Longitudinal metabolite and protein trajectories prior to diabetes mellitus diagnosis in Danish blood donors: a nested case-control study

Abstract

Aims/hypothesis: Metabolic risk factors and plasma biomarkers for diabetes have previously been shown to change prior to a clinical diabetes diagnosis. However, these markers only cover a small subset of molecular biomarkers linked to the disease. In this study, we aimed to profile a more comprehensive set of molecular biomarkers and explore their temporal association with incident diabetes.

Methods: We performed a targeted analysis of 54 proteins and 171 metabolites and lipoprotein particles measured in three sequential samples spanning up to 11 years of follow-up in 324 individuals with incident diabetes and 359 individuals without diabetes in the Danish Blood Donor Study (DBDS) matched for sex and birth year distribution. We used linear mixed-effects models to identify temporal changes before a diabetes diagnosis, either for any incident diabetes diagnosis or for type 1 and type 2 diabetes mellitus diagnoses specifically. We further performed linear and non-linear feature selection, adding 28 polygenic risk scores to the biomarker pool. We tested the time-to-event prediction gain of the biomarkers with the highest variable importance, compared with selected clinical covariates and plasma glucose.

Results: We identified two proteins and 16 metabolites and lipoprotein particles whose levels changed temporally before diabetes diagnosis and for which the estimated marginal means were significant after FDR adjustment. Sixteen of these have not previously been described. Additionally, 75 biomarkers were consistently higher or lower in the years before a diabetes diagnosis. We identified a single temporal biomarker for type 1 diabetes, IL-17A/F, a cytokine that is associated with multiple other autoimmune diseases. Inclusion of 12 biomarkers improved the 10-year prediction of a diabetes diagnosis (i.e. the area under the receiver operating curve increased from 0.79 to 0.84), compared with clinical information and plasma glucose alone.

Conclusions/interpretation: Systemic molecular changes manifest in plasma several years before a diabetes diagnosis. A particular subset of biomarkers shows distinct, time-dependent patterns, offering potential as predictive markers for diabetes onset. Notably, these biomarkers show shared and distinct patterns between type 1 diabetes and type 2 diabetes. After independent replication, our findings may be used to develop new clinical prediction models.

Keywords: Molecular biomarkers; Multi-omics; Polygenic risk scores; Temporality; Time-to-event prediction; Type 1 diabetes mellitus; Type 2 diabetes mellitus.

Fig. 1

Graphical representation of study set-up. Individuals with incident diabetes were selected based on a diabetes register capturing patients with type 1 or type 2 diabetes (DM) in the period 1977–2016. Only individuals with at least one DBDS inclusion sample collected in the period 2010–2016 and two additional plasma samples (archival sample or DBDS inclusion sample) collected in the period 2006–2016 were included as participants. The three plasma samples had to be donated at least 9 months apart to ensure ample time for molecular changes reflecting the development of diabetes to take place

Fig. 2

Biomarker-specific effect estimates for incident diabetes. Effect estimates obtained using mixed-effects models (fold change, FC) are shown for proteins (a), metabolites (a, b) and lipoprotein particles (b). All estimates are shown as point estimates and 95% CIs. Biomarkers with a significant interaction term are indicated by triangles that show the direction of the trend. Significant associations (FDR-adjusted p<0.05) are indicated by filled circles; non-significant estimates are indicated by open circles. bFGF, basic fibroblast growth factor; FA, fatty acids; GM-CSF, granulocyte macrophage colony-stimulating factor; IDL, intermediate-density lipoprotein; IP-10, interferon-induced protein 10; MDC, macrophage-derived chemokine; MUFAs, mono-unsaturated fatty acids; PlGF, placental growth factor; PUFAs, poly-unsaturated fatty acids; SFAs, saturated fatty acids; sVCAM-1, soluble vascular cell adhesion molecule-1; ULDL, ultra low-density lipoprotein; VEGF, vascular endothelial growth factor. The prefixes XS, S, M, L, XL and XXL refer to lipoprotein sizes from extra small to extremely large

Fig. 3

Per-year estimated marginal means for temporally changing biomarkers. Estimated marginal means are shown for proteins (a), metabolites (b) and lipoprotein particles (c) that showed a significant interaction between incident diabetes and time to end of follow-up (FDR-adjusted p<0.05) as assessed by ANOVA. Estimated marginal means are shown as point estimates and 95% CIs for the diabetes group (points with error bars) and the non-diabetes group (line with shaded area) for each year before the end of follow-up, i.e. time 0 (diabetes diagnosis for incident diabetes cases and the end of the study period for individuals without diabetes). Values have been z score-normalised to ease visualisation; hence one unit difference corresponds to one SD. The exact estimated marginal means are given in ESM Table 4. Significant estimates (FDR-adjusted p<0.05) are indicated by filled circles; non-significant estimates are indicated by open circles. Conc., concentration; IDL, intermediate-density lipoprotein; PG, phosphoglycerides; TG, triacylglycerol; ULDL, ultra low-density lipoprotein. The prefixes XS, S, M, L, XL and XXL refer to lipoprotein sizes from extra small to extremely large

Fig. 4

Parameter importance for the top 40 markers from each molecular dataset. Parameter importance for the top 40 markers for each molecular dataset as assessed by variable importance values from the surv-RF model using 100 trees over 1000 bootstraps and the percentage of models where p<0.1 for the marker estimate calculated from the boot-Poisson model with 1000 bootstraps. The rank within each combination of biomarker data types is shown in the heatmap. Markers are arranged according to molecular type and groups. Groups are coloured to assist distinction between marker groups. Variable importance has been multiplied by 10 to give a range of 0–1. AMI, acute myocardial infarction; bFGF, basic fibroblast growth factor; CAD, coronary artery disease; CKD, chronic kidney disease; FA, fatty acids; GIP, gastric inhibitory polypeptide; GLP-1, glucagon-like peptide-1; IMID, immune-mediated inflammatory diseases; IP-10, interferon-induced protein 10; MDC, macrophage-derived chemokine; MUFAs, mono-unsaturated fatty acids; NAFLD, non-alcoholic fatty liver disease; PlGF, placental growth factor; PUFAs, poly-unsaturated fatty acids; SFAs, saturated fatty acids; T1DM, type 1 diabetes; T2DM, type 2 diabetes; TSLP, thymic stromal lymphopoietin; ULDL, ultra low-density lipoprotein; VEGF, vascular endothelial growth factor. The prefixes XS, S, M, L, XL and XXL refer to lipoprotein sizes from extra small to extremely large