I am probably repeating myself, but I cannot emphasize enough that a common fallacy of medical research is the focus on pathologies. A recent example with respect to the "omnipotency" (that's what the Internet news could make you believe) of vitamin D comes from researchers at George Town University (GU. Press Release).
The scientists were able to replicate the results of previous studies, where high dose vitamin D supplementation had "significantly reduced development of estrogen receptor-positive (ER+) breast cancer", but for estrogen receptor-negative (ER-) breast cancer they found either no effect (lean mice) or even an increased rate of cancerous growth in the group of obese mice.
What's yet even more interesting that a similar contradiction was evident with respect to vitamin D's widely perpetuated beneficial effects on insulin resistance. In the pathologic model of the obese mice, vitamin D @ 15-25k IU per day was in fact able to ameliorate insulin resistance, in the lean, naturally insulin sensitive mice, however, insulin sensitivity was reduced by supplemental vitamin D.
As I've discussed it in the context of the most recent fish oil study (and will probably repeat for other "super nutrients"), in 99.9% of the cases the effect of - especially high dose - supplementation with ostensibly harmless "vitamins" or "nutrients" vary enormously depending on the subjects current nutritional and health status. To derive one-size-fits-it-all recommendations from individual studies and to transfer results obtained from a pathological model without further scientific investigations to a healthy one is careless and may turn out to be very dangerous.