Leveraging complementary multi-omics data integration methods for mechanistic insights in kidney diseases

Abstract

Chronic kidney diseases (CKDs) are a global health concern, necessitating a comprehensive understanding of their complex pathophysiology. This study explores the use of 2 complementary multidimensional -omics data integration methods to elucidate mechanisms of CKD progression as a proof of concept. Baseline biosamples from 37 participants with CKD in the Clinical Phenotyping and Resource Biobank Core (C-PROBE) cohort with prospective longitudinal outcome data ascertained over 5 years were used to generate molecular profiles. Tissue transcriptomic, urine and plasma proteomic, and targeted urine metabolomic profiling were integrated using 2 orthogonal multi-omics data integration approaches, one unsupervised and the other supervised. Both integration methods identified 8 urinary proteins significantly associated with long-term outcomes, which were replicated in an adjusted survival model using 94 samples from an independent validation group in the same cohort. The 2 methods also identified 3 shared enriched pathways: the complement and coagulation cascades, cytokine-cytokine receptor interaction pathway, and the JAK/STAT signaling pathway. Use of different multiscalar data integration strategies on the same data enabled identification and prioritization of disease mechanisms associated with CKD progression. Approaches like this will be invaluable with the expansion of high-dimension data in kidney diseases.

Keywords: Chronic kidney disease; Expression profiling; Nephrology; Proteomics.

Figure 1. MOFA and DIABLO integrative approaches applied to the C-PROBE cohort.

First step is training of the MOFA and DIABLO models. Second is the visualization of the variation and analysis of top-ranked features by both algorithms. Pathway analysis for top-ranked features is the third step. Validation of shared features in C-PROBE is the fourth step in this integrative approach.

Figure 2. Factors from MOFA model.

(A) Total percentage of variance explained by MOFA factors. (B) Data variance explained by each MOFA factor. (C) Kaplan-Meier (KM) survival curve using the value of MOFA Factor 2 reaching composite endpoint. (D) KM survival curve using the value of MOFA Factor 3 reaching composite endpoint. Log-rank test was used to determine significant differences in KM curves.

Figure 3. Expression levels and pathway enrichment of top 10 MOFA extracted features.

(A) The expression levels of urine proteomics that are top ranked by MOFA Factor 2. (B) Enriched pathways of the top 100 features extracted from MOFA Factor 2. (C) Protein-protein interaction network between complement components from 3 -omics data types. Gray-colored nodes depict features identified in a minimum of any 2 of the -omics data types.

Figure 4. Top 10 features, expression levels, and enriched pathways using DIABLO.

(A) Sparse partial least-squares discriminant analysis (sPLS-DA) plot for features identified by DIABLO. (B) Normalized expression levels of top 10 features identified by DIABLO in progressors and nonprogressors. (C) Pathway enrichment analysis for features identified by DIABLO model. KEGG pathway was the top pathway identified from RNA and proteins by DIABLO, while the lower panel, created using DIABLO, identified metabolites using MetaboAnalyst.

Figure 5. Consensus urinary proteins identified by MOFA and DIABLO.

(A) Venn diagram of intersection between top 100 urinary proteins ranked by MOFA Factor 2 and 34 urinary proteins identified by DIABLO. (B) KM curve of 1 shared urinary protein, complement C9, identified by both methods depicting higher concentration is associated with increased risk of progression to composite endpoint in the validation cohort. Log-rank test was used to determine significant differences in KM curves.

Figure 6. Validation of MOFA and DIABLO shared urinary proteins in independent C-PROBE samples.

(A) CoxPH survival model, representing the composite endpoint outcome, for one of the shared urinary proteins, complement C9, adjusted for baseline estimated glomerular filtration rate (eGFR), sex, and age at first visit. (B) C-index value of basic model (eGFR, age, sex, and albumin to creatinine ratio [ACR]) compared with models built by adding 8 shared urinary proteins to the basic model. The c-index or c-statistic is the most frequently used evaluation metric of survival models. The c-index value ranges from 0 (perfectly discordant) to 1 (perfectly concordant), and a c-index of 0.5 suggests that the model’s predictions are no better than random chance. *P < 0.05 based on the likelihood ratio test to compare the goodness of fit of the urinary protein model (eGFR + age + sex + ACR + urinary protein) and the basic model (eGFR + age + sex + ACR).