Background
Multiple sclerosis (MS) is an inflammatory autoimmune disease in the central nervous system [
1‐
4]. The prevalence of MS is rising globally [
1]. In the past decades, better understanding of the disease mechanisms of MS has led to several disease-modifying therapy development [
5]. However, current therapeutic treatments for MS still remain disappointing [
5]. It is necessary to explore the pathogenic mechanisms driving MS [
5]. It is known that MS is mediated by the interaction between genetic and environmental factors and well-powered genome-wide association studies (GWAS) have been explored to investigate the genetic risk factors in MS [
6].
The active form of vitamin D, 1,25-Dihydroxyvitamin D3 (1,25(OH)2D3), suppresses autoimmune diseases such as MS by reducing the production of proinflammatory cytokines such as IFN (interferon)-γ, IL (interleukin)-2, and IL-17, as well as enhancing the secretion of anti-inflammatory cytokines such as IL-4 and IL-10 [
7‐
11]. A recent retrospective case–control study showed that the effect of vitamin D3 on MS susceptibility was not mediated by regulatory T cells (Tregs) [
12]. Increased evidences suggest that decreased vitamin D levels is associated with an abnormal immune response in MS [
13,
14]. Low vitamin D levels have been shown to correlate with disease activity in various autoimmune disorders and neurological diseases including MS [
13‐
19]. Numerous epidemiological studies have strongly suggested that vitamin D insufficiency contributes to MS risk [
20]. One SNP, rs10766197 CYP2R1 which regulates serum 25-OH-vitamin D3 (25OHD) levels, is reported to relate with MS risk [
21,
22]. Thus, hypovitaminosis D was considered as a possible risk factor for MS, with multiple decades of research on the association of vitamin D status and MS [
23].
Mendelian randomization (MR) studies have been widely used to assess the causal link between an exposure and an outcome by an analytical method using genetic variants as instrumental variables [
13,
24‐
26]. Four previous MR studies used three or four serum 25OHD genetic instrumental variables (IVs) and suggested a link between predicted serum 25OHD and the risk of MS [
27‐
30]. MR studies have many limitations, including low statistical power. Thus, to improve statistical power, 20 serum 25OHD genetic IVs from the largest GWAS for serum 25OHD (n = 401,460) were used to update the effect of serum 25OHD levels on MS risk.
Discussion
Epidemiological evidence suggests an association between 25OHD deficiency and risk and disease progression of MS [
13,
20,
38‐
44]. This relationship is further proven by four previous MR studies using three or four serum 25OHD genetic IVs [
27‐
30]. In the first MR study in 2015, Mokry et al. [
28] identified an odds ratio [OR] of 2 by using four serum 25OHD genetic IVs obtained from a serum 25OHD GWAS (N = 33,996) [
45], as well as a MS GWAS involving 14,498 MS cases and 24,091 healthy controls from International Multiple Sclerosis Genetic Consortium (IMSGC) [
32]. In the second MR study in 2016, Rhead et al. [
29] identified an OR = 0.79 using three serum 25OHD genetic IVs. These samples were obtained from a serum 25OHD GWAS (N = 4501) [
46] and MS GWAS from members of Kaiser Permanente Medical Care Plan, Northern California Region (KPNC) (1,056 MS cases and 9,015 controls) and the Epidemiological Investigation of Multiple Sclerosis and Genes and Environment in Multiple Sclerosis MS case–control studies (6335 cases and 5762 controls). In the third MR study in 2020, Jacobs et al. [
27] identified an OR = 0.57 using six serum 25OHD genetic IVs obtained from a serum 25OHD GWAS (N = 79,366) [
47] and IMSGC discovery phase GWAS (14,802 MS, 26,703 controls). In the fourth MR study in 2021, Harroud et al. [
30] identified an OR = 0.72 using four serum 25OHD genetic IVs obtained from a serum 25OHD GWAS (N = 33,996) and IMSGC discovery phase GWAS (14,802 MS, 26,703 controls). In the present study, we identified an OR = 0.22 ~ 0.45 using 20 serum 25OHD genetic IVs from to date, the largest serum 25OHD GWAS (n = 401,460) [
31] and MS GWAS (14,498 MS cases and 24,091 controls) [
32]. Thus, our MR analysis further strengthens evidence for a causal link between 25OHD levels and MS risk.
