A low-tech solution might be to create the neighbor's IP address as a node, monitor it (even if only by ICMP), and then receive alerts when it's unavailable.
If the neighbor is protected by a firewall or ACLs so that it won't respond to ICMP or an agreed-on-and-secure-and-unique snmp-v3 solution, you could still monitor its status by watching traps and syslog entries from your router that connects to that neighbor. When the neighbor's interface is down or the address is unreachable, there should be logs and traps generated by that event. Ensure they are properly forwarded to you by NPM and you should have your alert.
The IP's in question are not part of the routing table, they are only the PtP link IP's(No loopbacks were used to source OSPF). We have several locations that have backup circuits, and the OSPF alert is to let us know when a link goes down, and since they are all Ethernet, unless the ISP switch/router drops, that particular interface will not be shown as down. We used to monitor based on traffic flow, if the backup circuit started passing X amount of traffic, we would get an alert, it was the only way the old system could monitor.
I will look into trapping, though, with routers and switches, its all or nothing, ASA's are the only ones I know of that you can filter traps with.
I think I figured out a work around which can double as a latency/packet loss alert/NCM auto switchover. As long as Orion can see the WAN Link IP, I can set that up as an ICMP only node and set thresholds for alerting, and timers for NCM.