Vitamin D supplementation differentially affects seasonal multiple sclerosis disease activity

Abstract Objectives Low ultraviolet‐B (UVB) radiation causes hypovitaminosis D, which is a known risk factor for multiple sclerosis (MS) and associated with MS disease activity. Our objective is to test whether vitamin D supplementation is most effective in lowering disease activity during the period of the year with low UVB radiation and consequently low serum 25‐hydroxyvitamin D3 (25(OH)D3) concentration. Methods Retrospective analysis of medical records from our outpatient department identified 40 MS patients with available data of at least 6 months before and during oral vitamin D supplementation. Serum 25(OH)D3 concentration was analyzed using immunoassay. UVB radiation data were provided by the local government. Annualized and quarterly relapse rates before and during vitamin D supplementation served as outcome parameters. Results During vitamin D supplementation (18,950 international units/week (mean, SD 3,397)), serum 25(OH)D3 concentration increased by 51 nmol/L and the UVB‐related seasonal variability in 25(OH)D3 levels ceased (rho = −0.13, p > .05). Furthermore, the annualized relapse rate decreased by approximately 50%. This was almost solely driven by the prominent reduction in the quarterly relapse rate in late winter/early spring, when 25(OH)D3 levels of nonsupplemented patients were the lowest. Conclusions Our study demonstrated the modulation of seasonal MS disease activity through vitamin D supplementation. Given the prominent reduction in the quarterly relapse rate in late winter/early spring, our data indicate a beneficial effect of supplementing MS patients with vitamin D, especially during this period of the year.

Ultraviolet-B (UVB) radiation (290-315 nm), which decreases with greater distance from the equator, is mandatory for the conversion of 7-dehydrocholesterol into previtamin D 3 (Webb, Kline, & Holick, 1988). Therefore, populations of mid-and high-latitude countries often lack a sufficient supply with VD and have a higher risk of developing MS (Kimlin, 2008). A retrospective data analysis conducted in France showed an association between regional variation in MS prevalence and UVB radiation (Orton et al., 2011). While UVB radiation is also thought to have a VD independent immunomodulatory effect (Breuer et al., 2014), it contributes to approximately 90% of the body's 25(OH) D supply (Federal Office for Radiation Protection (Germany), 2012).
Ascherio et al. reported a 57% lower relapse rate in patients with a higher 25(OH)D concentration (Ascherio et al., 2014). Furthermore, an analysis of the international MSBase registry data demonstrated that increasing latitudes away from the equator are associated with a shorter lag between seasonal UV radiation trough and relapse peak (Spelman et al., 2014). The relapse peak was recorded in early spring (March) and the relapse trough in autumn (October), which indicate an annual cyclical pattern of MS disease activity (Spelman et al., 2014). As UVB radiation is lowest during the winter months (Webb et al., 1988) and 25(OH)D 3 has a half-life of approximately 15 days (Jones, 2008), VD deficiency increases during winter. Consequently, hypovitaminosis D has the most substantial impact on the immune system in late winter/early spring.
To our knowledge no other study has yet examined the interactions among seasonal variations in UVB radiation, serum 25(OH)D 3 concentration, and MS relapse rates and the effect of VD supplementation on these associations. We aim to investigate whether MS relapse rates are higher, when UVB radiation and consequently 25(OH) D 3 levels are low. If this increase in MS disease activity is genuinely induced by hypovitaminosis D, then VD supplementation should be most effective during this period of the year.

| Patient group studied
Our retrospective medical chart analysis identified 40 MS patients with an observation period of at least 6 months before and after the initiation of VD supplementation. All patients were treated between April 2010 and June 2016 at the neurological outpatient department Neurocenter Peine, in Peine (latitude 52.3, longitude 10.2), Lower Saxony, Germany. Unless contraindicated, generally all MS patients from our outpatient department are recommended to take orally 20,000 international units (IU) of VD per week. This is well below the Food and Nutrition Board's upper tolerable intake limit recommendation of 4,000 IU/day (28,000 IU/week) (Institute of Medicine, Food and Nutrition Board, 2010). Patient consultation and documentation was conducted by MM. Data were retrospectively assessed by MM and AM. VD status was determined by measuring the serum 25(OH) D 3 concentration in a local laboratory using chemiluminescent immunoassay. If different immunotherapies were used in the predefined intervals (before and during VD supplementation), the MS therapy with the longest duration in each interval was defined as concomitant immunotherapy. Following the German MS guidelines, a relapse was defined as a worsening of clinical symptoms, which was not related to any other infectious disease or condition and lasted more than 24 hr (Gold, 2012).

