Well, why did you use 500 genes? How are we supposed to answer that for you? If it's arbitrary, you may as well state as much. Presumably, it's the top 500 based on effect size rather than significance, I'd hope. You may need to re-analyze with a defined threshold for log fold change or such to have a real answer, though you should really perform DE with such thresholds as your null hypothesis rather than filtering post hoc.