Not exactly something that's easy to confirm in humans. Of course the mice were genetically tweaked, it would be ... hard to measure an effect of a genetic factor without having experimental groups differing in that factor in a closely defined way.
Is the conclusion speculative? Absolutely. Then again, so is a lot of what we're discussing around here, and there is a reason the study made it into Nature (Impact Factor of 36).
Nature (Impact Factor of 36)
Ioannidis suggests that better journals produce less accurate research. A separate effect is that generalist journals are spread too a thin and can't competently referee. (eg, one editor can be highly biased towards his friends without the other editors being able to tell)
If it's worth saying, but not worth its own post, even in Discussion, it goes here.