Postpositive cum is rather unique in Latin in this regard (but not for PIE - see below), although there are some other postpositive uses found in Latin; they are well-known, e.g. Leumann mentions quo-ad (cf. ad-huc or ad-eo), see Fortson 2010b: 136 for more examples.
Basically, there are two approaches - either postpositive cum is an archaic holdover (communis opinio for decades - e.g. Clackson 2004 or s.v. com-, con-, co-; cum in de Vaan) or innovation (a new approach proposed, for instance, in Fortson 2010b).
After all, we usually don't reconstruct prepositions for PIE (see any textbook on PIE or historical linguistics, e.g. Beekes 2011: 245 or Fortson 2010a: 154, section 8.8), so it is quite common to trace IE prepositions back to PIE adverbs.
cf. Ivanov 1999 "It can be suggested that the Indo-European particle/adverbial element *kom could be used both as an enclitic or as a proclitic"
Fortson 2010a "From the comparative evidence it is not entirely clear whether these forms were prepositions, or rather occurred after their objects as postpositions; probably both patterns were current, though many researchers assume postpositional usage to be older. [...] In the other older IE languages [i.e. not Anatolian or Vedic - Alex B.], prepositions are the rule, although some (e.g. Old Persian, Greek, and Latin) evince limited postpositional use, like [...] Latin mecum."
Baldi 2017 explains this very clearly. When such particles (adpositions) combine with nomimals in PIE, they follow the noun they modify, whereas when they combine with verbs, they precede the modifying verb (p. 814). He also argues that data from other Italic languages (e.g. Umbrian), combined with Latin and Oscan data, "provide significant evidence for postpositional Proto-Italic" (p. 814).
see my other post
Since there are some prepositions in Latin that don't behave quite like typical prepositions, Weiss prefers to talk about adpositions instead (e.g. pp. 460-461).
NB: Weiss also adds that "[p]ostposition after pronominals is common with disyllabic forms like inter, propter, contra, etc." (footnote 45, p. 460).