Full text is open access at
http://www.cell.com/cell/fulltext/S0092-8674(17)30130-7 . Images and methods are linked via tabs near the top of the article.
Main points:
- This research involves mice, not humans in any meaningful capacity
- It uses a "diabetes model", which is not going to be the same as diabetes
- Fasting consisted of 50% of standard calories on Day 1 & 10% Days 2-4
- Fasting was followed by up to 10 days of unlimited normal rat chow
- Fasting cycles apparently continued for a bit over 10 weeks (6 cycles)
The first fatal flaw with this research is in the Methods section, which they've hidden on a separate page from the main body of their paper. They claim to have induced Type 1 diabetes with a single injection of streptozotocin, whereas it actually requires
multiple doses. They didn't give the mice Type 1 - they gave the mice hyperglycemia
Accordingly, the Type 1 results are already completely meaningless, and there's probably no need to look for further flaws regarding the Type 1 model.
"Type 2" mice were diabetic due to a genetic mutation which impacts leptin, and results in the mice grossly over-feeding themselves. Again, it doesn't sound horribly similar to how Type 2 typically occurs. These mice have elevated insulin at 2 weeks of age, obesity at 3-4 weeks, followed by severe depletion of beta cells (age?), and death by 10 months of age. The mice in the study were aged 10-12 weeks in the pre-treatment monitoring period, then assigned to the fasting or control group.
One potential problem with the Type 2 mice is that they were only 12 weeks old when the fasting and refeeding started, hence their beta cells were still in pretty good shape. Beta cell function was not directly assessed, but calculated based on fasting glucose and insulin levels. The mice were killed, presumably at 150 days after baseline at an age of 7.5 months. Accordingly, it was not shown if the diet resulted in longer life span than expected for that type of mouse, though one graph suggests that 45% of non-fasting mice died prematurely, aged 4.5-6.5 months. There's no mention of the deaths or the actual rates and ages in the text, which is rather suspicious.
The numbers they show for supposed proliferation of new beta cells is a bit dodgy, since the standard deviation is quite large - meaning the values for at least some mice were likely much, much lower.
The fasting mice, while doing better than the unlimited over-eaters, still weren't looking too good by the end. Day 90 shows control mice with no insulin, but the fasting mice are barely doing any better. Plus standard deviation isn't shown at that one time point, which is interesting to say the least, and measurements from day 120 and 150 are omitted, even though those days are included in other results. Fasting blood glucose is around 300 (16.7) around day 135, which is pretty dismal, and appears to be trending upward - no data for day 150. In general, glucose and insulin values are not shown at consistent intervals, meaning the researchers may have selected the days which made the fasting group look best. Values are also not clearly matched up with dietary periods (fasting or refeeding), despite the description claiming they are - there don't appear to be clear refeeding periods which should be twice as long as the fasting period, and all of the intervals look about the same.
Glucose tolerance showed the same rise and decline for the fasting rats as for the control rats, just with a lower starting point. Fasting rats showed a better reaction to insulin, but that doesn't support the suggestion that this diet potentially results in no need for insulin.
There's also no mention of correcting the threshold for statistical significance, to take into account the likelihood of false positives when making numerous comparisons.
At best, this is an extremely preliminary study. There is absolutely no way that anyone can speculate that it would be applicable to humans with Type 2, and it doesn't involve Type 1 at all, despite claims to the contrary. I rate this study a big fat "MEH"
