I’ve run into this before with password protected PDF files, but I suppose any kind of error that might cause a PDF to fail in text extraction (which would be needed for searching) could cause this.
You’re best bet is to go through your PDFs one by one and look for any errors like this. Once you find it, you’ll probably want to export and remove it (fully, with an emptied trash).
There does seem to be an issue where PDF files are being indexed too often. In theory that should only happen when they are first imported, and then never again unless you run the command to rebuild the index, or some activity necessitates doing so (like syncing with the iOS version).