The GS-35A tube is the liquid cooled version of the same tube. Some hams have fabricated their own version of a liquid cooling jacket to replace the fins on the GS-35B:

The designer of this jacket correctly highlights the challenges of selecting and maintaining the liquid medium since it is exposed to the full anode high voltage. Deionized water will require regular maintenance to maintain its insulating properties. Generally a liquid with lower maintenance requirements is selected with the tradeoff being lower specific heat.
While liquid cooling does reduce the in cabinet blower noise, the liquid medium will require cooling in the form of forced air, mechanical refrigeration, or other means. The result is that the noise can be relocated to an area away from the operator, but not entirely eliminated.
You will still require some forced air cooling in the cabinet in order to keep the seals of the tube intact.
With all of that being said, if noise reduction in the shack is the goal, I would design the amplifier to be blower cooled and remotely controlled. A few servo motors, some bandswitch relays and some remote metering seems much more practical and hassle free.