Vitamin D is a fat-soluble vitamin synthesized from a cholesterol derivative 1838. It exists in two forms: vitamin D₂ or ergocalciferol, which is produced by irradiation of ergosterol (provitamin D provided by the diet) by the action of ultraviolet (UV) radiation in the skin, and vitamin D₃ or cholecalciferol provided directly by foods or produced by the action of UV from cholesterol after transformation into 7-dehydrocholesterol 3839. In the liver, cholecalciferol is transformed into calcifediol or 25(OH)D, which enters the blood circulation, then, in the renal tubular cells, calcifediol is hydroxylated into calcitriol or 1,25-dihydroxyvitamin D (1,25(OH)D) which is the active form of vitamin D 3840.

Vitamin D is a steroid hormone 41 because of its mechanism of action which is exerted either directly on membrane receptors affecting extracellular and intracellular concentrations of Ca⁺⁺ via calcium channels and which define the nongenomic action, or by binding to nuclear receptors, which determines the genomic action 404243. In this second case, the vitamin D/receptor complex formed induces the synthesis of messenger ribonucleic acid (mRNA) which codes for a protein, Calcium Binding Protein (CaBP), responsible for the biological effect 4344. This type of action takes longer to be effective than the nongenomic action 42.

For a long time, the main role of vitamin D was considered to be the regulation of calcium and phosphate metabolism 16, in which bone was the main target organ and its action was considered to be limited to cell turnover by increasing the life span of osteoblasts by an anti-apoptosis effect 44. However, recent data suggest that muscles and the nervous system are also target organs of vitamin D.

First of all, several lines of clinical evidence suggest the existence of a link between vitamin D and muscle function. Cases of myopathy have been described in severe vitamin D insufficiency, responsible for rickets in children and osteomalacia in adults 4546474849. These severe forms of vitamin D insufficiency cause severe muscle weakness, usually proximal and involving the lower limbs 4849. Apart from these extreme cases, vitamin-insufficient myopathies are generally underdiagnosed due to the progressive and continuous onset of nonspecific clinical signs such as muscle pain, paraesthesiae or arthralgia, which are initially suggestive of a diagnosis of inflammatory rheumatic disease 49. Clinical signs may also not necessarily be related to muscle lesions as, in a series of 30 cases, Skaria et al. 47 showed that, although clinical signs were present in more than 95% of cases, only 30% of muscular biopsies revealed histological signs of vitamin D insufficiency-related myopathy. In case of severe vitamin D deficiency with osteomalacia, these signs are associated with widening of interfibrillary spaces, fatty infiltration, fibrosis and the presence of glycogen granules, with no signs of inflammatory reaction 4748. It has also been shown predominantly type II muscle fibres atrophy 50, while vitamin D repletion instead leads to an increase in relative fibres composition and in fibres area of type IIa muscle fibres 5152. It remains yet unclear if the increase in type II muscle fibre number is caused by new formation of type II fibres or a transition of already existing fibres from type I to type II 53.

Second, experimentation revealed that the genomic pathway of vitamin D action in muscle involves activation of 1,25(OH)D nuclear receptors that triggers the production of messenger RNA and the synthesis of proteins responsible for multiple phenomena such as calcium influx into the cell, membrane phosphate transport, phospholipids metabolism, and muscle fibre proliferation and differentiation 38394044. This genomic pathway of action of vitamin D also influences the polymorphism of VDR responsible for the nongenomic pathway of action 43. This nongenomic pathway has a complementary action to that of the genomic pathway either by activating a second messenger in the cell - cyclic AMP and/or diacylglycerol and/or inositol triphosphate and/or arachidonic acid - or by activating protein kinase C and the release of calcium into the cytosol 5455. This effect is responsible for the active transportation of calcium into sarcoplasmatic reticulum by Ca-ATPase increasing the calcium pool which is necessary for the successive attachments and detachments of myofilaments leading to sarcomeric shortening responsible for muscular contraction 56. Vitamin D therefore participates in the good functional equilibrium of fast-twitch type II muscle fibres, thereby preserving high muscle contraction speed and muscle power 38394041424356.

