Theoretically, you can convert it back. You'd have to write your own DOM traversal code and convert the HTML back to Markdown.
Generally Markdown is thought to be the human readable/writable source of the information that is converted to HTML for further markup and styling.
HTML can be much more complex than Markdown and can be indefinitely nested and partitioned into tags. This is why it's so questionable to write a general purpose converter which reliably converts HTML back to Markdown. Just imagine all the whitespace and paragraphs going bye-bye and possibly causing a terrible mess for the human eye.
My suggestion is: Unless you generate originating HTML yourself and know what it consists of, don't convert it back to Markdown. Keep the Markdown version all the time and convert to HTML when needed.