The difference is that in a "positive" regulator (the regulator regulates the positive side of the input voltage) like this an NPN is used in a common collector (emitter follower) configuration. This means that the voltage gain of the NPN from base to emitter (and note that the emitter is the output) is about 1x. This means that the NPN does not add gain to the feedback loop. This means that it is fairly easy to stabilize the loop.
When using a PNP the emitter will have to be at the input side making the circuit a common emitter. A common emitter has a gain depending on the load on collector. Here the circuit fed from this regulator is the load. So if the load changes, the gain changes ! This makes stabilizing the circuit much more challenging.
Also a common emitter has a negative gain meaning the output voltage is inverted compared to the input voltage. This means for the PNP version you also need to swap the + and - inputs of the opamp for the circuit to work.
Another difference is the minimum voltage drop needed for the regulator to work properly. In this case using a PNP will result in a lower dropout voltage compared to the NPN version.
I'd say that swapping the NPN for a PNP looks like a small change but really, it is not. It results in a fundamentally different circuit. Even if both circuits are still voltage regulators. With the PNP version you run a much higher risk of actually making an oscillator instead of a regulator.