| UVB radiation
We received the local (latitude 52.2, longitude 9.1) erythemal UVB radiation (mW/m²) data from the ministry of environment, energy, and climate protection of Lower Saxony. UVB radiation was recorded from April 2010 to June 2016. Mean UVB values were calculated for each month and each quarter of the year (January-March, April-June, July-September, and October-December).

| Endpoints
Efficacy endpoints were the individually analyzed quarterly (QRR) and annualized relapse rate (ARR). The ARR was calculated by dividing the total number of relapses by the observed time in years.
Calculation of the QRR was based on the monthly relapse rates, which were calculated by dividing the cumulative number of relapses by the total number of the respective month during the entire observation period. Afterward, the monthly relapse rates of each quarter were summed.

| Statistical analysis
We performed a descriptive analysis, which included all evaluable patients. Missing values were not imputed. The Kruskal-Wallis test was used to assess the different quarterly distributions of the year's UVB radiation. Regarding the comparison of serum 25(OH)D 3 levels before and during VD supplementation, we followed Hintzpeter et al.'s proposal and ran a seasonal deconvolution to control for the month of sampling (Hintzpeter et al., 2008). Using Spearman's correlation coefficient, we separately investigated the patients' data before and during VD supplementation to identify the effect of VD supplementation on the association between UVB radiation and serum 25(OH)D 3 concentration. The Wilcoxon signed-rank test was used to investigate the ARR and QRR before and during VD supplementation. Statistical significance was declared for p < .05 and illustrated in the figures using "*".

| Ethical approval
The present retrospective data analysis was in accordance with the ethic committee of the medical association of Lower Saxony, Germany.

| RESULTS
The baseline characteristics of the study population are presented in Table 1. The serum samples of 25(OH)D 3 measurements were obtained 2 years (mean, SD 1) after the initiation of VD supplementation. Patients received a mean VD dose of 18,950 IU per week (SD 3,397), which increased the mean serum 25(OH)D 3 concentration by 51.4 nmol/L ( Table 1). The year's UVB radiation differed significantly as higher values were recorded from April to September and lower values were recorded from October to March ( Figure S1).
Before VD supplementation, we found a strong correlation between the monthly UVB radiation at sampling and serum 25(OH)D 3 concentration (rho = 0.41, p < .01, number of 25(OH)D 3 measurements: 47). Consequently, serum 25(OH)D 3 levels of nonsupplemented MS patients showed seasonal variations exhibiting a peak from July to September ( Figure S1), which labels the end of the period of high UVB exposure ( Figure S1). This was the only period of the year during which serum 25(OH)D 3 levels of non-supplemented patients were above the lower limit of normal (50 nmol/L (Hintzpeter et al., 2008)). In contrast, during VD supplementation, serum 25(OH)D 3 levels remained steadily above 70 nmol/L and the seasonal fluctuations ceased (Figure 1). The correlation between UVB radiation and serum 25(OH)D 3 concentration was no longer significant (rho = −0.13, p = .33, number of 25(OH)D 3 measurements: 62).
In comparison to before supplementation, the ARR was approximately 50% lower during VD supplementation (Figure 2a

| DISCUSSION
Investigating the effect of VD supplementation on the interactions between seasonal variations in UVB radiation, serum 25(OH)D 3   (Table 1). Similarly, a prospective cohort study from Finland reported an increase of 56 nmol/L after supplementing patients with 20,000 IU of VD per week for 12 months (Soilu-Hänninen et al., 2012).

T A B L E 1 Baseline characteristics
In our study cohort, the ARR decreased by approximately 50% after the initiation of VD supplementation. In comparison, Burton et al.
reported a slightly lower reduction in the ARR (41%) (Burton et al., 2010). The patients included in our study were supplemented with VD regardless of their baseline serum 25(OH)D 3 concentration. Therefore, we cannot rule out that also patients with sufficient 25(OH)D 3 levels were enrolled. It is more likely, however, that this would have led to a reduced rather than an increased effect of VD supplementation.  The effect of VD supplementation on MS disease activity was most prominent in late winter/early spring (January-March), which is the period of the year with low UVB radiation and the lowest serum 25(OH)D 3 concentration. Interestingly, we did not observe an increase in disease activity from October to December, when UVB radiation is low as well and 25(OH)D 3 levels start to decline. An epidemiological study supports this finding demonstrating that the increase in relapse rate lagged 1.5 months behind the local UVB radiation (Tremlett et al., 2008). Therefore, UVB radiation of late summer might also influence relapses in the fourth quarter of the year. The approximately 15-day half-life of 25(OH)D 3 might play a key role in this lag (Jones, 2008). As our work was a purely retrospective study of medical records, we cannot provide seasonal immunological data to support our explanation for the lag in relapse rates. Immunological alterations induced by hypovitaminosis D such as suppression of inflammatory cytokines, stimulation of T-regulatory and Th2-cell differentiation, inhibition of B-cell development, and modulation of the Nf-kappa-B pathway (Szymczak & Pawliczak, 2016) might appear earlier than the clinical effects-in our case the increase in ARR. The lack of laboratory investigations is therefore the major weakness of our study and should be considered while interpreting the epidemiological data.

