Introduction
Chronic kidney disease (CKD) is a major global public health concern, underscoring the need for a comprehensive understanding of its progression and the development of effective preventive measures [1,2]. CKD is characterized by a gradual decline in renal function, which can eventually necessitate renal replacement therapies, such as dialysis or kidney transplantation [3]. Understanding the mechanisms underlying CKD progression is crucial for optimizing treatment strategies, improving patient outcomes, and reducing the burden of CKD-related complications [1,4].
A significant aspect of CKD study is the impact of vitamin D deficiency, a prevalent condition among CKD patients [5,6]. Vitamin D plays an essential role in maintaining calcium and phosphate balance, regulating immune responses, and exerting anti-inflammatory effects [7]. Deficiency in vitamin D can exacerbate CKD progression by increasing the risk of osteoporosis, secondary hyperparathyroidism, and vascular calcification, all of which contribute to further renal dysfunction [8,9]. However, the causal relationship between vitamin D levels and CKD progression has not been fully established, and evidence remains limited.
Previous studies have utilized observational designs to investigate the relationship between vitamin D deficiency and CKD progression [10,11]. In particular, our previous research demonstrated an association between vitamin D deficiency and renal events through cross-sectional studies, cohort studies, and propensity score matching (PSM) [12]. While these studies have provided valuable insights into the potential effects of vitamin D deficiency on CKD, they are inherently limited by the challenges associated with observational study designs, including confounding factors and reverse causation, which preclude the establishment of causality. To address these limitations and achieve more robust causal inference, Mendelian randomization (MR) has emerged as a robust analytical method [13].
Recent MR studies have investigated the potential causal association between circulating vitamin D levels and renal function. However, the findings have been inconsistent [14,15]. While some studies have suggested that higher genetically predicted vitamin D levels may adversely affect estimated glomerular filtration rate (eGFR) [14], other studies have demonstrated null associations [15]. Notably, the majority of these investigations were conducted in general population cohorts, with limited consideration of active vitamin D metabolites such as 1,25-dihydroxyvitamin D [1,25(OH)2D] or inclusion of patients with clinically confirmed CKD.
In this study, we examined MR analysis to investigate the causal link between genetically predicted vitamin D levels and the risk of CKD progression. Utilizing data from the Korean cohort study for outcomes in patients with CKD (Korean Cohort Study for Outcome in Patients With Chronic Kidney Disease [KNOW-CKD]), we assessed the causal effects of both 25-hydroxyvitamin D [25(OH)D] and 1,25(OH)2D on CKD progression.
Methods
Data source and study population
This study was designed to investigate the causal relationship between serum vitamin D levels (25(OH)D and 1,25(OH)2D) and CKD progression, using genetic variations (single nucleotide polymorphisms [SNPs]) as instrumental variables (IVs) under a MR framework. As shown in Fig. 1, the framework examines the association between SNPs and serum vitamin D levels and their subsequent causal effect on CKD progression, while accounting for potential confounders.
Following this approach, our study initially considered 2,426 individuals diagnosed with CKD. Among these, 2,238 participants were sourced from the KNOW-CKD cohort [16,17], a multicenter, prospective observational study that originally enrolled 2,388 CKD patients, including those with diabetic nephropathy (n = 309), hypertensive nephrosclerosis (n = 171), glomerulonephritis (n = 296), polycystic kidney disease (PKD; n = 364), and unspecified causes (n = 77). Additionally, 188 patients with biopsy-confirmed diabetic nephropathy were obtained from two medical institutions: the Human Biobank of Seoul National University Hospital (91 patients) and Kyung Hee University Medical Center (97 patients).
During the selection process, 364 individuals with PKD were excluded, along with 104 participants lacking genomic DNA and 63 cases without calculable eGFR slope due to loss to follow-up. Following rigorous genotype quality control (QC) procedures, genomic DNA samples from 1,895 individuals were retained, while 157 samples failed to meet QC standards. Additional exclusions included 521 and 1,233 individuals with missing information on the exposure variables 25(OH)D and 1,25(OH)2D, respectively.
