Spot instance requests are not cancelled after cluster "destruction"

I've noticed that when using spot instances, first a request is created and, when fulfilled, the associated instances are launched. Standard operating procedure on EC2. The problem is when the cluster is destroyed the spot request are still there (in an open state) and new instances may be assigned to them. I've experienced this behavior three times before starting to manually delete spot request when the cluster is no longer needed.

Anyone can confirm?