There's no explicit code on our side to handle that case yet, so I'm not sure what would exactly play out. Theoretically, it could cancel the client fetch side and assuming the proxy side was still working through the query, it would receive an early client error and go through its normal clean up.
The question being: will that clean up cancel the running query? I'd have to experiment to be sure. I'd probably start with a pg_sleep(long time) query and see what happens when it's aborted.