Estimation of historical control rate for a single arm de-escalation study – Application to the POSITIVE trial