There is a widespread Eurasiatic theory that puts all these families (except PIE) into one group, the case for common numerals for one and two seems more plausible. I also add Chukchi-Kamchadal family hare as it is considered close to PIE. I do not include word forms which seem not to be related.
PIE Korean Tungusic Burushaski Chukchi PAinu PKartvel Mongolian OTurkic e̯oinom hana ömen hen ənnen hine du̯oe̯ dur đöör tuu đor đuirim tuirem "second" Modern Tungusic languages | Modern Kartvelian languages Oroqen Manchu Even Megrel Svan Laz umun emu umūn đūr đuu̯e dūr žir đor žur
I did not include Uralic here even though it is usually considered related to PIE because the Uralic numeral for two is a cognate with PIE word for a pair (q̆eta̯), and the numeral for one seems to mean "non-paired" and come from the same root.
But if we consider the PIE word for four, q̆etu̯ores, we can analyze it as a compound q̆et-du̯or-es, that is "a pair of twos". Thus the most ancient PIE form of the word for two was possibly du̯or (possibly used with nouns with r/n stem?).