My academic path hasn't been exactly.... linear. I've made a rather large scientific transitions between various stages of my training, resulting in a publication record that has several themes. Overall this has been good for me, but it's not uncommon for me to be invited to give a talk, or review a paper, on a topic I haven't actively worked on in years. In the case of a talk I offer to present new and exciting data we have recently generated on our current lab obsessions, but reviews are a little different.
Generally I inform the AE that the topic is no longer something I follow closely, but that's rarely a deterrent. Having to chase down reviewers myself, I know that some reviewer pools are shallow and I would prefer someone with slightly dated experience over tangential experience. I'm not saying I can't contribute a solid review, but it does make me a little uncomfortable (or makes it more work) when my knowledge of the particular topic is >5 years old.
My question for today is: how far do go back into your scientific history to handle the review of a manuscript? If your training has all been along a similar theme, how far do you extend yourself?