Efficacy and safety of vitamin D supplementation in diabetic kidney disease: an umbrella review of systematic reviews and meta-analyses

1 Introduction

Diabetes mellitus imposes a growing epidemiologic burden. In 2021, the Global Burden of Disease (GBD) study estimated 529 million people living with diabetes (95% uncertainty interval [95% UI]: 500–564 million), with a global age-standardized prevalence of 6.1% (95% UI: 5.8–6.5) and a projection exceeding 1.31 billion people by 2050 (95% UI: 1.22–1.39 billion) (1). The International Diabetes Federation reported 589 million adults aged 20–79 years living with diabetes in 2024 and projected 853 million by 2050 (2). In South and Central America, the corresponding figure was 35.4 million in 2024, with an expected increase to 51.5 million by 2050; in Peru, adult prevalence was 6.4%, equivalent to 1,335,800 affected adults (3, 4). The renal burden attributable to diabetes represents a substantial component of this morbidity and mortality: a GBD 2021 analysis of chronic kidney disease due to type 2 diabetes estimated 107.6 million prevalent cases and 11.28 million disability-adjusted life-years (DALYs) in 2021, with sustained increases in mortality and DALYs since 1990; notably, Andean Latin America showed the greatest percentage change in age-standardized mortality rate across regions, placing Peru in an epidemiologic setting of direct clinical relevance to this question (5).

Vitamin D has been the subject of clinical investigation in diabetic kidney disease because of its involvement in regulation of the renin–angiotensin–aldosterone system, immune modulation, attenuation of tubulointerstitial inflammation, and effects on podocytes. Published clinical trials have evaluated both native forms (cholecalciferol, ergocalciferol, and calcifediol) and active forms and analogs (calcitriol, paricalcitol, and alfacalcidol), with outcomes including urinary measures of albuminuria and proteinuria, renal function estimated by glomerular filtration rate and serum creatinine, metabolic control, inflammatory markers, and clinical events (6, 7).

Systematic reviews published in recent years consistently indicate an antiproteinuric signal of heterogeneous magnitude associated with vitamin D supplementation, particularly with active analogs such as paricalcitol and calcitriol, whereas effects on glomerular filtration and serum creatinine are less consistent (8–10). A recent tertiary synthesis reported a possible reduction in proteinuria without a demonstrated effect on other renal parameters (11). This dispersion of results reflects recognizable features of the primary evidence base: pharmacologic differences between native forms and active analogs, populations with different stages of kidney disease, inconsistent reporting of renoprotective cointerventions, and urinary outcomes defined non-equivalently across studies. The accumulation of systematic reviews with partially overlapping inclusion criteria over a short time span justifies a hierarchical reading of the aggregated body of secondary evidence, capable of distinguishing consistent findings from those supported by individual reviews and of contextualizing the nutrient effect against the current renoprotective standard: renin-angiotensin system blockade, sodium-glucose cotransporter 2 (SGLT2) inhibitors, selected glucagon-like peptide-1 receptor agonists (GLP-1 RAs), and nonsteroidal mineralocorticoid receptor antagonists.

The main objective of this umbrella review was to critically evaluate and summarize the efficacy and safety of vitamin D supplementation in adults with diabetic kidney disease (DKD), including studies reported as diabetic nephropathy (DN). The review aimed to distinguish evidence for native vitamin D from evidence for active vitamin D analogs where the secondary literature allowed separable interpretation, without assuming pharmacologic equivalence or formal comparative ranking in the absence of direct or network evidence. The research question was operationalized as follows: adult population with DKD/DN; intervention with native vitamin D or active forms/analogs; comparator of placebo, no intervention, standard care, active comparators, or different formulations and doses; and primary outcome of reduced albuminuria or proteinuria, assessed by urinary albumin-to-creatinine ratio (UACR), urinary albumin excretion rate (UAER), or 24-hour proteinuria. Secondary objectives were to synthesize effects on estimated glomerular filtration rate (eGFR), serum creatinine, kidney replacement therapy, mortality, cardiovascular events, blood pressure, lipid profile, glycemic control, inflammatory markers, quality of life, and safety; describe evidence separately for native forms versus active analogs when possible; and analyze sources of heterogeneity by formulation, kidney disease stage, baseline 25-hydroxyvitamin D [25(OH)D] status, dose, treatment duration, and cointerventions. The a priori hypothesis was conservative: vitamin D might reduce albuminuria or proteinuria as a surrogate outcome, but the available secondary evidence does not yet allow confirmation of clinical renoprotection on hard outcomes or conclusions regarding comparative safety.