In epidemiological studies, the relationship between vitamin D and muscle function remains more controversial, as it has been inconsistently described 45. For instance, Bischoff-Ferrari et al. 57 observed, in a population of 319 community-dwelling subjects with a mean age of 75.9 years, that a 25(OH)D rate ≤ 12 ng/mL was significantly correlated with decreased leg extension strength, with a less intense effect in women compared to men. However, after adjustment for gender, age, body mass index and serum parathormone, this correlation was no longer significant 57. Annweiler et al. obtained similar results amongst community-dwelling older women aged 75 and older from the EPIDOS cohort 5859. They found a significant association of low serum vitamin D with low quadriceps strength 58 and handgrip strength 59 in the unadjusted model, but these associations were not significant anymore after adjustment for age, body mass index, number of chronic diseases, practice of regular physical activity, serum calcium concentration, creatinine clearance, and hyperparathyroidism 5859. In contrast, Mowe et al. 60, in a population of hospitalized subjects (n = 246) and subjects living at home (n = 103) between the ages of 70 and 91 years, showed that, regardless of the group considered, the serum 25(OH)D concentration was correlated with the grip strength of the non-dominant hand, difficulty climbing stairs, and regular physical activity. Finally, Kuczynski and Ostrowka 61 reported indirect evidence that low bone mineral density in osteoporotic elderly women presenting vitamin D insufficiency was associated with increased postural sway in the mediolateral plane.

Like observation studies, intervention studies have demonstrated discordant results concerning the effects of vitamin D supplementation on muscle function 2145. In a literature review published in 2003 and based on 33 clinical trials and a total population of 2,496 elderly subjects, only 3 trials showed a significant improvement of muscle strength and/or physical performance 21. In these 3 trials, the vitamin D supplement was associated with calcium. However, when trials presenting methodological bias were excluded, only one trial demonstrated a significant improvement. More recently, Annweiler et al. 45 conducted a systematic review which confirmed that the relationship between vitamin D and muscle function was controversial in clinical trials as some studies found a significant vitamin D-related improvement in physical performance, while others failed to show any effect of supplementation. These divergences highlighted the fact that the effects of vitamin D supplementation were directly correlated with the initial severity of vitamin D insufficiency 49. Vitamin D supplementation has also been reported to act significantly and specifically on so-called antigravity muscles 61. This action of vitamin D on muscle has been shown to play a role in maintenance of postural equilibrium. Dhesi et al. 62 reported that an intramuscular injection of 600,000 IU ergocalciferol in 70 subjects with a mean age of 76.6 ± 6.1 years, a history of falls and a 25(OH)D concentration ≤ 12 ng/mL versus an intramuscular placebo injection in a group of 69 matched subjects, significantly reduced postural sway. In this study, a 3% increase of the amplitude of sway was observed in the placebo group, while the amplitude of sway decreased by 13% in the intervention group. However, this study did not demonstrate any effect on muscle strength. Binder et al. 63 demonstrated that vitamin D and calcium supplementation significantly improved postural equilibrium tests.

The failure to demonstrate any positive effect of vitamin D on muscle performance could also be related to the duration of follow-up after starting treatment, which did not exceed 6 months in the majority of cases 2145, whereas the effect of vitamin D may be observed later. For instance, in the case of biopsy-documented myopathy, vitamin D supplementation restores muscle after a period of 6 to 12 months 46474849. Furthermore, the latest publications of experimental research on vitamin D receptors (VDR) suggest the existence of responders and non-responders to vitamin D. For example, Wang et al. 64 showed that a given VDR genotype corresponds to a given intensity of muscle strength, as these authors observed, in a population of 109 young women, that the AA homozygous genotype of ApaI VDR was associated with lower muscle strength than the aa or Aa homozygous genotypes. Similarly, the bb homozygous genotype of BsmI VDR was associated with lower muscle strength than the BB or Bb homozygous genotype. On the other hand, no difference was demonstrated between the various TaqI VDR genotypes 64.

