Study at a glance
What was the question?
Why do vitamin D studies reach different conclusions about autoimmunity? The review asks whether they measured the same exposure, disease stage and outcome before trying to explain the difference.
Who or what was studied?
Two investigations followed children at increased risk of type 1 diabetes. Two used inherited genetic variants to study vitamin D status and disease susceptibility. Additional brain and mouse research helped explain the limits of tissue and immune measurements.
What did the studies find?
The childhood associations differed. One genetic analysis supported a role for lower vitamin D status in multiple sclerosis susceptibility. The type 1 diabetes genetic analysis did not establish a clear overall effect. These studies used different instruments, populations and outcomes; the contrast alone does not prove that their true effects differ.
What does this mean?
A genetic tendency over life is not the same as starting supplements now. Timing and tissue context can matter, but they do not automatically explain every disagreement. A result that is not statistically clear does not prove zero effect. These findings do not establish a personal blood target, genetic treatment test or supplement plan.
Read the full article below ↓Vitamin D studies can disagree because they ask different questions, study different people or rely on different assumptions. This review compares prospective childhood observations with genetic analyses of multiple sclerosis and type 1 diabetes. It distinguishes demonstrated features of those investigations from proposed explanations for why autoimmune findings vary. For the shorter account, read Why Vitamin D Autoimmune Findings Differ: Study Summary.
First identify the effect being estimated
Measured vitamin D status in a child is not the same exposure as a lifelong genetic tendency toward a lower concentration. First persistent islet autoantibodies are not the same outcome as diagnosed type 1 diabetes or multiple sclerosis. A genetic interaction within one cohort is also different from a comparison of results between two diseases. Treating these as interchangeable creates disagreement that may partly reflect the question being asked. [1] [2] [10] [11]
The core evidence comprises four investigations: TEDDY, DAISY, Mokry’s multiple-sclerosis Mendelian randomization study and Manousaki’s type 1 diabetes Mendelian randomization study. Three additional assessed investigations illustrate regional expression and experimental context. Reuse of an investigation for another question does not make it an independent confirmation. The comparison is a bounded synthesis, not a pooled analysis or a systematic review claiming to explain all autoimmune variation.
Vitamin D Genetics introduces genetic influence without individual determinism. Here the more specific task is to identify what each genetic analysis can estimate and what assumptions it needs. The separate Common VDR Polymorphisms and Functional Implications: Full Evidence Review concerns receptor variation; receptor-associated differences should not be silently substituted for instruments selected to represent circulating vitamin D status.
TEDDY and DAISY: timing and measurement before interpretation
In TEDDY, Norris and colleagues compared 376 children with persistent islet autoimmunity and 1,041 controls from a population selected for increased genetic risk. Higher childhood-average 25-hydroxyvitamin D was associated with lower odds of the outcome: an odds ratio of 0.93 per five nanomoles per litre, with a 95% confidence interval of 0.89 to 0.97. An early-infancy analysis also reported an inverse association. These observations are relevant to early disease biology, but they did not assign supplements randomly. [1]
The childhood average included the first seropositive visit. It therefore cannot be described as exclusively before the initiation of autoimmunity. First detected antibodies also need not mark the first biological event. These distinctions limit a claim about a critical prevention window, even when the measurement precedes a clinical diagnosis. Timing is partly observed here and partly unknown; it should not become a complete explanation by assertion.
TEDDY examined interaction with a vitamin D receptor variant, rs7975232. The childhood interaction had a nominal probability of 0.0072 and a false-discovery-rate value of 0.06. The infancy analysis gave 0.0019 and 0.017 respectively. Correction for testing multiple possibilities matters: the two findings are not equally robust under that correction. An interaction is evidence that an association may vary by genotype in this dataset. It does not establish a clinical rule for tailoring supplements or predict an individual’s treatment response.
