I think it's not that meaningful to do that kind of statistical analysis due to really small sample size, including HG/TN group as well. I would conduct a randomized trial by enrolling patients with HG, but stratify by TN status.