Longitudinal proteomics over 15 years identified 845 protein biomarkers for incident diabetes (36.7% replicated in UK Biobank), compared to 342 from single timepoint analyses.
Cohort (n=5,322)
Yes
Does longitudinal protein measurement identify novel diabetes biomarkers compared to single timepoint measurement in a multiethnic cohort?
Longitudinal proteomic profiling identifies novel diabetes biomarkers and distinct biological pathways compared to single timepoint measurements.
Introduction and Objective: Circulating diabetes biomarkers have been identified with single timepoint proteomics. We studied repeated protein measurements in the same individuals over 15 years to reveal what additional information longitudinal data provides. Methods: We measured 2,943 Olink proteins in Multi-Ethnic Study of Atherosclerosis (MESA; N=5,322, mean age 61, 53% women) plasma samples from Exam 1 (2000-2002), 5 (2010-2012), and 6 (2016-2018). Associations with incident diabetes adjusted for age, sex, race/ethnicity, BMI, eGFR, total cholesterol, HDL-c, triglycerides, hypertension, and cholesterol medications were modeled using 1) Cox models with Exam 1 proteins and 2) time-updating Cox with Exam 1, 5, and 6 proteins and compared with single timepoint analyses in the UK Biobank (UKB). Results: We found 342 protein diabetes biomarkers in MESA single timepoint analyses (FDR0.05) Fig 1A, of which 87.1% replicated in the UKB (FDR0.05) Fig 1B; and 845 longitudinal proteins Fig 1A, of which 36.7% replicated Fig 1B. A similar proportion of proteins in both sets were casual in cis-MR (14% and 11%, respectively). The 550 unique longitudinal proteins were enriched for cell regulation and protein modification pathways, small molecule metabolic and catabolic pathways were enriched in the 277 shared proteins. Conclusion: Longitudinal versus single timepoint proteins identify novel diabetes markers which may reveal distinct biological disease pathways. Disclosure Z. Chen: None. M. Mi: None. J.L. Barber: None. G. Tiwari: None. C.D. Adams: None. S. Deng: None. P. Rao: None. D. Cruz: None. M. Sevilla-Gonzalez: Research Support; Current; Novo Nordisk. M. Goodarzi: None. X. Guo: None. K.D. Taylor: None. U. Tahir: Consultant; Ended; Amicus Therapeutics, Inc. A. Wood: Research Support; Current; Beef Checkoff. Consultant; Ended; Lundquist Institute for Biomedical Innovation. S. Rich: Advisory Panel; Current; Sanofi. Research Support; Current; Sanofi, Leona M. and Harry B. Helmsley Charitable Trust. Other - Consulting Associate Editor, Diabetes Care; Current; American Diabetes Association. R.E. Gerszten: None. J.I. Rotter: None. Funding NIDDK (DK127073), NHLBI (HL181500)
CHEN et al. (2026) conducted a cohort in Incident diabetes (n=5,322). Longitudinal proteomics vs. Single timepoint proteomics was evaluated on Incident diabetes. Longitudinal proteomics over 15 years identified 845 protein biomarkers for incident diabetes (36.7% replicated in UK Biobank), compared to 342 from single timepoint analyses.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: