I can't answer your question about the chances of extracting the data: that would depend on the nature of the corruption, and I have no way to know that. In the past when something similar happened, the user had tried to edit the document with an inappropriate tool, causing the first 64 bytes of data to be zeroed. In VDI that area happens to be static - so replacing that data with a patch tool repaired the file.
If you examine the clones with CloneVDI, does it agree that they are corrupt?
Perhaps it would be worth your while to view the first few hundred bytes of the file in a hex editor such as frHed
, and post a picture of the result here. That way we can see how much of the header is left.