Does Vitamin D Affect Risk of Developing Autoimmune Disease?: A Systematic Review

Abstract

Objectives

We evaluated the epidemiologic evidence that vitamin D may be related to human autoimmune disease risk.

Methods

PubMed limited to English from inception through April 2010 was searched using keywords: “vitamin D”, “autoimmune” and autoimmune disease names. We summarized in vitro, animal, and genetic association studies of vitamin D in autoimmune disease pathogenesis. We sorted studies by design and disease and performed a systematic review of: a) cross-sectional data concerning vitamin D level and autoimmune disease; b) interventional data on vitamin D supplementation in autoimmune diseases and c) prospective data linking vitamin D level or intake to autoimmune disease risk.

Results

Vitamin D has effects on innate and acquired immune systems and vitamin D receptor polymorphisms have been associated with various autoimmune diseases. In experimental animal models, vitamin D supplementation can prevent or forestall autoimmune disease. We identified 76 studies in which vitamin D levels were studied in autoimmune disease patients, particularly with active disease, and compared to controls. Nineteen observational or interventional studies assessed the effect of vitamin D supplementation as therapy for various autoimmune diseases (excluding psoriasis and vitiligo) with a range of study approaches and results. The few prospective human studies performed conflict as to whether vitamin D level or intake is associated with autoimmune disease risk. No interventional trials have investigated whether vitamin D affects human autoimmune disease risk.

Conclusions

Cross-sectional data point to a potential role of vitamin D in autoimmune disease prevention, but prospective interventional evidence in humans is still lacking.

Introduction

The cause of the breakdown in immune tolerance that allows for the development of immunity to self-targets in autoimmune diseases such as rheumatoid arthritis (RA), inflammatory bowel disease (IBD) and multiple sclerosis (MS) is unknown. It is hypothesized that environmental exposures, including factors that stimulate endogenous inflammation, trigger the development of autoimmunity in genetically susceptible individuals(1). Autoimmune diseases cluster within families and within individuals, with many individuals developing more than one autoimmune disease(2-4). Polymorphisms in several genes have been associated with increased susceptibility to multiple autoimmune diseases(2, 5-8). Autoimmune diseases also share epidemiologic risk factors such as cigarette smoking and crystalline silica exposure(9, 10). Many autoimmune diseases are characterized by activation of the adaptive immune system with associated innate immune cell activation leading to inappropriately elevated levels of widespread systemic inflammation, in particular tumor necrosis factor-α (TNFα) and interleukins-1 and -6 (IL-1 and IL-6), potent cytokines produced by macrophages and monocytes among other cell types.

The Autoimmune Diseases Coordinating Committee of the National Institutes of Health estimated that 23.5 million Americans were affected by one or more autoimmune diseases in 2005(11). This number appears to be growing and is almost certainly an underestimate(12). For unknown reasons, most autoimmune diseases are more common among women than men, although this is less true after menopause(13). Together, autoimmune diseases are the third leading cause of morbidity in the industrialized world and are a leading cause of mortality among women(14, 15). With the exceptions of celiac sprue and pernicious anemia, there are no current means for the prevention or cure of most autoimmune diseases. Treatment of autoimmunity often consists of corticosteroids, immunosuppressant agents and biologic agents that target anti-tumor necrosis factor and other inflammatory cytokines. The 2005 NIH report estimated that the annual direct and indirect costs of autoimmune diseases in the U.S. far exceed $100 billion annually.

Vitamin D, obtained from diet, supplements, or conversion of 7-dehydrocholesterol in the skin by ultraviolet-B radiation, is hydroxylated in the liver to 25-hydroxyvitamin D [25(OH)D], the major circulating vitamin D metabolite(16, 17), which is further synthesized by 1α-hydroxylase to the active form, 1,25-dihydroxyvitamin D [1,25(OH)₂D]. Current dietary recommendations are geared only to prevent quite low vitamin D levels. Circulating 1,25(OH)₂D has a very short half life and is tightly regulated by parathyroid hormone (PTH). Fibroblast growth factor 23 (FGF23), which is produced in osteoblasts, is also important in regulating 1,25 (OH)₂ production in the kidney(18). “Normal” levels for 25(OH) D at most laboratories are >20ng/ml. However, at 25(OH)D levels below 30ng/ml, PTH secretion is increased, suggesting that the current “normal” levels for vitamin D are inadequate.(19-21) Vitamin D deficiency using the current standard for serum25(OH)D level is relatively common. In one study, over 50% of inpatients at Massachusetts General Hospital had low 25(OH)D levels(22).

Vitamin D is an inexpensive and relatively safe nutritional supplement, widely held to have anti-inflammatory and immunomodulating effects. The purported health benefits of this inexpensive and available dietary supplement have received enormous attention in both the medical literature and the popular press(23-26). It has been widely hypothesized that vitamin D deficiency acts as an environmental trigger for the induction of autoimmunity, and that high-dose vitamin D supplementation could be preventive(27-34), yet the scientific evidence appears to conflict. The Agency for Healthcare Research and Quality recently performed a systematic review entitled, “Vitamin D and Calcium: A Systematic Review of Health Outcomes”, published in August 2009(35). The report investigated the published literature regarding serum 25 (OH) D or 1,25 (OH)₂ D concentrations and multiple disease outcomes, including autoimmune diseases. The review did not include studies investigating sunlight exposure as a source of vitamin D intake, nor did it include studies in which dietary intake of vitamin D was assessed without measurement of serum levels, as nutrient composition tables for vitamin D were thought to be inadequate. The report also excluded cross-sectional and retrospective case-control studies where the measure of exposure occurred after or concurrent with the outcome. After these exclusions, no studies were identified that addressed the relationship between vitamin D and incident autoimmune disease. We sought to perform an extensive systematic review of the published literature to evaluate the strength of all types of evidence linking vitamin D intake to the risk of incident autoimmune diseases.

Methods

We searched the PubMed database from inception through April 2010 restricted to English language and human studies with the following search terms “vitamin D AND”: “autoimmune disease”; “autoimmunity”, “rheumatoid arthritis”; “spondylitis”: “spondyloarthropathy”; “psoriatic arthritis”; “systemic lupus erythematosis”; “scleroderma”; “systemic sclerosis”; “myositis”; “dermatomyositis”; “polymyositis”; “vasculitis”; “polymyalgia rheumatica”; “type 1 diabetes”; “multiple sclerosis; “autoimmune thyroiditis“; “Graves”; “Hashimoto's”; “inflammatory bowel disease”; “Crohn's disease”; “ulcerative colitis”; “vitiligo”; “autoimmune hepatitis”; “Behcet's”; “uveitis”; “Addison's”. We excluded review articles, case reports and studies primarily related to bone metabolism or osteoporosis (unless they included relevant data on vitamin D levels). Studies on the topical and parenteral treatment of psoriasis and vitiligo with vitamin D analogs were beyond the scope of this review and treatment with vitamin D-related compounds has been extensively reviewed in a systematic manner (e.g. (36). In addition, primary biliary cirrhosis and celiac disease were not included in the PubMed search since loss of fat-soluble vitamins is common in these diseases and vitamin D deficiency is highly prevalent in cirrhotics(37).We also conducted hand searches of reference lists to add to our list of manuscripts.

We aimed to address the following questions: 1) What are the cross-sectional data linking vitamin D to autoimmune disease?; 2) What are the prospective data linking vitamin D to the risk of future autoimmune disease? 3) What are the interventional data concerning dietary and/or supplemental vitamin D and risk of autoimmune disease? We carefully reviewed the abstracts of the studies identified and selected those relevant to our questions to perform a systematic review of the strengths and weaknesses of this literature. We included and qualitatively summarized background basic science studies, both in vitro and in animal models, genetic studies, studies of vitamin D's effects upon biomarkers of systemic inflammation, as well as ecologic/geographic epidemiologic studies of the incidence or prevalence of autoimmune diseases according to latitude (potentially related to vitamin D from UV light).

To explore the potential for publication bias among studies of vitamin D intake and risk of developing autoimmune disease, a funnel plot was created for these studies by plotting the risk estimates (odds ratios or relative risks) of developing an autoimmune disease in each study (x axis) against the number of autoimmune cases in each study (y axis). (A pattern resembling a symmetrical inverted funnel is generally interpreted as showing no significant publication bias, whereas the absence of studies in the lower sections of an inverted funnel, where small studies would lie, implies the presence of publication bias.)(38)

Results

We identified 1446 potentially relevant studies through our literature search and sorted and reviewed them as in Figure 1. Two hundred and forty-two were immediately excluded as irrelevant to this review, leaving 1204 that were reviewed in further detail. Basic science and in vitro studies, genetic studies, ecological studies, and studies of biomarkers of inflammation, were reviewed and summarized in the text. The human epidemiologic studies were sorted into epidemiologic studies of vitamin D levels in existing autoimmune disease (Table 1), studies and trials of vitamin D supplementation in existing autoimmune disease (Table 2), epidemiologic studies of vitamin D level and risk of autoimmune disease (Table 3) and epidemiologic studies of vitamin D intake and risk of autoimmune disease (Table 4). We reviewed the evidence in each of the studies these tables in detail and summarized it in the text.

