Old Word, Excel, and PDF documents all had the same flaw. They are poorly salted and use RC4 with 40 bit equivalent encryption keys. I included code for generating rainbow table chains. I just tested the hash functions and I messed up the MD5-ing. So Word and Excel were broken. Also I had the wrong offset for Excel so this should be a little faster. If you downloaded them before you should download it again to get the corrected hash functions. You should also know that the 32 bit version is faster.
Just found out how you read a patent... the claims are the only thing you need to look at. Therefore I don't know of any patents that this is covered by. There's a patent that covers making and using a simple lookup table. Here's some prior art. You just need to understand that by "CRC" Jason Davis means the first 31 bits of a hash. 31 is the value picked for "M" from the patent and log base two of the key space is "N" from the patent. In Oct 8, 2005 Jason finished one that used M = 31, N = about 34.04. The only difference is that these are passwords instead of keys, but there is no novelty of saying we should use this on keys. Actually if you look at the LM hash they are the same. Anyway the good news is that this won't affect the LHTs because it is a different design that is much faster and smaller.