Mendelian randomization study not finding significant differences in COVID-19 outcomes based on vitamin D level. This study does not compare patients with deficiency/insuffiency/sufficiency, only providing ORs for increase in D levels. Authors note that their results do not apply to individuals with vitamin D deficiency.
Authors cite only 2 of the 25 vitamin D treatment studies (2 of 5 RCTs) at the time, including the only study reporting a negative effect. Authors indicate that they believe
[Murai] was a significant result, however that study used cholecalciferol with very late stage patients. In practice, calcifediol/calcitrol would be used due to the long delay in conversion of cholecalciferol, hence the study is not informative of either normal late stage treatment, or earlier treatment. That authors believe the study is important suggests a strong bias.
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.
Butler-Laporte et al., 6/1/2021, peer-reviewed, 16 authors.