try using the R package KEGGgraph to get a graph object from KGML files (check the methods parseKGML, KEGGpathway2Graph and getKEGGedgeData). Subsequently, you could use (as already suggested before) igraph for further analysis.

Not sure if there is an easy way to do this. Typically, KEGG enrichment is conducted the same way as GO enrichment (so, topology doesn't really matter).

That said, here are a couple suggestions:

1) KEGG pathways can be visualized here (somewhere there is also an option to upload fold-change values in order to color the pathway appropriately).

http://www.genome.jp/kegg/kegg2.html

2) There are some enrichment algorithms based upon topology (although I don't know for certain if they use KEGG)

http://genome.cshlp.org/content/17/10/1537

http://bioinformatics.oxfordjournals.org/content/25/1/75

I also developed the BD-Func algorithm to take positive versus negative regulation into consideration (although this is not based on the detailed graph characteristics):

https://peerj.com/articles/159/