Aims/hypothesis: Biomarkers are needed to accurately predict and monitor type 1 diabetes progression during the substantially heterogeneous presymptomatic period of the disease development. To address this concern, we studied temporal changes in the plasma and serum proteomes of < 5-year-old children with HLA-conferred risk for type 1 diabetes, by analysing longitudinal sample series that were collected at regular intervals between birth and diagnosis. Methods: Using mass spectrometry-based discovery proteomics, longitudinal sample series from children positive for multiple autoantibodies who had rapidly progressed to type 1 diabetes before 4 years of age (n=10) were analysed and compared with similar measurements from children, matched from age and gender who were either positive for single autoantibody (n=10) or autoantibody negative (n=10). Following analysis of the data with an additive Gaussian process regression model (LonGP), targeted proteomics was used to verify 11 biomarker candidates in a similar yet independent cohort of children who progressed to the disease within 5 years of age (n=31) and matched autoantibody negative children (n=31). Results: These data reiterated extensive age related trends for protein levels in young children. Further, by combining the utility of LonGP together with the targeted analysis in an extended cohort with more frequent sampling points, these analysis demonstrated that the serum levels of two peptides unique for apolipoprotein C1 (APOC1) were decreased after the appearance of the first islet autoantibody and remained relatively less abundant in children who later progressed to type 1 diabetes, in comparison to autoantibody negative children. Conclusions/interpretation: Through the comparative analysis of prospectively collected samples, our data indicate that the serum/plasma levels of APOC1 are decreased in islet autoantibody positive children who develop type 1 diabetes at a young age. In future studies, APOC1 could be used as an additional measurement to improve early prediction of type 1 diabetes and selection of subjects for intervention studies.

A TSQ Vantage Triple Quadrupole Mass Spectrometer (Thermo Scientific) was coupled with an Easy-nLC 1000 liquid chromatograph (Thermo Scientific) and used with a 20 mm x 100 µm i.d loading-column and a 150 mm x 75 µm i.d. analytical column, both packed with 5 µm Reprosil C18-bonded silica (Dr Maisch GmbH). A binary separation gradient was used. For Batches 2 and 3 this was from 5 to 21% B in 19 min, then to 36% B in 14 min, to 100% B in 4 minutes, with a 5 minutes isochratic period at 100% B and flow rate of 300 nl/min (Solvent A: 0.1% formic acid in MiliQ H2O and Solvent B: 80% acetonitrile, 0.1% formic acid in MiliQ H2O). The later was a modification from the gradient used for Batch 1, which differed in the duration of the first and second gradient steps, which were 28 and 22 minutes, respectively. Skyline software was used to manually inspect the SRM data to ensure and correct for appropriate peak integration. The peptide peak areas were normalized by dividing the total peak area of each peptide with the sum of the peak areas of endogenous alpha-1B-glycoprotein (A1BG) peptides (SGLSTGWTQLSK and ATWSGAVLAGR) in the same sample.

Non-fasting blood serum samples, were collected from children carrying human leukocyte antigen (HLA)-conferred genetic risk for type 1 diabetes in the DIPP study. The samples were from children who had been diagnosed with type 1 diabetes before 5 years of age (n=31,) together with samples from AAb- children matched by age and sex (n=31). The samples were collected between 2 and 60 months of age, with a median of nine samples per child. Serum samples were reduced with DTT, alkylated with iodoacetamide, then digested with trypsin. After desalting using SepPak 100 mg reversed phase SPE plates (SepPak C18, Waters, SKU: 186002321) and drying, heavy isotope-labeled synthetic peptides analogues (PEPotec, Thermo Fischer Scientific) for the protein targets were spiked (~10 fmol/µl) into the digests together with retention time standards (MSRT1, Sigma).