DAISY, reported by Simpson and colleagues, did not find a clear measured-status association with islet autoimmunity or progression to type 1 diabetes. For autoimmunity the hazard ratio was 1.12 per 20.27-nanomole-per-litre standard deviation, with a 95% confidence interval of 0.88 to 1.43. The progression estimate was 0.91 per 20.14-nanomole-per-litre standard deviation, with an interval of 0.68 to 1.22. These are different increments from TEDDY’s five-nanomole comparison. [2]
DAISY’s intake, status and progression analyses used different subsets. Denominator and scaling discrepancies in the report require caution about numerical precision. Its intervals do not establish an exact zero effect, but its results challenge a consistent protective interpretation of childhood observational evidence. Differences in population selection, exposure windows and outcome definitions are real study features. Whether any particular feature caused the contrast between findings remains an inference, not a result these two studies directly tested together.
For the underlying biological distinction between antibody detection and functional suppression, see Vitamin D and Autoimmune Initiation: Full Evidence Review. The question here is narrower: a positive association in one cohort and a non-significant association in another do not automatically demonstrate a statistically established difference between their effects. That requires an appropriate comparison, not simply contrasting their probability values.
What Mendelian randomization adds
Mendelian randomization uses genetic variants associated with an exposure as instruments for asking whether that exposure may affect an outcome. Because inherited variants precede disease, this can reduce some problems of reverse causation affecting measured blood status. It still depends on assumptions: the instruments must meaningfully relate to the exposure, avoid relevant confounding and influence the outcome through the exposure rather than an alternative route. The last problem is often described as horizontal pleiotropy. [10] [11]
A genetic instrument for lifelong differences in circulating 25-hydroxyvitamin D does not reproduce starting a supplement in adulthood. It cannot automatically identify the best age, a treatment dose, or the effect of correcting a short period of severe deficiency. Instrument strength, ancestry, sample selection and the scale of analysis also affect interpretation. A statistically precise genetic estimate can remain unsuitable as direct clinical dosing evidence.
Multiple sclerosis: supportive genetic evidence with assumptions
Mokry and colleagues used four vitamin-D-associated loci in their multiple-sclerosis analysis. A one-standard-deviation decrease in log-transformed 25-hydroxyvitamin D was associated with higher odds of multiple sclerosis. The fixed-effect estimate was 2.02, with a 95% confidence interval of 1.65 to 2.46. A random-effects analysis gave 2.07, with an interval of 1.45 to 2.96. These are odds estimates on a particular transformed exposure scale, not a doubling of an individual’s absolute probability after a specified fall in a routine blood test. [10]
The instruments explained about 2.44% of exposure variation, with an F statistic of 49.7. This supports relevance of the instruments but does not settle every causal assumption. There was heterogeneity between variant-specific estimates. Sensitivity analyses supported the overall interpretation, while leaving the usual limitations of genetic inference. No sensitivity test can prove that all alternative biological routes are absent.
The main paper, its material supplements and corrected figure were assessed. The published correction added an author and repaired the pathway illustration, including the DHCR7 arrow and ultraviolet-B labelling. This review uses the corrected interpretation and does not reproduce the old figure. The result supports an etiological role for genetically influenced vitamin D status in this multiple-sclerosis analysis; it is not a randomized trial showing that a later supplement prevents disease.
Type 1 diabetes: no clear overall genetic effect
Manousaki and colleagues used 69 variants in the primary type 1 diabetes analysis. The reported primary odds ratio was 1.093, with a 95% confidence interval of 0.855 to 1.396, per one-standard-deviation decrease in genetically instrumented, log-transformed 25-hydroxyvitamin D. It did not establish a clear overall causal effect. The interval still allows a range of effects, so the finding should not be converted into proof that vitamin D can never matter in type 1 diabetes. [11]
The MR-PRESSO global test indicated evidence of outlier-related pleiotropy, with a probability of 0.001. Removing identified outliers did not yield a significant association. That is relevant reassurance about this sensitivity analysis, but not proof that every instrument assumption holds. A null average estimate can also leave narrower questions unresolved, including nonlinear effects, particular developmental windows and effects too small to estimate precisely.
Those remaining possibilities are hypotheses, not benefits demonstrated by the null result. It would be circular to insist that every unfavourable study must have used the wrong timing or population, while accepting favourable studies without equivalent scrutiny. The same standards of exposure definition, precision and biological relevance should apply to both directions of finding.
