The current issue of RT has an entire article devoted to this, and in fact uses St. George as an example. Here's the pertinent stuff:
Another contender for fastest course in the West is the St. George Marathon, which drops 2,560 feet. But the steepest (and potentially fastest) of all is the now-defunct Pines to Palms Marathon in California, which virtually falls off a cliff with a total elevation loss of 4,920 feet.
In theory, descents that large should speed you up by four to eight minutes — a truly enormous assist. For Tucson, the gravity boost is almost precisely 4 minutes for 5-minute milers, 6 minutes for 7:30 milers, and 8 minutes for 10-minute milers. At St. George, it’s roughly comparable because, while that marathon has 400 more feet of downgrade, it also has one significant hill.
But two other factors make it unlikely that most runners will see such large gains.
One is simply that downhill courses can be rough on the quads. People in my running club always think they’re going to run PRs at Boston (a much more gentle downgrade) and only one has ever succeeded. Even if you train for them, downhills are deceptively tiring.
But a bigger problem is the fact that the only way to design a course with a big drop is by starting high. If you’ve not taken a couple days to acclimate, Tucson’s 4,730 feet is high enough to leave you out of breath. St. George starts even higher, at 5,240 feet.
Even if you’re acclimated, those are high enough to slow you down.
Altitude physiology researcher James S. Milledge studied this, publishing his results in the Oxford Textbook of Sports Medicine. The book is hard to find, but the results are summarized by South African exercise physiologist Timothy Noakes, in his treatise-length book, Lore of Running (4th Ed., pp. 574-580).
According to Noakes, Milledge examined the effect of altitude on running performance at distances ranging from 60 meters to the marathon. He found that at an altitude of 2,500 meters (8,250 feet), marathoners were slowed down by about 8 percent.
Noakes also describes a similar study from the Journal of Applied Physiology, in which a team led by kinesiologist Francois Peronnet of the University of Montreal conducted a mathematical analysis of the effect of altitude on running. Among other things, Peronnet’s team calculated that running a marathon in Mexico City (7,400 feet) would slow you down by 7.3 percent.
Both studies also found that up to elevations of about 8,000 feet, the amount of slowdown is directly proportional to the altitude. (Higher, it increases sharply.)
This means that even an elevation as low as 1,000 feet will slow you down. Specifically, each 1,000 feet above sea level will slow you down by about one percent.
Even more than beating up the quads, this is the Achilles heel of the “fast” downhill courses. In order to get the big drop, they have to start high. Even Steamtown averages 1,000 feet above sea level. Tucson averages 3,300 feet — high enough to pretty much negate the benefit of the downgrade. St. George is worse; its average elevation is about 4,500 feet — enough to convert a 3:00 marathon into a 3:08.