2 Methods

2.1 Protocol registration and reporting guidelines

The protocol for this overview of reviews was registered in the International Prospective Register of Systematic Reviews (PROSPERO) under identifier CRD420251250914 on December 10, 2025. The report was structured according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 (PRISMA 2020) recommendations for general systematic review components (12) and the Preferred Reporting Items for Overviews of Reviews (PRIOR) recommendations for overviews of reviews (13). The completed PRIOR checklist is provided as Supplementary Material 1.

All operational stages were designed to be performed by two independent reviewers. Discrepancies in study selection, data extraction, methodological quality assessment with A MeaSurement Tool to Assess systematic Reviews 2 (AMSTAR-2) (14), risk of bias assessment with Risk Of Bias In Systematic reviews (ROBIS) (15), and certainty of evidence assessment with Grading of Recommendations Assessment, Development and Evaluation (GRADE) (16) were resolved by consensus until agreement was reached. This rule was applied uniformly to preserve traceability and reduce arbitrariness in the synthesis.

2.2 Research question and eligibility criteria

The question was operationalized using the Population-Intervention-(Exposure)-Comparator-Outcome [PI(E)CO] format. The population comprised adults aged 18 years or older with diabetic kidney disease (DKD), including studies reported as diabetic nephropathy (DN), defined by clinical diabetic etiology or histologic confirmation when available. Reviews with mixed populations were accepted only if DKD/DN-specific data were extractable, or if at least 80% of participants had DKD. The intervention included any form of vitamin D: native forms, including cholecalciferol, ergocalciferol, and calcifediol; and active forms or analogs, including calcitriol, paricalcitol, alfacalcidol, doxercalciferol, eldecalcitol, and maxacalcitol. Eligible comparators were placebo, usual care, no intervention, active controls, different doses, or comparisons between formulations. The primary outcome was reduction in albuminuria/proteinuria, measured as urinary albumin-to-creatinine ratio (UACR), urinary albumin excretion rate (UAER), or 24-hour proteinuria. Secondary outcomes included estimated glomerular filtration rate (eGFR), serum creatinine, creatinine clearance, kidney replacement therapy, all-cause and cardiovascular mortality, major adverse cardiovascular events, blood pressure, lipid profile, glycemic control, inflammatory or oxidative markers, quality of life, and safety.

Systematic reviews with or without quantitative meta-analysis that described a search strategy, eligibility criteria, and synthesis methods were included. Reviews that combined randomized controlled trials (RCTs) with nonrandomized studies were eligible only if the RCT component was separable for causal inference. Narrative reviews, scoping reviews, rapid reviews, commentaries, editorials, abstracts without full text, reviews based exclusively on observational studies, and reviews of multicomponent interventions in which the isolated effect of vitamin D could not be discerned were excluded. When duplicate reviews addressing the same question were available, the one closest to the PI(E)CO framework, with separable RCTs, compatible outcomes, more complete coverage, greater recency, and lower relative risk of bias was retained as the anchor review; the others were retained as confirmatory, contextual, or historical evidence.

2.3 Search strategy

The search strategy was designed to identify systematic reviews and meta-analyses on vitamin D in diabetic nephropathy or diabetic kidney disease. MEDLINE/PubMed, Embase, the Cochrane Library, Web of Science Core Collection, and Scopus were searched from inception to April 2, 2026. No language restrictions were applied. The strategy combined three conceptual blocks: diabetic nephropathy or diabetic kidney disease; vitamin D, including native forms and active analogs; and systematic review, meta-analysis, or overview study design. Medical Subject Headings (MeSH), Emtree terms (Embase), and free-text terms adapted to each database were used.

The supplementary search included manual screening of reference lists from the included reviews, citation tracking of key reviews, searches in PROSPERO, and other complementary sources when appropriate. Full search strategies for each database are provided in Supplementary Material 2.

2.4 Study selection and data extraction

Identified records were imported into Rayyan QCRI, with duplicates removed before screening. Two reviewers independently screened titles and abstracts; records deemed eligible or uncertain by at least one reviewer proceeded to full-text assessment. At the full-text stage, disagreements were resolved by consensus or arbitration by a third reviewer. Reasons for full-text exclusion were explicitly recorded and are reported in the Supplementary Material.