Figure 1.

Flow diagram of literature search strategy and review.

Table 1.

Epidemiologic Studies of Vitamin D Levels in Existing Autoimmune Diseases

Autoimmune Disease	Geographic	Study Design	Subjects	Controls	Hormones studied	Association	Results	Year	Reference
Rheumatoid Arthritis	Turkey	Case-control	65 RA	40 Healthy controls	25(OH) D	No	P = 0.94 No difference in RA subjects vs. healthy controls, but significant decrease in subgroup with highest disease activity	2010	Turnahoglu et al, (116)
Rheumatoid Arthritis	North America	Case-only	266 early RA	-	25(OH) D	No	No multivariate associations of 25(OH) D with any disease measures with the exception of borderline association with rheumatoid factor positivity at enrollment (p = 0.05) No significant associations with disease activity after multivariate analysis	2010	Craig et al, (117)
Rheumatoid Arthritis, Psoriatic Arthritis, Ankylosing Spondylitis	Israel	Case-control	85 RA, 22 psoriatic arthritis, 14 AS	-	25(OH) D and PTH	No	Association between vitamin D level and ethnic origin but not disease activity (among other factors)	2009	Braun-Moscovici et al, (210)
Rheumatoid Arthritis	North America	Case-control (Nested in high risk cohort)	76 RA autoantibody positive, asymptomatic “at-risk” subjects	154 RA autoantibody negative “at risk” controls	25(OH) D	No	No association of autoantibody status and vitamin D level in individuals at high risk RA.	2009	Feser et al, (122)
Rheumatoid Arthritis	Italy, Estonia	Case-control	64 female Estonian RA, 53 female Italian RA	30 Estonian, 35 Italian age- and sex-matched	25(OH) D	Yes	Inverse correlation between levels and DAS28 scores among Italian patients in summer (r = -0.57, P< 0.0001) and Estonian patients in winter (r = -0.40, p< 0.05)	2007	Cutolo et al, (211)
Inflammatory Polyarthritis	Great Britain	Case-only	183 consecutive inflammatory polyarthritis < 6 months	-	25(OH)D, 1,25(OH)2 D	Yes	Signficant inverse relationship between 25(OH)D level at baseline and tender joint count, DAS28*, CRP, and HAQ* and between baseline 1,25(OH)(2)D and HAQ*.	2007	Patel et al, (121)
Rheumatoid Arthritis	Germany	Case-only	96 RA	-	25(OH) D and PTH	Yes	P <0.001 (with glucocorticoids), P < 0.01 (without glucocorticoids) Inverse correlation between vitamin D level and disease activity	1998	Oelzner et al, (212)
Rheumatoid Arthritis	Denmark	Case-control	29 RA, 21 SLE, 12 OA	72 Healthy controls	1,25(OH)2 D and 25(OH) D	No	No difference in vitamin D levels compared to controls	1995	Muller et al, (118)
Rheumatoid Arthritis	Finland	Case-only	143 female RA patients	-	1,25(OH)2 D 25(OH)D	Yes	63% patients had levels below normal limit during summer	1993	Kroger et al, (213)
Rheumatoid Arthritis	Denmark	Case-control	102 RA	38 Healthy subjects	25(OH) D; 24,25(OH) D; 25,26-OH D; 1,25(OH)2 D	Yes	P <0.01-0.001 (for 25(OH) D) 25(OH) D levels lower than in controls and significant inverse relation between level and functional class	1987	Als et al, (115)
Rheumatoid Arthritis	Great Britain	Case-control	30 RA	30 OA	1,25(OH)2 D	No	No difference between RA and OA and 1,25(OH)2 D levels did not correlate either with articular index or with sedimentation rate	1982	Bird et al, (119)
Rheumatoid Arthritis	Great Britain	Case-control	30 RA	30 OA	25(OH) D	No	No significant correlations between 25(OH) D and duration of arthritis or articular index	1980	Bird et al, (120)
Ankylosing Spondylitis and Psoriatic Arthritis	Germany	Case-only	76 AS, 120 PsoA	-	25(OH) D; 1,25(OH)2 D	Yes	P < 0.0005 for negative correlation between CRP and 25-OH D when combining AS and PsoA; P < 0.0005 for 25-OH level in PsoA versus AS	2009	Teichmann et al, (141)
Ankylosing Spondylitis	Turkey	Case-control	100 AS	58 Healthy controls	25(OH) D and PTH	No	P < 0.05 for 25(OH) D lower in cases than controls	2010	Mermerci Baskan et al, (137)
Ankylosing Spondylitis	Germany	Case-control	58 AS	58 matched healthy controls	25(OH) D; 1,25(OH)2 D; PTH	Yes	P < 0.05 for negative correlation between 1,25(OH)2 D and disease activity and TNF-alpha	2005	Lange et al, (140)
Ankylosing Spondylitis	Germany	Case-control	70 AS	45 matched healthy controls	25(OH) D; 1,25(OH)2 D; PTH	Yes	P < 0.01 for negative correlation between 1,25(OH)2 D and disease activity	2001	Lange et al, (139)
Ankylosing Spondylitis	Austria	Case-only	73 AS	-	25(OH) D and PTH	-	18% with 25(OH) D < 8 ng/ml, 73% with 25(OH) D less than 20 ng/ml	2001	Falkenbach et al, (214)
Ankylosing Spondylitis	Germany	Case-only	14 AS at entry and 15 months later	-	25(OH) D; 1,25(OH)2 D and PTH	No	Vitamin D levels did not differ significantly between baseline and follow-up in AS patients	1997	Lee et al, (215)
Ankylosing Spondylitis	Germany	Case-control	38 AS	52 controls	1,25(OH)2 D and PTH	No	No significant difference	1993	Franck et al, (138)
Systemic Lupus Erythematosus	Poland	Case-control	45 SLE	49 controls	25(OH) D	Yes	Lower 25(OH)D in cases than controls p = 0.0005, and antibodies to 1,25 (OH) D detected in 4 pts (8.9%, NS).	2010	Bogaczewicz et al, (127)
Systemic Lupus Erythematosus	Europe and Israel	Case-only	378 SLE	-	25(OH) D	Yes	R= -0.12, P = 0.018 for vitamin D levels and disease activity scores Negative correlation between 25(OH) D levels and disease activity in SLE patients	2010	Amital et al, (129)
Systemic Lupus Erythematosus	South Korea	Case-control	104 SLE	49 controls	25(OH) D	Yes	P = 0.03 for vitamin D insufficiency in SLE compared to controls, but did not correlate with SLE disease activity.	2010	Kim et al, (123)
Systemic Lupus Erythematosus	North America	Case-only	198 SLE	-	25(OH) D	Yes	R = -0.234, P = 0.002 for inverse correlation of vitamin D with disease activity	2010	Ben-Zvi et al, (128)
Systemic Lupus Erythematosus	Canada	Case - only	124 female SLE	-	25(OH) D and 1,25(OH)2 D	No	No significant association between low vitamin D levels and disease activity; 25(OH) D levels associated with season, glucocorticoid exposure, and serum creatinine	2010	Toloza et al, (132)
Systemic Lupus Erythematosus	Saudi Arabia	Case-control	165 SLE	214 volunteers	25(OH) D	Yes	P < 0.0001 for vitamin D deficiency in SLE versus controls	2009	Damanhouri, (124)
Systemic Lupus Erythematosus	North America	Case-only	181 SLE	-	25(OH) D	Yes	P = 0.018 for negative correlation between low 25(OH) D and SLE disease activity index (adjusted for age, season and white race)	2009	Wu et al, (130)
Systemic Lupus Erythematosus	North America	Case-only	38 pediatric SLE	207 healthy controls	25(OH) D; 1,25(OH)2 D; iPTH	Yes	P = 0.01 for low 25(OH) D low level and disease activity index scores	2009	Wright et al, (125)
Systemic Lupus Erythematosus	Brazil	Case-control	36 SLE	26 controls	25(OH) D; 1,25(OH)2 D; PTH	Yes	R = -0.65; P < 0.001 for 25(OH) D level and negative correlation with disease activity index	2009	Borba et al, (126)
Systemic Lupus Erythematosus	Spain	Case-only	92 SLE	-	25(OH) D	No	P = 0.08 for fatigue and vitamin D deficiency; no significant association with disease activity	2008	Ruiz-Irastorza et al, (133)
Systemic Lupus Erythematosus	U.S.	Case-only	37 SLE	-	25(OH)D	Yes	65% < 80 nmol/L and 20% <47.7 nmol/L. Above normal level correlated with low disease activity, but also with significantly higher dsDNA antibodies.	2008	Thudi et al, (131)
Systemic Lupus Erythematosus	Canada	Case-only	25 SLE	-	25(OH) D; 1,25(OH)2 D; PTH	No	1,25-OH lower in SLE patients using hydroxychloroquine compared to nonusers	2001	Huisman et al, (135)
Systemic Lupus Erythematosus	Denmark	Case-control	21 SLE, 29 RA, 12 OA	72 Healthy controls	1,25(OH)2 D and 25(OH) D	Yes	P = 0.0008 for decreased 25(OH) levels in SLE versus controls; no correlation with anti-DNA antibodies, sedimentation rate, or blood counts.	1995	Muller et al, (118)
Systemic Lupus Erythematosus and Dermatomyositis and Juvenile RA	North America	Case-only	17 pediatric SLE, 13 juvenile dermatomyositis, 83 JRA	-	25(OH) D; 1,25(OH)2 D; PTH	No	No significant differences between active and inactive stages of pediatric SLE with regards to vitamin D levels	1990	Reed et al, (134)
Undifferentiated Connective Tissue Disease	Hungary	Case-only	161 UCTD	59 controls	25(OH)D	Yes	25(OH)D significantly lower than in controls in summer and winter. Significant associations of low 25(OH)D with active manifestations and with evolution to diagnosed connective tissue disease within 2.3 years.	2008	Zold et al, (136)
Scleroderma	Italy	Case-only	108 scleroderma	-	25(OH) D	Yes	Vitamin D deficiency is associated with more severe disease P = 0.026 for longer disease duration, P = 0.014 for lower diffusing lung capacity, P = 0.037 for higher pulmonary artery pressure	2010	Carameschi et al, (147)
