I have a Gustard R26 DAC connected both with Ethernet Cable to my LAN and USB 2.0 QED cable to a MacBook PRO.
Audirvana Studio can be configured in both ways (UPnP or USB 2.0).
Which is the best solution in order to get maximum audio quality ?
Thanks all in advance
Oh this isn’t controversial at all. 
I have an optical Ethernet LAN DIY solution I think sounds excellent, other folks use direct connected USB and really love it.
I’d just say play around with inexpensive Ethernet and USB cables and see what sounds good to you.
Not too sure about maximum ‘audio quality’, but would say for maximum ease of use a direct USB connection would be better. UPnP and Audirvana aren’t really happy bedfellows in my experience.
@Ironz has a point. Manufacturers have done lots of UPnP variations, so it’s difficult for software developers to cope with all of them. That’s one reason my solution is DIY, so I know the UPnP software, which I installed, is plain vanilla and easy for Audirvana to communicate with.
Welcome aboard @Mahatma63 , USB is probably the easiest most robust way to go. As @Ironz says UPNP can be a crapshoot depending upon your device, I don’t remember anyone mentioning the Gustard specifically. @Jud is spot on give it a go both ways and decide what works best for your particular setup.
Have fun with it!
I suggest not connecting both simultaneously as there is a potential for grounding/earthing related noise due to the components being connected to different ground/earth potentials… Typically called “ground loops”…
That’s why you would use Cat6a Ethernet cable, for example, where there is no ground connection (at least if the cable actually meets the Cat6a spec).
Cat 6 UTP with twisted pairs
Very simple to tell whether your Ethernet cable is in spec (no ground connection): Just use a continuity tester you can get from a hardware store and see whether there is continuity between the cable terminations. If yes, no good; if no, you’re OK.
Doesn’t no continuity mean it’s broken?
You’re measuring continuity at the ends (but not the wires themselves), where some Ethernet cables have plastic (in which case you hardly need to), but others try to look fancy and have metal. What you want to see is whether the folks who are trying to look fancy have screwed up and left a ground path through those nice looking metal bits at the ends.
ok, get it, you’re testing between contracts at one end?
Both ends, not the actual contacts. 
Here’s a fancy looking cable with a bunch of metal cladding at the end. You want to test continuity between those bits of metal cladding (from one end to the other), not the actual contact wires.
Looking for the shield or shell connection to be open no continuity.
The cable insulation should be marked UTP if it’s an open shield cable
I do like the gold plated toslink cable ends 
really improves the sound
I see… at last!
Silky Smooth Sexy Sound 
This raises points about cables I’ve made before. First, where the cable looks fancy but it’s out of specification because there’s a ground connection, it may sound different because it’s conducting noise over ground. Of course if you hear a ground hum you know that’s bad, but just a little noise at the threshold of audibility can create a sensation of excitement or warmth. (In engineering Springsteen’s Born to Run album, Jimmy Iovine used something called the Aphex Aural Exciter - a noisemaker!)
Also, some people may be tempted to get Cat7 or Cat8, because 7 and 8 are higher than 6, obviously. But Cat7 and Cat8 are shielded, meaning a ground connection, which you don’t want. And the Cat6a spec is actually newer than either Cat7 or Cat8. (You want the unshielded flavor of Cat6a.)
The ground thing can go either way in my experience. If there is a problem you’re going to hear it usually. I’m a big fan of twisted conductors, we use a lot of this in RF and Instrumentation applications. Really reduces crosstalk and induction of spurious signals.
If your going to use grounded interconnects good idea to DVOM shield connection to ensure no major current flow attempting to complete a power circuit.