Data extraction was performed using a standardized form piloted in advance in Microsoft Excel. Minimum variables extracted from each review included first author, publication year, first author country when available, review type, databases searched, search date, protocol registration, number of primary studies, primary study designs, total number of participants, population criteria, vitamin D formulation, dose, route, duration, comparator, outcomes, pooled estimates with 95% confidence intervals (95% CIs), p value, statistical heterogeneity (I²), between-study variance (τ²) when available, meta-analytic model, variance estimation method, subgroup analyses, sensitivity analyses, publication bias assessment, renoprotective cointerventions, and adverse events. Extraction was performed in duplicate for the principal reviews, and discrepancies were resolved by consensus.

2.5 Assessment of methodological quality and risk of bias of the reviews

Methodological confidence in the included reviews was assessed with AMSTAR-2 (14). All 16 items were considered, with emphasis on the seven critical domains: prior protocol, comprehensive search, justification of excluded studies, risk of bias in primary studies, adequacy of meta-analytic methods, consideration of risk of bias when interpreting results, and assessment of publication bias. The global rating was applied strictly: high confidence if there were no critical flaws and at most one noncritical weakness; moderate if there was more than one noncritical weakness; low if there was one critical flaw with or without noncritical weaknesses; and critically low if there was more than one critical flaw. A more lenient operational classification could be retained only as a descriptive sensitivity analysis, not as the global AMSTAR-2 judgment.

ROBIS (15) was applied as a complementary tool for review-level risk of bias. The domains of eligibility, search and selection, extraction and appraisal, and synthesis of results were evaluated, with a global judgment for each review. Concordance between AMSTAR-2 and ROBIS was interpreted qualitatively, because AMSTAR-2 measures methodological confidence whereas ROBIS estimates the likelihood of bias in the review and its conclusions.

2.6 Assessment of overlap among primary studies

Overlap of primary studies across reviews was quantified using Corrected Covered Area (CCA) according to Pieper et al. (17). A citation matrix was constructed with rows representing unique primary studies and columns representing systematic reviews. CCA was calculated as (N−r)/(r×c−r), where N is the total number of occurrences of primary studies, r the number of unique primary studies, and c the number of reviews. The cutoffs used were <5% slight, 5–10% moderate, 11–15% high, and >15% very high.

The citation matrix included 49 unique primary studies and 115 occurrences across the 11 reviews considered for the overall corpus, yielding a CCA of 13.5%, consistent with high overlap. In the RCT-derived corpus, restricted to nine reviews with RCTs or separable randomized-trial data, 46 unique primary studies and 98 occurrences were identified, yielding a CCA of 14.1%, also consistent with high overlap. These values were treated as descriptive indicators of redundancy, not as a statistical test or an automatic exclusion criterion. The analytical consequence was to avoid pooling estimates from non-independent reviews and to prioritize an outcome-specific anchor-review strategy.

2.7 Data synthesis and quantitative reanalysis

The primary synthesis was structured by outcome and intervention type. No pharmacologic equivalence between native vitamin D and active analogs was assumed. UACR, UAER, 24-hour proteinuria, urinary protein-to-creatinine ratio (UPCR), and urinary protein concentrations were not indiscriminately pooled; they share a clinical direction, but they are not interchangeable metrics. The anchor review for each outcome was selected according to ordered operational criteria: proximity to the PI(E)CO framework, availability of RCTs or separable randomized-trial data, direct compatibility of the reported outcome, avoidance of double counting under high CCA, relative methodological quality and risk of bias, recency and completeness of coverage, and lower unexplained heterogeneity.

Anchor selection did not imply that non-anchor reviews were discarded; they were used to evaluate consistency, contextual applicability, and sensitivity of inference. Alternative anchor choices were considered qualitatively, but they did not alter the direction of the main conclusion because non-anchor estimates were either more heterogeneous, less outcome-specific, based on mixed designs, or dependent on post hoc sensitivity analyses.

