiles (Table 3). Each SD increase in PRS for BP was associated with an increased risk of hypertension (hazard ratio [HR], 1.24 [95% CI, 1.17–1.31]; P<0.001 for SBP; and HR, 1.23 [95% CI, 1.16–1.30]; P<0.001 for DBP) after adjusting for the aforementioned risk factors (Table 3).

Table 3.

Associations Between the PRS for BP and Hypertension Incidence

	Cases/controls	Unadjusted	Model 1	Model 2	Model 3
SBP
PRS categories
≤5	60/133	Reference	Reference	Reference	Reference
>5–50	643/879	1.56 (1.20–2.03)	1.58 (1.22–2.05)	1.61 (1.23–2.10)	1.60 (1.21–2.10)
>50–95	661/629	2.10 (1.62–2.73)	2.17 (1.67–2.82)	2.14 (1.63–2.80)	2.11 (1.60–2.79)
>95	61/56	2.27 (1.58–3.24)	2.42 (1.69–3.47)	2.43 (1.67–3.54)	2.44 (1.67–3.56)
PRS*		1.24 (1.18–1.31)	1.27 (1.20–1.34)	1.24 (1.17–1.31)	1.24 (1.17–1.31)
DBP
PRS categories
≤5	66/125	Reference	Reference	Reference	Reference
>5–50	624/872	1.29 (1.00–1.65)	1.34 (1.04–1.73)	1.25 (0.96–1.63)	1.26 (0.96–1.64)
>50–95	682/645	1.79 (1.39–2.31)	1.89 (1.46–2.43)	1.72 (1.32–2.24)	1.74 (1.33–2.28)
>95	53/55	1.76 (1.22–2.55)	1.92 (1.31–2.80)	1.76 (1.20–2.59)	1.77 (1.20–2.62)
PRS*		1.23 (1.17–1.30)	1.26 (1.19–1.33)	1.23 (1.16–1.30)	1.23 (1.16–1.30)

Multivariate‐adjusted Cox regression analysis was performed; the data are shown as hazard ratios (95% CIs). Model 1: adjusted for age, sex, and top 10 principal components. Model 2: adjusted for the variables in model 1+body mass index, smoking status, moderate to vigorous physical activity, and sodium intake. Model 3: adjusted for the variables in model 2+serum glucose and total cholesterol levels. BP indicates blood pressure; DBP, diastolic blood pressure; PRS, polygenic risk score; and SBP, systolic blood pressure.

^*PRS is expressed as a continuous variable for SD increment.

PRS for BP and Hypertension Onset

Compared with participants in the lowest 5 PRS_SBP percentiles, hypertension occurred a median of 4.5, 4.5, and 8.5 years earlier in the >5–50 (P=0.059), >50–95 (P=0.009), and >95 (P=0.007) percentiles, respectively (Figure 1A). For PRS_DBP, hypertension occurred a median of 5.0, 6.0, and 8.0 years earlier in the same percentile groups (P=0.247 for >5–50, P=0.014 for >50–95, P=0.027 for >95 percentile; Figure 1B).

Figure 1. Risk of incident hypertension by PRS categories for (A) systolic and (B) diastolic blood pressure.

The cumulative incidence of hypertension increased and hypertension occurred earlier in participants in the higher PRS percentiles than those in the lowest 5 PRS percentiles. PRS indicates polygenic risk score.

PRS for Hypertension Prediction

The AUC value for the incidence of hypertension on the basis of BP information was 0.750 (95% CI, 0.733–0.767). It increased to 0.754 (95% CI, 0.737–0.771) when PRS was added (ΔAUC, 0.004 [95% CI, 0.000–0.008]; P=0.026 for PRS_SBP; ΔAUC, 0.004 [95% CI, 0.000–0.007]; P=0.031 for PRS_DBP). The AUC value for hypertension incidence increased to 0.784 (95% CI, 0.767–0.800) when clinical risk factors were added to BP information (ΔAUC, 0.033 [95% CI, 0.024–0.044]; P<0.001) and further increased to 0.787 (95% CI, 0.771–0.803) after adding PRS_SBP (ΔAUC, 0.003 [95% CI, 0.000–0.007]; P=0.063) and to 0.787 (95% CI, 0.771–0.804) after adding PRS_DBP (ΔAUC, 0.004 [95% CI, 0.000–0.007]; P=0.031). Including both the PRS_SBP and the PRS_DBP (AUC, 0.788 [95% CI, 0.772–0.805]) did not improve predictive performance compared with including either alone.