Furthermore, Stein et al. 65 have suggested that the muscle effect of vitamin D insufficiency could be due to parathormone and not to a direct action of vitamin D on muscle. Vitamin D insufficiency triggers a series of reactions, including elevation of serum parathormone concentrations 3839404142. Serum parathormone appears to be an indirect tissue marker of vitamin D insufficiency that is more specific than the serum vitamin D concentration itself 65, as serum 25(OH)D has been demonstrated to be poorly correlated with the muscular tissue response 40. Furthermore parathormone has a muscle action that is independent of vitamin D 22. More specifically, studies in rodents have demonstrated that parathormone induces muscle catabolism 66, reductions in calcium transport (i.e., Ca-ATPase activity) and impairment of energy availability (with reduction in intracellular phosphate and mitochondrial oxygen consumption) and metabolism (including reduction in creatinine phosphokinase and oxidation of long-chain fatty acids) in skeletal muscles 67. This relationship between serum parathormone and muscles has been known for a long time in patients with primary hyperparathyroidism, whose clinical features comprise fatigue and muscle weakness 4042. These symptoms improve after parathyroidectomy 68. Furthermore, parathormone has been shown to predict falls 65 and muscle strength independent of 25(OH)D 69. The specific roles of vitamin D and parathormone on muscle are thus not fully elucidated 68.

Given the divergence in published results, it appears that vitamin D could affect neuromuscular function and fall risk in a way which does not involve only the muscle but also the CNS.

As in muscle, vitamin D acts according to genomic and nongenomic pathways 39404142. VDR have been demonstrated in some parts of the brain, especially in the hippocampus, hypothalamus, and limbic system but also in cortical, subcortical and spinal motor zones 707172737475767778. At the cellular level, these receptors are present on neurons and glial cells 4041424344454647484950515253545556575859606162636465666768697071727374.

Experimentally, in animals, vitamin D is involved in neurophysiology and regulates the metabolism of neurotransmitters including dopamine, acetylcholine, serotonin and gamma aminobutyric acid 7078, and the synthesis of certain growth factors such as Nerve Growth Factor (NGF) or Glial cell line-derived neurotrophic factor (GDNF) 7071727374757677. Vitamin D is also involved in the development and maturation of rodents brain 707175. In addition to this central action, vitamin D also acts on the peripheral nervous system, as a reduction of nerve conduction velocity has been reported in the case of severe vitamin D insufficiency 47.

Vitamin D is also involved in neuroprotection through immunomodulating, anti-ischemic and anti-oxidative properties. Indeed, trophic induction plays a neuroprotective role in cerebral ischemia 79, as well as an anti-neurodegenerative role for dopaminergic cells in models of Parkinson's disease 80. Moreover, it seems that vitamin D plays a part in the cerebral processes of detoxification by interacting with reactive oxygen and nitrogen species in rat brain and by regulating the activity of γ-glutamyl transpeptidase 81, a key enzyme in the antioxydative metabolism of glutathione. Concentrations around 0.1 to 100 nanomoles of 1,25(OH)D thus ensure an efficient protection of neurons against the direct effects of superoxyde ions and hydrogene peroxyde 80. Finally, VDR-dependent immunosuppressive effects, including increased concentrations of inflammatory cytokines, macrophages, polynuclears, as well as their sensitization to apoptotic signals, were described in the CNS 70. For illustration, in a model of mice with experimental allergic encephalitis, 1,25(OH)D inhibited autoimmune neurological processes 8283.

Vitamin D could also be vasculoprotective since vitamin D insufficiency has been associated with incident cerebrovascular disease 84. For instance, atherosclerosis is a systemic inflammatory disease related to vitamin D insufficiency 85. C-Reactive Protein is a marker of inflammation and atherosclerosis regulated by Interleukin-6 (IL-6) and Tumor Necrosis Factor-α (TNF-α) 86, which secretions dose-dependently decreased in presence of vitamin D 87. Furthermore, vitamin D insufficiency could be a contributing factor to hypertension - a major determinant of the development of cerebrovascular diseases - by the suppression of the renin-angiotensin system expression in the juxtaglomerular apparatus 88 and by an action on the arterial wall compliance 8889.

All together, these properties could stabilize the neurophysiologic function and explain why the lack of functional VDR in the brain of VDR-knockout transgenic mice models was responsible for behavioral disorders due not only to an increased level of stress but also to severe motor disorders 7378909192. For instance, the suppression of functional cerebral VDR in transgenic mice induced a decreased swimming capacity with fewer swimming movements, suggesting the essential role of vitamin D in motor control 90.