This study did not include de novo quantitative reanalysis of primary studies. The synthesis is based on the estimates reported by the anchor reviews selected for each outcome, with critical assessment of overlap using CCA, statistical heterogeneity using I² and τ² when available, and certainty of evidence using GRADE. The decision not to conduct a new meta-analysis was based on two explicit considerations: the high overlap between reviews (overall CCA 13.5% and RCT-derived corpus CCA 14.1%) prevents treating the estimates as independent confirmations, and additional pooling of primary studies already extensively captured by the anchor reviews would introduce analytical redundancy without adding new causal inference. A full primary reanalysis by outcome was beyond the scope of this overview.

2.8 Classification of credibility and certainty of the evidence

Certainty of evidence was assessed with GRADE for each outcome (16). Evidence from RCTs started as high certainty and was downgraded for risk of bias, inconsistency, indirectness, imprecision, and publication bias. GRADE constitutes the definitive certainty assessment of this study.

3 Results

3.1 Search results and analytical corpus

The selection process is summarized in Figure 1. A total of 813 records were identified through database searching: Scopus (n = 112), Embase (n = 74), PubMed (n = 52), Web of Science (n = 404), and the Cochrane Library (n = 171), with no additional records identified through other sources. No duplicate records were identified, so all 813 records proceeded to title and abstract screening. Of these, 793 were excluded at that stage. Thus, 20 full-text articles were assessed for eligibility, of which nine were excluded. Ultimately, 11 reviews met the inclusion criteria: nine contributed to the RCT-derived analytical corpus because they provided RCTs or separable randomized-trial data, and two were retained as contextual evidence due to mixed designs or incomplete separability (8–10, 18–25). Table 1 presents the main characteristics of the included reviews and the analytical role assigned to each within the synthesis.

Table 1. Characteristics of the included reviews and analytical role.

AMSTAR-2: CB = critically low. ROBIS: AP = some concerns. DKD, diabetic kidney disease; T2DM, type 2 diabetes mellitus; DN, diabetic nephropathy; RCTs, randomized controlled trials; MA, meta-analysis; NR, not reported; SR, systematic review. Detailed AMSTAR-2 and ROBIS matrices are presented in the Supplementary Material.

3.2 Synthesis of the primary outcome

The most consistent signal was observed in surrogate urinary outcomes. He 2022 was the anchor review for UACR and UAER because of outcome specificity, an RCT evidence base, and low-to-moderate heterogeneity. This signal supports reduced albuminuria, not proven clinical renoprotection. Estimates were reported as standardized mean differences (SMDs) when urinary metrics originated from heterogeneous scales, and as mean differences (MDs) when primary studies used a common metric. Wang 2019 provided favorable estimates for 24-hour proteinuria and UAER, but with extreme heterogeneity, and was therefore interpreted as a degraded secondary anchor. Xuan 2023 confirmed a potentially favorable direction, although its main results for proteinuria and creatinine depended on sensitivity analyses excluding an influential study. Table 2 summarizes effect direction, the available estimate, heterogeneity, GRADE certainty, and the main reasons for downgrading each outcome.

Table 2. Synthesis of effects, certainty, and operational classification by outcome.

95% CI , 95% confidence interval; I², statistical inconsistency; MACE, major adverse cardiovascular events; MD, mean difference; NE, not evaluable; NR, not reported; SMD, standardized mean difference; KRT, kidney replacement therapy; UACR, urinary albumin-to-creatinine ratio; UAER, urinary albumin excretion rate. GRADE constitutes the definitive certainty assessment of this study. The clinical interpretation column reflects the inference that can be sustained given the available secondary evidence and the documented overlap among reviews.

3.3 Renal function, hard clinical outcomes, and safety

No robust evidence of preserved eGFR or stable reduction in serum creatinine was identified. Hard outcomes, including kidney replacement therapy, all-cause mortality, cardiovascular mortality, and major adverse cardiovascular events, were not reported or synthesized sufficiently for inference. Safety was treated as an independent outcome domain. Across the included review-level syntheses, adverse events were sparsely and heterogeneously reported, and no robust anchor review was available for comparative safety. Clinically relevant events, including hypercalcemia, hyperphosphatemia, changes in serum calcium/phosphate, PTH suppression, nephrolithiasis, treatment discontinuation, serious adverse events, and differential safety between native vitamin D and active analogs, were not consistently synthesized. Therefore, absence of a consistent harm signal should be interpreted as insufficient ascertainment rather than evidence of safety. Table 3 summarizes the safety domains and the limits of review-level inference.

