The Daily Bolus

The weekend news has produced a rare mix of genuinely interesting diabetes science alongside the usual commercial background noise. Notably, some of the most consequential developments did not originate from device manufacturers or glucose-lowering drug pipelines, but from immune biology, metabolomics, and long-term disease-modification research.

Disease Modification: Immune and Cellular Approaches

NextCell ProTrans: Six-Year Durability Signal in Type 1 Diabetes

On 15 January 2026, NextCell Pharma AB released six-year follow-up data from its ProTrans study in newly diagnosed type 1 diabetes [1]. ProTrans is a single-infusion allogeneic mesenchymal stromal cell (MSC) therapy, designed to modulate immune activity rather than directly replace insulin or beta cells.

According to the company’s data, treated participants continue to demonstrate substantial preservation of endogenous insulin production, measured via stimulated C-peptide, well beyond the expected decline seen in untreated individuals. The separation from placebo appears durable over multiple years, suggesting a disease-modifying effect rather than a transient honeymoon extension.

This positions ProTrans alongside teplizumab as one of the very few interventions to show long-term biological impact on the autoimmune process in humans, albeit via a broader immunoregulatory mechanism.

Important caveats remain. The data are company-reported, sample sizes are small, and independent replication is still absent. Nonetheless, six-year follow-up in type 1 diabetes is rare, and the signal is difficult to dismiss as noise.

Reference:

[1] NextCell Pharma AB. Six-year data demonstrate durable disease-modifying effect of ProTrans in type 1 diabetes. Press release, 15 Jan 2026.

Physiology-First Data Science

Identifying Sub-Phenotypes in Type 1 Diabetes

An academic preprint released on 17 January 2026 applies modern machine-learning methods to CGM and laboratory datasets to identify distinct metabolic phenotypes within type 1 diabetes [2].

Despite similar HbA1c values, subgroups differed markedly in insulin sensitivity, glycaemic variability, and cardiometabolic risk markers. The authors argue that treating T1D as a single homogeneous condition may obscure clinically relevant physiology.

This has implications for automated insulin delivery systems, most of which still assume uniform insulin kinetics and glucose dynamics, despite mounting evidence to the contrary.

Reference:

[2] arXiv. Deep learning-based metabolic phenotyping in type 1 diabetes using CGM and laboratory data. Posted 17 Jan 2026.

The Diabettech Take

This weekend illustrates the widening gap between managing diabetes and changing diabetes.

On one side, we see relentless optimisation of interfaces: smarter reminders, better prediction of user behaviour, and increasingly polished software layers. These can reduce burden, but they do not alter underlying physiology.

On the other, NextCell’s ProTrans data and metabolomics-based risk stratification point toward something more uncomfortable for the industry: biology matters more than dashboards. Immune modulation, early metabolic derangement, and phenotypic heterogeneity are where genuine progress will come from — and they are slower, riskier, and far less commercial than subscription drugs or disposable sensors.

Diabettech remains sceptical not because innovation is bad, but because too much innovation is aimed at numbers rather than mechanisms. This week’s most interesting stories did not promise convenience. They hinted, quietly, at durability and variation.

And durability is the one thing diabetes care still lacks.