If you're going to report concentration for your analyte, and you want to use response factor calculated from the internal standard as an answer, then you probably should also use internal standard as a concentration. If you're just going to report the response factor as some sort of dimensionless number you get from a ratio of analyte and internal standard, then I suppose you can get away with constant volume.