Identification of human glucocorticoid response markers using integrated multi-omic analysis from a randomized crossover trial

  1. Dimitrios Chantzichristos  Is a corresponding author
  2. Per-Arne Svensson
  3. Terence Garner
  4. Camilla AM Glad
  5. Brian R Walker
  6. Ragnhildur Bergthorsdottir
  7. Oskar Ragnarsson
  8. Penelope Trimpou
  9. Roland H Stimson
  10. Stina W Borresen
  11. Ulla Feldt-Rasmussen
  12. Per-Anders Jansson
  13. Stanko Skrtic
  14. Adam Stevens
  15. Gudmundur Johannsson
  1. Department of Internal Medicine and Clinical Nutrition, Institute of Medicine at Sahlgrenska Academy, University of Gothenburg, Sweden
  2. Endocrinology, Diabetology and Metabolism, Sahlgrenska University Hospital, Sweden
  3. Department of Molecular and Clinical Medicine, Institute of Medicine at Sahlgrenska Academy, University of Gothenburg, Sweden
  4. Institute of Health and Care Sciences, Sahlgrenska Academy, University of Gothenburg, Sweden
  5. Division of Developmental Biology & Medicine, Faculty of Biology, Medicine and Health, University of Manchester, United Kingdom
  6. Clinical and Translational Research Institute, Newcastle University, United Kingdom
  7. BHF/University Centre for Cardiovascular Science, University of Edinburgh, United Kingdom
  8. Department of Medical Endocrinology and Metabolism, Copenhagen University Hospital, Denmark
  9. Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Denmark
  10. Wallenberg Laboratory, Department of Molecular and Clinical Medicine, Institute of Medicine at Sahlgrenska Academy, University of Gothenburg, Sweden
  11. Innovation Strategies and External Liaison, Pharmaceutical Technologies and Development, Sweden
11 figures, 4 tables and 4 additional files

Figures

Clinical and analytical part of the exploratory study and the replication step.

(A) Subjects with Addison’s disease (primary adrenal insufficiency, step 1) were studied in a random order during both physiological glucocorticoid (GC) exposure and GC withdrawal (step 2). …

Violin plots of serum and urinary glucocorticoids (GCs) during GC exposure and withdrawal.

Individual data and changes for morning (7 AM) serum cortisol and cortisone (A, B), and overnight (12 AM to 7 AM) urinary-free (U-free) cortisol and cortisone (C, D) from both interventions. Boxes …

Figure 3 with 4 supplements
Hypernetwork analyses of integrated ‘omic data.

(A) Hypernetworks differ from traditional networks in that edges can connect more than two nodes. Nodes are represented by black circles, edges by colored lines and surfaces. This demonstration …

Figure 3—figure supplement 1
Hypernetwork heat maps of integrated ‘omic data.

The analyses were performed to quantify groups of highly correlated ‘omic data sets. Hypernetwork heat maps show clusters of correlated peripheral blood mononuclear cell (PBMC) transcripts (total …

Figure 3—figure supplement 2
Transcriptome quality control measures.

(A) Peripheral blood mononuclear cell (PBMC) transcriptome probe-level log-intensity distribution. (B) Adipose tissue transcriptome probe-level log-intensity distribution. Density distributions …

Figure 3—figure supplement 3
Metabolome quality control measures.

Distribution before and following normalization of (A) gas chromatography-mass spectrometry (GC-MS) samples, (B) GC-MS metabolites, (C) liquid chromatography-mass spectrometry (LC-MS) samples, and (D

Figure 3—figure supplement 4
Chi-squared distance distribution from a range of dichotomization thresholds.

Hypernetworks were generated, dichotomizing the data using a range of cut-off r-values, from 0.1 to 0.9 by steps of 0.05, in order to generate the hypernetwork incidence matrix M. Overall …

Figure 4 with 1 supplement
An overlapping gene set in peripheral blood mononuclear cell (PBMC) and adipose tissue transcriptome can be used to classify glucocorticoid (GC) response.

These analyses were performed to depict the common predictive genes in PBMCs and adipose tissue. (A) Partial least squares discriminant analysis (PLSDA) showing complete separation of hydrocortisone …

Figure 4—figure supplement 1
Random forest distribution of minimal depth.