Scleroderma	Brazil	Case-control	10 juvenile scleroderma	10 matched controls	25(OH) D, iPTH	Yes	P = 0.04 for lower vitamin D levels compared with controls	2010	Shinjo et al, (143)
Scleroderma	Italy	Case-control	60 Scleroderma	60 controls	25(OH) D	Yes	P < 0.001 for lower vitamin D levels compared with controls, but no associations with disease features or skin score	2009	Calzolari et al, (142)
Scleroderma	France and Italy	Case-only	90 scleroderma	-	25(OH) D and iPTH	Yes	R = -0.17 (P = 0.04) for negative correlation between low vitamin D and disease activity score	2009	Vacca et al, (148)
Scleroderma	Israel	Case-only	60 scleroderma	-	25(OH) D, PTH	-	46% of scleroderma patients are vitamin D deficient	2008	Braun-Moscovici et al, (179)
Scleroderma	North America	Case-control	8 scleroderma	8 matched healthy controls	25(OH) D; 1,25(OH)2 D	No	Similar levels in cases and controls.	1991	Matsuoka et al, (144)
Scleroderma	Holland	Case-control	20 scleroderma	-	25(OH) D; 1,25(OH)2 D; 24-25(OH) D	No	Normal 25(OH) D and 24,25(OH) D levels in scleroderma; lower 1,25(OH)2 D in a subgroup with calcinosis	1985	Serup et al, (145)
Scleroderma	Holland	Case-control	25 scleroderma	92 controls	1,25(OH)2 D	No	P < 0.001 for higher 1,25(OH)2 D in scleroderma compared to controls	1984	Serup et al, (146)
Type I Diabetes Mellitus	India	Case-control	50 children within 1 week of diagnosis of T1D*	50 healthy children	25(OH) D	Yes	P < 0.009 for lower vitamin D in new onset diabetics	2009	Borkar et al, 8 (149)
Type I Diabetes Mellitus	North America	Case-control	46 new-onset T1D*, 110 established T1D,	153 control subjects; 106 first-degree relatives	25(OH) D	No	P = 0.87 No significant associations of reduced vitamin D level and T1D	2009	Bierschenk et al, (158)
Type I Diabetes Mellitus	Great Britain	Case-control	40 T1D,* 40 T2D*	41 non-diabetic controls	1,25(OH)2 D, PTH, Erythropoietin	Yes	P = 0.001 for median vitamin D level in cases vs. controls. Tubulointerstitial damage associated with low 1,25(OH)2 D	2009	Singh et al, (150)
Type I Diabetes Mellitus	North America	Case-only	128 Pediatric T1D*	-	25(OH) D	-	61% Vitamin D insufficient; 15% Vitamin D deficient	2009	Svoren et al, (216)
Type I Diabetes Mellitus	Qatar	Case-control	170 pediatric T1D*	170 healthy controls	25(OH) D, PTH	Yes	P = 0.009 for mean vitamin D level; 28.8% versus 17.1% severe vitamin D deficiency	2008, 2009	Bener et al, (217, 151)
Type I Diabetes Mellitus	Australia	Case-only	64 pediatric new-onset T1D*	-	25(OH) D	-	P = 0.001 (for associate acidosis) Low vitamin D in 42% with acidosis versus 5.6% without acidosis Acidosis may alter vitamin D metabolism or low vitamin D may contribute to presenting with ketoacidosis	2009	Huynh et al, (160)
Type I Diabetes Mellitus	Australia	Case-control	47 pediatric T1D*	94 Healthy controls	25(OH) D, 1,25(OH)2 D	Yes	P = 0.002 for 25(OH) D deficiency	2007	Greer et al, (152)
Type I Diabetes Mellitus	Sweden	Case-control	459 T1D* at diagnosi, 138 8 years later	208 - matched controls	25(OH) D	Yes	P < 0.0001 for vitamin D level at diagnosis; P = 0.04 for vitamin D level 8 years later	2006	Littorin et al, (155)
Type I Diabetes Mellitus	North America	Case-control	50 T1D*	63 T2D*	25(OH) D	No	P = 0.01 for lower 25-OH-D levels in T2D versus T1D* (adjusted for body mass index and age)	2006	Di Cesar et al, (156)
Type I Diabetes Mellitus	Italy	Case-control	46 T1D*	24 healthy controls	25(OH) D; 1,25(OH)2 D; PTH	Yes	P < 0.01 for vitamin D levels Lower vitamin D level in incipient nephropathy/microalbuminuria	1999	Verrotti et al, (157)
Type I Diabetes Mellitus	Germany	Case-control	49 new onset T1D*	42 healthy controls	25(OH) D; 1,25(OH)2 D	Yes	P < 0.01 for 1,25(OH)2 D at onset of T1D	1991	Baumgartl et al, (153)
Type I Diabetes Mellitus	Mexico	Case-only	22 T1D*	-	25(OH) D	Yes	P < 0.001 for low 25(OH) D in poorly controlled T1D*	1990	Arreola et al, PMID 2103709 (159)
Type I Diabetes Mellitus	Norway	Case-control	46 pubertal T1D*	191 Healthy controls	25(OH) D; 1,25(OH)2 D; 24,25(OH) D	Yes	P < 0.05 for 1,25(OH)2 D Relative decrease in 1,25(OH)2 D and increased 24,25(OH) D levels in T1D at puberty	1985	Rodland et al, (154)
Type I Diabetes Mellitus	Denmark	Case-only	74 T1D*	-	25(OH) D; 1,25(OH)2 D; 24,25(OH) D	-	P < 0.02 for 1,25-OH level during ketoacidosis; P < 0.01 for 25(OH) D in T1D groups with diabetic nephropathy	1983	Storm et al, (161)
Multiple Sclerosis	Holland	Case-control	36 MS	20 Healthy controls	25(OH) D	-	R= -0.359, P = 0.048 25(OH) D negative correlation with IgG index in MS	2010	Vogt et al, (162)
Multiple Sclerosis	Norway	Case-control	36 MS	38 other neurologic diseases	Serum and cerebrospinal fluid 25(OH) D	Yes	P = 0.0012 and 0.041 for cerebrospinal fluid-to-vitamin D serum ratio (lower in MS compared with other neurological diseases)	2010	Holmoy et al, (164)
Multiple Sclerosis	North America	Case-control	173 MS +9 transverse myelitis,	16 other neurologic diseases	25(OH) D	-	84% of all patients had insufficient levels Large numbers of patients with MS and transverse myelitis are deficient in vitamin D		Hiremath et al, (165)
Multiple Sclerosis	Argentina	Case-control	132 MS-various forms	60 Healthy controls	1,25(OH)2 D, 25(OH) D	Yes	P < 0.00001 for 25(OH) D and 1,25(OH)2 D levels (lower in relapsing-remitting MS during exacerbation compared with remission	2009	Correale et al, (167)
Multiple Sclerosis	Holland	Case-control	103 MS	110 Healthy controls	1,25(OH)2 D, 25(OH) D	Yes	Among women: for every 10 nmol/L increase in serum 25(OH)D level the odds of MS decreased 19% (OR = 0.81; 95% CI 0.69-0.95) for dose-dependent decreased odds of MS among women); r= -0.29; P = 0.02 for negative correlation between disability status and 25(OH) D levels in women only	2009	Kragt et al, (166)
Multiple Sclerosis	Finland	Case-control	23 MS	23 Healthy controls	25(OH) D and iPTH every 3 months for 1 year	Yes	P = 0.012 for inverse relationship between serum vitamin D level and MS clinical activity	2008	Soilu-Hanninen et al, (218)
Multiple Sclerosis	Great Britain	Case-control twin study	40 monozygotic and 59 dizygotic twins with MS	40 monozygotic and 59 dizygotic twins without MS	25(OH) D	No	No association with having MS (P = 0.4)	2008	Orton et al, (170)
Multiple Sclerosis	Holland	Case-control	267 MS	-	25(OH) D, 1,25(OH)2 D	Yes	P = 0.043 for high 25(OH) D and chance of remaining relapse-free	2008	Smolders et al,(172)
Multiple Sclerosis	Ireland	Case-control	29 MS	22 age- and sex-matched controls	25(OH) D; 1,25(OH)2 D; PTH	No	No differences between cases and controls	2007	Barnes et al, (169)
Multiple Sclerosis	Australia	Case-control	136 MS	272 controls	25(OH) D	Yes	OR = 3.07 (95% CI 1.37-6.90) for disability and vitamin D insufficiency	2007	van der Mei et al, (163)
Multiple Sclerosis	Finland	Case-control	40 MS	40 Controls	25(OH) D	Yes	P = 0.03 for lower vitamin D level during relapse versus remission	2005	Soilu-Hanninen et al, (171)
Multiple Sclerosis	Germany	Case-control	53 MS	415 Controls	25(OH) D	Yes	R2 = 0.8491 for vitamin D level; R2 = 0.7931 for brain lesions (two-fitted, third-order polynomial curves corresponding closely when 25(OH) D data lagged 2 months) Inverse correlation of gadolinium-enhancing lesions on MRI and vitamin D levels following seasonal fluctuations	2000	Embry et al, (168)
Hashimoto's Thyroiditis	India	Case-only	642 healthy individuals	-	25(OH) D	-	r - 0.08, P = 0.04 for vitamin D level inverse correlation with anti-thyroid antibodies	2009	Goswami et al, (177)
Crohn's disease	India	Case-control	34 Crohn's	34 irritable bowel syndrome controls	25(OH) D	Yes	Correlation coefficient -0.484, significance P<0.004 Lower vitamin D levels in Crohn's disease and association with severe disease activity	2009	Joseph et al, (173)
Crohn's disease and Ulcerative colitis	America	Case-only	130 young Crohn's, and UC*	-	25(OH) D, iPTH	No	P =0.97 in multiple regression for vitamin D level and disease activity	2006	Pappa et al, (176)
Crohn's disease	Japan	Case-control	33 Crohn's	15 healthy controls	25(OH) D, iPTH	Yes	P = 0.04 for vitamin D and disease activity in logistic regression	2004	Tajika et al(174)
Crohn's disease	Great Britain	Case-only	40 Crohn's	-	25(OH) D, 24,25(OH) D, 1,25(OH)2 D, PTH	Yes	P < 0.05 for low 25(OH) D levels in active disease	1985	Harries et al (175)
Vitiligo	Great Britain	Case-only	45 Vitiligo	-	25(OH)D	Yes	55.6% were insufficient (<30 ng/mL), and 13.3% were very low (<15 ng/mL).	2007	Silverberg, et al.(178)

