Oral Presentation AUS-oMicS 2025

Large-scale diaPASEF plasma proteomics reveals islet autoimmunity risk associations in pregnancy, cord blood, and infancy.   (#46)

Jumana Yousef 1 2 , Samantha J Emery-Corbin 1 2 , Vineet Vaibhav 1 2 , Helena Oakey 3 , Simon C Barry 4 , Maria E Craig 4 , Peter G Colman 4 , Jennifer J Couper 4 , Elizabeth A Davis 4 , Emma E Hamilton-Wiliams 4 , Leonard C Harrison 4 , Aveni Haynes 4 , Tony Huynh 4 , Ki Wook Kim 4 , Kelly J McGorm 4 , Grant Morahan 4 , William D Rawlinson 4 , Georgia Soldatos 4 , Rebecca L Thomson 4 , Jason Tye-Din 4 , Peter J Vuillermin 4 , John M Wentworth 4 , Megan AS Penno 3 , Laura F Dagley 1 2
  1. The Walter and Eliza Hall Institute of Medical Research, Parkville, VIC, Australia
  2. Department of Medical Biology, University of Melbourne, Parkville, VIC, Australia
  3. Robinson Research Institute, Adelaide Medical School, University of Adelaide, Adelaide, South Australia, Australia
  4. ENDIA Study Group, Australia

Background

This study employed an untargeted proteomic approach to analyse plasma samples collected during pregnancy, at birth (cord blood), and in infancy from mother-infant dyads in the Environmental Determinants of Islet Autoimmunity (ENDIA) nested case-control (NCC) study.

Aims

To identify protein biomarkers in maternal, cord blood, and infant plasma associated with islet autoimmunity (IA) risk in childhood.

Methods

The ENDIA study follows children with a first-degree relative with type 1 diabetes (T1D) from pregnancy through early childhood. The NCC study includes 54 ‘case’ children with persistent islet autoantibodies, matched 1:3 by age and sex with ‘controls’ who were antibody-negative at seroconversion onset in the case. We analysed 931 plasma samples via diaPASEF using nanoLC-MS/MS on a timsTOF Pro MS. Data were analysed with both DIA-NN v.1.8 (library-free) and Spectronaut v.19. The analyses employed weighted conditional logistic regression, linear models via limma, and paired tests at the point of seroconversion. We analysed cases versus controls at the time of seroconversion and conducted paired tests on a subset of cases to compare seroconversion against an earlier autoantibody-negative timepoint. Additionally, time-series analysis comparing cases versus controls was performed on longitudinal infant data.

Results

Logistic regression identified a small number of proteins with modest associations with case versus control status, while the paired analysis revealed 32 proteins significantly altered at seroconversion. Additionally, linear models and longitudinal trajectory analyses uncovered further significant proteins associated with IA risk including in pregnancy and cord blood. Subsequent pathway analysis highlighted metabolic, immune, and inflammatory alterations linked to IA. Ongoing work focuses on integrating findings from all statistical approaches and leveraging machine learning to refine and validate potential biomarkers for IA risk.

Conclusions

Plasma proteomics is advancing our understanding of IA risk by uncovering temporal protein dynamics, with implications for biomarker discovery and therapeutic development in T1D.