After applying these criteria, the final study population consisted of 1,217 participants for analyses involving 25(OH)D and 505 participants for 1,25(OH)2D. In total, 2,045 samples passed QC, retaining 7,763,720 SNPs for analysis. A detailed overview of the exclusion process is provided in Fig. 2.
We obtained both written informed consent and blood samples from all participants, and the Institutional Review Board of Seoul National University Hospital (C-1704-025-842) granted approval for the study.
Outcome variable: chronic kidney disease progression measured by estimated glomerular filtration rate slope
The eGFR in the KNOW-CKD cohort was calculated using the four-variable CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation [18]. The calculation of the eGFR slope was based on serum creatinine levels measured at various time points, including baseline, 6 months after enrollment, and at least one follow-up measurement between 2011 and subsequent follow-ups occurring every 1 to 7 years. Only participants with at least three eGFR measurements were included in the analysis.
We used a linear mixed model (LMM) in which “time” was included as both a fixed and a random effect. The fixed‐effect term for time (β1) estimates the average annual change in eGFR across the entire cohort. To accommodate individual variability, each participant was also allowed their own random intercept (u0i) and random slope for time (u1i). The random intercept captures how each person’s baseline eGFR deviates from the overall mean (β0), while the random slope captures how each person’s rate of eGFR change over time differs from the average. Formally, for subject i at time j:
The CKD progression for each individual is defined as their estimated annual eGFR slope,
expressed in mL/min/1.73 m2 per year. A more negative slope value indicates faster decline (i.e., more rapid CKD progression).
Exposure variables: serum 25(OH)D and 1,25(OH)2D levels
Baseline serum 25(OH)D levels were assessed using ADVIA Centaur Vitamin D Total Assay reagents (Siemens), and baseline serum 1,25(OH)2D levels were measured using the DIAsource 1,25(OH)2D-Vit D RIA Kit (DIAsource). The limits of detection for 25(OH)D and 1,25(OH)2D were 3.20 ng/mL and 1.4 pg/mL, respectively [17].
Genotyping and quality control
Genomic DNA was extracted from peripheral blood leukocytes collected in ethylenediaminetetraacetic acid-coated tubes to prevent coagulation. Genotyping was conducted using the KoreanChip (K-CHIP; Affymetrix Axiom KORV1.1, Thermo Fisher Scientific), a microarray specifically designed by the K-CHIP Consortium to optimize genetic studies for the Korean population by overcoming the limitations of multiethnic genome-wide arrays [21].
To ensure the accuracy and reliability of the data, QC procedures were performed in accordance with the K-CHIP QC protocol [21]. Samples were excluded if they failed to meet key criteria, including dish QC values below 0.82, call rates under 97%, or mismatches between reported and genetic sex. SNPs were removed if they had a call rate below 95%, a minor allele frequency (MAF) less than 1%, or deviated from Hardy-Weinberg equilibrium (p < 10–6). Following these QC steps, 745,176 autosomal SNPs were retained for further analysis.
Genotype imputation was carried out using minimac3, with phasing performed by shapeit_v2.12. The reference panel used for imputation was derived from phase 3 of the 1000 Genomes Project, which includes diverse global populations. Additional post-imputation QC excluded SNPs with low imputation quality (r2 < 0.7) or MAF <1%, resulting in a final dataset of 7,734,192 SNPs for analysis.
Statistical analysis
Association analysis
To investigate the association, both univariate and multivariate Cox proportional hazards regression analyses were performed to evaluate the relationship between serum 25(OH)D and serum 1,25(OH)2D levels and the risk of CKD progression. The multivariate model was adjusted for age, sex, cause of CKD, dietary protein intake (DPI), systolic blood pressure, serum FGF23, Klotho, history of diabetes mellitus, use of vitamin D supplements, and angiotensin receptor blocker (ARB) medication.
