Ron would really have needed his EFGR level and some of the other molecular markers to tune the prognosis. Even with that, prognosis is just a statistical guess - since many factors are not covered in whatever model you are looking at.

There are some interesting abstracts on the ASTRO 2014 conference site that deal with prognosis - including an alternate staging methodology for HPV+ oropharyngeal cancer developed with a RPA (recursive partitioning analysis) model.