I'd put my bet on the mechanism being simple heat. That never seems to be controlled-for in these high-energy non-ionizing RF experiments.
But I can't think what else could be involved other than thermal damage. Thinking about it, I also wonder how much thermal stress is required to produce damage. If only a small increment in thermal energy at a micro-level or at particular chemical bonds is what happens, it would be a challenge to detect and measure such tiny, localized changes. Then again if the effect (that is, RF agitating some chemical entity) is small enough maybe calling it "thermal" wouldn't be exactly right.
Will be very interesting to see how this line of research develops. In any case, a great question that requires a lot more work to decipher.
Contrast this with employees who worked closely with leaded gasoline, ionizing radiation, asbestos where the statistical signals are overwhelming.
For example, the mouse study showed alterations to myelin sheaths. Conceivably there could be long-term implications, as in leading to conditions like multiple sclerosis. Of course this sort of hypothesis remains to be tested. Late-appearing effects are notoriously difficult to establish as consequent to small insults years or decades before.
Remains to be seen where this line of research goes. I sure don't expect to see any outcome of it in the near future.
Could cell axons or other structures be acting like (very imperfect) antennas?