CJ,
#1. I simulated about 11 dB return loss @ 2.4 GHz -- L252, C253 & L253 form a LPF. Also, at these freqs you have to include the PCB traces and of course component parasitics.
#2. If you don't use the same vendor for the parts and the same P/N's (they used Murata C's & L's) and if you don't use the recommended layout, you will probably need different component values for best output power & eff. The device impedance is the complex conj of the 70 + j30 load impedance so it's 70 - j30 which is 82.86 ohms in par with 343 fF @ 2.4 GHz. The 18 pF caps are coupling caps but could be adjusted. I did a quick sim of the impedance matching parts (L251, C252, L261 & C262) and found 2.2 nH and 0.8 pF worked better. Of course this was only a simulation. It is best to keep the L's & C's the same for symmetry which is what I did. The match return loss was 18 dB at the node that feeds the T-LPF.
BR,
Eric Hooker
RF Consultant