There are two aspects to separating the different bands of cables. One is to prevent mains voltages from getting into telecomms / data circuits in the event of damage; this can be overridden if the data cables are insulated sufficiently for the highest voltage present. E.g. although a bell circuit might use 12V itself, if it is wired in a mains-rated cable there's another layer of adequate insulation between its conductors and the mains conductors and they can run together. Obviously this aspect doesn't apply to the powerline data adaptor because it doesn't have a cable of its own.
The second aspect is preventing interference. Data and comms interfaces are designed to be as resistant to incoming interference as possible, so that fields surrounding the cables don't disrupt the signal. But, where power and data run very close together, the field strengths can be high and some interference is likely. However, the signal encoding and modulation system used in the powerline adaptor is designed to work in an extremely electrically noisy environment, not just being near to mains cables but actually superimposed on the mains. So, it doesn't benefit from segregation because there's nothing worse to segregate it from.
They are a fudge, and will never offer the data performance of a dedicated data link. They do the best they can with the available connectivity, by sacrificing performance for interference-proofing.
