Unfortunately RTF has no watermarking concept, so the API call takes no effect there. Also I am not sure if it is possible to simulate watermark in RTF in general - by underlaying or overlaying an image.
Try to create a desired RTF document layout in OpenOffice etc avoiding non-standard application specific features and send it to us - we'll estimate if that kind of output is implementable in PD4ML.
MS Word is not the right tool in the case: its RTF is always full of proprietary extensions and redundant tags. And also you've sent us DOC, not RTF. We'll try to save the DOC as RTF to see if your markup survives.