Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Chronic kidney disease (CKD) affects roughly 10% of the global population and is strongly influenced by type 2 diabetes and genetic risk factors. It is typically evaluated using estimated glomerular filtration rate (eGFR) and albuminuria (UACR). Although genetic markers for each trait have been identified, their shared genetic architecture has remained unclear.WHAT THIS STUDY ADDS This study identifies 75 pleiotropic loci in the general population and six in individuals with T2D that influence both eGFR and UACR, and it maps these genes based on their directional effects.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY The findings reveal a map of common genetic risk marker clusters and biological pathways underlying eGFR and albuminuria, pointing to potential therapeutic targets for CKD and CKD with diabetes.Introduction Chronic kidney disease (CKD) is a debilitating condition and a hazard to public health. The prevalence of CKD is around 13% and increasing. 1 CKD is predicted to be the fifth leading cause of death by 2040, largely as a consequence of cardiovascular complications including hypertension, atherosclerotic vascular disease, heart failure, and dysrhythmia.2 This rise is further driven by aging of the population and the rapidly increasing prevalence of obesity, insulin resistance, and type 2 diabetes (T2D).1 The increasing number of persons with CKD is a global challenge that also carries an enormous financial burden.3 Gaining a better understanding of the genetic basis underlying CKD could enhance patient outcomes and reduce the burden of CKD, particularly in persons with T2D, on healthcare systems.Genome-wide association studies (GWAS) have broadened our understanding of the genetic basis underlying urine albumin-creatinine ratio (UACR) and estimated glomerular filtration rate (eGFR), where multiple genes have implications for the risk of CKD development.4–7 Through GWAS, genetic loci associated with UACR or eGFR have been successfully identified,6 7 with heritability estimates ranging up to 50% for albuminuria and 36–75% for eGFR, respectively.8 Advancing precision medicine within CKD and CKD with diabetes requires identifying homogeneous risk groups, and one way is through mapping shared genetic effects not only between clinically relevant subgroups5 but also between kidney function traits: UACR and eGFR.Previous studies report genetic overlap between eGFR and albuminuria9–11 but lack T2D stratification and do not use the updated creatinine-based eGFR race free equation (Inker et al)12 referred to as eGFR2021 from hereon. To our knowledge, this is the largest directional pleiotropy analysis using this updated equation and using a T2D subgroup stratification.Herein, we aimed to investigate if known eGFR loci (n=423) (a) replicate with the new eGFR2021 equation12 and (b) identify the pleiotropic loci that also associate with albuminuria measures, in an observational study comprising participants from the UK Biobank general and T2D populations.Materials and methods Study population The current study is a part of the larger precision medicine project entitled Redefining the EGFR-Albuminuria relationship in persons with or without type 2 Diabetes using GENetics (READ-GEN), which aims to investigate the relation between eGFR and UACR with respect to genetic (and other molecular) determinants, aiming to improve monitoring, prognosis, and treatment for CKD and CKD with diabetes.For the current investigation, we accessed the UK Biobank study population whose design has been described previously in detail (https://www.ukbiobank.ac.uk/). Briefly, UK Biobank study postal invitations were sent to 9.2 million adults aged 40–69 years who were registered with the UK National Health Service and lived across England, Wales, and Scotland. A total of 505 493 participants enrolled in this study with a response rate of 5.5%. The current cross-sectional study comprised 486 936 participants from the UK Biobank with genotype-phenotype13 information and a focus on the kidney function measures, UACR and eGFR2021.An overview of the Read-GEN study design is depicted in figure 1.Figure 1The overall READ-GEN study design. (A) Participants from the UK Biobank underwent genotyping and clinical phenotyping. (B) Data were quality controlled and normalized prior to analysis. (C) MLR models were used to replicate eGFR-associated loci (MLR 1) and assess their association with UACR (MLR 2), enabling identification of shared genetic loci. eGFR, estimated glomerular filtration rate; MLR, multiple linear regression; READ-GEN, Redefining the EGFR-Albuminuria relationship in persons with or without type 2 Diabetes using GENetics; SNP, single-nucleotide polymorphism; UACR, urinary albumin-to-creatinine ratio. Created with BioRender.com.The UK Biobank study was