Table 3. Safety outcomes and interpretability by vitamin D formulation.

CKD-MBD, chronic kidney disease-mineral and bone disorder; PTH, parathyroid hormone. The table summarizes availability and interpretability of safety domains at the review level; it is not a de novo adverse-event meta-analysis.

3.4 Overlap among primary studies

CCA showed substantial redundancy across reviews. The overall calculation was 13.5% and the calculation restricted to the RCT-derived corpus was 14.1%, both compatible with high overlap according to prespecified cutoffs. Therefore, the reviews should not be interpreted as independent confirmations of one another. Table 4 presents the overall and RCT-derived corpus CCA calculations, together with their interpretation and use in the synthesis.

Table 4. Overlap of primary studies using corrected covered area.

CCA, Corrected Covered Area. Formula: CCA = (N-r)/(r×c-r), where N corresponds to the total number of occurrences of primary studies, r to the number of unique primary studies, and c to the number of reviews. The cutoffs used were <5% slight, 5–10% moderate, 11–15% high, and >15% very high.

3.5 Methodological quality and risk of bias of the reviews

When AMSTAR-2 was applied strictly, all reviews were classified as having critically low confidence, mainly because of the absence of a traceable prior protocol and the lack of a justified list of full-text excluded studies. ROBIS showed a predominance of high risk of bias, concentrated in the synthesis domain because of clinical and methodological heterogeneity, mixed study designs, and non-equivalent urinary metrics. Item- and domain-level details are presented outside the main text to avoid table overload: AMSTAR-2 in Supplementary Material 6 and ROBIS in Supplementary Material 7.

3.6 Certainty of the evidence

GRADE was low for UACR and UAER, low to very low for 24-hour proteinuria, and very low or not evaluable for renal function, safety, and hard clinical outcomes. AMSTAR-2 and GRADE were interpreted as complementary rather than interchangeable judgments. Strict AMSTAR-2 ratings informed the risk-of-bias and interpretation domains, but they did not automatically determine the GRADE certainty for each outcome. UACR and UAER were retained as low rather than very low certainty because the anchor estimate was derived from randomized-trial evidence, was outcome-specific, showed low-to-moderate statistical heterogeneity, and had no major imprecision. Nevertheless, the evidence was downgraded for review-level methodological limitations, indirectness related to surrogate outcomes and heterogeneous DKD definitions, and possible publication bias. For 24-hour proteinuria, renal function, safety, and hard clinical outcomes, additional inconsistency, imprecision, sensitivity-dependent estimates, or absent synthesis justified very low or non-evaluable judgments. The complete GRADE profile by outcome is provided in Supplementary Material 8; key downgrading reasons are summarized in Table 5. The visual synthesis of effect direction and GRADE certainty by domain is presented in Figure 2, which distinguishes the favorable signal for albuminuria/proteinuria from the absence of robust evidence for renal function, hard outcomes, and safety.

Table 5. Main GRADE downgrading reasons by outcome domain.

GRADE, Grading of Recommendations Assessment, Development and Evaluation; KRT, kidney replacement therapy; MACE, major adverse cardiovascular events; UACR, urinary albumin-to-creatinine ratio; UAER, urinary albumin excretion rate.

3.7 Clinical applicability and cointerventions

Applicability to the contemporary DKD standard of care is limited. Reporting of angiotensin-converting enzyme inhibitors (ACE inhibitors) and angiotensin II receptor blockers (ARBs) was incomplete, and the use of SGLT2 inhibitors, GLP-1 RAs, and nonsteroidal mineralocorticoid receptor antagonists was absent or not systematically reported. Consequently, the antiproteinuric signal should be interpreted as evidence generated in a predominantly pre-SGLT2 inhibitor and pre-finerenone context. Therefore, vitamin D should be interpreted primarily within conventional indications such as correction of vitamin D deficiency or management of CKD-mineral and bone disorder, not as an established adjunctive renoprotective therapy for contemporary DKD care. Whether vitamin D provides incremental renal benefit on top of current standard-of-care therapy remains an untested hypothesis rather than a clinical recommendation. Figure 3 summarizes the clinical applicability problem: the evidence derives from small, short-term trials with predominantly surrogate outcomes and scant representation of the current therapeutic standard based on SGLT2 inhibitors, finerenone, and GLP-1 RAs.

