USB cables and Galvanic Isolation

USB of newer DACs is asynchronous, meaning incoming data clock and D/A conversion clock are independent.  Sound cannot be affected by the timing of the data, as it is with S/Pdif,  but still can be affected by injected electrical noise (that affect internal DACs clocking).  To limit amount of this electrical noise,  cables with data lines only (no power) are better, when power is not required (DAC has own supply).  Good shielding also plays role to limit ambient electrical noise pick-up.  Galvanic isolation separates two circuits electrically to prevent current flow between them.  It can be done with transformers or optocouplers, but in either case it is not perfect, since there is always very small capacitance between two sides of galvanic isolation, allowing very high frequencies to pass thru.  Nothing is perfect, but galvanic isolation eliminates most of electrical noise.  Another good step would be to reduce noise on the source (computer) side.  That might be difficult since some computers have horrible power supplies.

Yes, electrical noise can enter the DAC but is that really relevant if the DAC has a measure of -120 SINAD from the analog output?

Even if this -120dB SINAD was specified for the USB input signal I'm pretty sure they did not use for this specification a noisy computer with typical USB cable, but rather very quiet custom source (possibly battery operated) and inches long cable.  In addition, according to Benchmark FAQ, jitter effects should be reduced 25dB below noise floor since our ears are extremely sensitive to it.  It is because sidebands, that jitter creates, are not harmonically related to root frequency, hence very audible.
Exact quote:

It can be easily demonstrated that most people have the ability to detect and identify tones which are buried 25 dB or more below white noise (A-Weighted). Therefore, it is important to keep jitter induced side bands nearly 25 dB below the A-Weighted THD+N of the converter, otherwise the jitter may become audible.

and the link:

https://benchmarkmedia.com/blogs/application_notes/12142221-jitter-and-its-effects

nicktheknife, If you hear it without signal present, then it cannot be caused by the jitter.  Even with the signal present jitter wouldn't show as one frequency, but as added noise (lack of clarity etc).  It is alway possible that electrical noise injected into DAC by any cable becomes audible when multiple noise frequencies mix on any non-linear element (like transistor or diode) producing lower audible frequency (modulation).  This happens on analog side after D/A conversion.  Since it happens in one channel only it is likely a faulty part (like tube), but finding it might be difficult.  It is also remotely possible that right speaker cable (or speaker) is next to something that induces electrical noise into this channel only.

USB generates an unpredictable set of outcomes depending on the implementation of USB ports of the server and DAC

Asynchronous USB can affect sound only by the fact that it is connected to something, like computer (injecting noise).  The same applies to any other cable supplying digital data.  S/Pdif coax, connected to computer, can be far worse IMHO.  Ethernet might be slightly better, because of galvanic isolation, but still is not perfect.  Even Toslink, that has no electrical connection can make things worse by amplifying existing internal noise (slow edges).