作者
Katsiaryna Prystupa,Andreas Fritsche,Christian Binsch,Stefanie Lanzinger,Łukasz Szczerbiński,Adrian Janucik,Adam Jacek Kretowski,Oana‐Patricia Zaharia,Geronimo Heilmann,Nina Trinks,Maximilian Huttasch,Klaus Straßburger,Gidon Josia Bönhof,Alexander Strom,Rainer Guthoff,Syed M. Ijlal Haider,Alexander Dilthey,Tobias Marschall,Kálmán Bódis,Volker Burkart
摘要
BACKGROUND: Adult-onset type 1 diabetes exhibits substantial heterogeneity. We aimed to investigate whether this variability reflects distinct disease entities or different temporal stages of a shared autoimmune process. METHODS: In this prospective cohort study, we analysed data from adults (aged 18-69 years) in the German Diabetes Study who had clinically diagnosed type 1 diabetes within the preceding 12 months, and from individuals who were positive for glutamic acid decarboxylase antibody (GADA) but had originally been diagnosed with type 2 diabetes. We used unsupervised machine learning based on GADA status, fasting C-peptide, glucose, triglycerides, and HDL cholesterol to characterise clinical and metabolic type 1 diabetes subtypes. The resulting subtypes were compared for aetiological traits (polygenic risk scores for type 1 diabetes and type 2 diabetes, partitioned type 2 diabetes polygenic risk scores, human leukocyte antigen [HLA] profiles, and islet-directed autoantibodies), insulin sensitivity, glucagon-stimulated C-peptide trajectories as a measure of residual β-cell function, and complications during prospective follow-up. Reproducibility was assessed in adults with type 1 diabetes from the German Diabetes Prospective Follow-up Registry (Diabetes-Patienten-Verlaufsdokumentation [DPV]). FINDINGS: Analysed participants in the German Diabetes Study cohort were recruited between June 27, 2005, and Jan 20, 2025 (n=600), and were followed up for a mean of 6·1 (SD 4·7) years. We identified three type 1 diabetes subtypes: two GADA-positive subtypes, and one GADA-negative subtype. A similar clinical and metabolic subtype structure was observed in the DPV cohort, whose participants were recruited between 2000 and 2025 (n=483) and were followed up for a mean of 3·1 (SD 5·0) years. One GADA-positive subtype was characterised by lower BMI, higher insulin sensitivity, and severe β-cell dysfunction (classic type 1 diabetes), whereas the second showed higher BMI, lower insulin sensitivity, and better β-cell function (mixed type 1 diabetes). The two GADA-positive subtypes did not differ in autoimmunity or type 1 diabetes polygenic risk scores, whereas the GADA-negative subtype had lower genetic type 1 diabetes risk, despite no clear subtype-specific differences in HLA risk haplotypes. Longitudinal modelling indicated that the mixed type 1 diabetes group was positioned 6·1 years (95% CI 4·6-7·8) earlier along the β-cell function decline trajectory compared with the classic type 1 diabetes group. The risks of hypoglycaemia (log-rank p=0·0019), retinopathy (mixed type 1 diabetes vs classic type 1 diabetes: adjusted hazard ratio 4·93 [95% CI 1·71-14·24]), and cardiovascular autonomic neuropathy (log-rank p=0·028) differed across the groups. INTERPRETATION: Our results indicate that individuals with combined autoimmunity and lower insulin sensitivity present with clinical diabetes earlier than classic type 1 diabetes along the β-cell decline trajectory, identifying a potential timeframe for disease-modifying interventions in specific type 1 diabetes subtypes. FUNDING: BreakthroughT1D, German Research Foundation (DFG), German Center for Diabetes Research (DZD), Schmutzler Stiftung, German Federal Ministry of Health, Ministry of Culture and Science of the State of North Rhine-Westphalia, and German Federal Ministry of Education and Research. TRANSLATION: For the German translation of the abstract see Supplementary Materials section.