I’ve been following these studies for awhile. The biggest red flag for me is that the effect is constrained to male rates. This is exactly the sort of result you expect to see when not sufficiently controlling for multiple statistical analyses (i.e., unintentional p-hacking). I suspect a Bayesian approach with an appropriate prior on sex-sensitive biological cancer effects would not yield statistically significant results.
Some other broad concerns are outlined here: https://theness.com/neurologicablog/index.php/that-rat-cellp...