Yes, in version conn12p you can perform second-level analyses of graph theory measures by first defining your second-level model in the 'second-level results' tab (e.g. for a two-sample t-test select 'control' and 'patients' in the 'subject effect' list, and enter a [1,-1] contrast in the 'between-subjects contrast' field), and then clicking on the 'graph theory' button. On the new window you can then select the subset of ROIs that you wish to include in your network (by selecting the 'network of selected ROIs' option on the dropdown menu), select the graph-theory measure of interest (global efficiency, local efficiency, or cost), and the type of inferences (p-value threshold, FDR-corrected or uncorrected, one-sided or two-sided) you wish to obtain from your second-level analyses (in the example above, this will highlight any ROI that shows significant between-group differences in the selected graph-theory measure).
 
The only difference with version conn13a (which we are about to release) is that in the latter we have also added the ability to analyze mixed within- between- subject designs (e.g. if you had multiple within-subject conditions and wish to explore the condition x group interaction) for graph-theory measures (but if you are exploring resting-state connectivity you probably will not be needing this)
 
Hope this helps!
Alfonso