Critical extraction corrections were preserved: Shab-Bidar 2011 corresponds to native vitamin D3 with calcium in fortified yogurt drink, not calcitriol; Ahmadi 2013 should be recorded with three months of follow-up; Barzegari/Esfandiari 2019 should be treated as 50,000 IU/week for eight weeks pending definitive primary verification; and VITAL-DKD/de Boer 2019 should be considered indirect evidence unless a DKD-specific subgroup is extracted.

3.8 Analytical sensitivities and synthesis of results

Analytical sensitivity checks supported the same conservative conclusion. First, restricting the synthesis to reviews with randomized trials or separable randomized-trial data did not materially change the interpretation of UACR or UAER, because He 2022 remained the outcome-specific anchor. Second, excluding indirect evidence without a DKD-specific subgroup did not alter the inference for renal function or hard outcomes, which remained absent or insufficiently synthesized. Third, assigning greater weight to Xuan 2023 would not strengthen the conclusion, because its findings for proteinuria and creatinine were sensitivity-dependent after removal of an influential study. Fourth, separating native vitamin D from active analogs prevented a class-effect claim: native vitamin D evidence was limited and mainly narrative, whereas active analog evidence suggested a possible antiproteinuric direction but did not establish comparative efficacy or safety. Fifth, high overlap was managed by avoiding pooled interpretation of non-independent review estimates and by selecting one anchor review per outcome; reviews were not treated as independent confirmations of one another.

Across these scenarios, the conclusion remained unchanged: vitamin D supplementation is associated with a low-certainty favorable signal in surrogate urinary outcomes, without demonstrated preservation of eGFR, reduction in kidney replacement therapy, mortality, cardiovascular events, or reliable comparative safety.

4 Discussion

4.1 Main findings

This overview evaluated seven outcome domains in vitamin D supplementation for diabetic kidney disease or diabetic nephropathy (DKD/DN). Only surrogate urinary outcomes reached low GRADE certainty with a favorable direction: UACR, with SMD −0.24 (95% CI −0.39 to −0.09); UAER, with SMD −0.57 (95% CI −0.71 to −0.43); and 24-hour proteinuria, with MD −0.26 g/day (95% CI −0.34 to −0.17), the latter with extreme heterogeneity and dependent on sensitivity analysis in one of the contributing reviews. The domains of renal function, kidney replacement therapy, mortality, major adverse cardiovascular events, safety, and metabolic-inflammatory outcomes were graded as very low certainty or classified as not evaluable because of insufficient synthesis. The central pattern, therefore, is not one of clinical renoprotective effect, but rather a low-certainty antiproteinuric and antialbuminuric signal concentrated in surrogate outcomes. The fact that no domain reached moderate or high certainty does not represent an analytical failure: it reflects the information that a rigorous overview should reveal when the secondary literature has been more abundant than conclusive.

4.2 Comparison with other studies and clinical implications

The findings are consistent with the most recent individual systematic reviews and with the overview of meta-analyses published in 2025: vitamin D and its analogs may reduce proteinuria and albuminuria, but they do not show a consistent effect on eGFR, creatinine, or hard clinical outcomes (8–11, 20). The main difference relative to those syntheses lies not in the direction of the urinary effect, but in the critical appraisal of the aggregated body of secondary evidence using strict AMSTAR-2, ROBIS, overlap quantification with CCA, and outcome-specific GRADE certainty. He 2022 provided the most specific estimate for UACR and UAER; Wang 2019 extended the domain toward 24-hour proteinuria and inflammatory markers, with very high heterogeneity; Xuan 2023 confirmed a favorable direction, although some of its results depended on sensitivity analyses after removal of an influential study (8, 10, 20). The Chackochan 2025 overview served to evaluate external consistency of effect direction, but it was not incorporated as an analytical unit because it constituted tertiary evidence; its inclusion would have introduced redundancy with the systematic reviews already included in the corpus.

Clinical heterogeneity also limits formulation-specific inference. Native vitamin D compounds and active vitamin D analogs differ in pharmacologic activation, calcium-phosphate effects, and clinical indications, and they should not be interpreted as a single interchangeable intervention. However, the included review-level evidence did not allow reliable estimation of treatment modification by CKD stage, baseline 25(OH)D status, diabetes type, albuminuria severity, dose, or treatment duration. Therefore, the present overview can describe the direction of evidence by outcome domain, but it cannot define the optimal compound, dose, duration, or target subgroup.

