How are your C# or VB.net skills? I am thinking pre-processing the file might be one way. You could use a Script Task to read the file in reverse order until you found data that has already been imported. Depending on how much data you expect and how much memory you have available you could store the data as you reversed through the file and then write it in one shot when you found old data, or, you could not store the data as you reversed and just changed course to read the file in forward-order to the end when you found old data, writing a new file to disk as you went. Either way you would end up with a file that you could import straightaway using a Data Flow Task. It would be tedious but not terribly difficult if you have some .NET skills.

See the second example in this article to see how to read a file in reverse order. And the first example shows you how to write a file byte by byte. It is for .NET 4.0 but these techniques will work in 2.0, it's just that the 2.0 article does not have the nice example to show you how to read a file in reverse using Seek.

http://msdn.microsoft.com/en-us/library/system.io.filestream.seek(v=vs.100).aspx