In PDF, any line that starts with % is considered a ‘comment’ - similar to comments in C#, Java etc.
Now storing important metadata in comments is a bad practice and comments will be most likely stripped away by any PDF processing software. Instead metadata should be included as part of Cos/SDF object model (e.g. as custom entires in doc info dictionary), or ever better as embedded XML/XMP metadata stream.
Default version of PDFNet does not give access to comments (for efficency and other reasosns), but we could provide a customized version that will expose comments.