Distribution of minimal depth among trees from random forest generated in Figure 4B. Minimal depth equals the depth of the node closest to the root, which is capable of splitting participants into …

Figure 5 with 1 supplement
Integration of all circulating ‘omic data sets associated with glucocorticoid (GC) response.

These analyses were performed to lead to putative biomarkers of GC action. (A) Hypernetwork summary heat map of shared correlations between all circulating ‘omic elements (peripheral blood …

Figure 5—figure supplement 1
Heat map with clusters of circulating ‘omic data associated with glucocorticoid exposure identified using a correlation matrix.

Along the diagonal 11 clusters including peripheral blood mononuclear cell transcriptomic, miRNAomic and metabolomic data were identified (depicted in squares with their numbers). Color scale …

Replication of miR-122-5p as a putative biomarker of glucocorticoid (GC) action in the current biomarker discovery study.

Targeted replication of plasma miR-122-5p fold change from the current study population between subjects with Addison’s disease during GC exposure and GC withdrawal showed a significant …

Replication of miR-122-5p as a putative biomarker of glucocorticoid (GC) action in independent patient groups with different GC exposure.

(A) The expression of miR-122-5p was higher in subjects with rheumatoid arthritis and reduced GC exposure due to tertiary adrenal insufficiency after a short-term stop of the GC treatment (AI) than …

Appendix 1—figure 1
Integration of unsupervised peripheral blood mononuclear cell (PBMC) transcriptomic, metabolomic, and miRNAomic data in subjects between glucocorticoid (GC) exposure and GC withdrawal at 7 AM using similarity network fusion.

Integration of all circulating ‘omic data (A) simultaneously and (B) stepwise. Each matrix represents a subject-by-subject comparison using similarity network fusion of change in signal of ‘omic …

Appendix 2—figure 1
Gene ontology and interactome network model of differential gene expression in peripheral blood mononuclear cells (PBMCs) between glucocorticoid (GC) exposure and GC withdrawal.

Differentially expressed transcriptomic data in PBMCs was used to determine (A) the canonical pathways involved and (B) infer an interactome network model using BioGRID database, consisting of 2467 …

Appendix 2—figure 2
Correlated networks related to glucocorticoid (GC) action in the peripheral blood mononuclear cell (PBMC) and adipose tissue transcriptome.

The analyses were performed to identify similarities and differences in transcriptome (gene expression) associated to GC exposure in two different tissues. Network associations of GC response to …

Appendix 2—figure 3
Gene Ontology related to glucocorticoid (GC) action in the peripheral blood mononuclear cell (PBMC) and adipose tissue transcriptome.

(A) Numeric summary of biological pathways (Gene Ontology) associated with differential gene expression between GC exposure and GC withdrawal in the PBMC and adipose tissue transcriptomes. (B) …

Tables

Table 1
Clinical and biochemical tests assessed or collected immediately before each intervention and at 7 AM on the second intervention day during both interventions (n = 10).
9 AM before start of intervention
(first intervention day)
7 AM
(second intervention day)
Median (IQR)p-ValueMedian (IQR)p-Value
S-cortisol (nmol/L)HC43.2 (38.0–55.1)0.36298 (228–359)<0.001
Saline46.7 (43.1–61.7)44.4 (36.8–52.5)
Overnight U-free cortisol (µg/7 hr)*,HC678 (459–814)<0.001
Saline<0.01
S-cortisone (nmol/L)HC40.7 (29.5–49.2)0.7681.2 (60.5–94.9)<0.001
Saline42.1 (29.7–50.4)42.0 (28.9–47.3)
Overnight U-free cortisone (µg/7 hr)*,HC136 (106–151)<0.001
Saline<0.01
SBP (mmHg)HC122 (111–134)0.88123 (107–139)0.65
Saline127 (112–131)124 (101–137)
DBP (mmHg)HC76 (63–83)0.8869 (59–76)0.55
Saline72 (67–80)65 (60–70)
S-sodium (mmol/L)HC141 (139–142)0.85140 (138–141)0.97
Saline141 (138–142)140 (138–141)
S-potassium (mmol/L)HC4.4 (4.3–4.7)0.274.4 (4.3–4.8)0.26
Saline4.2 (4.0–4.5)4.4 (4.3–4.5)
P-glucose (mmol/L)HC5.0 (4.5–5.3)0.795.6 (5.2–5.8)0.08
Saline5.0 (4.6–5.1)5.2 (5.2–5.4)
Body weight (kg)HC72.2 (69.6–77.3)0.5372.8 (69.3–77.7)0.63
Saline73.5 (70.5–79.5)74.0 (70.5–79.2)
  1. *Overnight U-free cortisol and cortisone were collected from midnight to morning (12 AM to 7 AM) during physiological GC exposure (HC infusion) and GC withdrawal (saline infusion).

