historically the problem is that live attenuated virae do present the "wrong" sites ... so after one clinical study with babies some 30/40/50(?) years ago went terribly wrong (immunized arm with much severe rsv infection than usual) all development stopped for a long time. only with the biotech revolution we have the analytical capabilities to identify/model to target sites (though still with the usual failures of plasmid/vlp approaches) ...
... which is the other challenge: strength/persistence of immmune response as with any vaccine, see i believe nvax's "failed" maternal immunization program (supported by gates, but too weak response for what a commercial use would demand)