The principal report, material supplementary spreadsheets and correction were assessed. The correction concerned a background reference rather than revising the primary estimate. A proxy-excluded figure panel was inconsistent with the corresponding table, and the abstract total differed from the sum of its component sample counts. This review uses the principal table for the primary result and avoids a disputed total or an unsupported precise proxy-excluded estimate. Reporting discrepancies are interpretive limits, not evidence of misconduct.
Comparing diseases is not a formal test of different effects
Mokry’s multiple-sclerosis analysis supports an association under its instrumental-variable assumptions; Manousaki’s type 1 diabetes analysis does not establish a clear overall effect. That is a meaningful difference in available evidence. It is not, on its own, a formal demonstration that the true biological effects differ between diseases. Instruments, exposure resources, outcome datasets and modelling choices differ, and genetic resources can overlap across publications. No claim of wholly independent datasets or a common pooled effect is made here.
The practical conclusion is disease-specific interpretation. An MS susceptibility estimate should not become a claim about all autoimmune diagnoses. Likewise, a T1D null result cannot cancel every vitamin-D-related mechanism in another tissue. The issue is not whether one disease is a better example, but whether the evidence actually addresses the proposed generalisation.
Regional and experimental context sharpen the limits
Ramasamy’s healthy-brain study linked an MS-risk variant with CYP24A1 expression in particular regions. It measured a genotype-expression relationship, not active hormone production in affected lesions. That regional finding offers a concrete reason to avoid assuming that all tissues respond identically, but it does not prove why the two Mendelian randomization studies differ. Local Vitamin D Metabolism and Autoimmune Tissues: Full Evidence Review examines the local-metabolism evidence in detail. [7]
Driver’s receptor-deficient NOD experiments show another distinction: the recipient’s mineral and insulin-secretory context can affect clinical diabetes expression without a simple increase in transferable immune aggression. Whole-body receptor deletion is not modest genetically influenced variation in blood status. The rescue diet was not calcium-only, and four supplementary figures were unavailable for this review. The principal transfer and metabolic findings are used without claims about unassessed regulatory-cell function. [4]
Casteels tested functional suppression and disease transfer in calcitriol-treated mice within a cyclophosphamide-perturbed model. This is stronger functional evidence than counting a regulatory marker, but it does not identify a human genetic subgroup that benefits from supplements. The unavailable Figure 4 caption prevents precise assertions about that apoptosis panel. The functional-transfer findings remain model-specific, and the two NOD papers share research lineage rather than providing independent human replication. [5]
These examples connect to Vitamin D and Immune Tolerance and Vitamin D and the Timing of Immune Activation and Restraint: Full Evidence Review. They identify actual differences in what a system measures. They do not supply a universal retrospective explanation for every positive or negative autoimmune study.
A defensible interpretation of variation
The evidence demonstrates different exposures, populations, endpoints, regional expression patterns and experimental contexts. It suggests, but does not fully test, additional explanations involving developmental timing and nonlinear responses. Genetic interaction within TEDDY, disease susceptibility from Mendelian randomization and functional suppression in mice should each retain their own meaning. None supplies an interchangeable measure of immune benefit.
Clinical treatment questions require their own evidence, as discussed in Vitamin D, Immune Tolerance Mechanisms and Autoimmune Outcomes: Full Evidence Review. This comparison supports qualified genetic evidence for MS susceptibility, uncertain prospective islet-autoimmunity associations and no clear overall T1D effect in the selected genetic analysis. It does not justify genotype-guided supplementation, a universal autoimmune prevention claim or treating non-significance as proof of no possible biology.
Scientific sources and access
Evidence assessed on 5 November 2026. This is an original synthesis of selected research, not a reproduction of the papers or a systematic review. Material source-access limits are identified in the analysis.
[1] Norris and colleagues. Plasma 25-Hydroxyvitamin D Concentration and Risk of Islet Autoimmunity. Diabetes. 2018 Jan;67(1):146 to 154. DOI: 10.2337/db17-0802. PMID: 29061729.
