To add to that, this chart shows increasing noise factor with frequency for several American UHF triodes:
I’d guess, and it’s only a guess, that the PC88 curve would be proximate to that of the 6BA4. I doubt that its gain as an RF amplifier would be sufficient that PC86 mixer contribution to the overall noise factor could be neglected, so on its own it would need to be somewhat better than 10 dB at 500 MHz and similarly better than 15 dB at 900 MHz.
The 10 to 15 dB range previously shown for the PC88, PC86-based UHF tuner probably means 10 dB at the bottom of Band IV and 15 dB at the top of Band V.
The frequency effect is also observable for the VHF tuners, in that the Band III noise factors are higher than those for Band I.
Cheers,