They're just optimized and characterized differently.
Transistors intended for digital use focus on optimizing and characterizing switching speed, how well current is blocked when off, and how well current is conducted when on, and other relevant parameters relevant for full on/off operation. They also don't care so much about noise or stability and linearity in the ohmic region.
Analog transistors are optimized and characterize for operation the ohmic region. They spend all their time operating halfway between on and off so they don't care about switching times or how well the transistors blocks or conducts when fully off or on. Instead, they care more about bandwidth, low noise and stable, linear, predictable, consistent behaviour in the ohmic region.
For example, since digital operation is full on or off, it means that if you make it very very efficient when fully on you can use less silicon. But this doesn't work for analog operation since operation in the ohmic region means you have to dissipate the same amount of power in the transistor no matter what for the circuit you use it in. It needs to dissipate that power to do its job, so you can't cheat and use less silicon by making it conduct better.
You can press one into service as the other, but since it may not work as well since it's not optimized for it and you may not have sufficient data to determine if it's appropriate since the relevant parameters where not characterized.