That is a difficult question to answer.
Some journals (the very top journals) just won't accept any inferences/conclusions that you make from in silico pathway analyses alone. You will require validation results performed in the wet-lab to back up the findings. Other journals will fully accept such conclusions. The benefit of using KEGG, however, is that it already has a good World reputation and is curated, which means, at least from what I understand, that the pathways in KEGG are screened by human eyes and have experimental backing.
The way that people bring in things like KEGG into their writing is through the use of the conditional tense (could, would, ought to, etc.), and through statements like "these findings will require further validation". However, using statements like this, you're already implying that you're targeting a journal of impact factor <10.
Enrichment analysis on pathways also allows you to derive P values that provide more evidence that a particular pathway will be disturbed in your condition of interest. If a KEGG pathway contains 20 genes, all of which are down-regulated in your study, then it's a good sign that this particular pathway will be affected, and the enrichment P value would be virtually nil / zero.
Just my own 2 cents!