Some clinical data in humans appear to support the hypothesis of a favorable action of vitamin D on cognitive function, especially attention, as Yaffe et al. 93 observed, in a population of 8,333 women over the age of 65, that cognitive performance on frontal and attentional tests were lower in women with a low BMD or vertebral fractures, establishing a link between post-menopausal osteoporosis - related to vitamin D insufficiency - and cognitive decline. Although the hypothesis of a simple temporal relationship is possible in this study, the hypothesis of an action of vitamin D on cognitive function is highly likely 94. In particular, epidemiological studies revealed lower serum 25(OH)D concentrations in subjects with Alzheimer disease than in healthy subjects 9596. In addition, emerging analytical studies have brought new evidence 94. For instance, Wilkins et al. 97 found a significant positive association between the serum 25(OH)D levels and the scores at the Clinical Dementia Rating and at the Short Blessed Test in 80 older subjects aged 65 and over, living at home (40 subjects with AD and 40 non-demented subjects). Additionally, Przybelski et al. 98 and Oudshorn et al. 99 highlighted an association with the Mini Mental Status Examination (MMSE) score. Similarly, Llewellyn et al. demonstrated among 1,766 non-demented subjects or with Mild Cognitive Impairment aged 78 years on average that the lowest 25(OH)D concentrations, the highest risk of pathological Abbreviated Mental Test score 100. In line with this, Annweiler et al. showed a 2-fold risk of global cognitive impairment (Pfeiffer's Short Portable Mental State Questionnaire) among 752 older women (mean age 82 years) 101. Finally, Buell et al. 102 showed among 318 participants (mean age 73.5 years, 72.6% women) that 25(OH)D insufficiency was associated with more than twice the odds of all-cause dementia and of Alzheimer disease. In contrast, two studies found no significant association 103104. First, Jorde et al. have unsuccessfully explored the linear association of 25(OH)D with 6 specific cognitive functions (working memory, episodic memory, speed of information processing, language, executive functions and intelligence) in 148 older subjects with hyperparathyroidism (mean age 62 years, 46% women) 103. Second, McGrath et al. found no significant positive logistic association between the quintiles of serum 25(OH)D concentrations and several specific cognitive tasks among 4,747 adults between 20 and 59 years (Symbol-digit Substitution Coding Speed: attention and episodic memory; Serial Digit Learning Trials To Criterion: working memory) 104.

From a prospective perspective, Slinin et al. 105 highlighted a trend for an independent association between lower 25(OH)D levels and odds of cognitive decline by Modified Mini Mental State score among 1,604 men enrolled in the Osteoporotic Fractures in Men Study and followed for an average of 4.6 years. Additionally, Llewellyn et al. 106 showed that low 25(OH)D levels were associated with substantial decline in MMSE score among 858 adults aged ≥65 years studied over a 6-year period.

Literature review shows that the choice of confounders is essential and could explain the divergences in results. Analyses should thus take into account a list of covariates such as depression or serum parathormone concentrations.

First, depressive mood is associated with both cognition and vitamin D. Indeed, depression by itself can mimic dementia - when people are depressed, they can have difficulty concentrating, which leads to forgetfulness - or is often part of dementia, or may cause by itself executive dysfunction 107. Additionally, a relationship between vitamin D deficiency and anxio-depressive disorders is likely since low serum 25(OH)D concentrations are closely associated with active mood disorders 70 and have been proposed as the missing link between seasonal changes in photoperiod and seasonal mood swings 70. In line with this, one clinical trial supported the efficacy of vitamin D supplementation on mood disorders 108. Finally, accounting for depression is of primary importance while exploring the involvement of vitamin D-related cognitive functioning in locomotor function as depressed people are usually less active and loose muscle mass as well as sensorimotor performance 70.