[2] Simpson and colleagues. No association of vitamin D intake or 25-hydroxyvitamin D levels in childhood with risk of islet autoimmunity and type 1 diabetes: the Diabetes Autoimmunity Study in the Young (DAISY). Diabetologia. 2011 Nov;54(11):2779 to 2788. DOI: 10.1007/s00125-011-2278-2. PMID: 21858504.
[4] Driver and colleagues. Calcium insufficiency accelerates type 1 diabetes in vitamin D receptor-deficient nonobese diabetic (NOD) mice. Endocrinology. 2011 Dec;152(12):4620 to 4629. DOI: 10.1210/en.2011-1074. PMID: 21952242.
[5] Casteels and colleagues. 1,25-Dihydroxyvitamin D3 restores sensitivity to cyclophosphamide-induced apoptosis in non-obese diabetic (NOD) mice and protects against diabetes. Clinical and experimental immunology. 1998 May;112(2):181 to 187. DOI: 10.1046/j.1365-2249.1998.00568.x. PMID: 9649179.
[7] Ramasamy and colleagues. Genetic evidence for a pathogenic role for the vitamin D3 metabolizing enzyme CYP24A1 in multiple sclerosis. Multiple sclerosis and related disorders. 2014 Mar;3(2):211 to 219. DOI: 10.1016/j.msard.2013.08.009. PMID: 25568836.
[10] Mokry and colleagues. Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study. PLoS medicine. 2015 Aug;12(8):e1001866. DOI: 10.1371/journal.pmed.1001866. PMID: 26305103.
[11] Manousaki and colleagues. Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study. PLoS medicine. 2021 Feb;18(2):e1003536. DOI: 10.1371/journal.pmed.1003536. PMID: 33630834.
Mokry and colleagues. Correction: Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study. PLOS Medicine. 2016. DOI: 10.1371/journal.pmed.1001981. PMID: 26963823.
Manousaki and colleagues. Correction: Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study. PLOS Medicine. 2021. DOI: 10.1371/journal.pmed.1003624. PMID: 33914743.
Related reading
Effect of Prenatal Vitamin D and Selenium Supplementation on Minipuberty in Male Offspring of Women with Autoimmune Thyroiditis
Background/Objectives: Minipuberty represents the second phase of physiological activation of the reproductive axis and may play a role in postnatal genital development.
Original paper ↗Read more →Vitamin D Status in Patients at the Department of Internal, Autoimmune, and Metabolic Diseases: A Descriptive Cross-Sectional Study
The aim of this study was to examine the levels of vitamin D (25(OH)D) among patients hospitalized in 2022 at the Department of Internal, Autoimmune, and Metabolic Diseases.
Original paper ↗Read more →High frequency of vitamin D receptor gene polymorphism FokI in Brazilian Type 1 diabetes mellitus patients with clinical autoimmune thyroid disease
The aim of this study was to evaluate the relationship between VDR FokI polymorphism (rs10735810), thyroid autoimmunity and thyroid dysfunction (TD) in Brazilian T1DM.
Original paper ↗Read more →Type 1 Diabetes and Other Autoimmune Diseases: Epidemiology, Pathophysiology and Screening
A comprehensive review of literature on T1D and associated AID was conducted, with the aim of drawing informed conclusions relevant to clinical practice.
Original paper ↗Read more →The Relationship Between Autoimmune Disorders and Multiple Sclerosis: Clinical Insights and Therapeutic Approaches
This study investigated the prevalence of AIDs among Polish patients with MS (PwMSs) and its potential effects on disease characteristics.
Original paper ↗Read more →Altered Expression of Vitamin D Metabolism Genes and Circulating MicroRNAs in PBMCs of Patients with Type 1 Diabetes: Their Association with Vitamin D Status and Ongoing Islet Autoimmunity
This study aimed to decipher differences in the expression of genes associated with vitamin D metabolism in T1DM patients and to ascertain if there is a correlation between serum 1,25(OH)2D3 levels and the expression of these genes.
