Simply put, we can't make conclusions of human diet outcomes based on studies conducted in mice.
But can we even do so in mice?
Maybe not.
It seems like there are 3 huge confounders for mouse diet (and other) studies.
The first is that a mouse's living arrangements (eg. noise, bedding, light, the pH of their water, and more) can markedly affect study outcomes.
The second is fecal-oral eating. Yes, mice eat each others' feces, and if we're talking the impact of diet on health, knowing which mouse ate how much of whose feces and what said feces contained might matter.
And lastly there's this tweet by Emily Deans who was recounting Ronald Kahn's experiences that noted that strains of genetically identical mice have different metabolic responses to identical diets (which may speak to the living arrangement issue, or to differing microbiomes, or both).
All this to say, in free-living humans, diet studies are anything but simple, and even in captive mice, the control of confounding variables is challenging.