Second, vitamin D belongs a complex biological system, and its insufficiency causes an elevation of serum parathormone 109. Patients with primary hyperparathyroidism usually exhibit cognitive disorders 109110, that could be reversed after parathyroidectomy 110. Moreover, in the Helsinki Ageing Study, high parathormone concentrations indicated an independent 2-fold risk for a five-year cognitive decline 111. The systemic microvascular disease involving cerebral vasculature together with hypercalcemia have been proposed to result in disruption of the blood brain barrier and accumulation of calcium deposits in brain tissue, leading to cognitive impairment 111. In vitro studies have also shown that parathormone increases intracellular calcium concentration and causes cell deterioration in the rodent hippocampal neurons 112. Furthermore, individual differences in the cell membrane ability to resist calcium influx have been hypothesized to cause the well-known but poorly understood variability of clinical symptoms in patients with hyperparathyroidism 111.

Anyway and to the best of our knowledge, the association of hypovitaminosis D with global cognitive impairment persist after adjustment for these both covariables.

This association of vitamin D with global composite cognitive scores has been recently explained by executive function and processing speed impairments 106113. Amongst 1,080 subjects (mean age 75 years, 76% women) free of neuropsychiatric disorders (epilepsy, schizophrenia, bipolar disorder, mental retardation, brain tumors, Human Immunodeficiency Virus), Buell et al. found a significant positive linear association between serum 25(OH)D concentrations and scores in tests exploring executive functions (Trail Making Test: flexibility) and speed of information processing (Digit Symbol Coding) 113. In addition, Llewellyn et al. 106 found a substantial decline on Trail-Making Test B among 858 adults 65 years or older enrolled in the InCHIANTI study and followed for an average of 5.2 years. Executive functions include all heterogeneous cognitive processes required in the regulation of cognitive activity during the treatment of complex and/or new and/or conflictual tasks 114. These frontal and attention functions are precisely those which enable us to adapt our behaviors - such as walking - to expected or unforeseen situations of daily living. They are therefore of prime importance for determining posture, navigation abilities and locomotor performance. For instance, they have direct impact on selection of postural control strategies when older adults encounter specific temporal and environmental constraints which could place them at risk for falls 114115116.

From an epidemiological point of view, vitamin D insufficiency is very frequent in the elderly and is dependent on the presence or absence of a history of falls 120121. The prevalence of vitamin D insufficiency is estimated between 40% and 50% in non-fallers over the age of 65 and up to 70% in fallers 65120121. It has also been demonstrated in institutionalized elderly that fallers had lower serum vitamin D concentrations than non-fallers 121.

In addition, the majority of data published over the last 15 years demonstrated the existence of a significant effect of vitamin D and/or calcium supplementation on fall reduction 1617. It has indeed been shown that vitamin D supplementation (800 IU/day) either alone or in combination with calcium (500-1200 mg/day) allows a very marked reduction in the number of falls in the same individual but also in the number of fallers, with a reduction of up to 50% 161718. A 2004 meta-analysis confirmed that simple vitamin D supplementation, regardless of its type but at a dose of 800 IU/day, allowed a mean reduction of the fall rate by 22%, with a maximum effect of 53% when combined with oral calcium 16. This meta-analysis also showed that the number of subjects needed to treat to prevent one fall was 15. Furthermore, the most recent meta-analysis by Bischoff-Ferrari et al. 17 demonstrated that vitamin D supplementation of at least 700UI per day might reduce the risk of falls amongst older adults by 19%. In addition, a minimum serum vitamin D concentration of 60 nmol/L could result in a 23% fall reduction, whereas lower concentrations had no effect on the number of falls 17.

In addition to vitamin D-related phosphocalcic regulation, the vitamin D-related fall rate reduction induces a fracture rate reduction. A 2005 meta-analysis on the antifracture effect of vitamin D supplementation based on 12 clinical trials combining a total of 19,114 women over the age of 60 and living at home showed a significant reduction of the relative risk of hip fracture by 26% and other non-vertebral fractures by 23% 22. This antifracture effect was only observed for a vitamin D supplementation of 700 to 800 IU per day. A similar result was observed in frail institutionalized elderly subjects 65. In contrast, the Cochrane Systematic review concluded that there was no reduction in fracture rate related to vitamin D supplementation alone 18, while combined calcium and vitamin D supplementation reduced significantly the incidence of fractures in older adults living in institutionalized care facilities 18, which was confirmed by two 2007 meta-analyses 122123. In line with this, a third 2007 meta-analysis concluded that calcium with or without vitamin D may reduce the total fracture risk by 12% 41. Finally, Bischoff-Ferrari et al. 23 most recently demonstrated in a 2009 meta-analysis of high-quality double-blinded randomized clinical trials - including 42279 adults aged 65 and older - the protective action of oral supplemental vitamin D against nonvertebral fractures with a dose dependant effect. This prevention was effective whether in community-dwelling or institutionalized older individuals, and was interestingly independent of additional calcium supplementation 23.