Original paper ↗Read more →Type 1 Diabetes Autoimmunity: To What Extent Vitamin D is Beneficial
Moreover, the coupled therapy and clinical trials that have investigated the use of vitamin D as a therapeutic intervention for autoimmune disorders were also examined, highlighting both successes and limitations.
Original paper ↗Read more →The Footprints of Poly-Autoimmunity: Evidence for Common Biological Factors Involved in Multiple Sclerosis and Hashimoto’s Thyroiditis
Although the biological implications of these data need to be further investigated, we have highlighted the relevance of studies comparing different autoimmune pathologies for the understanding of the core concepts of autoimmunity.
Original paper ↗Read more →Combination therapy with saxagliptin and vitamin D for the preservation of β-cell function in adult-onset type 1 diabetes: a multi-center, randomized, controlled trial
Here, we conducted a multi-centre, randomized, controlled trial to assess the β-cell preservation effects of saxagliptin alone and saxagliptin combined with vitamin D as adjunctive therapies in adult-onset autoimmune type 1 diabetes.
Original paper ↗Read more →Combined Use of Vitamin D and DPP-4 Inhibitors as a Potential Adjuvant Treatment Strategy to Enhance the Efficacy of Novel Beta-Cell Replacement Therapies for Type 1 Diabetes
Emerging evidence suggests that vitamin D and dipeptidyl peptidase-4 (DPP-4) inhibitors exert synergistic immunomodulatory, anti-inflammatory and antioxidant actions.
Original paper ↗Read more →Effect of Calcitriol on Bone Turnover and Osteocalcin in Recent-Onset Type 1 Diabetes
We investigated the effect of supplementation with calcitriol on bone turnover in recent-onset type 1 diabetes (T1D).
Original paper ↗Read more →Immunomodulatory Effect of Vitamin D and Its Potential Role in the Prevention and Treatment of Type 1 Diabetes Mellitus: A Narrative Review
The aim of this narrative review is to present the molecular mechanisms for the vitamin D immunomodulatory effect as well as review human clinical studies on the use of vitamin D as adjuvant therapy in type 1 diabetes mellitus.
Original paper ↗Read more →Leptin and IGF-1 in Infancy Are Associated With Variants in DHCR7 and CYP2R1 That Relate With Type 1 Diabetes and 25OHD
Vitamin D has been variably implicated in risk of developing type 1 diabetes based on cohorts of at-risk individuals.
Original paper ↗Read more →Prevalence and characteristics of overweight/obesity in children newly diagnosed with type 1 diabetes in Kuwait
Overweight/obese children were older at diagnosis compared to non-obese children (median age 9.3 vs.
Original paper ↗Read more →Combined Therapy of Vitamin D3-Tolerogenic Dendritic Cells and Interferon-β in a Preclinical Model of Multiple Sclerosis
For this purpose, we analyzed the efficacy of a combined therapy based on the use of vitamin D3 (VitD3)-tolDC plus interferon beta (IFN-beta) in MS.
Original paper ↗Read more →COVID-19 and Multiple Sclerosis: A Complex Relationship Possibly Aggravated by Low Vitamin D Levels
COVID-19 can evolve into a severe disease associated with immediate and delayed sequelae in different organs, including the central nervous system (CNS).
Original paper ↗Read more →Cryopreserved vitamin D3-tolerogenic dendritic cells pulsed with autoantigens as a potential therapy for multiple sclerosis patients
Tolerogenic dendritic cells (tolDC) have been postulated as a potent immunoregulatory therapy for autoimmune diseases such as multiple sclerosis (MS).
Original paper ↗Read more →Disability progression in multiple sclerosis: a latent class analysis of predictors
We aimed to identify patterns of disability progression and their determinants to improve individualized risk assessment and support clinical decision-making.
Original paper ↗Read more →Linking Mechanisms of Vitamin D Signaling with Multiple Sclerosis
Environmental triggers often work via signal transduction cascades that modulate the epigenome and transcriptome of cell types involved in the disease process.
Original paper ↗Read more →Vitamin D as a Potential Therapy for Multiple Sclerosis: Where Are We?
