Actually, the cell line issue might not be so bad or rather all cell line work is much worse than you all imagine.
No cell lines actually resemble the real tissue in gene expression if you say, do an rna screen and see what actuallg gets expressed. They are changed by line establishing process, use of various media, number of passages aka how long they’ve been in culture from thawing a line till the test (cant be too little or too much, requires line to recover after being frozen), person who’se culturing (some people tend to starve cells and passage full plates and have different parameters of work even despite some labwide protocols), adherence to the plate and growing in flat monoculture, getting plates out too much or watching them too lonv under the microscope. Some effects are temporary and not so serious. Some effects are rather serious. Like the fact that flat surface cell lines will never be a good proxy for the actual tumor growing in 3d in the actual human. That’s why people experimenting with spheres, multucultures with other cells, organelle cultures, animal models, and eventually all the clues from all the models add up to something that could be a theory of some mechanism. And then in most cases when applied to humans in real life it might not work or be harmful.
Now, the skin cell in cell media was likely arrayed and had features similar to breast cancer cells so it might have been okay to use for this super imperfect approximation.
Not sure if this particular instance. In vitro is only a part of it and peer reviewed journals differ in their authority ratings, with big boss journals like cell requiring you do deliver crazy amounts of different models to make any statement. Reading those papers is like reading house ads, scientists kind of learn to cut through bs