^*DAS28= Disease Activity Score 28 joint assessment; HAQ= Health Assessment Questionnaire; PsOA= psoriatic arthritis; T1D= type 1 diabetes; T2D= type 2 diabetes; UC= ulcerative colitis

Table 2.

Studies and Trials of Vitamin D Supplementation in Existing Autoimmune Disease

Autoimmune Disease	Population	Study Design	Subjects	Controls	Vitamin D analog	Follow-up Period/Dura tion of Study	Results	Year	Reference
Rheumatoid Arthritis	Yugoslavia	Open-label trial	19 RA on DMARDs	-	oral alphacalcidiol (2 ug/d)	3 months	Clinical improvement in the majority of 89%	1999	Andjelkovic et al, (182)
Rheumatoid Arthritis	Sweden	Double-blind trial	25 RA treated	24 RA placebo	oral calciferol (100,000 IU/d)	1 year	67% Objective and subjective improvement; lower sedimentation rates and higher hemoglobin levels	1973	Brohult et al, (181)
Psoriatic Arthritis	Hungary	Open-label trial	10 PsoA	9 PsoA placebo	oral alphacalcidol (0.25 ug twice/d)	6 months	P = 0.048 for disease activity at 6 months	2009	Gaal et al(183)
Psoriatic Arthritis	North America	Open-label study	10 PsoA	-	Oral 1,25(OH)D 2ug/d	6 months	P < 0.01 for tender joint count	1990	Huckins et al, (184)
Systemic Lupus Erythematosus	Spain	Open-label; observational	60 SLE	20 SLE	various doses cholecalciferol or calcidiol liquid	2 years	P = 0.87 and P = 0.63 for disease activity and damage indices; P = 0.001 for fatigue based on a visual analogue scale. No effect on SLE activity.	2010	Ruiz-Irastorza et al, (185)
Scleroderma	Holland	Randomized, double-blind, placebo controlled trial	7 scleroderma, 20 morphea	-	Oral calcitriol 0.75 ug/d for 6 months plus 1.25 ug/d for 3 months	9 months with 6 months followup	No significant difference between placebo and calcitriol with regards to skin score	2000	Hulshof et al, (190)
Scleroderma	Macedonia	Open-label trial	3 morphea	-	Oral calcitriol 0.5-0.75 ug/day	4-6 months with 6 months followup	Significant clinical improvement with regards to joint mobility, skin flexibility, skin induration	1999	Caca-Biljanovska et al, (188)
Scleroderma	Holland	Open-label trial	7 pediatric linear scleroderma	-	Oral calcitriol 0.25-0.75 microgram/day per dose-escalating protocol	3-10.5 months	5 out of 7 patients with good to excellent improvement of skin lesions	1999	Elst et al, (189)
Scleroderma	Holland	Open-label trial	3 generalized morphea	-	Oral calcitriol 0.5-0.75 ug/day	3-7 months	Improved mobility of joints and increased skin extensibility	1994	Hulshof et al, (187)
Scleroderma	France	Open-label trial	11 scleroderma p	-	Oral calcitriol at a mean dose of 1.75 ug/day	6 months to 3 years	Significant improvement of clinical findings	1993	Humbert et al, (186)
Type I Diabetes Mellitus	China	Randomized controlled trial	17 latent autoimmune diabetes (LADA)	18 LADA	1 alpha-OH D3 0.5 ug/day (+/- Insulin)	1 year	P < 0.01; 70% versus 22% maintained fasting C-peptide 1 alpha-OH D3 plus insulin therapy can preserve beta-cell function in patients with LADA	2009	Li et al, PMID 19488999 (191)
Type I Diabetes Mellitus	Italy	Randomized controlled trial	34 recent-onset T1D* receiving calcitriol plus insulin	33 recent-onset T1D receiving nicotinamide plus insulin	Calcitriol 0.25 ug on alternate days	1 year	P < 0.03 for reduction in insulin requirement at 3 and 6 months but not 9 and 12 months Calcitriol has a modest effect on residual beta-cell function and only temporarily reduces the amount of insulin required	2006	Pitocco et al, (192)
Multiple Sclerosis	Canada	Randomized controlled trial	25 MS	24 MS	up to 40,000 IU/day over 28 weeks, then 10,000 IU/day for 12 weeks	52 weeks (Phase I/II dose-escalation trial)	P = 0.09 for annualized relapse rate (0.26 versus 0.45) treatment group patients appeared to have fewer relapse events although not statistically significant	2010	Burton et al,(195)
Multiple Sclerosis	North America	Observational study	40 MS on either low dose cholecalciferol or high dose ergocalciferol	9 MS without vitamin D	cholecalciferol (< 800 IU/day) or ergocalciferol (50,000 IU/day) for 7-10 d then weekly or biweekly doses)	4 to 12 months	P = 0.01 for 25(OH) D increase in patients switching from low to high dose supplementation; effect on relapse rate not studied	2009	Hiremath et al (165)
Multiple Sclerosis	Canada	Open-label trial	12 MS patients in an active phase	-	Daily calcium + increasing doses of vitamin D3 (from 28,000 to 280,000 IU/week)	28 weeks	P = 0.03 for decreased number of gadolinium-enhancing lesions per patient Vitamin D did not affect disease progression or activity but the number of gadolinium-enhancing lesions per patient decreased	2007	Kimball et al(194)
Multiple Sclerosis	North America	Open-label trial	15 cases of relapsing-remitting MS	-	Oral calcitriol 2.5 ug/day	48 weeks	33% at baseline versus 29% at both 24 and 48 weeks (for gadolinium-enhancing lesions). On-study exacerbation rate (27%) was less than baseline.	2005	Wingerchuk et al, (219)
Multiple Sclerosis	Israel	Open-label trial	5 relapsing-remitting MS	-	Alphacalcidol 1.5 ug/day	6 months	3 patients stable, 1 improved, 1 relapsed	2003	Achiron et al, (220)
Multiple Sclerosis	U.S.	Observational	16 MS	-	Calcium, Magnesium and vitamin D as cod liver oil cod liver oil at the rate of 5,000 i.u./day	1-2 years	Fewer exacerbations than expected historically p<0.01	1986	Goldberg et al,(193)
Inflammatory Bowel Disease	Eastern Europe	Open-label trial	17 Crohn's taking 1,25(OH) 2 D	19 Crohn's taking 25(OH) D	Supplemental (1,25 OH versus 25(OH) D)	12 months	P < 0.05 (for disease activity) 1,25(OH)2 D has a more prominent beneficial effect on disease activity compared with 25(OH) D	2009	Miheller et al, (196)

