Daily Endocrinology Research Analysis

Type 1 diabetes (T1D) is an autoimmune disease in which T cells mediate the elimination of the insulin-producing beta cells in the pancreatic islets, resulting in the need for exogenous insulin. Studies of T1D in both patients and the nonobese diabetic (NOD) mouse model of the disease illustrate that beta cell-specific CD8+ T cells are central contributors to the beta cell destruction characterizing the disease. Compared to HLA-A and HLA-B, relatively little is known about the participation of HLA-C-restricted T cells in T1D. To tackle this question, we developed and characterized an NOD-based model of spontaneous T1D that transgenically expresses HLA-C03:04, a common and enriched allotype in T1D patients. Using an unbiased screen of an exhaustive peptide library comprised of 8- to 11-mer peptides derived from the key autoantigen insulin, islet-infiltrating T cells from the mice were found to recognize the insulin A-chain peptides A11-19 and A13-21 presented by HLA-C03:04. Guided by these findings, T-cell lines were established from the islets of HLA-C03:04-positive human donors with T1D, including a donor with demise at onset of T1D. These human islet-derived T cells also responded to the conserved A11-19 and A13-21 peptides. The presence of HLA-C03:04-restricted insulin-specific T cells in both mouse and human islets suggests the participation of peptides presented by HLA-C molecules in T1D pathogenesis. This work also demonstrates the utility of the mouse model in identifying human disease-relevant HLA-C-restricted epitopes and suggests a general strategy for the exploration and manipulation of HLA-C-restricted T cells in autoimmune diseases.

Medical foundation models, pre-trained on large-scale unlabelled data, show strong performance and data efficiency when adapted to various clinically relevant applications. However, how pre-training data shape the generalisability and fairness of these models remains unexplored. Here we address this using two cohorts from Moorfields Eye Hospital (UK) and the Shanghai Diabetes Prevention Program (China), each containing 904,170 fundus photographs for model pre-training. Using identical pipelines, we train parallel foundation models using individual cohort and evaluate them on downstream tasks with publicly available datasets and held-out data from each site. The parallel models show competitive performance to data that differ substantially from their pre-training data. Nevertheless, we observe fairness gaps over age subgroups, whereas sex and ethnicity show minimal impact. These results demonstrate the good generalisability of retinal foundation models and indicate that pre-training demographic attributes shape fairness differently, highlighting the importance of domain-specific, fine-grained data curation for efficient foundation model development.

OBJECTIVES: Primary aldosteronism (PA) is a common cause of secondary hypertension. To address the specificity limits of immunoassays, we developed and validated an LC-MS/MS method for urinary aldosterone and tetrahydroaldosterone and established method-matched reference intervals for excretion in 24 h urine. METHODS: Urinary aldosterone, tetrahydroaldosterone and their glucuronidated metabolites were extracted in the presence of internal standards using offline solid-phase extraction (SPE), followed by enzymatic hydrolysis to release the glucuronidated fraction. Subsequently, aldosterone and tetrahydroaldosterone were analyzed by online SPE in combination with LC-MS/MS. Reference intervals were established based on 24 h urine samples from 265 individuals participating in the Lifelines Cohort study. RESULTS: Intra- and inter-assay imprecision ranged from 2.0-12.3 % for aldosterone, and 1.3-6.3 % for tetrahydroaldosterone. The lower limits of quantification were 0.44 nmol/L and 0.10 nmol/L, respectively. Recoveries ranged from 97-106 %, calibration was linear, with correlation coefficients greater than 0.999, and no carry-over was observed. Total aldosterone concentrations measured by LC-MS/MS were consistently higher than those obtained by radioimmunoassay. In the reference population, 24 h urinary excretion ranged from 5.4-76.7 nmol/24 h for aldosterone and 21.4-269.9 nmol/24 h for tetrahydroaldosterone. CONCLUSIONS: This validated LC-MS/MS assay, together with a method-matched normative dataset, enables standardized urinary aldosterone profiling and defines reference intervals that will help improve the interpretability of results in the biochemical diagnosis of PA.