In theory you can, but I am not sure if Pega has tested - you will need to test it out yourself. Confluent is a complete streaming platform (including Kafka and a lot of other services). Check the Confluent website for details.
We can Leverage Kafka's K-table as the topics persistence hub and connect the Pega via RESTful API exposed on top of those K-tables. Hope that should address the need with HA set up and within the efficiency threshold (Throughput PoV). Please try and share your view.