Yes, the difference is simply whether the geometry is microstrip or CWPG.
Magnitude of edge coupling is quite small: typically on the order of ~10%, at design-rule spacing, for nominal-impedance traces (it increases somewhat for higher impedance and vice versa), so for example you can have a microstrip at 50Ω, bring in pours all around, and it drops to only 45, maybe 40Ω. Or the trace width (to maintain 50Ω) shrinks only slightly (whatever it is, 5-20% ballpark) for CPWG vs. microstrip.
There is some effect on radiation. Obviously, the trace is visible from the outside, radiating fields couple to it; the trick is, the coupling is normally quite small. It does vary with frequency and trace dimensions (width, height above substrate, electrical length), but generally it's in the... -40dB or better sort of range, I think? That is, say: comparing the radiation from a trace over ground plane, to a dipole of equal dimensions but the ground plane removed. Despite the trace being line-of-sight visible, most of the field lines terminate into the ground plane, or circle around very near the trace. The effect with ground brought in (CPWG) is likewise fairly modest. I *think* stronger than (but still ballpark comparable to) the edge-coupling (self-impedance) change, but, I haven't seen a reference to this exactly. (There is a NIST(?) microstrip radiation test paper you can find, however, which illustrates the radiative coupling much better than my wave-of-the-hand.)
In short, it's a small enough effect that, unless you really need that little (10 or 20%) edge in performance -- don't worry about it.
Commercial applications these days, rarely if ever use shielding (cans or enclosures), as the radiation from 3.3V LVCMOS traces, with up to 100MHz clock speeds and couple-ns rise times, is modest enough, at least if you aren't running it around and across a big whole board, or between connectors and cables; and LVDS (generally speaking; exact examples including LVDS per se (EIA-644), USB (High Speed and up), PCIe, HDMI..) is used widely for faster rates. As the name suggests, Low Voltage Differential Signaling both uses less signal strength (fractional to 10s mA, 100s mV) and is routed as differential pairs, canceling out noise (and radiation!) and making the small signal level practical for fast and reliable digital transmission. The main exception is when working at serious RF -- microwave frequencies, usually for a radio as such (WiFi etc.), where a can (even without the cover placed) may be needed for adequate shielding of internal elements.
For something like an RF amplifier, it might not matter, or the shield can is mandatory for various reasons (interference, coupling to other stages that would otherwise cause oscillation, receiver images, etc., or undesired radiation amount/pattern/etc.) and so the minor effect of CPWG is irrelevant compared to the major shielding of the can/enclosure (>100dB is achievable fairly easily).
Tim