If the S/PDIF to AES converter uses a (good) transformer, I don't see why it would necessarily pose a problem. There even are some potential advantages, but whether the tradeoff of more complicated signal path vs. these advantages are worth it depends on the entire chain.
Having a good transformer in the path will offer two advantages:
1) isolate the incoming signal if not already isolated
2) filter RF noise if the transformer is well designed to operate only in the appropriate band
3) jack up the signal swing. S/PDIF is designed for 1Vpp when terminated, wheras AES can go much higher (up to 10Vpp). This of course leads to improved CMRR and higher noise immunity, and this gain is had for free when using the correct transformer and termination.
On the other hand, cheap transformers have high parasitic capacitance and bandwidth limitations: often a regular Ethernet 'pulse' transformer is used which is great for Ethernet speeds, but S/PDIF works at lower frequencies too and therefore they're not as great. Choosing the correct transformer is therefore rather important and only a few companies make them. Scientific Conversion, has some of the best, but hard to find and at a premium price. For AES to S/PDIF one will anyway need a custom transformer dedicated to the job, or using two and an active buffer.
I'm not sure if true 75ohms is really that important TBH, especially since I haven't seen any PCB's that accurately follow the transmission line similarly all the way to the receiver IC. This is especially true with products where the I/O module is separate from the mainboard, where 2 PCB layers alone are enough to get the job done, but not maintain the impedance. It's not that it cannot be done (otherwise measuring gear or even a DSTV decoder would never work), it's just that nobody bothers.
AES uses XLR which definitely is not 110ohms, and many folks prefer AES over BNC S/PDIF which (can) be full 75ohms.