Selection of instrumental variables (genome-wide association study)
The eGFR slope was estimated using an LMM with random intercepts, implemented through the MIXED procedures in SAS software, version 9.4 (SAS Institute, Inc.). Based on the KNOW-CKD cohort, we performed a genome-wide association study (GWAS) to identify SNPs to be used as IVs for serum 25(OH)D (n = 1,217) and 1,25(OH)2D (n = 505).
The GWAS analysis was performed using PLINK software (version 1.9) [22] applying an additive genetic model that assumes the genetic effects of two risk alleles are equivalent to twice the effect of a single risk allele [23]. Linear regression models were used for the association analysis, with adjustments for covariates, including age, sex, and the first 10 principal components (PC1–PC10), to account for population stratification. This adjustment mitigates confounding due to ancestral background differences among participants. As a sensitivity analysis, we further adjusted the GWAS for additional potential confounders—namely primary cause of CKD, DPI, systolic blood pressure, serum FGF23, Klotho levels, history of diabetes mellitus, vitamin D supplement use, and ARB therapy—alongside age, sex, and PC1–PC10.
To ensure robustness, potential batch effects from genotyping were evaluated and controlled where necessary. All statistical tests were two-sided, and SNPs meeting the predefined p-value threshold of <5 × 10–5 were selected for further MR analysis. A total of 93 SNPs for 25(OH)D and 55 SNPs for 1,25(OH)2D were identified (Fig. 2). This moderate threshold has been used in previous studies to optimize the balance between statistical power and the number of valid IVs [24,25]. We next applied linkage disequilibrium clumping with an R2 threshold of 0.1 and a 500-kb window to ensure independence of signals. This reduced our candidate lists to 89 and 93 SNPs (Fig. 2). We then harmonized alleles and positions with the CKD progression GWAS summary statistics, excluding palindromic variants with intermediate minor allele frequencies to avoid strand alignment ambiguity, as well as any SNPs absent from the outcome dataset. To minimize weak instrument bias, variants with an association p ≥ 0.05 for CKD progression were discarded. Finally, we screened the remaining SNPs for horizontal pleiotropy by manual review of known trait associations and removed any with evidence of pleiotropic effects. After these sequential filters, the final instrument sets comprised 18 SNPs for 25(OH)D and 27 SNPs for 1,25(OH)2D (Fig. 2).
We also examined previously reported trait associations for each selected SNP using the GWAS Catalog (https://www.ebi.ac.uk/gwas/) to assess potential biological relevance and identify possible pleiotropic variants. Most SNPs were not directly associated with CKD-related phenotypes, but some were located near genes involved in metabolic or immune-related pathways. These findings were considered in the interpretation of MR results and highlighted the exploratory nature of the instrument selection.
Mendelian randomization analysis
To investigate the causal relationship between vitamin D levels and CKD progression, MR analysis was conducted using the selected genetic variants as IVs.
To investigate the causal effect of vitamin D on CKD progression, we performed a one-sample MR analysis in the KNOW-CKD cohort, following STROBE-MR (Strengthening the Reporting of Observational Studies in Epidemiology using MR) guidelines [26]. Individual-level data allowed us to estimate genetic associations with both circulating vitamin D (serum 25(OH)D and 1,25(OH)2D) and eGFR slope within the same population. We first identified SNPs associated with each vitamin D via GWAS (p < 5 × 10–5), then selected IVs based on their strength (F-statistic > 10) to exclude weak instruments. To address potential horizontal pleiotropy and reinforce the validity of our causal inference, we applied six complementary MR methods—inverse variance weighted (IVW), radial IVW, MR-Egger, weighted median, penalized weighted median, and simple median. Our framework (Figs. 1, 2) rests on three core assumptions: 1) relevance: each IV is robustly associated with vitamin D; 2) independence: IVs are uncorrelated with confounders of the vitamin D–CKD progression association (controlled via population-structure adjustment and GWAS sensitivity analyses including clinical covariates); and 3) exclusion restriction: IVs influence CKD progression solely through their effect on vitamin D, with no alternative biological pathways. Descriptions of each MR method are provided below.
