More interested in primary endpoints that fail then new indications that work. It apprears that primary endpoints that fail with secondary endpoints that work tends to be more by chance than actual efficacy in the endpoint.