PRS for BP and Risks of CVD and CKD

During a median follow‐up of 14.0 (interquartile range, 14.0–14.0) and 14.0 (interquartile range, 13.0–14.0) years, 171 and 542 incident CVD and CKD events were reported, respectively. Both the >5–50 and >50–95 PRS_SBP percentiles were associated with a higher CVD risk than the lowest 5 percentiles (HR, 2.89 [95% CI, 1.10–7.61]; P=0.031 for the >5–50 PRS_SBP percentile; and HR 2.73 [95% CI, 1.04–7.22]; P=0.042 for the >50–95 PRS_SBP percentile; Table S2). However, the highest 5 PRS_SBP percentiles were not consistently associated with CVD incidence owing to wide CIs. The >50–95 PRS_SBP percentiles were associated with a lower CKD risk than the lowest 5 percentiles (HR, 0.69 [95% CI, 0.48–0.98]; P=0.041; Table S3). There were no significant associations between PRS_SBP and CKD in the other percentiles. PRS_DBP percentiles were not associated with the risk of CVD or CKD (Tables S2 and S3). A 1‐SD increase in PRS_SBP or PRS_DBP was not significantly associated with CVD or CKD (Tables S2 and S3).

Sensitivity Analysis

Analysis using the PRS calculated using Lassosum yielded comparable results. Higher PRS percentiles were associated with higher SBP and DBP (both P<0.001) and a family history of hypertension (P=0.010 for PRS_SBP and P=0.003 for PRS_DBP). The differences between the highest 5 and lowest 5 PRS percentiles were 5 and 2 mm Hg for SBP and DBP, respectively. Lassosum analyses results are presented in Tables S4 through S6.

Subgroup Analysis

Interactions between the PRS_SBP and clinical risk factors were assessed using a likelihood ratio test. Overall, there was no significant interaction between PRS_SBP and clinical risk factors, with the exception of MVPA (Figure 2). MVPA exhibited significant interaction with the relationship between PRS_SBP and incident hypertension and a greater risk in the group with <150 min/wk than in those with ≥150 min/wk.

Figure 2. Forest plot of incident hypertension in the different subgroups.

The HRs (dots) and 95% CIs (horizontal bars) are estimated separately for each subgroup. The x axis represents the HRs on a natural logarithmic scale. BMI indicates body mass index; HR, hazard ratio; and MVPA, moderate to vigorous physical activity.

Discussion

In the current study, PRS_SBP and PRS_DBP were independently associated with higher hypertension prevalence, incidence, and earlier onset in the general Korean population. Incorporation of the PRS for BP in the clinical information modestly improved the predictive value for incident hypertension. Despite variations in associations between PRS_SBP and incident CVD or CKD, the PRS_DBP was not associated with these conditions. The genetic risk for incident hypertension was attenuated by MVPA. Our findings indicate that the PRS for BP can identify individuals with a higher genetic predisposition to hypertension, allowing earlier prevention in Northeast Asian populations.

Some variants associated with BP phenotypes are enriched in or specific to Asian populations. ³² While GWAS identified genes shared with European ancestry, such as FGF5 and ATP2B1, we identified Asian population–specific genes such as ALDH2. When comparing our findings with those from large‐scale studies conducted in Japan, ²⁷ China, ³³ and Europe, ³⁴ we observed the highest number of replicated signals in Japan, followed by China and Europe. Differences in linkage disequilibrium patterns and risk allele frequencies across ethnic populations limit the generalizability of European population–derived PRS to other populations. ¹⁸ , ³⁵ A recent study using GWAS summary statistics from Biobank Japan demonstrated that the PRS for BP was associated with prevalent hypertension in a Japanese cohort, whereas a European population–derived PRS was not. ¹¹ In a previous study conducted among Europeans, an increase of 1 SD in PRS_SBP or PRS_DBP was associated with a 2.60 mm Hg increase in SBP and a 1.47 mm Hg increase in DBP, respectively. ³⁶ In another study conducted on a European cohort, the risk of incident hypertension increased 1.42‐fold and 1.41‐fold, respectively, per 1‐SD increase in PRS_SBP or PRS_DBP. ¹⁰ In the Chinese population, a 1‐SD increase in the genetic risk score based on 22 SNPs was associated with a 14% increase in the odds of incident hypertension. ³⁷ The effect size of the PRS for BP in our study was between that reported by studies of European and Chinese cohorts. To the best of our knowledge, this is the first study to report an association between the PRS for BP and incident hypertension risk in an Asian population. By using millions of variants, including Asian‐specific genes and long‐term follow‐up data, our findings contribute to an accurate assessment of the effects of PRS on hypertension risk in Northeast Asian populations.

Genetic predisposition for BP starts to influence BP as early as 3 to 5 years of age and becomes more pronounced with age. ¹⁴ In our study, individuals in the highest 5 PRS_SBP and PRS_DBP percentiles developed hypertension a median of 8.5 and 8.0 years earlier than those in the lowest 5 percentiles, respectively. This aligns with recent studies linking higher PRS_SBP and PRS_DBP percentiles to increased risk of early‐onset hypertension. ¹⁰ Despite hypertension awareness and treatment rates of 67% and 63%, respectively, in South Korean adults aged ≥20 years, these rates dropped to <50% in adults aged <50 years and decrease further in younger adults. ³⁸ BP is generally not measured in adolescence or young adulthood. Therefore, encouraging young adults with a high genetic risk for hypertension to monitor their BP and initiate early and targeted preventive and therapeutic interventions is crucial. The PRS can capture early‐onset disease risk, even in the absence of traditional risk factors, motivating individuals with a higher genetic risk to initiate lifestyle modifications or therapeutic interventions. ³⁹ However, further research is warranted to determine whether PRS‐based stratified screening or lifestyle interventions in young adults can ameliorate hypertension and reduce the incidence of its complications.