^*DMARDs= disease-modifying anti-rheumatic drugsPsOA= psoriatic arthritis; T1D= type 1 diabetes; LADA= latent autoimmune diabetes

Table 3.

Epidemiologic Studies of Vitamin D Level and Risk of Developing an Autoimmune Disease

Autoimmune Disease	Population	Study Design	Subjects	Controls	Hormone studied	Confounders considered	Results	Year	Reference
Rheumatoid Arthritis	Holland	Case-control	79 subjects who donated blood and later developed RA. Samples from 3 time points: 1, 2, and 5 years before symptoms	79 controls matched for age, sex, and time of sample donation	25(OH) D	Age, sex, time sample donation	No association between vitamin D level and preclinical RA (1, 2, 5 and more years before symptoms)	2006	Nielen et al, (197)
Multiple Sclerosis	North America	Case-control	257 military recruits with banked blood who later developed MS	514 controls matched for age, sex, race/ethnicity, and dates of blood collection	25(OH) D	Age, sex, race/ethnicity, and dates of blood collection	OR 0.59, 95% CI 0.36-0.97, for the risk of MS among whites. Inverse relation is particularly strong before age 20. No association among Blacks and Hispanics	2006	Munger et al, (198)

Table 4.

Epidemiologic Studies of Vitamin D Intake and Risk of Developing Autoimmune Disease

Autoimmune Disease	Population	Study Design	Subjects	Controls	Vitamin D Intake	Follow-up Period/Duration of Study	Confounders Considered	Results	Year	Reference
Rheumatoid Arthritis	North America (Nurses’ Health Study)	Prospective cohort	722 who developed RA	186,389	Dietary and supplement intake	Up to 22 years	Age, all women, race, age at menarche, oral contraceptive use, parity, duration of breastfeeding, menopausal status, postmenopausal hormone use, cigarette smoking, latitude of residence at age 15 (North, Middle or South U.S.), physical activity, body mass index	No association; highest quintile intake RR 1.0 (95%CI 0.8, 1.3)	2008	Costenbader et al, (200)
Rheumatoid Arthritis	North America (Iowa Women's Health Study)	Prospective Cohort	152 who developed RA	29,386	Diet and supplement intake	11 years	Age, all women, all postmenopausal,	Higher intake at baseline associated with lower risk up to 11 years later: RR 0.67 (95%CI 0.44,1.00)	2004	Merlino et al, (199)
Systemic Lupus Erythematosus	North America (Nurses’ Health Study)	Prospective Cohort	190 who developed SLE	186,389	Diet and supplement intake	Up to 22 years	Age, all women, race, age at menarche, oral contraceptive use, parity, duration of breastfeeding, menopausal status, postmenopausal hormone use, cigarette smoking, latitude of residence at age 15 (North, Middle or South U.S.), physical activity, body mass index	No association; highest quintile intake RR 1.4 (95%CI 0.8, 2.3)	2008	Costenbader et al, (200)
Type I Diabetes Mellitus	Finland (Diabetes Prediction and Prevention Study)	Birth Cohort	165 children who developed TID or advanced beta-cell autoimmunity	3723 infants	Maternal vitamin D intake and at 1 and 3 months post partum.	Mean of 4.3 years	familial diabetes and genetic risk factors for T1D, sex, gestational age, maternal age, maternal education, delivery hospital, route of delivery, number of earlier deliveries and smoking during pregnancy	Maternal intake of vitamin D from food RR 1.25 (95%CI 0.80,1.95), and from supplements RR 1.05 (95%CI 0.95-1.16) for risk of advanced beta cell autoimmunity/type 1 diabetes in offspring	2010	Marjamaki et al, (208)
Type I Diabetes Mellitus	Italy	Case-control	83 T1D ages 0-14	166 age and sex-matched, same region	Vitamin D administration during lactation as baby (recall)	Recall as adults	medical history of severe infections nor history of surgical operations	OR 0.31 (95% CI 0.11, 0.8) for vitamin D administration during lactation and risk type 1 DM	2007	Tenconi et al, (202)
Type I Diabetes Mellitus	Sweden	Cohort-retrospective and prospective	8.7% at 1 year and 8.9% at 2.5 years + for glutamic acid decarboxylase or islet antigen-2 autoantibodies	18,886 infants	Questionnaire: Vitamin D supplementation during pregnancy, at 1 year and 2.5 years	2.5 years	familial type 1 diabetes, maternal education, maternal age, delivery type, weight increase from birth, breastfeeding duration, introduction of cow's-milk protein, fish intake	OR 0.71 (95% CI 0.52-0.96) for reduced islet-cell autoimmunity at 1 year, but OR 1.25 (95%CI 0.91, 1.73) at 2.5 years. Surrogate measure for type 1 diabetes used.	2007	Brekke et al, (221)
Type I Diabetes Mellitus	North America (Diabetes Autoimmunity Study in the Young)	High-risk cohort	16 children who developed islet-cell autoimmunity	206 children without islet-cell autoimmunity	Maternal dietary and supplemental vitamin D intake during pregnancy-recalled after birth	Average 4 years	HLA genotype, family history of type 1 diabetes, presence of GDM, and ethnicity	HR 0.37 (95% CI 0.17-0.78) for increased maternal vitamin D intake and risk for islet-cell autoimmunity. Surrogate measure for type 1 diabetes used.	2003	Fronczak et al, (207)
Type I Diabetes Mellitus	Norway	Case-control	545 cases (children mean age 10.9)	1668 controls population (children mean age 9.3)	Questionnaire: vitamin D supplement use during pregnancy and the first year of life (recalled)		maternal use of cod liver oil during pregnancy, child's use of cod liver oil during the first year of life, duration of exclusive breastfeeding, child's age at introduction of solid foods, maternal education, maternal smoking during pregnancy, maternal age at delivery, child's number of siblings, type 1 diabetes among child's siblings or parents, and the child's age and sex	OR 0.98 (95% CI 0.73, 1.31) for maternal use of vitamin D ≥ 5 times/week vs. none during pregnancy and OR 0.97 (0.73, 1.29) for vitamin D use ≥ 5 times/week during first year of life	2003	Stene et al, (203)
Type I Diabetes Mellitus	Finland	Birth cohort	81 developed T1D	10366 newborns	Vitamin D supplementation during first year of life	31 years	Sex, maternal parity, gestational and maternal age, maternal education, social status, birth weight, and growth rate in infancy and suspected rickets	RR 0.12 (95% CI 0.03-0.51) for regular versus no supplementation	2001	Hypponen et al, (206)
Type I Diabetes Mellitus	Norway	Case-control	78 cases	980 controls	Questionnaire: Maternal itamin D intake during pregnancy (recall)	-	Age, sex, breastfeeding and maternal education	OR 1.27 (95 % CI 0.70, 2.31) for vitamin D supplementation in 1st year life	2000	Stene et al, (204)
Type I Diabetes Mellitus	Europe (7 centers)	Case-control	820 cases	2335 controls-population-based, age-group matched	Questionnaire or interview: recalled infancy vitamin D intake	-	Age, birth weight, duration of breast feeding, maternal age and study center	OR 0.67 (95% CI 0.53, 0.86) for vitamin D supplementation in infancy	1999	EURODIAB Substudy 2 study group, (222)
Multiple Sclerosis	North America (Nurses’ Health Study)	Prospective Cohort study	173 MS cases	92,253 women	Dietary and supplemental vitamin D	Up to 20 years	Age, all women, smoking and latitude at birth	RR 0.67 (95% CI 0.40,1.12); p trend 0.006; for highest quintile	2004	Munger et al,(201)