granted ethical approval by the North West Multi-center Research Ethics Committee (approval number: 11/NW/0382, 16/NW/0274, or 21/NW/0157 as applicable, as the committee renewed approval in 2011, 2016, and 2021). All participants provided informed consent. This research was conducted under application number 71 699.Genetic variants (single nucleotide polymorphisms) investigated We selected a recently published GWAS meta-analysis study (Stanzick et al)7 to identify known eGFR-associated loci. The study was selected following the criteria: studies predominantly including participants of European descent and not using the updated eGFR2021 equation. We gave preference to the largest and most recent European GWAS study.7Stanzick et al7 identified 424 single nucleotide polymorphisms (SNPs) associated with eGFR. To further explore the genetic associations between eGFR and UACR, we were able to extract 423 of 424 eGFR SNPs from the UK Biobank. Most of the extracted genetic data pertained to individuals of European ancestry (94%). The participants in the UK Biobank were classified as Europeans if the individuals were British, Irish, or from any other European background (Field ID: 21000).The UK Biobank extracted DNA from blood samples and genotyped at the Affymetrix Research Services Laboratory.13 UK Biobank investigators performed extensive genotyping quality control, including checks for Hardy-Weinberg equilibrium.13 Genotype calling was conducted using Affymetrix Power Tools with custom algorithms developed for the UK Biobank, including methods to resolve poorly performing variants and handling rare alleles. Relatedness among participants was evaluated using the Kinship-based INference for Genome-wide association studies kinship estimator, with relatives up to the third-degree flagged in the dataset.13Phenotype data In addition to genetic data, we retrieved key phenotype information from the UK Biobank, including age (Field ID: 21022), sex (Field ID: 31), systolic blood pressure (SBP) (Field ID: 4080), serum creatinine (Field ID: 30700), urine albumin (Field ID: 30500), urine creatinine (Field ID: 30510), body mass index (BMI) (Field ID: 21001), and glycated hemoglobin (HbA1c) (Field ID: 30750). Data from serum creatinine measures, age, and sex were used to calculate eGFR 202112 and data from urine albumin and urine creatinine were used to calculate UACR.Following analysis in the pooled UK Biobank population, we performed a subanalysis among persons with T2D. A participant was considered to have T2D if the following criteria were met: (a) answering YES to ‘Diabetes diagnosed by a doctor’ (Field ID: 2443), (b) age of diagnosis was above 35 years (Field ID: 2976), and (c) no prescription of insulin treatment within 1 year of diagnosis (Field ID: 2986). Participants were excluded from the T2D subgroup if they had missing data on age of diagnosis or reported having gestational diabetes (Field ID: 4041), as previously described.11The UK Biobank collected phenotype data using standardized protocols during initial assessment visits.14 Age was calculated based on the date of birth and the date of assessment center visit. Sex was acquired from the National Health Service central registry and, in some cases, updated by participant self-report. BMI was calculated as weight in kilograms divided by height in meters squared (kg/m²) using measurements obtained during the baseline assessment visit. HbA1c was measured using high-performance liquid chromatography on Bio-Rad Variant II Turbo analyzers and reported in mmol/mol according to The International Federation of Clinical Chemistry standards. Urine albumin was measured by immunoturbidimetric analysis, and creatinine in urine was measured by enzymatic analysis, both using a Beckman Coulter AU5400. Serum creatinine was assessed enzymatically using a Beckman Coulter AU5800 platform. SBP was measured twice using a validated Omron device and averaged. These procedures ensured consistent assessment across the cohort. Further details on measurement protocols and assay quality are available.14Statistical analysis In both the overall population and the T2D subgroup, we excluded individuals with missing genotype or phenotype data. Furthermore, we excluded values for data elements that exceeded four SD from the mean ( µ±4 σ). Data preprocessing and filtering are presented in online supplemental figure 1 for the overall population and online supplemental figure 2 for the T2D subgroup. Given the right-skewed distribution of the UACR data, we used natural log-transformed values for further analysis. To ensure comparability, we also log-transformed eGFR2021. The UK Biobank performs comprehensive quality control on genomic data prior to making it available, ensuring minimal genotyping errors in the dataset.13 14SP110.1136/bmjdrc-2025-005863.supp1Supplementary dataPrior to