This has to be a quite dirty procedure and I think that the route you have in mind is promising.

The sugar-aa conjugates can be expected to be soluble in absolute ethanol or acetonitrile. Therefore a HILIC extraction might work for you.

1. Dilute and your sample with cold acetonitrile (to 80-90%)
2. Remove the precipitate by centrigugation (filtration)
3. Apply the supernantant on a HILIC sorbent (eg ZIC-HILIC SPE). Then the non-polar components and salt (NaCl) will just pass through while polar are trapped.
4. Elute the polar compounds with water/buffer
5. Continue with derivatisation etc (or inject a small volume on a HILIC column for direct analysis ie MS, this works extra good if the analytes can be eluted from the SPE using >50% ACN)

6-10 validation... check that the analytes does not co-precipitate in step 1

Good luck!