Basic Science in vitro and in vivo studies

Through binding to the vitamin D receptor (VDR), the lipid-soluble active 1,25(OH)₂D regulates an array of genes, many involved in inflammation and acquired and innate immune responses(27, 39-42). VDRs are found at high levels on dendritic cells, T and B lymphocytes and macrophages. The function of these cells is profoundly affected by binding of activated 1,25 (OH)₂D(43-47). The 1α-hydroxylase that converts 25(OH)D to its active form is expressed in the kidney, in activated macrophages, dendritic cells and other tissues(29, 48, 49). This form of the enzyme, unlike that in the kidney, is not regulated by parathyroid hormone, but rather inducible by factors including interferon gamma and is downregulated with dendritic cell maturation(50). The expression of VDRs on resting CD4+ T cells increases 5 fold with T cell activation(51). 1,25(OH)₂D inhibits the expression of interleukin(IL)-2, an important growth factor for T lymphocytes, and suppresses the secretion of Th1 cytokines IL-12, interferon gamma, and tumor necrosis factor, while increasing IL-4, IL-5 and IL-10, leading to the development of a Th2-skewed T cell population.(52-58) The addition of 1,25(OH)₂D to CD4+ T cells also inhibits the expression of IL-6, a co-factor stimulating Th17 cells, important in the development of autoimmunity(59, 60). When added to SLE-patient derived B cells in vitro, 1,25(OH)₂D inihibits autoantibody production(61).

In vitro, 1,25(OH)₂D inhibits the differentiation of monocytes into dendritic cells and blocks the stimulatory effects that T cells have on them(58, 62, 63). Instead, 1,25(OH)₂D promotes monocyte differentiation into macrophages, prevents them from releasing inflammatory cytokines and chemokines(64), and reduces their capacity to present antigens to lymphocytes by decreasing MHC-II molecule cell surface expression(39, 65). 1,25(OH)₂D-VDR transcriptional signaling also exerts anti-inflammatory effects through the down-regulation of the prostaglandin (PG) pathway and cyclooxygenase-2 (COX-2)(66) and tolerizing effects through capacity to convert CD4 T cells into IL-10-secreting T regulatory cells, suppressing the proliferation of responder T cells.(67). 1,25(OH)2D interacts with VDRs on osteoblasts, stimulating expression of the receptor activator of nuclear factor kB (NFkB) ligand (RANKL)(27). Immature dendritic cells that have differentiated from monocytes in the presence of 1,25(OH)₂D respond poorly to inflammatory chemokines that regulate dendritic cell maturation and migration to lymph nodes(45). Experimental data show that 25(OH)D can inhibit pro-inflammatory cytokines such as IL-6 and TNF-α, decrease serum levels of C- reactive peptide (CRP), and upregulate production of the anti-inflammatory cytokine interleukin-10 (IL-10)(44). 1,25(OH)₂D downregulates dendritic cell production of IL-12 and augments IL-10, important in the development of T regulatory (Treg) cells(27, 58). As dendritic cells are central to the maintenance of both protective immunity and self-tolerance(68, 69), vitamin D's influence on their maturation and function could have consequences on autoimmune disease risk and/or progression. However, as the pathogenesis of autoimmune disease itself is still unclear, many of pathways implicated suggest a potential role of vitamin D insufficiency in disease progression, but it is not clear to what role in the triggering of autoimmune disease vitamin D intake or deficiency play.

In vivo, supplementation with 1,25(OH)₂D forestalls the development of inflammatory arthritis, autoimmune encephalomyelitis (a model for MS), type I diabetes and autoimmune thyroiditis in experimental animal models(70-74). Treatment with a low calcemic vitamin D analog had a prophylactic as well as therapeutic effect on a murine model of Th1-like colitis(75). Administration of 1,25(OH)₂D or its analogs to non-obese diabetic mice modulates the expression of chemokines and cytokines and prevents diabetes(76). Vitamin D receptor knock-out mice develop severe diarrhea, rectal bleeding, and marked body weight loss, leading to death in 2 weeks. Thus, vitamin D deficiency, it is thought to compromise the mucosal barrier and increase susceptibility to mucosal damage and potentially the risk of inflammatory bowel disease(77).

Evidence in Humans: Associations of Vitamin D with Circulating Biomarkers of Systemic Inflammation

Cross-sectional studies in healthy and ill populations suggest potential favorable effects of vitamin D—as measured by circulating 25(OH)D, sun exposure, or dietary or supplement intake—on inflammatory biomarkers(78-80). Inverse associations between serum vitamin D levels and serum CRP concentrations have been found in patients with diabetes mellitus, atherosclerotic vascular disease, inflammatory polyarthritis,and prolonged chronic illness(35, 81, 82). Increased TGF-β serum levels were observed when vitamin D was given for a 6 month period to 16 subjects with MS(83). TNF-α, IFN-γ, and IL-13 levels, however, did not change after supplementation. In a trial involving 200 overweight subjects enrolled in a weight loss program, participants were randomized to vitamin D (83 ug/d) or placebo in a double-blind manner for 12 months and serum TNF- α levels did decrease more in the vitamin D group than in the placebo group marker (10.2% compared with 3.2% decrease; p 0.049)(84). In a recent interventional study, 324 adults were assigned to 20,000 IU of vitamin D per week, 40,000 IU vitamin D per week or placebo. Multiple cytokines, including interleukin (IL)-2, -4, -5, -10, -12, -13, -17, intercellular adhesion molecule-1, interferon-gamma, monocyte chemotactic protein-1, and high CRP, were measured at the start and end of one year. No significance differences in changes levels of any of these cytokines was detected, nor was there any indication of a polarization of the T cells towards a Th2 dominant type(85).

Ecologic Associations implicating Vitamin D in Autoimmune Disease

Evidence comes from ecologic observations that several autoimmune diseases, including inflammatory bowel disease, MS, and type I diabetes and RA are more prevalent at Northern latitudes where sun exposure is reduced(86-92). The strongest ecologic evidence linking vitamin D with autoimmune disease risk is for MS (93). An increased prevalence of MS at Northern, compared to Southern, latitudes has long been observed and a strong inverse correlation of MS incidence with UV light exposure is also seen(91, 94). MS has also been associated with birth during the winter compared to other seasons of the year and it is hypothesized that this could reflect low maternal vitamin D during pregnancy(95). Lastly, seasonal variation in MS relapses detected by MRI has been observed, with increased flares occurring during the winter compared to summer months(96).

Genetic Polymorphisms in Vitamin D Pathway Genes Associated with Autoimmune Diseases

Polymorphisms in the VDR gene have been associated with increased risk of multiple autoimmune diseases, including Hashimoto's thyroiditis, IBD, Graves’ disease, RA, SLE, primary biliary cirrhosis, autoimmune hepatitis, Addison's disease, vitiligo, celiac disease and type I diabetes, as well as MS in humans(97-112). A vitamin D response element (VDRE) is found in the promoter region of the HLA DRB1*1501 allele, an allele strongly associated with MS susceptibility pathogenesis in Caucasians(112). However, not all of the polymorphisms associated with autoimmune diseases in past studies have known functional consequences and the strengths of the associations vary. Moreover, not all of these associations have been replicated.

