Firstly I'm assuming buildings 1 - 4 are chained due to cabling because obviously that is not great in terms of redundancy if for example building 4 went and took out 1 - 3 with it. Assuming it is cabling then if you were to use multiple areas you couldn't put each of those buildings (1 - 4) into it's own area because each area needs a direct connection to the backbone area.
So it's really a question like you say of using one big area or one area per building with the above proviso.
Personally I would use areas simply because if everything needs to go back to the DC and there are no local resources then it seems inefficient to me to have to propagate the LSAs for every network across the entire campus. Not only is this unnecessary in your setup but it is also an overhead the network and network devices simply don't need.
So I would put buildings 1 - 4 into an area and then the rest of the buildings each have their own area. If you have planned the IP addressing so you can summarise from each building then you simply advertise these summary ranges to the DC.
And as all traffic has to go back to the DC you only need a default route to be injected per area as there is only one way in and out of each area.
That is the way I would approach it but I haven't done design for a while so I am sure there are other ways to do it and hopefully others may chime in :)
Jon