McNemar's Test and sample size: how to estimate discordant pairs?

This post was in the biostats forum, but a quick review of the previous McNemar's topics shows most were posted in here.

I want to use McNemar's test to determine if diagnostic test A or B has better sensitivity then gold standard test C. I have estimates on the sensitivity,specificity and prevalance of test A and B, but I am not sure on how to determine the estimated discordant pairs that I would expect. My local biostatistician does not know the answer to this and I do not see an answer in the forums.

Any help pointing me to the right resource or showing me the right calculation to estimate discordant pairs would be so greatly appreciated.