Mendelian randomization studies compare the estimated effect of SNPs associated with variation in vitamin D levels on the health outcomes in large numbers of patients. For more background on Mendelian randomization studies and their limitations see [nature.com].

For reasons why Mendelian randomization may fail in this case, see [nutrition.bmj.com]. Authors suggest that it may come down to the use of 25(OH)D concentration in serum as a less than ideal proxy for vitamin D status of cells involved in the immune response. For most other purposes, it may not matter much that unbound (free) 25(OH)D is the better predictor of vitamin D deficiency and the resulting unfavourable outcomes. But for the MR analysis, the genetic instrument is strongly dominated by variation in the GC gene which modulates the concentration of vitamin D-binding protein (VDBP) in blood and thereby indirectly the concentrations of 25(OH)D and 1,25-dihydroxy vitamin D. Thus, the common GC alleles rs4588A and rs7041T are both associated with much lower than average vitamin D concentrations. In contrast, directly measured unbound (free) vitamin D concentrations are minimally affected by these alleles, if at all.

[Grant] suggest that the primary reasons for Mendelien randomization failure include that the total SNP-induced variation in 25(OH)D has often been less than assay variance, and that genome-wide association studies of SNP effects have been made on the full range of 25(OH)D levels, while the data is non-linear with a significant percentage in the low and high plateaus of the outcome relationships.