Back

Plasma protein biomarkers predict both the development of persistent autoantibodies and type 1 diabetes 6 months prior to the onset of autoimmunity: the TEDDY Study

Nakayasu, E. S.; Bramer, L.; Ansong, C.; Schepmoes, A.; Fillmore, T.; Gritsenko, M.; Clauss, T.; Gao, Y.; Piehowski, P.; Stafill, B.; Engel, D.; Orton, D.; Moore, R.; Qian, W.-J.; Sechi, S.; Frohnert, B.; Toppari, J.; Ziegler, A.; Lernmark, A.; Hagopian, W.; Akolkar, B.; Smith, R. D.; Rewers, M.; Webb-Robertson, B.-J.; Metz, T. O.

2022-12-11 endocrinology
10.1101/2022.12.07.22283187 medRxiv
Show abstract

Type 1 diabetes (T1D) results from an autoimmune destruction of pancreatic {beta} cells. A significant gap in understanding the disease cause is the lack of predictive biomarkers for each of its developmental stages. Here, we conducted a blinded, two-phase case-control plasma proteomics analysis of children enrolled in the TEDDY study to identify biomarkers predictive of autoimmunity and T1D development. First, we performed untargeted proteomics analyses of 2,252 samples from 184 individuals and identified 376 regulated proteins. Complement/coagulation, inflammatory signaling and metabolic proteins were regulated even prior to autoimmunity onset. Extracellular matrix proteins and antigen presentation were differentially regulated in individuals with autoimmunity who progressed to T1D versus those who maintained normoglycemia. We then performed targeted proteomics measurements of 167 proteins in 6,426 samples from 990 individuals and validated 83 biomarkers. A machine learning analysis predicted both the development of persistent autoantibodies and T1D onset 6 months before autoimmunity initiation, with an area under the receiver operating characteristic curve of 0.871 and 0.918, respectively. Our study identified and validated biomarkers highlighting pathways affected in different stages of T1D development.

Matching journals

The top 10 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.