Dear Abraham
I may misunderstand your question, but from my feeling, what you aim at doesn't really make a lot of sense.
Fist of all, VBM yields local volume estimates (or, in the case of unmodulated images, tissue-class probabilities), not cortical thickness. If you want to analyze cortical thickness within your signfiicant volumetric changes, you probably need to use a surface-based package rather than SPM (though tools have been developed for SPM as well: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2330066/, they are apparently not publicly released). Be aware, though, that these usually entail sueface-based normalization and hence may end up in a slightly different space than your VBM data.
When asking about extracting local volume estimates for each subject at the significant VBM differences for a post-hoc ROI-analysis, I would advice against it. This would be the classical "double-dipping", as you have a biased selection of your voxels of interest. Plotting the effect in SPM should be the method of choice for visualizing an effect.
Finally, if you want to assess volume in predefined ROIs, you can use SPM add-ons like the marsbar or the Anatomy Toolbox (for histological ROIs). In this case, you would extract the values from the modulated GM image for each subject. While most tools provide mean rather than sum measures, this doesn't matter, you just need to multiply by the size of your ROI.
Best
Simon