I'll readily grant this can reasonably be construed as suggestive but not "a lot of accumulating evidence", my bad for phrasing it improperly. To me this line of research is raising interesting questions warranting further digging, in a "hm, I wasn't expecting that result going by the received wisdom" way.
Following the sporadic posts of Type 2 patients in online forums self-reporting success at getting off conventional treatment protocols, what appears to come up in common among them is what is currently considered (by the ADA and similar outfits in other nations) radically aggressive methodology to blood glucose control; <100 mg/dL post-prandial is not considered out of line in that crowd, and they avoid spikes as much as possible. A very small number have been at it for 20+ years and well into their 60-70's, and are reporting no long-term disabilities commonly associated with advanced stages of the metabolic disorder. They tend to be in very low (<50 mg/day) carb or even ultra-low (<20 mg/day) carb regimens, or constant ketosis, or varying fasting cycles, with varying amounts of exercise, or a combination of all of the above, with lot of nuances thrown in by each individual. A lot of what they practice directly goes against published large institutional guidelines, but it is really difficult to argue against the end results in their comprehensive blood panels and other bio-markers, so it is a fascinating case to me of the impact of an N=1 / DIY / Quantified Self ethos that increasingly more people are embracing with the aid of increasingly better technology. I find it really exciting that increasingly more laypeople are empirically "science'ing the shit out of this" (to paraphrase "The Martian") with a continuous hypothesis-test-adjust loop upon themselves. It is definitely not science by any conventional means, but as haphazard as it is, it is yielding in a startling number of cases exceptionally better results than the average and mean Type 2 patient experience.