analysis, we standardized the independent variables age, sex, and SBP. To account for population stratification, the first 10 genetic principal components provided by the UK Biobank (Field ID: 22009) were included as covariates in all regression models. To validate the 423 loci identified by Stanzick et al,7 we applied multiple linear regression using two adjusted linear models. In the first model, ln(eGFR2021) was the dependent variable, with the eGFR-associated SNPs and covariates (age, sex, SBP, and genetic principal components 1–10) as independent variables. In the second model, we assessed whether the newly identified eGFR2021-associated loci were also associated with UACR, using ln(UACR) as the dependent variable and the same set of covariates. We applied a nominal threshold of p≤0.05 to both models. We applied the Bonferroni correction (BC) with a threshold of Pb≤1.18 × 10–4 for the second regression model to account for multiple testing. The Bonferroni corrected threshold was determined as Pb≤αm, where α was set to 0.05 and m was the number of analyses performed. We used the same approach when performing the analysis on the T2D subgroup with the addition of the independent variables BMI and HbA1c. In analyses of the T2D subgroups, we applied a Bonferroni corrected threshold of Pb≤6.58 × 10–4.To assess potential selection bias due to missing data used in multiple linear regression model one and two, we performed a descriptive comparison between individuals with eGFR data and individuals with UACR data using standardized mean differences (SMD), where values <0.2 were considered indicative of negligible imbalance.15We also calculated the percent difference in mean levels per added effect allele (EA) using the formula: %difference=(ebeta−1)×100. We performed all analyses using R and Python.Sensitivity analysis To assess whether the association between genetic variants and UACR was independent of eGFR, we conducted a sensitivity analysis for loci that were Bonferroni-significant in the UACR model. For each of these loci, we ran a second regression model with UACR as the dependent variable, including the same covariates (age, sex, SBP, and genetic principal components 1–10 (including BMI and HbA1c for T2D subgroup)) and additionally adjusting for baseline eGFR 2021. This adjustment accounts for the fact that all tested loci were originally identified based on their association with a previous equation used to determine eGFR. We incorporated a nominal association threshold of p value ≤0.05 for the sensitivity analysis.Power calculations We performed power calculations for each associated SNP using Quanto (V.1.2.4) as previously. 14 All analyses were conducted under a continuous trait model for independent individuals. We assumed a gene-only hypothesis, applied an additive model of inheritance, and used a two-sided significance level of 0.05. For variant-specific inputs, we entered the EA frequency along with the population-specific mean and SD of ln(eGFR2021) and ln(UACR), both of which were estimated from UK Biobank data.Expression quantitative trait locus analysis To provide functional annotation of identified loci, kidney-specific expression quantitative trait locus (eQTL) analyses were performed using the Human Kidney eQTL Atlas (Susztak Lab; Sheng et al, Nature Genetics).16 SNPs that were associated with eGFR and UACR in the main analysis were queried to assess their association with gene expression in microdissected kidney compartments, including tubules and glomeruli.Results Clinical characteristics Descriptive characteristics of the study population are presented in table 1. The overall UK Biobank cohort after quality checks included 432 451 participants (online supplemental figure 1) with a mean age of 56.5±8.1 years; 54.2% were women. The mean SBP was 140±20 mm Hg, serum creatinine 0.81±0.16 mg/dL, and eGFR2021 94.7±12.8 mL/min/1.73 m². UACR data were available for 137 355 participants after exclusions in the overall cohort (31.8% of the overall population), with a median UACR of 1.05 (25%, 75% range: 0.65–2.04) mg/mmol. In the T2D subgroup (n=16 265), the mean age was 60.6±6.5 years and 35.3% were women. The mean SBP was 144±18 mm Hg, serum creatinine 0.85±0.22 mg/dL, and eGFR2021 92.1±15.4 mL/min/1.73 m². UACR data were available for 8853 participants (online supplemental figure 2), corresponding to 48.5% of the T2D subgroup, with a median UACR of 1.55 (25%, 75% range: 0.86–3.57) mg/mmol. The T2D subgroup had a mean BMI of 31.6±5.6 kg/m² and a HbA1c of 51.5±12.2 mmol/mol.Table 1Clinical characteristics of the overall and type 2 diabetes study participants from the UK Biobank studyCharacteristicsOverall populationType 2 diabetesParticipants, n432 45116 265Female, n (%)234 193 (54.2)5743 (35.3)Age (years)56.5±8.160.6±6.5SBP (mm Hg)140±20144±18Serum creatinine (mg/dL)0.81±0.160.85±0.22eGFR (mL/min/1.73 m²)94.7±12.892.1±15.4UACR (mg/mmol)1.05 (0.65–2.04)1.55 (0.86–3.57)BMI (kg/m²)27.4±4.831.6±5.6HbA1c (mmol/mol)36.2±6.751.5±12.2Data are mean ± SD. Albuminuria (UACR) measures are depicted in median and IQR. