Maximum frequency is set by acceptable losses, and "breakup" caused by additional non-TEM0 modes. Wiki is probably erroneous. Maybe their source explains in better detail (anyone got that edition handbook?), or maybe it's also erroneous.
Losses are determined by construction, generally with larger build having lower loss, and also smoother metal faces (especially silver plated), and higher-Q dielectric (or the lack thereof -- as with foam, spiral or other sparse supports). The design choice is generally to use the largest line you can, up to the TEM cutoff, given cost, loss budget, and other considerations like available space, flexibility, etc.
"Cutoff" is not that it stops transmitting signals, but that it stops being as well-behaved of a medium -- specifically what happens is, as higher order (waveguide) modes come into play, their impedances and velocities are all different, and so the signal arrives at its destination over a range of times -- it becomes smeared out, dispersed; hence the property is called dispersion (when velocity depends on frequency).
Tim