And class II also:
https://www.cancernetwork.com/immuno-oncology/significance-mhc-ii-immunotherapy
Patient-specific variations in MHC-II molecules have as much, if not more, of an impact on the mutations that arise in tumors as MHC-I variations
And, to confirm your statement about MHC-1:
Immune escape strategies aimed to avoid T-cell recognition, including the loss of tumor MHC class I expression, are commonly found in malignant cells.
But it seems to me this is the game you have to play. Somehow, until you figure out to fix the immune system itself in T1 diabetes, the Beta cells you transplant must somehow avoid detection by the immune system, whether it be by physical barriers (not solved yet) or by T-cell avoidance games. In both cases, cancer risk becomes an issue—although, in this case, physical barriers, if we could ever solve them, become useful in stopping cancer spread.
However, there is another, totally different way to approach it: reprogram patient cells to regress to stem cells (we know how to do that), then to turn into insulin-generating cells. I have always thought it was the right way to go because it will never require anti-reject meds—but it is much more expensive (must be done per patient), and does not solve the issue of diabetes’ auto-immune future attacks on these beta cells (you still have to take immuno-suppressants for that). The interesting thing is that the second step has now been successfully accomplished in this study, by the very same UCSF team.