    One of the 10 subjects was not included in the analysis because of a problem during sample collection.

  2. Below the limit of detection.

    DBP: diastolic blood pressure; GC: glucocorticoid; HC: hydrocortisone; IQR: interquartile range; P: plasma; S: serum; SBP: systolic blood pressure; U: urinary.

Table 2
Summary of differentially expressed ‘omic elements in association with response to glucocorticoids.
‘Omic data setTotal number of ‘omic elementsNumber of significant elements (p<0.05)Number of significant elements (FDR < 0.05)
PBMC transcriptome28,86944263997
Adipose tissue transcriptome28,86935203115
Plasma miRNAome25299
Serum metabolome1643814
  1. FDR: false discovery rate; miRNA: microRNA; PBMC: peripheral blood mononuclear cell.

Table 3
Validation of the predictive genes from the current exploratory study against previous studies examining GC response in cellular systems.
Study titleGEO #PMIDNAUC
(95% CI)
OOB*
AUC
OOB* error rate‡ (%)
Dexamethasone effect on epidermal keratinocytes in vitroGSE2648717095510 (Stojadinovic et al., 2007)200.70
(0.51–0.89)
0.8030
Dexamethasone effect on GC-resistant and -sensitive lymphoblastic leukemia cell linesGSE2215221092265 (Carlet et al., 2010)240.71
(0.52–0.90)
0.7829
In vivo GC effect on non-leukemic peripheral blood lymphocytesGSE2277921092265 (Carlet et al., 2010)160.88
(0.63–1.0)
0.966
Osteosarcoma cell line response to activation of specific GC receptor alpha isoformsGSE671117682054 (Lu et al., 2007)
22174376 (Jewell et al., 2012)
600.96
(0.89–1.0)
0.993
GC effect on lens epithelial cellsGSE304016319822 (Gupta et al., 2005)120.83
(0.63–1.0)
0.7217
  1. *OOB data, the bootstrapping approach of Random Forest, ensures that every tree is built using ~63% of the available data, leaving ~ 37% that can be used for a validation test.

    AUC up to 0.96 demonstrates a high probability of correctly classifying a randomly selected sample from each study.

  2. OOB error rate = prediction error using the OOB validation data.

    The gene set that classified both PBMC and adipose tissue transcriptomes in relation to GC exposure with fold change in the same direction (see Figure 4B – 59 genes) was validated by further testing in five other publicly available studies of GC action in cellular systems.

  3. AUC: area under the curve of the receiver operating characteristic; CI: confidence interval; GC: glucocorticoid; GEO: Gene Expression Omnibus; GEO #: study number deposited with GEO; N: study number size; OOB: out-of-bag; PBMC: peripheral blood mononuclear cell; PMID: PubMed ID number of the manuscript describing the data.

