I don't think any "nginx + apache VS nginx" (or "nginx +apache VS apache", or any multiple servers vs single server) statistics do any good or give any kind of trustworthy information. The reason is that you can have nginx + apache set up in a million ways (well, not quite, in about three or four ways) and performance wise it all boils down to what it serves.. That's why this kind of stats aren't done frequently (if at all).
You should take a look of what each web server does best and fit it to your needs.. There are setups of nginx proxying to apache for php (which, personally I don't like), setups where nginx is used as a load balancer and apache (or ligthttpd, or even nginx) as the actual servers, setups of nginx serving CDN and apache (or another server) serving anything else, or even more complex setups.
So you see, each setup would give another set of stats, again, depending on the content served and connections... The simple "Hello world.php" would not be of any values in this cases.. The one thing that is clear in all stats (if you didn't yet see this) is that Apache needs quite a bit of extra memory to handle many concurrent connections (which is because of it's design).. That's it, for all the rest you should really do your own testing (like vl23 already suggested) on your one application.. Or compare them server vs server (which, like I said, are easy to find on Google) on a category that matters to you.
Anyway, I've searched for some discussions that cover "nginx as a reverse proxy for apahce/mod_php VS nginx + php-fpm":
https://groups.drupal.org/node/27082
http://serverfault.com/questions/298...oxy-for-apache
I don't see anything new in there, but you know, maybe you would like to browse over them...
Hope it help..