Among 115 newly identified serum 25OHD genetic variants, 20 effectively independent 25OHD genetic instrumental variables were extracted from MS GWAS summary statistics. We did not identify any significant pleiotropic variant among the selected 20 25OHD genetic instrument variants in the MS GWAS datasets by pleiotropy analysis. Of note, five MR analysis methods demonstrated that 25OHD genetic instrument variants firmly influence the risk of MS through 25OHD but not through other pathways. We found that increased 25OHD levels due to genetic changes were significantly associated with reduced MS risk. For every 1 SD, genetically increased 25OHD levels could reduce MS risk 0.940, 0.835, 0.781, 1.484, and 0.913 using MR-egger, weighted median, inverse variance weighted (multiplicative random effects), simple mode and weighted mode, respectively. These results suggest a causal association between genetically increased 25OHD levels and MS risk.
Both MR-egger and inverse variance weighted methods demonstrated a significant heterogeneity among the 20 25OHD genetic instrument variants in the MS GWAS datasets. Thus, we mainly used the random effect model to estimate the MR effect size. Inverse variance weighted (multiplicative random effects) showed a reduced trend of MS with 1 SD genetically increased 25OHD levels (beta = − 0.781, 95% CI: [− 1.105, − 0.457], p = 2.34E−06). MR-egger, weighted median, simple mode and weighted mode also proved a causal link between 25OHD levels and MS risk. In addition, a single SNP effect analysis and leave-one-out sensitivity analysis for the effect of the 20 serum 25OHD SNPs on MS suggests that our results are robust with no obvious bias based on the effect of the single 25OHD SNP on MS.
25OHD deficiency is becoming an increasing problem worldwide. A recent finding led to a proposed link between 25OHD deficiency and autoimmune diseases [
48]. It is reported that 25OHD deficiency is associated with MS risk [
39]. Apart from the maintenance of healthy calcium metabolism, Vitamin D and its active product (1,25-dihydroxyvitamin D) have additional roles in immune and central nervous system cell homeostasis [
38].
To date, there is evidence to suggest there are some benefits of vitamin D for patient-important outcomes among people with MS [
40]. However, vitamin D did not affect recurrence of relapse and worsening of disability in patients with MS [
40]. However, in clinical trials, while doses of up to 40,000 IU/day were tested and appeared safe as add-on therapy for short-term periods, chronic and high-dose therapy can lead to life-threatening complications related to vitamin D toxicity, including renal failure, cardiac arrythmia and status epilepticus [
43].
There were several strengths to this MR study. First, this MR design was based on three principal assumptions and we used different methods to prove three principal assumptions. Second, we used the large-scale serum 25OHD GWAS dataset (n = 401,460) and MS GWAS summary statistics (14,498 MS cases and 24,091 controls) [
31,
32]. With a large-scale GWAS, it is easy to find a causal association between 25OHD levels and MS risk. Third, the two GWAS from European ancestry reduce the influence of population stratification. Fourth, using independent statistical methods, we proved 20 independent genetic variants to be the effective instrumental variables. Fifth, we used five MR analysis methods including MR-egger, weighted median, inverse variance weighted (multiplicative random effects), simple mode, and weighted mode. Critically, all five MR methods proved the causal association between 25OHD levels and MS risk. Lastly, our results were robust, with no obvious bias based on the different methods.
Of course, our MR study had some limitations. First, our MR analysis is only based on European ancestry. Therefore, our conclusions need to be investigated in other ancestries. Second, the GWAS dataset from IMSGC is based on clinically diagnosed MS [
32]. Different diagnostic criteria may produce different results. Third, among 115 newly identified serum 25OHD genetic variants [
31], only 20 25OHD genetic instrumental variables were effective and independent. Fourth, although we proved that increased serum 25OHD levels could reduce MS risk, the mechanisms by which 25OHD reduced MS risk is still unclear. Fifth, the genetically increased serum 25OHD level does not necessarily reflect the active form of vitamin D, 1,25-Dihydroxyvitamin D3 (1,25(OH)2D3), in the serum [
49]. Sixth, while our MR studies may provide the strongest evidence for a causal link between vitamin D levels and the risk of MS, our study does not address the role of vitamin D on disease activity in patients with established MS. Therefore, our study must expand to explore the causal association of vitamin D levels with the activity of MS disease using other databases.
Although MR study offers several advantages, it also has many limitations [
50]. These limitations include pleiotropy, low statistical power, canalization, population stratification (which is confounding by ancestry), etc. [
50]. To detect and correct for pleiotropy, probably the most challenging to address, a robust method MR-egger intercept was used in our MR study. To improve the statistical power, we used the, to date, largest large-scale GWAS for serum 25OHD [
31] and MS [
32]. To eliminate population stratification, both 25OHD GWAS and MS GWAS were chosen from European descent. Although we try to minimize the shortcomings, maybe there are still some limitations. Thus, our results need be reproduced in randomized controlled trials (RCTs).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.