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2021 race free equation. UACR was available in a subset of participants (overall, n=137 335; type 2 diabetes, n=8853). BMI in the overall population had information on n = 430 689 participants and HbA1c had n = 411 340.BMI, body mass index; eGFR, estimated glomerular filtration rate; HbA1c, glycated hemoglobin; SBP, systolic blood pressure; UACR, urine albumin creatinine ratio.Furthermore, all SMDs were <0.2, except for BMI in the overall population (SMD=0.205), which showed a marginal difference. Overall, this suggests no meaningful imbalance between individuals with eGFR and/or UACR measurements (online supplemental tables S1 and S2).SP210.1136/bmjdrc-2025-005863.supp2Supplementary dataeGFR2021 and UACR associated loci (overall population)In the overall population (n=432 451), after adjusting for age, sex, SBP, and genetic principal components 1–10, a total of 422 loci were nominally associated with eGFR2021 (p≤0.05, figure 2). When analyzing UACR as the dependent variable in the second multiple linear regression model, 75 loci of the 422 loci were nominally associated (p≤0.05) with UACR, and after BC, 12 loci were significant (Pb ≤ 1.18×10−4, figure 2).Figure 2A flow chart of the number of identified loci with significance thresholds for the overall population. The figure illustrates the two-stage multiple linear regression approach used to evaluate the association between 423 previously reported loci for eGFR2021 and UACR in the overall UK Biobank population. In the first model, 422 loci were nominally associated with eGFR2021 (p≤0.05). In the second model, which used UACR as the dependent variable, 75 loci remained nominally associated (p≤0.05), and 12 remained significant after Bonferroni correction (Pb≤1.18 × 10–4). eGFR2021 was calculated using the CKD-EPI 2021 race-free equation. UACR and eGFR were natural log-transformed prior to analysis. CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR2021, estimated glomerular filtration rate calculated using the 2021 CKD-EPI race-free equation; UACR, urine albumin-creatinine ratio.The effect sizes for the significant loci associated with both eGFR2021 and UACR in the overall population are visualized in a scatter plot (figure 3). Since eGFR2021 and UACR were natural log-transformed prior to analysis, the beta coefficients represent the change in natural log-transformed eGFR2021 or UACR per EA. We also calculated the percent difference in mean levels per added EA. The result for the overall population is presented in the online supplemental tables S3–S5.Figure 3Scatter plot of shared genetic loci (with effect estimates) for eGFR versus UACR in the overall population. This plot shows beta coefficients for 75 SNPs associated with both eGFR2021 and UACR (p≤0.05). The x-axis represents the effect size on ln(eGFR2021), and the y-axis represents the beta effect on ln(UACR). Black dots represent loci with significance (p≤0.05); blue dots highlight the 12 loci significant after Bonferroni correction in the second multiple linear regression model (Pb≤1.18 × 10–4). Gene names are annotated for the significant loci after Bonferroni correction. CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR2021, estimated glomerular filtration rate calculated using the 2021 CKD-EPI race-free equation; SNP, single nucleotide polymorphism; UACR, urine albumin-creatinine ratio.In the first quadrant (figure 3, top right), we observed 22 loci where both eGFR2021 and UACR exhibited positive effects, with 19 loci showing association at p value ≤0.05, and three significant after BC. In the second quadrant, we found 11 loci associated with lower eGFR2021 and higher UACR. Out of these, nine loci were nominally associated (p≤0.05) before, while two were significant after BC. One of the top loci identified in this quadrant was DGAT2 rs3060 (C allele) on chromosome 11. For UACR, the beta coefficient was 0.025 (95% CI 0.013 to 0.036; p=2.15 × 10−5) with a percentile difference in mean per EA of 2.52% and for eGFR2021, the beta coefficient was −0.0026 (95% CI –0.0036 to –0.0017; p=8.49 × 10−8) with a percentile difference in mean per EA of –0.26%. In the third quadrant, representing negative effects on both traits (lower eGFR2021 and lower UACR), we found 35 loci, with 29 associated at p value ≤0.05, and six significant after BC. Finally, we observed seven loci in the fourth quadrant which exhibited a higher eGFR2021 and a lower UACR. Of these, six loci were nominally associated at p value ≤0.05 while one was significant after BC.eGFR2021 and UACR loci in the T2D subgroupIn the T2D subgroup (n=16 265), 423 loci were analyzed for eGFR2021 association, and 76 loci were nominally associated (p≤0.05; figure 4). In the second multiple linear regression model with UACR as the dependent variable, six loci showed associations (p≤0.05; figure 4). One locus showed significance after BC (Pb ≤ 6.58×10-4). The associated loci in the T2D subgroup were visualized in a scatter plot, showcasing the effect sizes for each locus on ln(eGFR2021) and ln(UACR) (figure 5). The results for the T2D subgroup and the calculated percentile difference in mean per added EA for the T2D subgroup is presented in online supplemental tables S6–S8.Figure 4A flow chart of the number of identified loci with significance thresholds for the T2D group. The figure outlines the two-stage multiple linear regression framework applied to evaluate the association between 423 previously reported eGFR loci and UACR among individuals with T2D in the UK Biobank. In the first regression model, 76 loci were nominally associated with eGFR2021 (p≤0.05). In the second model, using UACR as the dependent variable, six loci were associated (p≤0.05), and one remained significant after Bonferroni correction (Pb≤6.58 × 10–4). Both outcome variables, eGFR2021 and UACR, were natural log-transformed prior to analysis. CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR2021, estimated glomerular filtration rate calculated using the 2021 CKD-EPI race-free equation; T2D, type 2 diabetes; UACR, urine albumin-creatinine ratio.Figure 5Scatter plot of shared genetic loci (with effect estimates) for eGFR versus UACR in the T2D group. The scatter plot shows beta coefficients for loci associated with both eGFR2021 and UACR in individuals with T2D. The x-axis displays beta coefficients for ln(eGFR2021), and the y-axis displays beta coefficients for ln(UACR). Black dots represent loci significant at p value ≤0.05, and blue dots are loci significant after Bonferroni correction in the second multiple linear regression model (Pb≤6.58 × 10–4). Gene names are annotated for all loci. CKD, chronic kidney disease; CKD-EPI, Chronic Kidney Disease Epidemiology Collaboration; eGFR2021, estimated glomerular filtration rate calculated using the 2021 CKD-EPI race-free equation; T2D, type 2 diabetes; UACR, urine albumin-creatinine ratio.In the first quadrant (figure 5, top right), we observed one nominally associated locus. In the second quadrant, we identified two loci. In the third quadrant, we found three loci, including the BC significant locus GCKR rs1260326 (EA=C), located on chromosome 2. For UACR, the beta coefficient was −0.082 (95% CI −0.12 to −0.049; p=1.29 × 10−6) and for eGFR2021, the beta coefficient was –0.0045 (95% CI –0.0086 to –0.00045; p=2.95 × 10−2) with a percentile difference in mean per added EA of –7.88% and –0.45%, respectively. In the fourth quadrant, no loci were identified.Comparing the loci identified in the T2D subgroup with the overall population, two nominally associated T2D loci were specific only for this subgroup (GSTA2, USP2-AS1). These loci are presented in table 2.Table 2Loci nominally associated with both eGFR and UACR in the type 2 diabetes subgroupSNP rsIDChrGeneEA/OAEAFeGFR2021UACREffect (mean % change)P valueEffect (mean % change)P valuers64588686GSTA2T/C0.65−0.0058 (−0.58)5.5×10−30.045 (4.6)9.2×10−3rs1089235811USP2-AS1A/G0.56−0.005 (−0.50)1.4×10−20.039 (3.9)2.0×10−2Model effect estimates are based on log transformed eGFR and UACR levels whereas % change is the percent difference on mean levels of eGFR and UACR per added effect allele. Model is adjusted for age, sex, systolic blood pressure, BMI, HbA1c, and the first 10 genetic principal components.Chr, chromosome; EA, effect allele; EAF, effect allele frequency; eGFR, estimated glomerular filtration rate; OA, other allele; SNP, single nucleotide polymorphism; UACR, urinary albumin to creatinine ratio.Sensitivity analysis In total, 12 loci reached significance after BC in the overall population, while one reached significance in the T2D subgroup after BC. After adjusting for eGFR 2021, all tested loci remained nominally associated (p≤0.05), indicating that their association with UACR was independent of eGFR, supporting the robustness of the observed associations. Full results are provided in online supplemental table S9.Power estimate We estimated the statistical power for all associated loci; 26.7% (20/75) in the overall population and 0% (0/6) in the T2D subgroup had a power above 80% for both analyses regarding ln(eGFR 2021) and ln(UACR) associations. Focusing specifically on the Bonferroni-significant loci, 100% (12/12) in the overall population and 0% (0/1) in the T2D subgroup met the power threshold for both analyses. These results suggest that while many nominal associations may be underpowered, most Bonferroni-significant associations are supported by sufficient