I'm not an expert on whether and by how much certain kinds of training can shift the substrate utilization ratios at marathon pace, but I would like to point out one thing: for most runners (skinny; reasonably fit), running 20-22 miles at ~1min per mile slower than marathon pace WILL NOT fully deplete your glycogen stores.
The ratio of fat to carbs that you burn is proportional to the relative intensity (% VO2 max) that you are running. If you're running a 5k (nearly 100% VO2 max), over 95% of your energy is coming from carbohydrates. But at an easy pace (say 65% of VO2 max), you're only getting about half of your energy from carbs; the rest is from fat. The relationship is not linear; it looks more like this:
http://i.imgur.com/9zvRsas.png
For details on this see these two papers:
Rapoport, B. I., Metabolic Factors Limiting Performance in Marathon Runners. PLoS Computational Biology 2010, 6 (10), e1000960.
Romijn, J. A.; Coyle, E. F.; Sidossis, L. S.; Gastaldelli, A.; Horowitz, J. F.; Endert, E.; Wolfe, R. R., Regulation of endogenous fat and carbohydrate metabolism. The American Journal of Physiology 1993, 265 (3 Pt 1), E380-E391.
But anyways, the bottom line is that most distance runners carry enough carbs to go more like 27-29 miles at an easy pace. If you want to deplete glycogen in training, you must either a) run after fasting, perhaps overnight—many runners do this unintentionally—or b) run at a high percentage of your VO2 max (i.e. faster than just easy or moderate).
One final thing to keep in mind is that you will start to feel the effects of glycogen depletion far before you actually RUN OUT of stored carbs. This is probably because of a central-governor-like pre-emptive pacing strategy, but whatever the reason for it, you'll start to slow down (and, presumably, start to get a carb-sparing training stimulus) when your carb reserves drop below about 30%, IIRC.