You should also add a filter to crawl the domain where the PDFs are, then (at
the page you input the sites) chose the mirroring mode as "Get separated
files" intead of "Download website(s)". It will be something like:

-* +philmat.oxfordjournals.org/* +*.full.pdf

This way HTTrack will crawl the HTML pages, but only save the PDF files it
finds. The MIME type filters only work for pages that where already scheduled
for download.