Back

Longitudinal Proteogenomic Analysis Reveals Mechanistic Insights into the Progression from Prediabetes to Type 2 Diabetes

Singh, A.; Ganslmeier, M.; Tutino, M.; Park, Y.-C.; Machann, J.; Schick, F.; Peter, A.; Lehmann, R.; Wang, Y.; Cheng, Y.; Sandforth, L.; Schuth, S.; Seissler, J.; Perakakis, N.; Schwarz, P. E. H.; Szendrödi, J.; Wagner, R.; Solimena, M.; Schürmann, A.; Kabisch, S.; Pfeiffer, A. F. H.; Bornstein, S. R.; Blüher, M.; Stefan, N.; Fritsche, A.; Preissl, H.; Schwartzenberg, R. J. v.; de Angelis, M. H.; Roden, M.; Bocher, O.; Zeggini, E.; Birkenfeld, A. L.

2026-02-16 endocrinology
10.64898/2026.02.13.26346161 medRxiv
Show abstract

Prediabetes and type 2 diabetes (T2D) are metabolic disorders characterized by insulin resistance and {beta}-cell dysfunction. To understand the molecular mechanisms driving the transition from prediabetes to T2D, we performed a longitudinal proteogenomic analysis on 458 participants from the Prediabetes Lifestyle Intervention Study (PLIS). We identified 185 plasma proteins to be differentially expressed between conditions, 36 of which predict future T2D-onset. Integrating genetic data from 321 individuals, we generated a genome-wide protein quantitative trait loci (pQTL) map, identifying 86 differential and 700 shared cis-pQTLs between prediabetes and T2D. Mediation analysis revealed 60 putative causal links connecting allele-driven plasma protein expression to clinical traits, identifying body fat distribution, insulin resistance, and {beta}-cell function as central drivers of pathogenesis. Collectively, these findings highlight specific proteins underlying disease progression and substantiate the view that prediabetes and T2D are not distinct conditions, but rather stages on a unified metabolic spectrum.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.