That is a good question, and I too would like to know.
The newer versions of Exchange support ReFS, and for the most part I don't see why SmarterMail wouldn't work with ReFS unless they are using Extended Attributes, Short Names, Object IDs, Transactions, or Hard Links.
Even if SM doesn't use any of those and does support ReFS I personally would be hesitant to use it for anything other than Hyper-V as it is created specifically for it and StorageSpaces (for which it works really well in both cases), as it is known to have issues with SAN, iSCSI, or even Raid Controllers not in pass-through mode, leaving the potential for total data loss in those scenarios. Specifically for SM I'm not sure if the performance boost from using 4K or 8K blocks would be enough to offset the performance loss for calculating checksums for every .hdr, .eml, and .grp file that is written. For Mail Servers with a high volume of traffic I could see it potentially resulting in a performance hit to get that extra layer of self-healing resiliency...but if you have the spare CPU cycles that may be a non-issue.
It will be interesting to find out what SmarterTools has to say about it.