Calibra convert to htmlz then back to mobi corrupts toc

[twoten]

I have tided up a book in HTML converted from PDF exported to Mobi.
All appears OK with a nested TOC showing chapter titles then beneath this word definitions.

I wanted to edit this further so converted back to HTMLZ with the intention of editing and converting the finnished item back to MObi again.

The problem I have is when I convert trhe file back to HTMLZ the ''word definitions' have lost there links. I am sure this happens when converting back to the HTMLZ file as all the links appear to be OK in the Mobi file.

I have looked at the first HTML file and all the classes match for the 'word definitions' but the html file from the Mobi file has changed the calibre classes have changed.

Does anyone know if it's possible to convert back to the HTMLZ file without changing the classes. I have tried disabling all the setting in the Calibra conversion but still have a problem.