statistical power. Full results are reported in online supplemental table S10 (overall population) and online supplemental table S11 (T2D subgroup).eQTL analysis Kidney-specific eQTL analysis identified regulatory effects for 24 out of the 77 (75 in overall population and the two loci specific to the T2D population) SNPs associated with eGFR and UACR. Of these, five loci showed significant associations in tubular tissue, seven in glomerular tissue, and 12 loci demonstrated effects in both compartments. The results are present in online supplemental table S12.Discussion Our study identified a total of 422 loci nominally associated with race free eGFR 2021 in the overall UK Biobank population. Of these loci, 75 were also nominally associated with UACR, and 12 of these loci were significant after BC. These results indicate a shared genetic architecture for eGFR2021 and UACR measures in the overall UK Biobank population. In the T2D subgroup, 76 loci associated with eGFR2021 and six were nominally associated with both eGFR2021 and UACR, with one locus remaining significant after BC. While some genetic loci were common between the overall population and individuals with T2D (four loci), others were specific to T2D (two loci; GSTA2, USP2-AS1), reflecting a potentially different genetic architecture for diabetic and non-diabetic CKD.We observed pleiotropic associations at several loci. Among the 75 loci nominally associated with both eGFR2021 and UACR in the overall population and six loci in the T2D subgroup, the direction of effects varied across quadrants, reflecting (negative to positive) effect estimates for eGFR2021 versus UACR. Identifying loci associated with both eGFR and UACR is particularly informative, especially when the associations follow different directions of effects (eg, a variant associated with lower eGFR and higher UACR). This suggests a shared pathophysiology contributing to both impaired kidney function and increased glomerular damage, which are two hallmarks of progressive CKD.Some pleiotropic patterns are consistent with prior studies. For example, our recent study demonstrated pleiotropic patterns, with key albuminuria risk genes like CUBN (participating in urinary albumin re-uptake in the renal tubules),17 which also influence eGFR.11 18 Specifically, the albuminuria-increasing risk variants were associated with normal kidney function (eGFR). Furthermore, one study tested the association of eGFR loci with UACR more than a decade back.9 This study used data from the CKDGen Consortium19 and the CARe Renal Consortium and found one SNP (rs17319721 in the SHROOM3 gene (EA=A)) to be associated with both traits.9 More recently, a multitrait GWAS applying a joint analysis of eGFR and UACR identified 15 loci associated with albuminuria, the majority of which were novel, underscoring the importance of shared genetic architecture and pleiotropic mechanisms between these traits.10 While our approach differs, both strategies highlight shared biology between kidney function and albuminuria. Of the loci identified in this multi-trait analysis, only rs700753 was assessed in our study; it showed association with eGFR but not UACR, suggesting potential heterogeneity across populations or analytical approaches.Building on this, our analyses highlight additional pleiotropic loci of potential clinical relevance. Among these, one locus identified in the overall population was DGAT2 rs3060 (EA=C), which showed a pattern of lower eGFR2021 and higher UACR, suggesting a potential role in kidney function regulation. The locus was both Bonferroni-significant and had adequate statistical power. Notably, this SNP has been shown to influence the effectiveness of treatment with niacin in reducing liver fat in people with dyslipidemia.20 Despite being in a non-coding region, this variant may still exert functional effects, likely affecting gene regulation and pharmacodynamics.20 Furthermore, DGAT2 is involved in triglyceride synthesis,21 and dysregulated triglyceride metabolism has been implicated in kidney disease.22 This may suggest a metabolic link between DGAT2 variation and impaired kidney function. In line with this, DGAT2 inhibition has been explored as a potential therapeutic strategy for improving lipid metabolism in metabolic disorders, although its effects appear to be context-dependent and species-dependent, suggesting that DGAT2-related pathways may also be relevant in kidney disease progression.23 24 Supporting a potential regulatory mechanism, kidney-specific eQTL analysis showed that rs3060 influences DGAT2 expression in both tubular and glomerular compartments, as well as UVRAG expression in glomeruli. This suggests that the observed associations with eGFR and UACR may be mediated through altered gene expression in key kidney structures.In the T2D subgroup, it is notable that fewer loci