As far as I know, data merge can only add text to a frame and only if it's in a field in the record.
I would bet, though, that if the alt text can be included in the meta data of the images, one of the smart folks over in scripting can come up with a way to bring it in after the merge is complete.
Australia doesn't have section 508 so I'm not quite sure what is required by law... but if it's a line of type such as a description, should be do-able.
1) If the images have a description or a title in the IPTC metadata, the data merge frame containing the image could simply have a caption containing either the description or title on another layer (that is possible from captions) and then that layer is toggled on/off depending on what output is required
2) Much the same as above, but instead of using the IPTC metadata, this uses an additional field in the datamerge database. A textbox is put above the image (again, in another layer) and that textbox contains the description from the datamerge database. Just as in the previous solution, the layer is toggled on/off depending on what output is required.
Or am I barking up the wrong tree and need a layman's definition of what section 508 requires?