Parallel Multi-Omics in High-Risk Subjects for the Identification of Integrated Biomarker Signatures of Type 1 Diabetes

Abstract

Background: Biomarkers are crucial for detecting early type-1 diabetes (T1D) and preventing significant β-cell loss before the onset of clinical symptoms. Here, we present proof-of-concept studies to demonstrate the potential for identifying integrated biomarker signature(s) of T1D using parallel multi-omics.

Methods: Blood from human subjects at high risk for T1D (and healthy controls; n = 4 + 4) was subjected to parallel unlabeled proteomics, metabolomics, lipidomics, and transcriptomics. The integrated dataset was analyzed using Ingenuity Pathway Analysis (IPA) software for disturbances in the at-risk subjects compared to controls.

Results: The final quadra-omics dataset contained 2292 proteins, 328 miRNAs, 75 metabolites, and 41 lipids that were detected in all samples without exception. Disease/function enrichment analyses consistently indicated increased activation, proliferation, and migration of CD4 T-lymphocytes and macrophages. Integrated molecular network predictions highlighted central involvement and activation of NF-κB, TGF-β, VEGF, arachidonic acid, and arginase, and inhibition of miRNA Let-7a-5p. IPA-predicted candidate biomarkers were used to construct a putative integrated signature containing several miRNAs and metabolite/lipid features in the at-risk subjects.

Conclusions: Preliminary parallel quadra-omics provided a comprehensive picture of disturbances in high-risk T1D subjects and highlighted the potential for identifying associated integrated biomarker signatures. With further development and validation in larger cohorts, parallel multi-omics could ultimately facilitate the classification of T1D progressors from non-progressors.

Keywords: biomarker signature; biomarkers; diagnosis; early prediction; integrated analysis; lipidomics; metabolomics; multi-omics; network prediction; omics; prognosis; proteomics; signaling pathways; transcriptomics; type 1 diabetes.

Figure 1

The impact of thresholding strategy on the integrative analysis of parallel multi-omics datasets. Bar graph representation of features remaining in each omics dataset after different levels of selection stringency for feature selection (fold-change cutoff value) used for input into the integrative analysis shown as percentages of the total identified features (see Table 1 for absolute numbers).

Figure 2

Independent analyses in the individual datasets from multi-omics reveal modest differences in the transcriptome, metabolome, and lipidome of high-risk T1D subjects compared to healthy controls with only a few significantly different features. A comparison between high-risk T1D subjects (red/pink) and healthy controls (green; n = 4 each), using each omics-type dataset pooled from each group, was performed by t-test with correction for multiple comparisons with the Holm–Sidak method. Shown are all miRNAs, lipids, and metabolites that had significant differences between the two groups (p < 0.05).

Figure 3

The ability to predict diseases and functions and alterations in canonical pathways and upstream regulators in high-risk T1D subjects decreases with increasing stringency of feature selection. Proteomics and metabolomics datasets with feature selection at different cutoff values were interrogated for predicted involvement in various immune inflammatory processes/pathways using the “Comparison Analysis” module of IPA software. (A) Diseases and functions predicted to be most affected in the high-risk T1D subjects compared to healthy controls by independent analyses in the proteomics and metabolomics datasets. (B) Canonical pathways analysis with increasing cutoff values for feature selection in the proteomics dataset. (C) Upstream regulator analysis in the proteomics, metabolomics, and lipidomics datasets, without curation for any specific physiological function at the indicated cutoff values, showing the total number of upstream regulators predicted to be activated and inhibited. Results in A and B are shown as heatmaps with the orange color indicating activation and its intensity the magnitude of the z-value (statistical score that accounts for the directional effect of change for functions or molecules in the experimental datasets; https://go.qiagen.com/IPA-transcriptomics-whitepaper, accessed on 16 February 2021). Results in C are shown as concentric Venn diagrams showing the number of activated (orange) or inhibited (blue) upstream regulators predicted in the analyses at the indicated cutoff values; no predictions were possible at cutoff values of 2.0 or above.

Figure 4

Cause-effect relationship analysis provides mechanistic insight into T1D pathogenesis based on integrated multi-omics datasets. (A–C) Three-tier diagrams (networks), predicted by the “Regulator Effects” analysis module in IPA software using the integrated quadra-omics dataset with decreasing stringency of feature selection threshold (i.e., fold-change cutoff). Each network shows the predicted upstream regulators (top row), selected molecules with significant differential expression as identified in the actual data (middle row), and the immune functions/processes predicted to be affected (bottom row). (A) A network that was consistently predicted at cutoff values of 1.1, 1.2, and 1.3 showing activation of the inflammatory response. (B) A network predicted at the cutoff values of 1.1 and 1.2 showing upregulation of CD4+ T-lymphocyte proliferation and ROS generation. (C) A network predicted only at the cutoff value of 1.1 showing activation of cellular movement of T-lymphocytes, granulocytes, and phagocytes. Marker key—Triangle: kinase; horizontal-oval: transcription regulator; vertical-oval: transmembrane receptor; diamond: enzyme; square: cytokine; horizontal ellipse: metabolite; circle: other; octagon: function. Color key—orange: predicted activation, blue: predicted inhibition. Connecting line color key—orange: activation; blue: inhibition; and gray: not predicted.

Figure 5

Integrated molecular network analysis in combined quadra-omics datasets highlights the potential for identifying integrated biomarker signatures of T1D. The integrated quadra-omics dataset was analyzed in IPA using the “Molecular Network Prediction” module with a focus on immune/inflammatory processes. The integrated global network shown in (A) was obtained at the cutoff value of 1.1 and was curated to reduce complexity and highlight nodes with known involvement in T1D, namely, NF-κB, TGF-β, VEGF, arachidonic acid, arginase, and the microRNA Let-7a-5p (Let-7) family. Individual contributions to this network by each omics-type dataset are highlighted in full color and some nodes annotated (for emphasis) with the rest of the network faded in the background (for clarity): (B) proteomics, (C) small transcriptomics, (D) metabolomics, and (E) lipidomics. Network elements are represented by symbols of various marker shapes for identification and different colors for activation or inhibition. Marker shape key—horizontal-oval: transcription regulator; vertical-oval: transmembrane receptor; diamond: enzyme; square: cytokine; vertical-rectangle: G-protein coupled receptor; broken-lined vertical-rectangle: ion channel; and horizontal-diamond: peptidase. Marker color key—orange: predicted activation, blue: predicted inhibition. Connecting line color key—orange: activation; blue: inhibition; yellow: findings are inconsistent with the state of the downstream molecule; and gray: not predicted. High resolution versions of these networks are provided in the Supplementary Figure S2.

Figure 6

Key features identified in molecular networks relevant to immune/inflammatory processes in high-risk T1D subjects and the impact of the fold-change cutoff value on their identification. Comparison analysis in IPA software was applied to individual networks with scores at 10 or above, and key molecules were inferred based on curated integrated molecular networks generated at the indicated cutoff values (see Supplementary Figure S3). (A) Key molecules predicted to be activated in high-risk T1D subjects compared to healthy controls. Most prominent features, such as NF-κB, TGF-β, VEGF, IL-1, and TNF, were consistently identified in networks predicted at all cutoff values ≤2.0; none were yielded at cutoff 3.0. (B) Key molecules that were predicted to be negatively regulated (inhibited) also decreased with increasing the threshold cutoff values. Orange color indicates features predicted as activated and blue color as inhibited.