There is significant evidence on this front, but also significant counter- examples and statistical noise that suggests it is possibly a false correlation..

Note that one of the counter arguments in the link misses the point of the lithium hypothesis. They point out that the supposed preventative dose of lithium is tiny compared to the therapeutic dose. The hypothesis is that lithium is a micro-nutrient necessary in tiny amounts for good brain health. Over-loading this nutrient would overcome metabolic differences or radically shift the balance of certain metabolic processes. But, the point still stands that the statistical correlation is pretty weak, if one avoids cherry picking data and accounts for confounding variables like access to health care.

I don't personally have an opinion; no one knows why lithium works for bipolar in the first place, so the hypothesis that it is a micro-nutrient is plausible on its face.