Association between polygenic risk scores for hypertension and low-density lipoprotein cholesterol with incident chronic kidney disease

Data source and study population

UK Biobank was a large prospective cohort study of more than 500,000 participants, aged 40 to 69 years, at 22 assessment centers in England, Wales, and Scotland from 2006 to 2010. The aim of the cohort study was to investigate the association of exposures and genetic risk factors with various chronic diseases by collecting a wide range of phenotypic and genotypic data. A detailed description was previously provided [13]. At baseline, self-reported questionnaires regarding sociodemographic, physical, lifestyle, and medical conditions were completed by patients. They also provided biospecimen samples, and anthropometric measurements were made. All participants provided written informed consent, and the study protocol was approved by the Northwest Multi-center Ethics Committee (IRAS project: ID299116). All procedures followed were in accordance with institutional guidelines. The study protocol and baseline characteristics are publicly accessible on the study website. The data are available to all researchers upon request (www.ukbiobank.ac.uk).

Among the 486,133 participants with PRS values in the cohort, patients with underlying CKD were excluded defined as those with (i) an estimated glomerular filtration rate (eGFR) of <60 mL/min/1.73 m², or (ii) a urinary albumin-creatinine ratio (UACR) of >30 mg/g at enrollment, or (iii) CKD or end-stage kidney disease (ESKD) diagnosis according to International Classification of Diseases, 10th Revision (ICD-10) codes (n = 96,225). We further excluded participants with a history of malignancy (n = 33,918), non-European descent (n = 20,329), and incomplete measurements of covariates (n = 89,768) (Fig. 1).

Figure 1.

Organization of study participants by PRSs.

CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HDL-C, high-density lipoprotein cholesterol; HTN, hypertension; LDL-C, low-density lipoprotein cholesterol; PRS, polygenic risk score; SBP, systolic blood pressure; UACR, urine albumin-to-creatinine ratio.

Genotyping and polygenic risk scores

PRSs are weighted sums of individual risk alleles derived from genome-wide association studies (GWASs), representing an individual’s genetic predisposition to a specific trait [11]. Genotyping of single-nucleotide polymorphisms, imputation, and quality control were carried out by the UK Biobank as previously described, and details are described in the Supplementary Methods (available online) [14]. In 2022, the UK Biobank released a set of PRSs for 53 diseases and quantitative traits, which are available online [15]. UK Biobank released two sets of PRSs. The standard PRS was calculated for all UK Biobank individuals from multiple external GWAS sources, and the enhanced PRS was calculated for a testing group of 104,231 individuals from within the UK Biobank by training on the UK Biobank subgroup and external data. In this study, we used the standard PRSs for HTN and LDL-C.

Main exposures

The primary exposure of interest was the PRSs for HTN (PRS HTN) and LDL-C (PRS LDL-C). The individual association of each PRS with the risk of CKD was assessed using continuous modeling. In addition, to evaluate the combined association of the genetic predisposition to HTN and LDL-C, participants were divided into four groups: 1) the ‘reference group,’ with both PRS <the median, 2) the ‘high LDL-C-risk group,’ those with PRS LDL-C ≥the median and PRS HTN <the median, 3) the ‘high HTN-risk group,’ those with PRS LDL-C <the median and PRS HTN ≥the median, and 4) the ‘combined high-risk group,’ those with both PRSs ≥the median.

Study outcome

The primary outcome was incident CKD, which was identified according to ICD-10 codes in any primary-care data, hospital-inpatient data, and death-registry records, or according to Office of Population Censuses and Surveys Classification of Interventions and Procedures version 4 (OPCS-4) codes in hospital-inpatient data. The UK Biobank also included general practice data for about one-third of the enrolled participants. Thus, we created an additional subcohort by using eGFR and albuminuria data. In this subcohort, CKD was strictly defined as an eGFR <60 mL/min/1.73 m² or a UACR >30 mg/g on two or more occasions at least 90 days apart.

