Delabialisation in Germanic

I have collected some examples of *kW, *gW and *gWH losing their
labiality when adjacent to an *o in the ancestor of Germanic (a similar
loss next to an *u is well-known and less interesting). They include
such items as *gWolbHo- > *kalBa-, *dHogWHo- > *DaGa-, *kWól(h1)so- >
*xalsa-. It has struck me that the *o in all these cases seems to be
"Rasmussenian", i.e. an element of infixal origin. There is no
delabialisation in pronouns such as *kWod > *xWat- or generally before
thematic *-o-, cf. snoigWHo- > *snaiGWa-, or in the o-preterite (Gmc.
*kwaþ- < *-gWot-). The only clear counterexample I can think of is
*warma- if from *gWHormo-, but in the light of Gk. tHermó-, Arm. jerm,
Alb. zjarr one would expect *gWHermó- > PGmc. *GWerma-, which casts
doubt on the significance of this exception.

Greek doesn't seem to display such a rule. To be sure, we have boukolos
from *kWolh1o-, but *gWHon-ah2 > pHone: 'carnage', *kWolh1-o-> polos. I
wonder whether anything of the kind has been observed in Celtic or
Italic. Is it thinkable that the distinction between infixal *O and
"plain" *o survived long enough to cause dialect-specific changes?

Piotr