I went from the science news article back to the actual scientific paper at http://care.diabetesjournals.org/content/diacare/early/2017/06/13/dc16-2121.full.pdf to better understand what they found in terms of these type 1’s who are still producing insulin.
First some background so the numbers make sense. They used an “ultra-sensitive ELISA” test to look for c-peptide. For this test, they set 1.17 pmol/L as the minimum detection level: anything below was considered “no measurable c-peptide.” To calibrate your understanding of this number, less than 600 pmol/L is considered proof of diabetes, and less than 200 pmol/L is considered severe insulin deficiency. (These are fasting numbers: if you drink glucose, that can stimulate a weak but not completely failed pancreas to desperately produce a bit more insulin and c-peptide.) Now this ultra-sensitive ELISA test had a coefficient of variation of 3.8% intra-assay and 5.5% inter-assay at a measurement of 37 pmol/L. Which basically means the errors are small compared to the levels they are measuring.
Now for the measurements. They had 46 type 1s who had some remaining insulin production: the average amount was 28.6 pmol/L plus or minus 15. That’s about 1/6th or 1/7th of the amount that’s considered a severe insulin deficiency. So it’s quite a meager amount. There were 67 patients who had no measurable insulin production (recall, this means less than 1.17 pmol/L.)
So when the science news article says “almost half of patients produce insulin” it’s actually about 40% of these type 1 patients, and even while fasting so not challenging themselves with a glucose load, on average they were producing about 1/20th of the amount of insulin that would be considered proof of diabetes.
So yes, 40% of the type 1s in that study had some beta cells available for regeneration, but they’d have to multiply their beta cell population by more than 20-fold to get out of the diabetic category. For the other 60%, they are worse off by another factor of 25 (or more.)
I’d say the paper was interesting because it investigates why some 40% didn’t have their beta cells more or less completely wiped out. But in any case, for a cure they would be needing to multiply their beta cell population by large factors, like 100 or 1000. I don’t know whether that’s something that would regenerate all by itself, or whether there would need to be 2 interventions, one to stop the autoimmune attack and a second one to stimulate/facilitate reproduction of beta cells.