Potential Transcriptional Biomarkers to Guide Glucocorticoid Replacement in Autoimmune Addison's Disease

Åse Bjorvatn Sævik; Anette B. Wolff; Sigridur Björnsdottir; Katerina Simunkova; Martha Schei Hynne; David William Peter Dolan; Eirik Bratland; Per M Knappskog; Paal Methlie; Siri Carlsen; Magnus Isaksson; Sophie Bensing; Olle Kämpe; Eystein S Husebye; Kristian Løvås; Marianne Øksnes


J Endo Soc. 2021;5(3) 

In This Article

Patients and Methods

Study Design

In a two-step approach, we investigated the expression of GC-regulated genes in different settings of GC replacement in patients with AAD: first, after a high-dose stress test with intravenous 100 mg HC, and second, at normal morning cortisol compared to very low morning cortisol levels. In the second step, normal morning cortisol was achieved through near-physiological replacement with continuous subcutaneous hydrocortisone infusion (CSHI) and very low morning cortisol levels through an over-night (> 15 hour) fast from conventional oral hydrocortisone replacement treatment (OHC).

In the first step, we wanted to explore the immediate effects of high-dose HC exposure on 3 selected genes. In the second step, we compared expression levels of a selected large range of GC-regulated genes in patients with AAD with normal morning cortisol versus very low morning cortisol levels.

Step 1: Gene Expression in Response to 100 mg Hydrocortisone

Patients. Ten patients (50% females) with verified AAD were recruited from the National Registry for Addison's disease (ROAS). Their median daily cortisone acetate dose was 37.5 mg/day (12.5–50 mg/day), equivalent to 30 mg/day HC (10–40 mg/day). Blood samples were collected at 8 to 9 AM (time point 0), after having abstained from cortisone acetate treatment for 18 hours, and repeated at 2 (2h), 4 (4h), and 6 (6h) hours after intravenous infusion of 100 mg HC. Samples were anonymized. All subjects provided written informed consent and ethical permission was granted prior to study start (Norway REK Vest no. 014.03).

RNA Purification From Blood. Whole blood was sampled into Tempus Blood RNA tubes at time points 0, 2h, 4h, and 6h, and RNA extraction from leukocytes performed on the 6100 Nucleic Acid PrepStation (Applied Biosystems, USA). Amount and quality of the extracted RNA was verified by the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, USA) and the Agilent 2100 Bioanalyzer (Agilent Technologies, USA). All RNA samples were qualitatively adequate with RNA integrity numbers between 6.9 and 9.7.

Global Transcriptional Analysis in Blood Using RNA Expression Microarray. Initially, RNA samples from time points 0 and 2h from all 10 patients were included. Microarray experiments were performed using the Applied Biosystems 1700 Expression Array system. An amount of 800 ng of total RNA from each sample was reverse transcribed, amplified, and DIG-labeled (DIG-dUTP; Roche, Germany), using the Applied Biosystems NanoAmp RT-IVT Labeling Kit. The amount (82–147 μg) and the quality of the DIG-labeled cRNA was controlled by both NanoDrop spectrophotometer and Agilent 2100 Bioanalyzer. All samples except one passed the qualitative analysis, and both samples from this individual were removed. Hence, we proceeded with 18 samples. Then, 20 μg of DIG-labeled cRNA was hybridized to the Applied Biosystems Human Genome Survey Microarray v.1.0. The chemiluminescent signal detection, image acquisition, and image analysis of the microarrays were performed on the Applied Biosystems 1700 Chemiluminescent Microarray Analyzer. In addition to the QC report generated by the AB1700 scanner software, the integrated R software was used to control data integrity graphical outputs. Data was loaded onto BioArray Software Enviroment (BASE), and PANTHER was used to annotate the probe IDs.

Among the 150 most upregulated genes in the global transcription analysis, we selected 3 for further investigation (Supplementary appendix 1[15]): FK506 binding protein 51 (FKBP5), matrix metallopeptidase 9 (MMP9), and delta sleep inducing peptide immunoreactor (DSIPI; also known as TSC22D3 and GILZ). FKBP5 was selected as it is an important short-loop feedback inhibitor of GC action;[16]MMP9 because mounting evidence points at its potential as a biomarker in multiple conditions, including cardiovascular[17,18] and inflammatory diseases;[19,20] and DSIPI for being a multi-tissue biomarker of GC action.[21]

Real-time PCR Verification of Differentially Expressed Genes Following 100 mg HC Intravenously. Real-time PCR (rt-PCR) verification of the microarray-data at time points 0, 2h, 4h, and 6h was done for FKBP5, MMP9, and DSIPI using the High-Capacity RNA-to-cDNA Kit and commercially available Taqman-probes. Glyceraldehyde 3-phosphate dehydrogenase (GADPH) was applied as housekeeping gene. The PCR amplification protocol was denaturation at 95 °C for 10 minutes, followed by 40 amplification cycles each at 95 °C in 15 seconds and 60 °C for 1 minute. All samples were run in triplicates on the 7900HT PCR system. The ΔΔCt method was applied to calculate differences between the time points.[22]

Hormone Analysis. Venous blood was collected at time points 0 and 2h for analysis of serum cortisol and plasma ACTH. The hormone analyses were done using immunoassay kits from Diagnostic Products Corp (Los Angeles, CA, USA; Siemens, Cat# L5KCO2, RRID:AB_2877715, and Siemens, Cat# L5KAC2, RRID:AB_2877714) at Haukeland University Hospital, Bergen, Norway, both assays with a coefficient of variation (CV%) <10%.