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Recombinant DNA reagentGeneChip WT PLUS Reagent KitAffymetrix IncOtherP/N 703174 Rev. 1
Commercial assay or kitHuman Gene 1.0 ST arrayAffymetrix Inc
Commercial assay or kitHuman Gene 1.1 ST arrayAffymetrix Inc
Commercial assay or kitHuman Gene 2.0 ST arrayAffymetrix Inc
Commercial assay or kitmiRCURY LNA Universal RT microRNA PCR, Polyadenylation, and cDNA Synthesis KitExiqon
Commercial assay or kitmiRCURY RNA Isolation Kit-BiofluidsExiqon
Chemical compound, drugSolu-CortefPfizer Inc
Software, algorithmAgilent Masshunter ProfinderAgilent Technologies, IncOtherVersion B.08.00
Software, algorithmSPSSSPSSRRID:SCR_002865
Software, algorithmRR Project for Statistical ComputingRRID:SCR_001905
Software, algorithmRstudioRstudioRRID:SCR_000432
Software, algorithmMetaboAnalystRRRID:SCR_016723
Software, algorithmModuland algorithmhttps://www.linkgroup.hu/docs/ModuLand-ESM1-v5.pdf
Software, algorithmCytoscapeCytoscapeRRID:SCR_003032
Software, algorithmQlucoreQlucore Omics Explorerhttps://www.qlucore.com/bioinformatics
Software, algorithmRobust Multi-Array Average algorithmhttp://www.molmine.com
Software, algorithmChromaTOFLECOhttps://www.leco.com
Software, algorithmMATLAB R2016aMathworkshttps://www.mathworks.com
Software, algorithmRoche LC softwareRoche Molecular Systems, Inc
Software, algorithmNormFinderAarhus University Hospital, DenmarkRRID:SCR_003387
Software, algorithmNIST MS 2.0 softwareNISThttps://chemdata.nist.gov
OtherLightCycler 480 Real-Time PCR SystemRoche Molecular Systems, IncRRID:SCR_020502
OtherAgilent 1290 Infinity UHPLC-systemAgilent Technologies, Inchttps://www.agilent.com
OtherAgilent 2100 Bioanalyzer systemAgilent Technologies, IncRRID:SCR_018043
OtherAgilent 6550 iFunnel Q-TOF LC/MSAgilent Technologies, IncRRID:SCR_019433
OtherENCODEStanford UniversityRRID:SCR_015482
OtherUCSC Genome BrowserUCSCRRID:SCR_005780
OtherTarBaseDIANA ToolsRRID:SCR_010841
OthermiRecordsBiolead.orgRRID:SCR_013021
OtherTargetScanWhitehead Institute for Biomedical ResearchRRID:SCR_010845
OtherBioGRIDTyersLab.comRRID:SCR_007393
OtherIngenuity Pathway AnalysisQiagenRRID:SCR_008653

Additional files

Supplementary file 1

Analysis of 'omic datasets.

(1) S file 1a. Transcriptomic analysis of peripheral blood mononuclear cells and adipose tissue from 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (2) S file 1b. miRNA with differential expression in plasma samples from 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (3) S file 1c. Gas chromatography-mass spectrometry (GC-MS) of serum samples from 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (4) S file 1d. Liquid chromatography-mass spectrometry (LC-MS) of serum samples from 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (5) S file 1e. The set of 59 genes with fold changes in the same directions in peripheral blood mononuclear cells (PBMCs) and adipose tissue transcriptomic data sets from 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (6) S file 1f. The order of the 11 clusters of ‘omic data in the correlation matrix. (7) S file 1g. Analysis of all ‘omic sets at 7 AM (morning of the second intervention day) treated with hydrocortisone and saline (control). (8) S file 1h. Biological pathways associated with differential gene expression in PBMCs between hydrocortisone and saline treated patients at 7 AM on the second intervention day. (9) S file 1i. The central gene in each network module arranged in hierarchical order as ranked by network centrality score. (10) S file 1j. Causal network analysis of the differential gene expression in PBMCs associated with glucocorticoid action.

https://cdn.elifesciences.org/articles/62236/elife-62236-supp1-v1.xlsx
Transparent reporting form
https://cdn.elifesciences.org/articles/62236/elife-62236-transrepform-v1.pdf
Reporting standard 1

CONSORT 2010 Randomised Trial Checklist.

https://cdn.elifesciences.org/articles/62236/elife-62236-repstand1-v1.pdf
Reporting standard 2

CONSORT 2010 Flow Diagram.

https://cdn.elifesciences.org/articles/62236/elife-62236-repstand2-v1.pdf

Download links