Surprsingly I could not find information on this. Many lenses have vignetting profiles available. Vignetting changes if you use a cropped sensor, or, in case of zoom lenses, can change with zoom.
Does the T-stop as measured by manufacturers (those that state the it) or others (such as DxOMark, who don't mention vignetting) account for it? Since vignetting is not uniform, this could affect meaningfully using it to adjust exposure.
In the same vein - vignetting changes with aperture, but is it usually proportional? If you observe this vignette chart on this page http://3d-kraft.com/index.php?option=com_content&view=article&id=127 (Ctrl+F = Fall off / Vignetting), different lenses have wildly different rates at which the fall off changes. How does that play into using the T-number?
Finally, if we forego my concerns about the variable effective transmission with aperture and zoom, I had a very amateur question about the T-stop. For the lenses that don't write the T-stop on the aperture ring (i.e. most lenses under $10,000), how do you use it when closing the aperture? Let's say I have a lens that's nominally f/2.8, but the T-number is f/3.1. If I close the aperture to f/4, would the resulting T-number be simply (4+0.3)=4.3, because it's the logarithm, and the EV value is just shifted? Or is it more complex than that?