Step 2: Gene Expression in Normal Versus Very Low Morning Cortisol Levels

Patients and Treatment Modalities. Patient selection and study design has been described in detail elsewhere.[23] In short, a prospective, randomized, cross-over study was conducted, comparing 12 weeks treatment with OHC versus CSHI. The 2 treatment periods were separated by a minimum 8 weeks washout, in which the patient followed his or her usual GC replacement. At the end of each of the 12-week treatment period, the patients returned to the hospital for blood sampling at 8 AM.

For the present study, we used the 2 treatment modalities as a means of controlling morning GC exposure. When treated with OHC, patients were asked to take their final HC dose before 5 PM the day before testing and were practically cortisol depleted upon testing. When treated with CSHI, however, patients continued to receive HC infusions, meaning the blood samples were collected shortly after the simulated morning cortisol peak. For the rest of this paper, we therefore refer to the 2 different GC replacement situations as very low cortisol (>15 hour fasting from OHC) and normal cortisol (CSHI).

Our study cohort consisted of 27 patients with AAD, among whom 18 patients were from Norway and 9 from Sweden (Table 1). The original study cohort included 5 additional patients who were not included here due to lack of RNA samples. All participants provided written informed consent, and ethical approval was granted in both countries before study start (EudraCT #2009-010917-61; NCT 01063569).

Transcriptional and Hormonal Analysis in Blood. All patients provided whole blood, serum, and plasma samples before and after 12 weeks in each treatment arm. Serum cortisol was analyzed by liquid chromatography mass spectrometry[24] and plasma ACTH by chemiluminescent immunometric assay (Immulite 2000; Siemens AG, Munich, Germany; Siemens Cat# L2KAC2, RRID:AB_2783635) with a CV% of ≤10.1% and ≤8%, respectively.

RNA extraction from leukocytes was done using the Tempus Spin RNA Isolation Kit (Applied Biosystems). The cDNA was made using the RT2 First Strand Kit (Qiagen). A customized version of the RT2 Profiler PCR Array Human Glucocorticoid Signaling kit (Qiagen) was then used in order to profile expression of genes of relevance to GC activity. Two of the controls in the panel occurred in triplets, and we replaced the 4 spare copies with SPP1, ARNTL, ARNTL2, and MMP9 (Version 1). In a pilot analysis of version 1, we found 6 genes with gene expression levels below the threshold limit; these were replaced by genes previously shown to be significantly altered in patients with adrenal insufficiency (Version 2; ADM, MMP12, CASP8, EN-RAGE, RETN, CXCL1).[5] The complete overview of the 93 genes included in Version 2 is listed in the Supplementary Appendix (S2).[15]

We analyzed the gene expression levels in 2 samples from each of the 27 patients, collected at the end of each 12-week treatment period. The assays were run according to the protocol of the manufacturer on the 7900HT PCR system (Applied Biosystems). Data was analyzed with the ΔΔCt method using the mean of 4 housekeeping genes as reference (GADPH, HPRT1, RPLP0, and B2M).[22]

Statistics. Descriptive data are presented as median with range. In step 1, normalized signal levels of each probe from the global transcription analysis were log2 transformed and then quartile normalized (Limma package, R software). To account for any negative intensity values produced by the normalization steps, all values had the lowest negative value added to make all intensity values zero or greater, which would then allow for fold change calculations to be carried out. A positive Log2 fold change value represents an increase in fold change where a value of 1 is equal to a doubling of the original value. Next, we performed t tests analysis along with a false discovery rate (FDR) adjustment. The 150 most upregulated genes were ranked according to log2 fold change after excluding genes with FDR above 0.01. Fold change for rt-PCR results was calculated as the ratio between the final over the initial gene expression level. We performed Spearman's correlation (IBM SPSS Statistics) for the relationship between hormone and gene expression levels. Statistical significance was set to 0.05. In step 2, we employed the Wilcoxon signed rank test (IBM SPSS Statistics) to compare normalized gene expression values in normal versus very low morning cortisol, presented as z-score where a score closer to 0 suggests even distribution between the groups. The effect size was calculated by dividing the z-score by the square root of the number of observations (ie, 2 observations for each case, N = 54). To minimize the risk of type 1 error, the significance value was set to P < 0.01. For calculation of fold change, the normalized gene expression value at normal serum cortisol was divided by the very low cortisol for each patient. In both steps, plasma ACTH levels exceeding the upper reference limit (278 pmol/L) were plotted as 278 pmol/L and the lower reference limit (1.1 pmol/L) was plotted as 1.1 pmol/L.