Cross-sectional Studies of Vitamin D in Existing Autoimmune Diseases (Table 1- supplemental material)

Circulating 25(OH)D, reflecting all sources of vitamin D exposure with a half-life of 2-3 weeks, has been used in epidemiologic studies as a comprehensive and stable indicator of vitamin D status(17, 113). Some studies have assessed 1,25(OH)2D levels, but as the half-life of 1,25(OH)D is only 4 hours and this metabolite dependent on fluctuating calcium need, these results are harder to interpret (113). Studies of vitamin D levels comparing populations with and without existent autoimmune diseases have been conducted around the world and with somewhat conflicting results. In RA, for example, two past studies revealed lower levels of 25(OH)D in RA patients than in healthy matched controls(114, 115), but five studies did not find such a difference (116-119, 120 ). Patel and colleagues found a strong inverse association between baseline levels of serum 25(OH)D in patients with newly diagnosed early inflammatory polyarthritis (45%of whom were classified as having RA at 1 year) and baseline disease activity, as assessed by tender joint counts, RA disease activity scores (DAS28) and health assessment questionnaires (HAQ) scores(121). For each 10-ng/ml increase in 25(OH)D, they found a decrease in the DAS28 of 0.3 and in the C-reactive peptide level of approximately 25%. At 1 year, only significant inverse association between higher baseline vitamin D levels lower HAQ scores. In a large group of individuals at increased risk for RA, however, plasma 25(OH) D concentrations were not associated with the presence of RA-related autoantibodies(122).

In SLE, at least six case-control studies have now demonstrated a lower level of 25(OH) D in SLE cases than in matched controls(118, 123-127). Several studies have found that lower levels of vitamin D correlate with more active SLE (125, 126, 128-131), but several others have not confirmed this(118, 123, 132-134). One study showed that 25(OH)D level was lower in hydroxychloroquine users than non-users and hydroxychloroquine is known to inhibit the synthesis of vitamin D(135). A case-control study from Hungary reported lower 25(OH)D levels in individuals with undifferentiated connective tissue disease (before starting medications) than controls in summer and winter seasons, and demonstrated that levels were lower in those with more active manifestations and among the 35 who developed into a diagnosed connective tissue disease within 2.3 years of follow-up(136). A study involving 113 children with SLE, juvenile RA or juvenile dermatomyositis did not find an abnormal prevalence of 25(OH)D or 1,25(OH)D deficiency, although the study's primary goal was to assess osteocalcin levels(134).

There have been four case-control studies comparing 25(OH) D and 1,25 (OH) D levels in patients with ankylosing spondylitis to healthy controls. A study of 100 AS patients and 58 controls in Turkey found that 25(OH) D levels were lower in patients than controls(137), but an earlier study of 38 AS patients and 52 controls in Germany had not detected a significant difference(138). Two studies by Lange have found that 1,25(OH)D levels are lower in ankylosing spondylitis patients and negatively correlated with disease activity, thought possibly related to the high prevalence of osteoporosis seen in this disease(139, 140). Seventy-six ankylosing spondylitis patients had significantly lower 25(OH) levels than did 120 psoriatic arthritis patients, and a significant negative correlation between C-reactive peptide levels and 25(OH) D levels was reported in the combined cohort of these patients in a study by Teichmann(141).

We found eight past studies addressing vitamin D levels in patients with scleroderma. Of the five case-control studies, two demonstrated lower 25(OH) D levels in patients with scleroderma than controls(142, 143), and three did not(144-146). In two case-only analyses, it was found that 25(OH)D levels were, not surprisingly, lower in those with more severe underlying scleroderma, more longstanding disease, more disease activity, more pulmonary hypertension and lower diffusing lung capacity(147, 148).

In type I DM, there have been at least nine studies documenting lower 25(OH) D levels in established or new onset type I diabetes patients, compared to matched controls(149-157), and only one published study that did not confirm this(158). It has also been shown that 25(OH)D levels are lower in individuals with poorly controlled type I diabetes(159), ketoacidosis(160, 161), incipient nephropathy or tubulointerstitial damage(150, 157).

In MS as well, there have now been several studies reporting that 25(OH) D levels are lower in cases with MS than in healthy controls (162-168), and two that have not(169, 170). Lower vitamin D levels have been found in MS patients especially during the summer, compared to winter months(171), in those with progressive as opposed to relapsing-remitting forms of the disease(172), and associated with increased disability in MS(163, 166), and clinical activity and risk of relapse(162, 167, 168, 172).

Two case-control studies have reported lower 25(OH)D levels in patients with Crohn's disease compared to controls(173 , 174), and a three of four studies have found that levels are lower in those with active inflammatory bowel disease(173-176). The only study of the association between vitamin D level and thyroid autoimmunity was conducted in India where a large group of over 600 healthy individuals were screened for anti-thyroid antibodies. A weak but significant inverse correlation was found between vitamin D level and titer of anti-thyroid peroxidase autoantibodies(177). One cross-sectional case-only study of 25(OH) D levels in patients with vitiligo reported apparently low levels, without controlling for season(178). We found no epidemiologic studies of vitamin D levels in existing pernicious anemia, vasculitis, adult inflammatory myositis, polymylagia rheumatica, Addison's disease, uveitis, or Behcet's disease.

In sum, many, but not all case-control studies, circulating levels of 25(OH)D and occasionally 1,25 (OH)₂D have been found to be lower in subjects with various autoimmune diseases than in matched healthy controls. The evidence pointing to lower vitamin D levels in existing disease is not equally strong for all autoimmune diseases and some diseases have yet to be studied. The differences between the many similar studies of a specific disease may relate to their different sizes and the heterogeneity in their statistical power, as well as to adjustment for confounding factors. Taken together, it is also apparent from these studies that ethnic origin (179), season of the year (180), and disability (163, 166), have strong effects on circulating vitamin D levels and it is not clear that all of these cross-sectional studies controlled for these important confounders. Additionally, disease activity and disability, glucocorticoid use(132), serum creatinine(132), microalbuminuria(157), ketoacidosis(160, 161), and use of hydroxychloroquine(135) influence circulating vitamin D levels. Hydroxychloroquine prescribed in several autoimmune diseases may inhibit the conversion of 25(OH) D to 1,25(OH)₂D(135). Thus, while interesting, the major limitation of these cross-sectional studies is that reverse causation is very likely. Low vitamin D level may be the consequence, not the cause, of active autoimmune disease.

Vitamin D Supplementation and Effects on Existing Disease Activity (Table 2)

Several small open-label studies and interventional trials have tested the effects of supplementation with vitamin D, or its analogs, on the activity of established autoimmune diseases. In a double-blind trial conducted in 1973, oral calciferol (100,000 IU/d) for one year was administered to 25 subjects who were compared to 24 subjects who received a placebo. The intervention group had improved hand strength, decreased morning stiffness and need for analgesics/anti-inflammatory medications and decreased erythrocyte sedimentation rate (ESR). (181) In a more recent study, high dose 1-alpha (OH) D3 (alphacalcidiol) reduced pain and CRP levels in 19 RA subjects in a 3 month open-label trial(182).

Two small open-label studies of 6 months each have shown some promise for oral alphacalcidol (0.25 ug twice/d and 1,25(OH)D in reducing disease activity and tender joint counts in psoriatic arthritis(183 , 184). In an open-label observational study of 60 SLE patients in Spain who took vitamin D3 supplementation for two years, significant improvement was seen in subject fatigue, as measure by a visual analog scale, but not SLE disease activity measures(185). Ongoing NIH-sponsored trial is investigating vitamin D supplementation on disease activity and interferon-related cytokine activation (“the interferon signature”) in SLE.

Five small trials have now studied the effects of oral calcidiol supplementation on scleroderma, morphea or linear scleroderma(186-190). While earlier open-label studies had shown some benefit, in particular for morphea, the most recent randomized trial found that calcitriol was not more effective than placebo in patients with morphea, and the scleroderma group in that trial (n=7) was too small to any draw conclusions(190). On the other hand, randomized trials in Italy and China have now shown that vitamin D supplementation in individuals with latent autoimmune diabetes (autoantibodies highly associated with the development of type 1 diabetes) can forestall the development of type 1 diabetes, although it only temporarily reduces the amount of insulin required(191, 192).

We found five small studies or trials of vitamin D supplementation for MS. Fewer exacerbations of disease were seen among 16 MS patients given daily cod liver oil vitamin D for a year in an open-label study in 1986(193). In another open-label study of calcium and escalating doses of vitamin D supplementation in 12 MS patients, there did not appear to be an effect on disease progression or activity but the number of gadolinium-enhancing lesions per patient appeared decreased after just 12 weeks(194). In the largest randomized controlled trial yet, 25 MS subjects were randomized to doses of up to 40,000 IU/day over 28 weeks, followed by 10,000 IU/day for 12 weeks over 52 weeks and 24 MS subjects were randomized to placebo. While the treatment group appeared to have fewer relapse events, the difference was not statistically significant (p=0.09) possibly due to insufficient statistical power(195). Thus far, the studies have been small and inconclusive and larger scale trials are called for.

