It depends on the size of the board. If the propagation delay is less than the 1/2 risetime it won't make much difference for signal ringing but still can make a difference for RF emissions.
If you read the ST documents and follow the 2 layer Gerber examples and have experience doing this before, then go ahead with 2 layer. However, at extra copper costs, 4 Layer gives higher signal integrity at max speed with 1ns risetimes and lower impedance with the same size tracks and gaps but now the gnd plane is 1/2 the height so 1 mm track and gap may drop from 85 to 50 ohms (est.)
Compliance for VDD = 3.3 V, LQFP176, TA =+25 °C, fHCLK = 168 MHz, conforms
to IEC 61000-4-2 when all the design rules are followed for capacitance and layout. (level 2 is 1kV ESD)
If you need to learn about EMI immunity, here's a useful ref from TI.
Reducing track impedance usually results in better immunity and signal integrity for nanosecond rise times but with 2 layers it is possible to pass or fail but easier to pass and route on 4 layers. It also depends on microcode routines that affect spectrum.