I don't think there is a general solution, at least not one implemented in Java. FOP reads XML files, not Doc - you could try saving the Doc files in XML format, and have FOP works on that. I have never seen the Word XML format, though, so I don't know how much content and formatting is exposed through that. One other option might be to use the Jakarta POI library to read the Word file, and then generate the PDF with a library like iText. That would be a lot of implementation work, though, except maybe for the simplest Word documents.
I tried using POI to programmatcally translate Word Doc's to HTML. The Excel functionality is pretty mature but the Word functionality is unusable (it looks like they're having trouble finding someone to support that code). I've been using OpenOffice.org to do the job. OOo can generate PDF's, so maybe it is an option for you. The OOo API is non-trivial but there's tons of examples in the API download and more examples and discussion in the support forum.