Yep, you can definitely do this, though it'll require some fiddling and some understanding of how miniaudio works. You'll need to set up a chain of nodes:

source sound > reverb > ma_sound

To do this, you can load your source sound any way you like. You can load it directly from a ma_decoder, or you can use ma_resource_manager to load if you want background loading, or you can even pre-decode into a raw audio buffer. Whatever you choose, once you have your data source you can create a ma_data_source_node which becomes the first node in the chain. Once you have that, you can initialize a ma_reverb_node and wire up your ma_data_source_node to it. Then you can plug that into your ma_sound object.

Your ma_sound object will need to be initialized differently to what you're doing now. Use ma_sound_init_ex() and pass in a ma_sound_config. In that config, leave the file path and data source as unset, and set the channelsIn member of the config to 2 (since that is the number of output channels from the reverb node). Once your ma_sound is initialized, wire up the output of your reverb to the input of your ma_sound.

Feel free to drop more questions here if needed. As a starting point to get you going, consider first setting up your ma_data_source_node and get that working at a basic level (wire it up straight to the endpoint - ma_engine_get_endpoint()). Then plug in reverb, and then finally ma_sound for spatialization.

Also, if spatialization is the only part of ma_sound you care about, you could also wire up straight to the ma_spatializer object. However, this is not a node, so if you want to use the node graph technique you'd need to wrap it in a custom node. You can also bypass all of the node stuff entirely if you want to go even lower level - lots of options and flexibility depending on how low level you want to go.