I am not talking about plausible sounding medical/scientific rationalization. I do believe there are fundamentally sound subgroup analysis. For example, squamous/non-squamous subgroup is different from metastasis/no-metastasis. I guess that's what separates this board from others :-).