This kind of thing shows up all over the place in all fields of research. It’s why for example although you can see immune system differences at a population level between patients with disease X and healthy controls, you can’t in any way use it as a biomarker for an individual who might have disease X.

[1] https://link.springer.com/article/10.1186/s13293-026-00976-9...

My experience interacting with transgender people is that transgender men and cisgender women have a lot of things in common and visa versa, so I would expect this to follow that pattern, but it would be interesting to see it studied.