Statistical analysis

Baseline characteristics are presented as numbers and percentages for categorical variables, means with standard deviations (SDs) for normally distributed variables, or medians with interquartile ranges for variables with a skewed distribution. Cause-specific hazard models were used to examine the risk of incident CKD for an individual’s PRSs in total cohort and subcohorts categorized according to PRS. The occurrence of death before the development of CKD was considered a competing risk of incident CKD [16]. Three models were used, with sequential adjustment for confounding factors. Model 1 contained no adjustments. Model 2 was adjusted for age, sex, body mass index (BMI), socioeconomic status, Townsend deprivation index score, alcohol use, smoking status, handgrip strength, salt addition to food, and comorbidities (diabetes mellitus [DM], cardiovascular disease, chronic pulmonary disease, and liver disease). Lastly, model 3 was further adjusted for eGFR, triglyceride concentration, high-density lipoprotein cholesterol (HDL-C) concentration, and use of medications, such as BP- and lipid-lowering drugs. The results are presented as hazard ratios (HRs) and 95% confidence intervals (CIs). Subgroup analyses were conducted to examine the effect modification of key characteristics, such as sex, age, DM, smoking status, BMI, and use of BP- or lipid-lowering medications. The proportionality assumption of the Cox model was ascertained by cumulative incident function and Schoenfeld residuals [17]. Cumulative incident CKD events were derived using the cumulative incidence function for a competing risk, and the difference between curves was analyzed using the Gray test [18].

For sensitivity analyses, the risk of incident CKD according to PRS was assessed in the subcohorts of patients utilizing eGFR and albuminuria data. An additional sensitivity analysis was performed to adjust for physical activity and dietary factors by using metabolic equivalents, total sodium intake, and total energy intake. We further conducted multiple imputations by chain equation with missing covariates [19]. All analyses were performed using Stata (version 17; StataCorp LLC) or R (ver. 4.3.2; R Foundation for Statistical Computing) software. A two-tailed p-value less than 0.05 was considered statistically significant.

Baseline characteristics

The baseline characteristics of the participants stratified according to the median PRSs are described in Table 1. Overall, the median values of PRS LDL-C and PRS HTN were –0.05 and –0.07, respectively (Supplementary Fig. 1, available online). Among the 245,893 patients, the mean age of the patients was 55.7 years, and 49.8% were male. The mean baseline eGFR was 95.9 ± 11.6 mL/min/1.73 m², and the mean BMI was 27.3 ± 4.6 kg/m². Overall, 9,406 (3.8%) had a history of DM and 12,038 (4.9%) had a history of cardiovascular disease. Compared with the reference group, the combined high-risk group exhibited a higher mean systolic BP (SBP; 141.1 ± 19.1 mmHg vs. 136.0 ± 18.5 mmHg), LDL-C concentration (145.4 ± 34.1 mg/dL vs. 131.5 ± 30.1 mg/dL), and HTN (28.6% vs. 16.2%). In this group, a larger proportion of individuals were treated with lipid-lowering (21.8% vs. 12.4%) and antihypertensive (19.8% vs. 10.0%) medications compared to the reference group. No clinically relevant differences were observed in other baseline variables among the four groups (Table 1). Baseline SBP and LDL-C values were positively correlated to PRS HTN (γ = 0.17) and PRS LDL-C (γ = 0.28), respectively (Supplementary Fig. 2, available online).

Table 1.

Baseline characteristics according to combined PRSs for LDL-C and HTN

Relationship between polygenic risk scores and incident chronic kidney disease

During a median follow-up of 13.7 years, 7,771 participants (3.2%) experienced CKD events, with an overall incidence rate of 2.38 per 1,000 person-years. There were 1,722 (2.8%), 1,778 (2.9%), 2,165 (3.5%), and 2,106 events (3.5%) of incident CKD, and the corresponding incidence rates were 2.12, 2.15, 2.63, and 2.61 per 1,000 person-years in the reference group, high LDL-C-risk group, high HTN-risk group, and combined high-risk group, respectively. In the subcohort with strictly defined CKD using general practice data, 7,626 cases (7.3%) of incident CKD occurred during a median follow-up of 13.6 years, with an incidence of 5.47 cases per 1,000 person-years. There were 1,552 (6.0%), 1,604 (6.1%), 2,269 (8.6%), and 2,201 cases (8.5%) of incident CKD in the four groups, respectively, with incidence rates of 4.49, 4.55, 6.44, and 6.39 per 1,000 person-years (Table 2, Supplementary Fig 3, available online). Independent associations of PRS HTN and PRS LDL-C with incident CKD were analyzed using a competing-risk model (Table 3). After full adjustment for confounding factors, a higher PRS HTN was associated with a significantly higher risk of incident CKD (HR per 1–SD higher PRS, 1.07; 95% CI, 1.05–1.10). However, the PRS LDL-C was not significantly associated with the risk of incident CKD (HR per 1–SD higher PRS, 1.00; 95% CI, 0.98–1.03).