BP is influenced by genetic, environmental, and demographic factors and their interactions. ¹ In a recent study, cardiopulmonary fitness positively affected SBP across the genetic risk distribution, with individuals in the highest decile of PRS_SBP exhibiting greater benefits. ⁴⁰ This is consistent with the results of the present study, in which increased incident hypertension due to high genetic predisposition was attenuated by MVPA. These results further emphasize that genetically determined high BP is not deterministic and can be mitigated by exercise. Khera et al ⁴¹ reported that coronary artery disease risk varies with smoking, body mass index, physical activity, and diet, even among individuals with similar genetic risks. Although we did not identify significant interactions between the PRS for BP and other important hypertension contributors such as body mass index, sodium intake, or smoking, interpretation requires caution because these lifestyle factors were measured only at baseline.

Some studies have shown that adding a genetic risk score for BP to clinical risk models provides little additional value. ⁴² , ⁴³ However, more recent studies using PRS have reported improved prediction of hypertension when the PRS for BP is included. ¹⁰ , ¹⁴ The modest improvement in the predictive performance observed in our study after adding PRS information to the clinical risk factors should be interpreted in the context of elevated clinical risk, as our cohort consisted of a middle‐aged population. Although clinical risk factors change and often worsen over time, the PRS for BP offers insights into lifetime hypertension risk. In a recent study, the PRS for hypertension was associated with both prevalent and incident hypertension across ages 21 to 80 years. ²¹ Because the PRS for BP is independent of traditional risk factors but complements them, it could be used alone or in combination with traditional risk factors to improve the accuracy of estimating lifetime hypertension risk. Considering that our study demonstrated a higher incidence of hypertension with increasing PRS percentiles, the utility of the PRS is expected to be particularly high among individuals at substantially increased risk.

Large prospective cohort studies have reported that the PRS for BP is associated with increased CVD risk. ¹⁰ , ¹⁴ However, the effect size of the genetic risk score for BP on kidney outcomes is smaller than that for cardiovascular and cerebrovascular diseases, ⁴⁴ and associations between genetic risk for BP and CKD have been inconsistent. ⁴ , ¹⁴ In the present study, neither PRS_SBP nor PRS_DBP was associated with the incidence of CVD or CKD. Several factors may explain these discrepancies. First, some SNPs may have pleiotropic effects on CVD risk factors, which could lead to different associations with CVD than the BP effect would be expected to cause. For example, the ALDH2 gene is associated with increased BP but has a protective effect on cholesterol, reducing the risk of coronary artery disease. Second, CVD and CKD are multifactorial and complex; the PRS for BP is just 1 of many risk factors potentially limiting its overall effect. Third, the risk of vascular complications varies depending on hypertension treatment and the degree of BP control, which may offset the CVD and CKD risks attributed to the PRS for BP. Finally, because the incidence of CVD and CKD was low in the current study, the possibility of false negatives cannot be ruled out. Further studies with larger sample sizes are needed to clarify associations among PRS, BP, and hypertension‐related complications.

Limitations

The present study had some limitations. First, because we included only participants of Northeast Asian descent, the findings may have limited generalizability to other populations. Although we used GWAS data from the largest Korean and Japanese cohorts with detailed phenotypic information, additional discovery and validation data are required to elucidate the complex links between genetic risk and hypertension. Second, Biobank Japan is a hospital‐based disease‐ascertained cohort, whereas KoGES is population based; however, this difference is unlikely to significantly bias GWAS results. Third, we lacked data on antihypertensive medication types, doses, and adherence, which prevented us from assessing treatment responses using PRS percentiles. Fourth, some misclassification may have occurred because BP was measured biennially and was not measured via 24‐hour ambulatory monitoring. Finally, in this study, 140/90 mm Hg was used as the threshold for hypertension in accordance with Korean and European guidelines, ³⁰ , ⁴⁵ although lower BP thresholds for defining hypertension have been proposed. ⁴⁶ Further studies are needed to evaluate the effects of PRS on hypertension under the newly defined criteria.

Conclusions

A high PRS for BP is associated with an increased risk of prevalent and incident hypertension in a Northeast Asian population. The PRS for BP, whether used independently or alongside traditional clinical risk factors, can identify individuals at greater risk of developing hypertension and thereby enable early detection, prevention, and treatment to reduce the burden of hypertension and the incidence of associated complications.

Sources of Funding

This research was supported by the Soonchunhyang University Research Fund (20240405).

Disclosures

None.

Supporting information

Data S1

Tables S1–S6

Figures S1–S4