I believe the CFI where I was based said that descending quick (cruise speed almost) should generate enough friction on the leading edge to keep it from icing in those conditions.
Let's say you're cruising at FL50 at ISA-5°C. The static air temperature would therefore be +0.1°C. According to a very handy application which I use, the total air temperature at 105KIAS will be about +1.7°C; if you increased speed to 150KIAS it would be +3.5°C and at 180KIAS it would be +4.9°C.
So I suppose that a little ice formed at +0.1°C on the leading edge
might be persuaded to melt and detach with an increase to 180KIAS - the vibration and buffeting of the average spamcan might also help to shake it loose. On the other hand, of course TAT is really only accurate at the stagnation point, so that nicely melted ice might simply dribble back and freeze again on a cold-soaked aerofoil...
??