Driver and sequencer are connected by TLM push/pull port in UVM. If driver is getting response from interface and needs to be driven to sequence then same port can be used to send response back to sequence.
For that response can be sent by using seq_item_export.put(rsp) in driver and sequence will use get_response(rsp) . By this way request and response flow would be independent.
response can also be sent immediately back to sequencer after request by using item_done(rsp) .
Even uvm provide rsp_port in uvm_driver which is analysis port in sequencer which be used to send response back to any other component using analysis port imp.