Table 2.

Outcome event rates according to groups stratified by median value of PRS LDL-C and PRS HTN

Table 3.

LDL-C and HTN PRSs and incident CKD events

To examine whether a stronger genetic predisposition for LDL-C and HTN synergically contributes to the risk of CKD, we analyzed the combined association of PRSs for HTN and LDL-C with the risk of CKD (Table 4, Fig. 2). Compared with the reference group, the high LDL-C risk group did not have an increased risk of incident CKD (HR, 1.03; 95% CI, 0.97–1.10). However, the high HTN-risk group was associated with a 13% increased risk of CKD (HR, 1.13; 95% CI, 1.06–1.20). Notably, the combined high-risk group did not exhibit a further increased risk of CKD compared with those with a higher PRS HTN alone.

Table 4.

Hazard ratios for incident CKD according to PRS groups

Figure 2.

Cumulative incidence function of incident CKD according to HTN and LDL-C PRS groups.

CKD, chronic kidney disease; HTN, hypertension; LDL-C, low-density lipoprotein cholesterol; PRS, polygenic risk score.

The significant association observed between PRS HTN and CKD risk suggests that genetic predisposition to HTN may play a more direct role in kidney function deterioration compared to LDL-C. It is also important to note that the lack of association between PRS LDL-C and CKD does not rule out LDL-C’s contribution to CKD risk. Instead, it indicates that the genetic predisposition to LDL-C may have limited predictive value for CKD. Our findings showed no significant synergistic effect, as individuals with high PRSs for both HTN and LDL-C did not exhibit a greater CKD risk than expected from PRS HTN alone. This highlights that the two genetic predispositions likely act through independent pathways without interaction. From a clinical perspective, these findings underscore the importance of HTN management for CKD prevention, particularly in populations with a genetic predisposition to HTN, while the direct focus on LDL-C levels may be less impactful in this regard.

Sensitivity analysis

Similar results were observed in the subcohort derived from general practice data. The adjusted HRs per 1–SD higher PRS HTN and PRS LDL-C were 1.21 (95% CI, 1.18–1.24) and 1.02 (95% CI, 1.00–1.05), respectively (Table 3). In the combined analysis, the corresponding HRs were 1.05 (95% CI, 0.98–1.12), 1.39 (95% CI, 1.31–1.49), and 1.45 (95% CI, 1.36–1.55) for the high LDL-C-risk group, the high HTN-risk group, and the combined high-risk group, compared with the reference group (Table 4). However, there was no notable synergistic increase in CKD risk when comparing the high HTN-risk group with the combined high-risk group (Supplementary Table 1, available online). Additional sensitivity analyses, in which we adjusted for total sodium intake, total energy intake, and physical activity, and analysis with multiple imputation datasets yielded similar results (Supplementary Tables 2 and 3, available online).

Our sensitivity analysis suggests a consistent and significant association between higher PRS HTN and CKD risk underscoring the pivotal role of genetic predisposition to HTN in CKD development. However, the predisposition to higher LDL-C levels had minimal impact on CKD development.

Subgroup analysis

We also examined whether potentially modifiable factors influenced the association of PRSs HTN and LDL-C with incident CKD. The p for interaction in each case was >0.05, suggesting that these relationships were not influenced by sex, age, DM, cardiovascular disease, smoking status, BMI, or medication use (Fig. 3).

Figure 3.

Subgroup analysis for the association of PRS HTN and LDL-C with incident chronic kidney disease.

All models were adjusted for age, sex, body mass index (BMI), socioeconomic status, Townsend deprivation index, alcohol status, smoking status, handgrip strength, salt added to food, comorbidities (diabetes mellitus [DM], cardiovascular disease [CVD], chronic pulmonary disease, and liver disease), estimated glomerular filtration rate, triglycerides, high-density lipoprotein cholesterol, medications such as blood pressure (BP)-lowering and lipid-lowering drugs and top four principal components.

aHR, adjusted hazard ratio; CI, confidence interval; HTN, hypertension; LDL-C, low-density lipoprotein cholesterol; PRS, polygenic risk score.