Obviously, the biggest difference between the two studies is that DEFUSE 2 was a cohort study and MR-RESCUE was a randomized clinical trial. DEFUSE 2 used NIHSS change as an outcome, MR-RESCUE used the modified Rankin Scale score. These things really do matter.
The methods of determination of "territory at risk" also differed between studies, with DEFUSE 2 performing a trace-method of both MRI diffusion and MRI perfusion imaging, called RAPID. MR-WITNESS used a block-permutation algorithm using CT and/or MRI data to determine mismatch.
The times to enrollment in MR-RESCUE were long (5.5 hours or so) biasing towards the null. A similar time to MRI in DEFUSE 2 was about an hour shorter.
It is overall difficult to compare the studies. There are many inherent biases in interventional cohort studies that are difficult to capture, and given the relatively small sample sizes being used, differences in effectiveness of reperfusion, location of embolus, etc. can exert substantial impacts on the results. Finally, the actual techniques used to determine whether or not there was a mismatch differed between studies.
One thing we have always known about endovascular therapy is that picking the right patients is crucial. The issue with that is there is no consensus on which technique should be used to do that. It's constantly evolving along with the devices we use to perform the clot extraction.
At our facility we use infarct core volume and ASPECTS score to determine candidacy, and we are not currently using MR-P or CT-P in most circumstances. However, I've seen enough DWI reversibility to have misgivings about even that sometimes.
The final word has certainly not been uttered as it pertains to patient selection for endovascular therapy, but MR-RESCUE and DEFUSE 2 are different enough in their approach that I would not call them contradictory to one another.
