Case-Control Studies with Contaminated Controls