I am planning on putting my Blender model into Unity and turning it into a VRM, but since it’s already rigged and baked with separated meshes for eyes, lashes, teeth, tongue, I was planning on generating the needed blend shapes separately instead of trying to combine them all into one mesh. I’ve heard this causing headaches for others when it comes to Unity not reading more than one mesh’s blend shapes (ie, only moving the face or teeth but not both) but haven’t found the solution for it.
You might get more traction with this question on the Asset Importing and Exporting forum: Unity Engine - Unity Discussions