Background Low serum 25(OH)D3 (vD) in early disease may be associated with worse prognosis in multiple sclerosis (MS). Determinants of serum vD concentration are established in healthy individuals, but poorly studied in people with MS (pwMS). Unpicking the determinants of vD concentration in pwMS may have important preventative and therapeutic implications.

Methods We recruited a study cohort of pwMS from the UK MS Register (n=315) and matched controls (n=232). We measured serum 25(OH)D3 and genotyped six vD-associated SNPs. We developed a weighted genetic risk score (GRS) using these SNPs and validated it in 373,357 UK Biobank participants. We then assessed predictors of vD levels in our cohort and constructed multivariable models to assess the influence of GRS.

Results The GRS was well-validated, being strongly associated with vD status in the Biobank cohort (p<2x10-16). In our study cohort, taking vD supplementation, having MS, lower BMI, increased age and supplementation dose were associated with higher vD levels (false discovery rate, FDR<5%). In multivariable models adjusting for these confounders, GRS was strongly associated with vD level in non-supplementers (p=0.004), but not in supplementers (p=0.47).

Conclusions Our findings suggest vD supplementation is the most important determinant of vD level in pwMS, and genetic determinants may play a relatively minor role.