However, some negative results appear to contradict these previous findings, as they failed to demonstrate any significant fall or fracture reduction 9101112131415 (Table 1). These mixed results could be due to potential confounders. Firstly, the vitamin status appears to be essential, as vitamin D insufficiency, defined according to a serum cut-off value ranging between 10 and 30 ng/mL of 25(OH)D, is more often associated with a significant effect 8. Secondly, the daily dose of vitamin D is decisive and must be at least 800 IU per day. For instance, in Jackson et al.'s study 13, the dosage was only half that recommended to obtain an effect on the risk of falls. Thirdly, subjects must comply with treatment. In Jackson et al.'s study 13, negative results were obtained by intention-to-treat analysis, but in this study, only 59% of women presented good compliance with vitamin D and calcium treatment, defined by the authors as taking 80% or more of the prescribed treatment. When the analysis was limited to women with good compliance with treatment, the effect on reduction of hip fractures was significant with a 29% reduction of the fracture rate. Calcium and vitamin D supplementation was also associated with a 26% reduction of the fall rate for women with no history of falls. Fourthly, the initial health status of elderly subjects seems also decisive, as it directly influences the risk of falls and complications 124. As an example, in Cochrane Systematic review, the effect of combined vitamin D and calcium on fractures was solely shown in institutionalized subjects 18. Ageing, either physiological or pathological, is a process which modifies the individual's health status. At the population level, it results in the formation of a heterogeneous group in terms of health status 11124125126 comprising a subgroup of high-risk subjects with an altered state of health due to multiple diseases, with functional limitations and impaired adaptation capacities and a high risk of falls 124125126. The mixed conclusions could also depend on selection of studies for inclusion in the meta-analyses 161721. As an example, a negative study was excluded from the last meta-analysis because patients were "in an unstable health state" although it was not an initial exclusion criterion 17127. It should also be noted that several studies showed that vitamin D2 was less effective than vitamin D3 in humans 128129130. In addition, the absence of effect of vitamin D supplementation on fractures could depend on the type of fracture considered 101112131415. Finally, fall was usually not the primary outcome in these studies and assessment of fall frequency was not optimal 101112131415.

BMD: Bone mineral density; CNS: Central nervous system; APA: Anticipated postural adjustments; 25(OH)D: 25-hydroxyvitamin D; UV: Ultraviolet; 1,25(OH)D: 1,25-dihydroxyvitamin D; mRNA: Messenger ribonucleic acid; CaBP: Calcium Binding Protein; OR: Odds ratio; VDR: Vitamin D receptor; NGF: Nerve Growth Factor; GDNF: Glial cell line-derived neurotrophic factor; MMSE: Mini Mental Status Examination.

CA serves as a consultant for Ipsen Pharma company. He has no relevant financial interest in this manuscript. MMO reports no conflict of interest. He has no relevant financial interest in this manuscript. AMS serves as a consultant for Ipsen Pharma company. She has no relevant financial interest in this manuscript. GB reports no conflict of interest. He has no relevant financial interest in this manuscript. BF reports no conflict of interest. He has no relevant financial interest in this manuscript. OB serves as a consultant for Ipsen Pharma company. He has no relevant financial interest in this manuscript.

CA has full access to the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analyses. Study concept and design: CA and OB. Acquisition of data: CA and OB. Analysis and interpretation of data: CA, OB, MMO, AMS, and BF. Drafting of the manuscript: CA and OB. Critical revision of the manuscript for important intellectual content: MMO, AMS, GB, and BF. Obtained funding: not applicable. Administrative, technical, or material support: CA and OB. Study supervision: OB. All authors read and approved the final manuscript.