There are a couple of different threads on the whole question of autoformers vs. direct-coupled approach for Mac power amplifiers - most of the arguements are very much parallel to such other unsolveable audio debates, especially the transformer vs. transformerless one.
But a couple of points specifically about McIntosh . . . the main thing is, the autoformer and autoformerless amps aren't two totally different design approaches, in fact within a given generation, the rest of the circuit design is virtually identical.
Virtually all solid-state amplifiers exhibit a point where, as the load impedance goes down, distortion starts rising fairly quickly - this is due mainly to intrinsic semiconductor physics. An amplifier designer can make some choices that affect at what impedance this starts to happen, but as with everything else in life . . . there are tradeoffs. An autoformer can be used to alter this curve as well, and it seems that McIntosh usually prefers the set of tradeoffs associated with the autoformer over some other choices they could make in their output stages.
The MC7200 and its ancestor, the MC2002, are the oddballs in the Mac amplifier line, in that they are higher-end amps that don't use autoformers. My guess is that these amps were engineered at a time where many in the market felt that an amplifier should have a very high advertised "damping factor", and McIntosh couldn't deliver a high number with their current autoformer technology - so direct-coupled amps were developed to satisfy this. However, McIntosh has revised some of their feedback techniques with the autoformer amps, and their recent (autoformer) amps have damping-factor specifications similar to direct-coupled amps like the MC7200, so they probably feel that a high-end direct-coupled amplifier is no longer a necessary part of their line.