1) IVW: This method combines the effect estimates from genetic instruments, weighting them by the inverse of their variances. It is a highly efficient and statistically robust method when horizontal pleiotropy is not present. However, its assumption of no intercept in the model may reduce its reliability when pleiotropic effects are significant. 2) MR-Egger: MR-Egger introduces an intercept term to the model, accounting for potential directional pleiotropy. It allows for variation in pleiotropic effects across different genetic instruments. While it is effective in handling heterogeneity, MR-Egger typically has lower statistical power than the IVW method. 3) Weighted median: This method calculates consistent causal effect estimates even when up to 50% of the genetic instruments are invalid, as long as the assumption of InSIDE (Instrument Strength Independent of Direct Effects) holds true. 4) Penalized weighted median: An extension of the weighted median approach, this method adds penalties for SNPs that show significant deviations. By reducing the influence of outliers, it strengthens the robustness of causal estimates, ensuring more reliable results. 5) Simple median: The simple median method estimates the causal effect by calculating the median of individual SNP effect estimates. While it may be less precise compared to the weighted or penalized weighted median methods, it remains a useful alternative when only a small number of valid instruments are available. 6) Radial IVW: The radial IVW method is a modification of the standard IVW approach that uses radial regression to identify and adjust for outlier SNPs contributing disproportionately to heterogeneity. It is particularly effective in addressing complex horizontal pleiotropy, thereby improving the accuracy and reliability of causal estimates [13]. In this study, the radial IVW method was applied to identify potential outliers that may contribute to horizontal pleiotropy and heterogeneity, in order to reduce the biasing effects of outlier SNPs.
Additionally, we incorporated the MR Pleiotropy RESidual Sum and Outlier (MR-PRESSO) method as a part of the MR analysis to evaluate the causal effects of vitamin D levels on CKD progression. MR-PRESSO is a robust statistical tool that identifies and adjusts for horizontal pleiotropy [27], which occurs when genetic variants used as IVs influence the outcome through alternative pathways unrelated to the exposure of interest. By mitigating biases caused by pleiotropic effects, this approach enhances the validity and reliability of causal inferences, ensuring more accurate conclusions regarding the role of vitamin D in CKD progression.
Results
Supplementary Tables 1 and 2 (available online) present the baseline characteristics according to serum vitamin D status. Patients with 25(OH)D levels <15 ng/mL had a significantly shorter follow-up duration (3.9 years vs. 4.0 years, p = 0.02), a more rapid decline in eGFR (–2.6 mL/min/1.73 m2 per year vs. –2.3 mL/min/1.73 m2 per year, p < 0.01), and lower serum 1,25(OH)2D concentrations (22.2 pg/mL vs. 25.6 pg/mL, p < 0.01) compared to those with levels ≥15 ng/mL. Similarly, patients with 1,25(OH)2D levels <25 pg/mL exhibited shorter follow-up duration (2.5 years vs. 3.0 years, p = 0.03), faster eGFR decline (–2.6 vs. –2.3, p = 0.01), and lower 25(OH)D levels (16.7 ng/mL vs. 19.3 ng/mL, p < 0.01) compared to those with higher levels. In both comparisons, lower vitamin D status was significantly associated with markers of renal metabolic disturbances, including anemia, hypocalcemia, hyperphosphatemia, lower high-density lipoprotein cholesterol, and elevated intact parathyroid hormone levels. According to Supplementary Table 3 (available online), the most common causes of CKD among patients with measured 25(OH)D levels (n = 1,217) were PKD (29.8%), diabetes mellitus (25.4%), and glomerulonephritis (24.3%). Meanwhile, among those with measured 1,25(OH)2D levels (n = 505), glomerulonephritis was the leading cause (39.0%), followed by diabetes mellitus (24.9%) and hypertension (17.8%).
