Unfortunately, there is no "proper" way to do what you are trying to do. You could have your working directory on a shared drive and just notify your reviewer when you are ready for them to start their review process, but that side-steps the accountability by not having each development/review iteration officially logged within TFS. This means you should check in your work and let the reviewer do their job, then continue on in that fashion to make any changes requested by the reviewer, check in, and get another code-review.
For completeness I will mention my suggestion from my comments here as well.
My suggestion would be to create a self-contained, short-lived development branch where you will do your development and have your code reviewed. Then once the development and reviewing has been completed to satisfaction, that branch can be merged back up and destroyed. This provides a much cleaner and safer approach. 1) It reduces the clutter in the history within TFS. 2) It prevents multiple unnecessary automated builds/tests/etc... from being triggered.
In your comment you suggest that this changes the "structure of your branching methods". I don't see how doing this changes anything in any way that matters. Your merge would be just like your final development check-in except that by this time all reviews have been completed and you are performing a single, clean check-in. It will still contain all of your check-in and review information, however instead of a cluttered chain of check-ins, you will have a single collapsed node which contains every single thing that was done for that particular task.
I would check with your manager, your code reviewer and/or anyone you have that is in charge of TFS and creating/maintaining your TFS policy. This approach really doesn't change anything in regard to how the rest of your process works. You would have simply abstracted your development cycle to a self-contained environment. The second you perform your merge you are right back into your normal process as you have it now.