In a study of 5,630 adults with type 1 diabetes, the ones whose pancreas still made a little of its own insulin had fewer dangerous blood-sugar crashes, fewer emergency cases of acid building up in the blood, and less eye damage over the next decade. The benefit held even when their average blood sugar was the same as everyone else's.
That last part carries the finding. The protection did not come from running tighter numbers. It tracked with how much insulin the body could still produce on its own, a quantity measured by a small protein called C-peptide ↗.
C-peptide is the offcut of insulin production. The pancreas builds insulin as a larger molecule, proinsulin, then snips out a connecting piece and releases it alongside the finished hormone, one molecule of C-peptide for every molecule of insulin. Injected insulin contains none of it. So the amount of C-peptide in the blood is a clean readout of how much insulin a person is still making themselves, and nothing else. In type 1 diabetes, where the immune system destroys the insulin-making beta cells, that number is usually low. But it is rarely zero, and it varies a lot from person to person.
What the cohort showed
The data come from the Scottish Diabetes Research Network Type 1 Bioresource, or SDRNT1BIO, a population cohort of people with type 1 diabetes who were at least 16 when they enrolled. Their diabetes had lasted a median of 21 years. The team, led by Joseph Mellor and Helen Colhoun, measured C-peptide from a single untimed blood draw using a Roche immunoassay sensitive down to 3 picomoles per litre, then followed the group for a median of 10.8 years through linked electronic health records. They published the results online June 11 in Diabetes Care ↗.
Three outcomes moved with C-peptide, all in the protective direction. Higher residual insulin output was linked to fewer cases of diabetic ketoacidosis, the acute emergency where a lack of insulin lets acids flood the blood (1,015 events, p less than 0.0001). It was linked to fewer episodes of severe low blood sugar bad enough to land someone in hospital (539 events, p = 0.0055). And it was linked to less new diabetic retinopathy, the eye-vessel damage that is a leading cause of blindness in working-age adults (1,172 events, p less than 0.0001).
The key test was whether this was just a roundabout way of measuring blood-sugar control. It was not. The three links survived adjustment for HbA1c, the standard measure of average blood sugar over the prior few months. Two people with identical HbA1c still had different risks depending on how much insulin they were making on their own. In a smaller group of 407 people with repeat measurements, a later C-peptide reading still tracked with ketoacidosis risk, so this is not only about where someone started.
Where the signal stopped
Not everything moved. C-peptide showed no association with cardiovascular disease (714 events, p = 0.9) or with death (566 events, p = 0.8). The benefit, on this evidence, sits on the acute-crisis and microvascular side of the ledger, not the heart-attack-and-survival side. An observational cohort cannot prove cause either way, and the authors frame the work as characterizing associations, not establishing them.
Still, the pattern lines up with a clear mechanism. A trickle of the body's own insulin responds to glucose minute by minute in a way that injected doses cannot. It blunts the glucose swings that drive an acid crisis or a severe low.
That is the case for protecting beta cells rather than only replacing their output. It is the logic behind teplizumab, the Provention Bio antibody now sold as Tzield, which the FDA approved in 2022 to delay the onset of type 1 diabetes by blunting the immune attack on the islets. The SDRNT1BIO results put long-term outcome numbers under that strategy. The amount of insulin a person still makes is not a rounding error on the way to full insulin dependence. Across a decade, residual C-peptide tracked with who ended up in the emergency room with ketoacidosis and whose retinas held up.