October -December October -December
Before VD supplementation, patients also demonstrated a peak of relapse rates from July to September, which was not paralleled by 25(OH)D 3 deficiency. This relapse peak points to additional triggering factors of disease activity aside from UVB radiation and serum 25(OH) D 3 concentration. Oqawa G et al. demonstrated in a Japanese cohort of MS patients that relapse rates had two peaks over the year: one in the winter and one the in summer (Oqawa et al., 2004). The summer peak in relapse rates was explained by higher temperatures leading to immunological modulations, in particularly to a change in leukocyte nitric-oxide production (Beenakker et al., 2001;Oqawa et al., 2004). A complementary immunological explanation from a recently conducted study are the summer troughs of melatonin production, which ultimately lead to the induction of pathogenic Th17-cell differentiation, inhibition of protective T-regulatory cells, and deactivation of IL-10 promotors (Farez et al., 2015). Further studies are warranted to clarify these hypotheses as possible reasons for the increase in relapse rates during summer.
Retrospective studies are generally limited in their validity.
However, we have no evidence for biases often associated with retrospective study designs, for example, selection or information bias. To check for the selection bias, we compared the baseline characteristics of included patients (≥6 months follow-up, n = 40) with the baseline characteristics of all other MS patients from the outpatient department (n = 65). In this comparison, we did not find significant differences between both patient groups regarding age, gender, MS phenotype, ARR, and score on the expanded disability status scale. An information bias is unlikely because patient consultation and documentation was conducted by only one physician (MM). Also, a regression to the mean effect appears unlikely as the initiation of VD supplementation was not driven by disease activity, which is supported by the inclusion of 15/40 patients who were relapse free before the start of VD supplementation. Furthermore, the correlation analysis between UVB radiation and serum 25(OH)D 3 concentration needs clarification. As indicated by the included data points, some patients had more than one observation in this correlation analysis. Consequently, some individuals possibly contributed more than others to the observed effect. However, as UVB radiation contributes to approximately 90% of the body's 25(OH)D 3 supply (Federal Office for Radiation Protection (Germany), 2012), we conclude that seasonal UVB radiation has a stronger impact on serum 25(OH)D 3 levels than the effects caused by each individual patient.
Our conclusion is supported by a correlation analysis between the mean values of serum 25(OH)D 3 levels and the mean concurrent UVB radiation, which precludes the inclusion of more than one observation per patient (prior VD supplementation: rho = 0.42, p = .01, n = 35; during VD supplementation: rho = −0.24, p = .19, n = 32). Lastly, immunotherapies differed between observation periods, which might have a major effect on the ARR as our outcome parameter. However, our multivariate cross-sectional linear regression analysis proved that VD supplementation has a significant effect on MS disease activity.
Altogether, our data indicate that VD supplementation has beneficial effects on MS disease activity and that these effects are higher during seasons with low UVB radiation and consequently insufficient 25(OH)D 3 supply. Therefore, our data argue for an intensified VD supplementation in MS patients, especially during winter and early spring.
Regarding the seasonality of VD efficacy, upcoming VD supplementation studies should adjust their efficacy analysis for the different seasons of the year. Furthermore, future research should focus on the identification of additional triggering factors of MS disease activity.
Ideally-as is the case with VD-these factors could be modulated therapeutically. In addition to standard medication, this would allow supportive therapies adjusted for the specific immunological demands during different seasons of the year.

ACKNOWLEDGMENTS
We thank A. Hainsch (Trade and industrial inspection agency of state of Lower Saxony-Air Quality Network Lower Saxony, Germany) for providing the regional climate data.

STATEMENT
On behalf of all authors, the corresponding author states that there is no conflict of interest. This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.