The biological plausibility of reduced albuminuria is reasonable. Vitamin D receptor activation modulates the renin-angiotensin-aldosterone system, tubulointerstitial inflammation, oxidative stress, and podocyte integrity--mechanisms consistent with reduced urinary albumin leakage (6, 7). Mechanistic plausibility, however, is not equivalent to clinical causality. Causal inference for hard renal outcomes requires temporal consistency, dose-response effects, adequate counterfactual frameworks, pragmatic trials with sufficient power for clinical outcomes, and, where applicable, triangulation with Mendelian randomization. In the available body of evidence, none of these requirements are met for outcomes beyond albuminuria and proteinuria; the very low GRADE certainty for renal function and clinical events directly reflects this limitation.

Similarly, reduction in albuminuria or proteinuria should not be interpreted as disease modification unless it is accompanied by evidence of slower eGFR decline, reduced kidney failure, fewer cardiovascular events, or lower mortality. In the present evidence base, the favorable direction is confined to urinary surrogate outcomes and does not establish clinical renal protection.

From a clinical standpoint, the current evidence does not justify recommending vitamin D as a specific renoprotective therapy to prevent eGFR decline, kidney replacement therapy, mortality, or cardiovascular events in DKD/DN. Vitamin D should be interpreted primarily within conventional indications such as correction of vitamin D deficiency or management of CKD-mineral and bone disorder, not as an established adjunctive renoprotective therapy for contemporary DKD care. The available urinary signal was generated in treatment contexts with incomplete ACE inhibitor/ARB reporting and little or no systematic representation of SGLT2 inhibitors, finerenone, or GLP-1 receptor agonists. Consequently, whether vitamin D provides incremental renal benefit on top of current standard-of-care therapy remains an untested hypothesis rather than a clinical recommendation (26, 27).

4.3 Study limitations

This study has several limitations. First, an overview inherits the quality of the included reviews; here, strict AMSTAR-2 classified all 11 reviews as critically low confidence, making any derived synthesis necessarily conservative. Second, the available corpus is based on small RCTs with short follow-up and predominantly surrogate outcomes; biologically real but clinically modest effects on renal function or hard outcomes may fail to reach statistical significance in low-powered aggregate meta-analyses, a limitation that GRADE captures through the imprecision domain but does not solve. Third, primary studies not included in systematic reviews are excluded by design from an overview. Fourth, small-study effects and excess significance tests, as reported by the source reviews, have low power when meta-analyses include few primary studies. Fifth, the search cutoff of April 2, 2026 means that later reviews were not incorporated. Sixth, the overall CCA of 13.5% and the RCT-derived corpus CCA of 14.1% indicate high overlap; therefore, conclusions should be interpreted as a hierarchical synthesis of redundant secondary evidence rather than repeated independent confirmation. Additional limitations include possible publication bias and selective reporting of favorable surrogate urinary outcomes in small vitamin D trials, limited follow-up for evaluating CKD progression, and the inability of the available review-level evidence to identify an optimal formulation, dose, treatment duration, or patient subgroup most likely to benefit.

5 Conclusions and recommendations

The clinically supportable conclusion is limited but useful: vitamin D or its analogs show a low-certainty antiproteinuric and antialbuminuric signal in adults with DKD, including studies reported as DN, but the evidence remains confined to surrogate urinary outcomes and does not establish preservation of renal function, reduction in hard renal events, mortality, cardiovascular events, or comparative safety. Future primary research should prioritize pragmatic, multicenter RCTs with adequate power, stratified by baseline 25(OH)D status, formulation (native versus active analogs), CKD stage, albuminuria category, and concomitant use of RAAS blockers, SGLT2 inhibitors, and finerenone, with sufficient follow-up to assess eGFR slope, renal progression, cardiovascular events, and calcium-phosphate safety. Until such evidence becomes available, vitamin D should not be promoted as a specific renoprotective intervention in DKD/DN; correction of a nutritional deficiency and protection of the kidney are distinct clinical hypotheses and should be treated as such.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fneph.2026.1883351/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Maria Amicone, Federico II University Hospital, Italy

Reviewed by: Yerbolat Iztleuov, West Kazakhstan Marat Ospanov State Medical University, Kazakhstan

Suba A., Dr. M.G.R Education and Research Institute, India