In the association analysis using the Cox proportional hazard model, lower serum 25(OH)D and 1,25(OH)2D levels were significantly associated with an increased risk of CKD progression in both univariate and multivariate analyses (Fig. 3).
After rigorous data QC, a total of 93 SNPs were identified for 25(OH)D and 55 SNPs for 1,25(OH)2D, all meeting the significance threshold of p <5e-5 (Supplementary Tables 4, 5; available online). To visualize the genome-wide association results, Manhattan plots for 25(OH)D and 1,25(OH)2D are provided in Supplementary Figs 1 and 2 (available online), highlighting SNPs that exceeded the suggestive or genome-wide significance thresholds.
IVs associated with 25(OH)D and 1,25(OH)2D levels, excluding potential pleiotropic SNPs, are summarized in Tables 1 and 2. Variables included in the tables are chromosome number, chromosomal position, SNP name, function, nearest gene, allele, MAF, beta coefficient, standard error (SE), p-value, and F-statistics.
We validated the three core assumptions of MR as follows: 1) Instrument strength: All SNP instruments had F-statistics >10 (Tables 1, 2), indicating strong association with vitamin D levels. (2) No confounding of SNP–exposure: In sensitivity GWAS, we adjusted for age, sex, PC1–PC10, cause of CKD, DPI, systolic blood pressure, FGF23, Klotho, diabetes history, vitamin D supplement use, and ARB therapy. The SNP–vitamin D effect sizes and directions remained essentially unchanged compared with the primary model (Supplementary Tables 6, 7; available online), indicating that our instruments are unlikely to be confounded. (3) No horizontal pleiotropy: MR-Egger intercepts were non‐significant and MR-PRESSO global tests were non‐significant for 25(OH)D and 1,25(OH)2D, respectively (Table 3), indicating no evidence of directional pleiotropy.
Table 3 presents the results of sensitivity analyses conducted to assess horizontal pleiotropy in the MR analysis of the association between serum vitamin D levels (25(OH)D and 1,25(OH)2D) and CKD progression. One of the core assumptions of MR, the exclusion restriction assumption, requires that the IVs influence the outcome only through the exposure. Horizontal pleiotropy represents a key violation of this assumption and must therefore be evaluated to ensure the validity of causal inference. In the MR-Egger regression, the intercept was estimated to be 0.008 for 25(OH)D (SE = 0.060, p = 0.90) and 0.059 for 1,25(OH)2D (SE = 0.054, p = 0.28), indicating no statistically significant evidence of directional horizontal pleiotropy. Similarly, the MR-PRESSO global test yielded non-significant p-values for both 25(OH)D (p = 0.99) and 1,25(OH)2D (p = 0.91), suggesting no presence of outlier SNPs contributing to pleiotropy. These findings indicate that the IVs used in this study are unlikely to be biased by horizontal pleiotropy and support the validity and robustness of the causal estimates derived from the MR analyses.
Table 4 presents the results of MR analyses examining the causal effect of genetically predicted serum vitamin D levels on CKD progression in the KNOW-CKD cohort. MR sensitivity analyses using the additional confounder-adjusted SNPs are summarized in Supplementary Table 8 (available online) and likewise demonstrate effect estimates concordant with the primary results.
Individuals with genetic variants associated with higher serum 25(OH)D levels exhibited a slower decline in eGFR compared to those without such variants. In the IVW analysis, each unit increase in genetically predicted 25(OH)D level was associated with an attenuated annual decline in eGFR by approximately 0.246 mL/min/1.73 m2 per year (β = –0.246, SE = 0.093, p = 8.54E-03), indicating a statistically significant negative association.
The radial IVW method, which accounts for potential outlier SNPs, yielded the same point estimate (β = –0.246) but with a markedly smaller SE (0.044) and a substantially more significant p-value (p = 2.13E-08). This suggests that the adjustment for influential outliers may have contributed to the improved precision and statistical significance of the causal estimate. Although consistent directions of effect were observed in the penalized weighted median, weighted median, and simple median methods, their p-values (p = 7.81E-02, p = 8.50E-02, and p = 7.45E-02, respectively) did not meet the conventional threshold for statistical significance. The MR-Egger method also showed a similar trend (β = –0.276, SE = 0.242, p = 2.70E-01), though not significant.
