I second @Alex- 's idea, just couple of notes:
1. If firewall from which you want to check IP_A is managed from the same management, you can use management to connect to firewall over SIC (cprid_util). If return value of ping is 0, IP_A is reachable, otherwise not reachable.
2. Create both NAT rules manually and save both NAT rule UIDs. One of NAT rule will be always disabled, second NAT rule will be always enabled. Depending if IP_A is reachable or not, first NAT rule with already known UID will be disabled, second NAT rule with already known UID will be enabled and vice versa.
3. script will be run on management every XY minutes and do needed action once change is detected, including policy push.
Kind regards,
Jozko Mrkvicka