reached nominal association (six loci, online supplemental table S8). This may be due to the smaller population and that the analyses by Stanzick et al7 did not stratify for T2D. One BC significant pleiotropic locus to highlight, located in the third quadrant (lower left) in figure 5, is GCKR rs1260326 (C allele), located on chromosome 2. These results suggest that the C allele of rs1260326 in GCKR may be linked to lower eGFR2021 and UACR, playing a potential role in kidney function and metabolism. Interestingly, the glucokinase regulatory protein (GCKR) gene, and specifically the rs1260326 variant, is a well-studied genetic locus with significant effects on metabolism.25 26 This SNP leads to a proline-to-leucine substitution (P446L) and is known for its pleiotropic effects, particularly influencing lipid and glucose metabolism. Specifically, the T allele of rs1260326 is associated with increased levels of triglycerides, C reactive protein, and lower levels of insulin and fasting glucose.25–27 This makes it a critical marker for metabolic syndrome and T2D risk. Supporting a potential regulatory mechanism, kidney-specific eQTL analysis demonstrated that rs1260326 influences the expression of multiple genes, including NRBP1 and ATRAID, in both tubular and glomerular compartments, supporting a pleiotropic role through gene regulation in key kidney structures.While some loci influencing eGFR2021 and UACR are shared between the overall and T2D populations, we identified two loci (GSTA2, USP2-AS1) with nominal association effects only in T2D, indicating a potential T2D-specific role in kidney function and UACR (table 2). Supporting a potential regulatory mechanism, kidney-specific eQTL analysis demonstrated that rs6458868 influences GSTA2 expression in both tubular and glomerular compartments. GSTA2, a gene involved in glutathione-mediated detoxification and oxidative stress responses, may therefore contribute to kidney function through gene regulatory mechanisms.28 These findings highlight the need for further investigation into these loci to understand their functional impact and therapeutic potential in diabetic CKD.This study has notable strengths and limitations that warrant discussion. One of the key strengths lies in comparing the association of existing eGFR loci with UACR, leading to the identification of multiple pleiotropic loci. The large sample size combined with the homogenous phenotyping of kidney function measures enhances the reliability of our findings. Additionally, the ability to stratify the analysis by T2D status allows for a focused examination of T2D-specific loci, providing novel insights into the genetic relationship between UACR and eGFR2021.However, there are some limitations to consider. The study population consists predominantly of relatively healthy individuals with early-stage T2D and normal or near-normal kidney function, where the GFR estimating equations exhibit relatively limited accuracy, further limiting the generalizability of findings to persons with moderate-to-advanced CKD.12 29 In addition, the definition of T2D was based on a composite of self-reported diagnosis, age at onset, and timing of insulin initiation, which may introduce some degree of misclassification. Furthermore, medication use, including agents known to affect kidney function and albuminuria (eg, antihypertensive or glucose-lowering therapies), was not explicitly accounted for and may introduce residual confounding. Further research in more advanced disease stage populations may help to validate and extend these findings. In addition, all analyses were conducted within a single cohort (UK Biobank), and independent replication was not performed, which may limit the generalizability and robustness of the findings. Furthermore, while almost all the Bonferroni-significant loci had sufficient statistical power, a notable proportion of the nominally significant loci did not meet the 80% power threshold, which may limit the robustness of the suggested associations.In conclusion, the READ-GEN study identified 75 pleiotropic loci nominally associated with eGFR2021 and UACR in the overall population and two loci specific to the T2D subgroup, advancing our understanding of the genetic heterogeneity underlying CKD pathobiology. This study highlights the potential for improved risk stratification for individuals at risk of CKD in accordance with the precision medicine in diabetes initiative.30 Studies investigating albuminuria loci and their association with eGFR are also pending. Several shared kidney function loci warrant deeper investigation and validation in high-risk CKD populations, especially those with high effects and/or mechanistic relevance, including DGAT2 rs3060, linked to lipid metabolism and kidney function, and GCKR rs1260326, which may connect metabolic traits and kidney health in T2D.