Notably, more robust and consistent associations were identified between 1,25(OH)2D levels and CKD progression. The radial IVW method demonstrated a significant protective association, with each unit increase in genetically predicted 1,25(OH)2D level corresponding to a slower eGFR decline by approximately 0.256 mL/min/1.73 m2 per year (β = –0.256, SE = 0.057, p = 8.84E-06). Similar statistically significant negative associations were confirmed by the IVW, MR-Egger, and weighted median methods (p = 5.50E-04, 3.52E-02, and 1.13E-02, respectively), and the simple median method also yielded consistent findings (β = –0.268, SE = 0.082, p = 8.76E-03). These findings suggest that genetically predicted vitamin D levels may influence the rate of kidney function decline in CKD, with particularly robust and consistent associations observed for 1,25(OH)2D.
Fig. 4 illustrates scatter plots comparing the estimated effects of SNPs on serum vitamin D levels with their effects on CKD progression, and analogous plots for the confounder-adjusted instruments are shown in Supplementary Fig. 3 (available online).
Discussion
This study confirmed a significant negative association and causal relationship between serum vitamin D levels and CKD progression, as demonstrated by Cox proportional hazards analysis and MR. According to the MR findings, genetically predicted higher serum 25(OH)D levels were associated with a slower decline in eGFR. While the IVW method showed statistical significance, other methods such as the penalized weighted median, weighted median, and simple median demonstrated consistent negative associations that did not reach conventional levels of significance. In contrast, more robust and consistent evidence was observed for 1,25(OH)2D. All MR methods demonstrated significant negative associations with CKD progression. These findings support the possibility that vitamin D may play a protective role in the progression of CKD.
Our previous research, utilizing cross-sectional, cohort, and PSM analyses, confirmed a significant association between lower 25(OH)D levels and an accelerated decline in renal function among CKD patients [12]. Another cohort study conducted in Italy demonstrated that individuals with lower baseline 25(OH)D levels had a significantly higher risk of CKD progression and progression to end-stage renal disease [28]. Similarly, a longitudinal study in Europe reported the reno-protective effects of vitamin D, showing that supplementation with active vitamin D analogs was associated with reductions in proteinuria and a slower decline in eGFR [29]. Additionally, randomized controlled trials conducted in North America have explored the role of 1,25(OH)2D, the active form of vitamin D, in CKD patients [30]. These trials have shown that reduced levels of 1,25(OH)2D are associated with higher risks of inflammation, oxidative stress, and proteinuria, all of which contribute to kidney damage. Systematic reviews and meta-analyses incorporating data from cohort and interventional studies across Europe, Asia, and North America have further supported these findings [31,32]. These analyses consistently report that vitamin D deficiency is linked to adverse renal outcomes, including faster CKD progression, increased proteinuria, and heightened cardiovascular risks in CKD patients.
Additionally, previous MR studies conducted in general European populations have yielded conflicting results. One study identified significant inverse associations between genetically predicted levels of both 25(OH)D and 1,25(OH)2D and eGFR, suggesting a potential nephrotoxic effect of elevated vitamin D levels [14]. In contrast, a large-scale two-sample MR study involving over 440,000 individuals of European ancestry reported no evidence of a causal association between 25(OH)D and eGFR or CKD progression [15]. More recently, an MR study evaluating multiple micronutrients found no significant relationship between genetically predicted vitamin D levels and CKD or acute kidney injury, although a positive association with PKD was observed [33]. These divergent findings may reflect methodological variability, including differences in study populations, genetic instruments, and the vitamin D metabolites analyzed. Notably, most prior MR studies have focused exclusively on 25(OH)D and were conducted in general population cohorts, thereby limiting their clinical relevance for individuals with established CKD. In contrast, the present study incorporated both 25(OH)D and its active metabolite, 1,25(OH)2D, within a clinically diagnosed CKD cohort, providing more specific and clinically applicable causal inferences.