Disease-modifying medications, which mainly aim to suppress such aberrant immune response, have significantly improved MS treatment.
Original paper ↗Read more →Why Vitamin D Autoimmune Findings Differ: Study Summary
A short guide to differing childhood and genetic findings, separating inherited vitamin D tendencies from treatment effects and personal advice.
Read more →Original sources
Plasma 25-Hydroxyvitamin D Concentration and Risk of Islet Autoimmunity
[1] Norris and colleagues.
Publication details
[1] Norris and colleagues. Plasma 25-Hydroxyvitamin D Concentration and Risk of Islet Autoimmunity. Diabetes. 2018 Jan;67(1):146 to 154. DOI: 10.2337/db17-0802. PMID: 29061729.
No association of vitamin D intake or 25-hydroxyvitamin D levels in childhood with risk of islet autoimmunity and type 1 diabetes: the Diabetes Autoimmunity Study in the Young (DAISY)
[2] Simpson and colleagues.
Publication details
[2] Simpson and colleagues. No association of vitamin D intake or 25-hydroxyvitamin D levels in childhood with risk of islet autoimmunity and type 1 diabetes: the Diabetes Autoimmunity Study in the Young (DAISY). Diabetologia. 2011 Nov;54(11):2779 to 2788. DOI: 10.1007/s00125-011-2278-2. PMID: 21858504.
Calcium insufficiency accelerates type 1 diabetes in vitamin D receptor-deficient nonobese diabetic (NOD) mice
[4] Driver and colleagues.
Publication details
[4] Driver and colleagues. Calcium insufficiency accelerates type 1 diabetes in vitamin D receptor-deficient nonobese diabetic (NOD) mice. Endocrinology. 2011 Dec;152(12):4620 to 4629. DOI: 10.1210/en.2011-1074. PMID: 21952242.
1,25-Dihydroxyvitamin D3 restores sensitivity to cyclophosphamide-induced apoptosis in non-obese diabetic (NOD) mice and protects against diabetes
Published in Clinical and experimental immunology. 1998 May;112(2):181 to 187. DOI: 10.1046/j.1365-2249 (1998).
Publication details
[5] Casteels and colleagues. 1,25-Dihydroxyvitamin D3 restores sensitivity to cyclophosphamide-induced apoptosis in non-obese diabetic (NOD) mice and protects against diabetes. Clinical and experimental immunology. 1998 May;112(2):181 to 187. DOI: 10.1046/j.1365-2249.1998.00568.x. PMID: 9649179.
Genetic evidence for a pathogenic role for the vitamin D3 metabolizing enzyme CYP24A1 in multiple sclerosis
Published in Multiple sclerosis and related disorders. 2014 Mar;3(2):211 to 219. DOI: 10.1016/j.msard (2013).
Publication details
[7] Ramasamy and colleagues. Genetic evidence for a pathogenic role for the vitamin D3 metabolizing enzyme CYP24A1 in multiple sclerosis. Multiple sclerosis and related disorders. 2014 Mar;3(2):211 to 219. DOI: 10.1016/j.msard.2013.08.009. PMID: 25568836.
Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study
[10] Mokry and colleagues.
Publication details
[10] Mokry and colleagues. Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study. PLoS medicine. 2015 Aug;12(8):e1001866. DOI: 10.1371/journal.pmed.1001866. PMID: 26305103.
Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study
[11] Manousaki and colleagues.
Publication details
[11] Manousaki and colleagues. Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study. PLoS medicine. 2021 Feb;18(2):e1003536. DOI: 10.1371/journal.pmed.1003536. PMID: 33630834.
Correction: Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study
Published in PLOS Medicine (2016).
Publication details
Mokry and colleagues. Correction: Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study. PLOS Medicine. 2016. DOI: 10.1371/journal.pmed.1001981. PMID: 26963823.
Correction: Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study
Published in PLOS Medicine (2021).
Publication details
Manousaki and colleagues. Correction: Vitamin D levels and risk of type 1 diabetes: A Mendelian randomization study. PLOS Medicine. 2021. DOI: 10.1371/journal.pmed.1003624. PMID: 33914743.