Lastly, one open-label trial in Eastern Europe has compared the effects of supplementation with 25(OH)D to those of 1,25(OH)D for Crohn's disease and found that 1,25 D supplementation was associated with a significantly greater decrease in disease activity (assessed by Crohn's Disease Activity Index) (196).

Studies to date have mainly been small and most open-label, and have include a range of vitamin D preparations, doses and durations. However, it appears that vitamin D may have an adjunctive role in amelioration of RA, MS, type I diabetes, and perhaps SLE and Crohn's disease. This evidence addresses the question of whether vitamin D has an effect upon established disease, but not upon the risk of developing an autoimmune disease de novo.

Epidemiologic Studies of Vitamin D Level and Risk of Developing Autoimmune Disease (Table 3)

In the hierarchy of epidemiologic studies, stronger data for association with susceptibility is afforded by prospective studies. Additionally, prospective studies are able to address the potential role of vitamin D in the pathogenesis of autoimmune disease. The relation between vitamin D serum or plasma level—measured by circulating 25(OH)D——and the incidence of autoimmune diseases has been examined in only two prospective observational epidemiologic studies of RA and MS, but not in other autoimmune diseases. Serum 25(OH)D level was not related to future RA risk in a Dutch case-control study examining blood bank samples from 79 individuals who developed RA compared to 79 healthy controls, matched on age, sex, and time of blood donation(197). To address the question in MS, Munger and colleagues used banked blood samples from Department of Defense military recruits, 257 of whom had later developed MS. Higher serum levels of 25(OH)D were associated with significantly lower risk of incident MS, but only among whites, not Blacks nor Hispanics, and the effect was most pronounced for individuals under age 20(198).

Epidemiologic Studies of Vitamin D Intake and Risk of Developing Autoimmune Disease (Table 4)

Prospective epidemiologic data on vitamin D intake and the risk of developing autoimmune diseases have been systematically reviewed and summarized in Table 4. The diseases that have been studied include RA, SLE, type 1 DM and MS. For RA risk, the data have been somewhat conflicting. In the Iowa Women's Health Study, higher baseline vitamin D intake was associated with decreased risk of subsequent RA(199). In the Nurses’ Health Study, a prospective cohort study involving over 180,000 women with multiple dietary assessments over 20 years, intake of vitamin D was not related to risk of developing either SLE or RA(200). However, in the same cohort, an inverse relationship was discovered between higher intake of vitamin D from supplements, but not overall dietary intake and lower risk of incident MS up to 20 years later(201).

Several observational studies have addressed the association between maternal vitamin D intake in pregnancy, lactation, or intake in infancy and early childhood and the risk of developing type 1 diabetes. A meta-analysis of four case-control (202-204) was performed in 2008(205)omitted from table 4 due to redundancy). Pooled data from the case-control studies showed that the risk of type 1 diabetes was significantly reduced in infants who were supplemented with vitamin D compared to those who were not supplemented (pooled odds ratio 0.71, 95% CI 0.60 to 0.84). One large birth cohort study prospectively assessed vitamin D supplementation during the first year of life (as part of a ricket's prevention study), and confirmed cases of type 1 diabetes in the children up to 31 years later. They found a strongly protective effect of regular vitamin D supplementation in the first year of life: RR 0.12 (95% CI 0.03-0.510 (206).

Surrogate markers for the development of type 1 diabetes, glutamic acid decarboxylase or islet antigen-2 autoantibodies, which are highly associated with type 1 diabetes risk, have been used in a few prospective studies. Two birth cohorts of infants at high genetic risk of type 1 diabetes have investigated relationships between maternal intake of vitamin D during pregnancy and the development of this islet cell autoimmunity. A high maternal dietary intake of vitamin D in foods, but not supplements, assessed after pregnancy by questionnaire, was with associated with increased risk of islet cell autoantibodies among the children followed for an average of four years in an American birth cohort(207). A larger prospective birth cohort in Finland has reported no association between maternal vitamin D intake (also assessed post-partum by questionnaire) and risk of type 1 diabetes or islet cell autoantibodies in their high genetic risk offspring, followed up to 4.3 years, however(208).

Our funnel plot for the studies examining the relationship of vitamin D dietary intake and the risk of the four autoimmune diseases that have been studied, is shown in Figure 2. Given the dearth of studies, we have included studies of all four autoimmune diseases with substantial heterogeneity in design, exposure assessment and timing in relation to onset of disease/autoantibodies and outcomes. The plot appears slightly asymmetric, displaying potential publication bias with few small non-protective studies. However, given the small number of studies, no conclusions can be drawn.

Figure 2. Funnel plot of the risk estimates (odds ratios or relative risks) of developing autoimmune disease associated with vitamin D intake (none vs. supplementation or highest vs. lowest level of intake).

Sample size of the study refers to the number of autoimmune disease cases in each study. The appearance of an inverted funnel, where small studies of positive and negative studies have been published argues against a potential publication bias.

Potential Causes of Conflicting Results

While the development of an autoimmune response to self-antigens underlies all of these diseases, it is likely that vitamin D is not uniformly involved in the pathogenesis of each and every autoimmune disease and is more important in the etiology of some versus others among them. It is clear that not all autoimmune diseases have been equally studied and for most there is a dearth of evidence addressing the potential effects of vitamin D on autoimmune disease susceptibility. Past case-control studies may have been biased by subject recall, cohort studies may have been underpowered for small effects, and potential confounders such as socioeconomic status may not have been consistently well addressed. Prospective observational studies that rely on dietary intake are hampered by subjects’ ability to accurately recall food intake and by inadequate means of assessing vitamin D content of all foods. The inconsistency in epidemiologic data regarding vitamin D intake and risk of autoimmune disease has led to some debate and may result in part from the fact that oral vitamin D intakes in many of the studied populations were too low to produce significant effects, as well as limited within-population variability in intakes. The difference between high and low oral intakes in observational studies is generally only 300-400 IU/d. Given that a vitamin D₃ intake of 1 μg [40 IU] increases circulating 25(OH)D by ~1 nmol/L,(209) such an increment would be expected to raise 25(OH)D by ≤10 nmol/L.

Misclassification of vitamin D status due to incomplete or single exposure assessments could also contribute to null or inconsistent findings. The critical window for vitamin D intake or level to affect risk of developing an autoimmune disease is not known and may vary according to disease: maternal intake during pregnancy, early life intake and intake during different phases of adulthood may not be equally relevant to different diseases. Possible health benefits of vitamin D may be offset by other components in dairy products, which are main dietary sources of vitamin D. Disentangling independent effects of vitamin D and calcium on risk of autoimmune disease in observational studies is difficult due to their high correlation in countries—including the U.S.—where milk is fortified with vitamin D. The use of vitamin D supplements in a trial setting would eliminate this problem.

Conclusions

Understanding of the pluripotent immunomodulating and anti-inflammatory effects of vitamin D is advancing. Despite the in vitro and animal evidence for vitamin D's potential to decrease systemic inflammation and prevent autoimmune disease in humans, there remains insufficient human data to firmly support the hypothesis that vitamin D intake is related to the risk of developing autoimmune disease. Data from laboratory studies and cross-sectional and observational epidemiologic investigations suggest a potential protective effect for vitamin D in autoimmune disease susceptibility.

The human epidemiologic and interventional evidence, however, is weak. Despite the multitude of cross-sectional and case-control studies that demonstrate lower levels of vitamin D among individuals with autoimmune diseases than healthy controls, these studies are not able to address causality. Reverse causality, that the disease process of an autoimmune disease, even in early disease, could lower circulating vitamin D concentration is a likely possibility in these studies. Prospective epidemiologic data on vitamin D status and whether intake level has any effect on the incidence of autoimmune diseases are limited or conflicting. There have been no large randomized trials of high-dose vitamin D supplements for the primary prevention of autoimmune diseases in a general population.

Confirming or refuting that elevated dietary vitamin D intake reduces systemic inflammation and/or the risk of incident autoimmune disease is critical, as no prophylactic therapy currently exists for these diseases. The growing enthusiasm for vitamin D supplementation, however, underscores the need for timely and rigorous testing of this hypothesis before use becomes so prevalent as to render it impossible. To understand the effects of moderate-to-high dose vitamin D supplementation, a high-quality, double-blind randomized controlled trial with a large sample size and long duration and designed specifically to include an assessment of these outcomes in a general population is necessary.