Although patients with CKD, whether in the pre-dialysis or dialysis stage, are commonly treated with active vitamin D analogs or cinacalcet, genetically predicted vitamin D levels may still offer clinically meaningful insights. This study provides biological evidence supporting a potential causal role of endogenous vitamin D deficiency in CKD progression. While vitamin D-related genetic variants are not yet directly applicable in routine clinical practice, they may help identify individuals at higher risk of deficiency and guide the development of more personalized supplementation strategies.
Therefore, rather than suggesting immediate changes to current treatment guidelines, our findings may serve as a foundation for genotype-based risk prediction and therapeutic optimization in CKD management. Notably, this study utilized an MR analysis based on GWAS data, which effectively addresses the limitations of observational studies, such as confounding factors and reverse causation. This methodological approach enhances the accuracy of causal inference, contributing to a deeper understanding of the observed differences in effect sizes and statistical significance compared to previous research.
A possible mechanism underlying the relationship between vitamin D levels and CKD progression has been suggested. Vitamin D plays a crucial role in maintaining calcium and phosphate homeostasis, which is essential for kidney function in CKD [34]. Dysregulation of vitamin D metabolism in CKD can lead to secondary hyperparathyroidism, vascular calcification, and exacerbation of renal damage [35].
Additionally, vitamin D modulates immune responses by suppressing pro-inflammatory cytokines such as interleukin 6 and tumor necrosis factor alpha, thereby exerting anti-inflammatory and anti-fibrotic effects [36,37]. These properties are particularly significant in CKD, where chronic inflammation contributes to progressive kidney damage. Furthermore, vitamin D inhibits the renin-angiotensin system, reducing intraglomerular hypertension and proteinuria, which are key drivers of CKD progression [38]. Oxidative stress is another critical factor in CKD pathophysiology. Vitamin D has been shown to enhance the expression of antioxidant enzymes and reduce reactive oxygen species, thereby mitigating oxidative damage [39,40].
From a genetic perspective, variations in genes involved in vitamin D metabolism and function may influence its protective effects in CKD. Previous GWAS have identified genes such as CYP2R1, GC, and CYP27B1 that are associated with vitamin D metabolism [41–43]. These genetic variations may regulate the interaction between vitamin D levels and CKD progression by affecting the synthesis, activation, and degradation pathways of vitamin D. As such, they play a critical role in modulating the relationship between vitamin D levels and kidney function in CKD patients. The significant negative association observed in this study between vitamin D levels and CKD progression is likely attributable to these molecular, physiological, and genetic mechanisms. These findings emphasize the biological importance of maintaining adequate vitamin D levels in CKD patients and underscore the need for further research considering genetic factors.
