The tully-fisher relation is an empirical relationship IIRC used to calculate distances to galaxies. You don't know that the tully-fisher relation applies in exactly the same way to a class of galaxies that are characteristically different from those galaxies which provide the dataset for calibrating the distance (IIRC, this is done by using cepheid variable stars). By necessity, UDGs are an extrapolation of the tully-fisher relationship. Suppose the tully-fisher is off by two digit percent due to UDGs being strange -- then it's possible that our calculations about the galaxies "dark matter content" could be completely wrong.