There is no work in progress on this front that I am aware of.
you can see there are 2 scenarios. One is to interpret w:instrText such as
Using xml Syntax Highlighting
TOC \o "1-3" \h \z \u </w:instrText>
Parsed in 0.000 seconds, using GeSHi
and generate a TOC from that. It wouldn't be particularly hard to add support for some subset of TOC options.
The other would be to interpret a complete TOC generated by Word (which my immediate inclination would be to avoid, but which might be better if you are trying to copy exactly what Word does)
The approach I'd take is to add/modify a template in docx2fo.xslt which matches a TOC field, and either processes that in a different mode, or hands processing off to an extension function. I guess if a style inherits from a heading style, it also needs to be included in the TOC, which suggests an extension function.
There is some code in package org.docx4j.model.fields which you may want to look at.