Almost a WAG, but re your first question here is return_question which may be worth checking:

The difference may be rooted in how you're handling times in the script. Are you, for example, using any construct that might be converting (or, "failing to convert") some dynamically created element in the newsletter to localtime?

As an aside, any html "latest update" field would be pretty worthless if dynamic in the sense of using the time of the current request as the time of the latest update; in fact, one might regard that kind of specious date as damned near fraud (I do! YMMV). Yes, there are quite a lot of entities that do just that, but widespread practice doesn't make it legit, IMO.

Just for clarity, which I seem to be having difficulty with just now) I have no problem with a "latest update" that's tied to content change -- for example, the latest revision of specific content in the db used as source of the page's content, or the last time the hmtl content was actually changed).