Several limitations should be acknowledged when interpreting the findings of this study. First, a key assumption in MR analysis is that genetic variants influence CKD progression exclusively through their impact on vitamin D levels. Although the MR-Egger intercept was utilized to address potential pleiotropy, unmeasured functional effects of genetic variants may independently affect renal function [44]. Moreover, genetically predicted vitamin D levels may only partially reflect overall vitamin D status. Genetic instruments do not capture non-genetic and modifiable factors such as dietary intake, sun exposure, physical activity, comorbidities, or medication use, all of which can influence circulating vitamin D levels and may act as residual confounders in the context of CKD progression. Therefore, although MR is designed to reduce confounding and reverse causation, it cannot eliminate all sources of bias, and its findings should be interpreted within the valid scope of genetic inference. Notwithstanding these theoretical limitations, we performed additional GWAS-adjusted MR sensitivity analyses controlling for age, sex, cause of CKD, DPI, systolic blood pressure, serum FGF23, Klotho, diabetes history, vitamin D supplementation, ARB use, and the first 10 principal components. The directions and magnitudes of SNP–vitamin D associations and the MR estimates on eGFR slope remained essentially unchanged, underscoring the robustness of our results. Second, this MR analysis was limited to assessing the linear effects of circulating vitamin D levels in CKD patients. A larger dataset would be required to perform nonlinear MR analyses, particularly to investigate potential nonlinear relationships between vitamin D levels and CKD progression. Third, this study includes findings linked to Weak Instrument Variables in MR analysis, necessitating cautious interpretation [45]. The validity of the instrument depends on the strength and precision of the association between genetic IVs and the risk factor, which in this case is vitamin D. To identify SNPs associated with 25(OH)D and 1,25(OH)2D levels, a suggestive genome-wide significance threshold (p < 5 × 10–5) was employed due to the moderate sample size and exploratory nature of the analysis. While this approach improved instrument availability, it may have increased the likelihood of including false-positive variants and weakened the overall instrument strength. Future studies with larger sample sizes and stricter thresholds may help confirm and extend the present findings. Fourth, pleiotropy, defined as a single genetic variant influencing multiple traits, remains a critical factor in MR analysis. If pleiotropy is present, it may act as a confounding factor, distorting causal relationships. While multivariate MR analysis is recommended to address confounding, this study did not incorporate such an approach [46,47]. Instead, pleiotropy was carefully assessed using various methods to identify and evaluate potential sources of bias. Fifth, despite confirming a causal relationship between serum 1,25D levels and CKD progression through MR analysis, the study’s limited sample size restricts the generalizability of the findings. Larger, more comprehensive studies are necessary to validate these results and explore more nuanced effects or subtle variations. Additionally, the analysis was based on baseline levels of 25(OH)D and 1,25(OH)2D, without considering longitudinal changes in vitamin D levels over time. Finally, this study employed a one-sample MR design using individual-level data from the KNOW-CKD cohort. While this approach helps minimize bias due to sample heterogeneity, it has limited generalizability to external populations. In addition, the genetic instruments were selected using a suggestive genome-wide significance threshold (p < 5 × 10–5), which excluded key vitamin D metabolism genes such as CYP2R1, DHCR7, GC, and CYP24A1, potentially limiting the biological plausibility of the findings. Nevertheless, the consistency of results across various sensitivity analyses supports the robustness of our findings. Future studies using two-sample MR designs with larger, publicly available GWAS datasets and stronger genetic instruments are warranted to validate and extend these results.
Nevertheless, this study is significant as the first to use MR analysis based on genetic variants associated with serum vitamin D levels to investigate their causal relationship with CKD progression in a Korean CKD patient cohort. A major strength of this study is its validation of the causal relationship between serum 25(OH)D levels and CKD progression, demonstrating consistency with prior research. Furthermore, it is the only study to date to evaluate the relationship between serum 1,25(OH)2D levels and CKD progression. Future studies with larger sample sizes and more extensive data from Korean CKD patients are required to further explore significant SNPs and refine the evaluation of causal relationships.
The causal relationship between vitamin D and CKD progression was evaluated using an MR approach, which reduces confounding and reverse causation. Unlike traditional observational studies, MR utilizes genetic instruments to approximate randomized exposure, strengthening the inference of causality. The consistent findings across multiple MR methods, especially for 1,25(OH)2D, support a potentially protective causal role of active vitamin D in CKD progression.
This study elucidates the causal relationship between serum 25(OH)D and 1,25(OH)2D levels and CKD progression through MR analysis based on GWAS data. The findings demonstrate significant negative associations between genetically predicted vitamin D levels and CKD progression, highlighting the potential protective role of vitamin D in maintaining kidney function. These results suggest that proper management of vitamin D levels may be beneficial for CKD patients and provide critical evidence for the development of vitamin D-related therapeutic strategies. Furthermore, addressing vitamin D deficiency as a public health issue may require preventive measures to improve vitamin D status not only in CKD patients but also in the general population.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
