Glucocorticoid activation by 11β‐hydroxysteroid dehydrogenase enzymes in relation to inflammation and glycaemic control in chronic kidney disease: A cross‐sectional study

Summary Objective Patients with chronic kidney disease (CKD) have dysregulated cortisol metabolism secondary to changes in 11β‐hydroxysteroid dehydrogenase (11β‐HSD) enzymes. The determinants of this and its clinical implications are poorly defined. Methods We performed a cross‐sectional study to characterize shifts in cortisol metabolism in relation to renal function, inflammation and glycaemic control. Systemic activation of cortisol by 11β‐HSD was measured as the metabolite ratio (tetrahydrocortisol [THF]+5α‐tetrahydrocortisol [5αTHF])/tetrahydrocortisone (THE) in urine. Results The cohort included 342 participants with a median age of 63 years, median estimated glomerular filtration rate (eGFR) of 28 mL/min/1.73 m2 and median urine albumin‐creatinine ratio of 35.5 mg/mmol. (THF+5αTHF)/THE correlated negatively with eGFR (Spearman's ρ = −0.116, P = 0.032) and positively with C‐reactive protein (ρ = 0.208, P < 0.001). In multivariable analysis, C‐reactive protein remained a significant independent predictor of (THF+5αTHF)/THE, but eGFR did not. Elevated (THF+5αTHF)/THE was associated with HbA1c (ρ = 0.144, P = 0.008) and diabetes mellitus (odds ratio for high vs low tertile of (THF+5αTHF)/THE 2.57, 95% confidence interval 1.47‐4.47). Associations with diabetes mellitus and with HbA1c among the diabetic subgroup were independent of eGFR, C‐reactive protein, age, sex and ethnicity. Conclusions In summary, glucocorticoid activation by 11β‐HSD in our cohort comprising a spectrum of renal function was associated with inflammation and impaired glucose control.


| INTRODUC TI ON
Glucocorticoids (GCs) are steroid hormones that play a critical role in regulating energy metabolism, inflammation, cardiovascular function and behavioural processes. In excess, GCs drive metabolic disease, insulin resistance and cardiovascular disease. 1 Critical to the actions of GCs are their pre-receptor metabolism by the 11beta-hydroxysteroid dehydrogenase (11β-HSD) type 1 and type 2 enzymes. 2 11β-HSD1 is a bidirectional enzyme that primarily converts inactive GC (cortisone) to its active counterpart (cortisol) and is widely expressed throughout the body. 3 In contrast, 11β-HSD2 solely inactivates GCs and is primarily expressed in mineralocorticoid sensitive tissues such as the kidney.
Together, these determine local tissue exposure.
Several studies have described a correlation between loss of renal function and GC activation by 11β-HSD enzymes, measured as a rise in the ratio of cortisol/cortisone or a rise in the ratio of their respective metabolites (tetrahydrocortisol [THF]+5α-tetrahydrocortisol [5αTHF])/ tetrahydrocortisone (THE). 4, 5 Due to limitations of measuring steroid metabolism in this way, the relative contribution of each 11β-HSD enzyme to the reported shift in cortisol metabolism is not known. Loss of renal expression of 11β-HSD2 and diminished cortisol inactivation likely plays a role. 6 However, in vitro and animal models of uraemia propose concurrent 11β-HSD1 upregulation and associated cortisol generation. 7,8 Chronic kidney disease (CKD) entails a pro-inflammatory state. 9 Pro-inflammatory cytokines are potent inducers of 11β-HSD1 activity, promoting local GC activation both in vitro and in vivo. 10,11 Upregulation of 11β-HSD1 expression occurs in several human inflammatory conditions with a strong correlation between systemic measures of inflammation and increases in cortisol activation. 12,14 Existing studies in CKD have not evaluated inflammation as a driver of 11β-HSD1 activity or GC activation.
The inflammatory activation in CKD contributes to insulin resistance. 15,16 Insulin resistance manifests from the early stages of renal impairment, and diabetes mellitus co-exists in one third of people with CKD. 16,18,19 Disturbances in glucose metabolism in the context of CKD are independently associated with cardiovascular disease, accelerated loss of renal function and death. 16,18,19 Elevated 11β-HSD1 activity has been implicated in metabolic disease and insulin resistance. 23,24 Suppression of 11β-HSD1 activity protects against insulin resistance in murine models, including models of uraemia, and has improved HbA1c in Phase II clinical trials for type 2 diabetes. 7,26,27 Whether elevated 11β-HSD1 activity is involved in insulin resistance in the setting of human CKD has not yet been explored.
Given the central role of inflammation in regulating cortisol activation via 11β-HSD1, we investigated its association with shifts in the cortisol-cortisone equilibrium in human CKD. Furthermore, we examined whether augmented GC activation in CKD relates to impaired glucose handling. To test this, we performed a cross-sectional study with 342 participants representing a broad spectrum of renal function. Using urine (THF+5αTHF)/THE as a validated indicator of systemic 11β-HSDmediated GC activation, we examined its correlation with renal function, inflammation and glycaemic control in CKD. 5

| Baseline assessment
Study participants' demographic information, past medical history and concomitant medications were taken from medical records.
Diabetes mellitus was defined by history of diet-controlled diabetes, treatment with anti-diabetic medications or HbA1c greater than 48 mmol/mol.

| Urinary steroid measurements
The balance of systemic (ie, total body) cortisol-cortisone interconversion by 11β-HSD enzymes was taken as the ratio calculated from the urinary excretion of (THF+5αTHF)/THE. This methodology is well established as reflective of 11β-HSD1 activity and has previously been validated in the literature, 30,31 including the use of early morning spot urine samples. 5,34 Spot urine samples were collected at morning clinic visits and stored at −80°C until analysis. Steroids were extracted from 400 µL of urine based on a previously reported protocol. 35 The urine was A Waters Xevo mass spectrometer coupled to an Acquity uPLC with an electrospray ionization source in positive ionization mode was utilized in these experiments. THE, THF and 5αTHF were separated on a BEH C18 1.7 µm 5 cm column at 60°C with the mobile phases methanol and water both with 0.1% formic acid. Gradient profile was starting at 30% methanol, hold for one minute then a linear gradient to 60% methanol at 5 minutes, followed by washing and re-equilibration steps. Steroids were identified by comparison to authentic reference standards, (THE, THF and 5αTHF purchased from Steraloids, Newport, RO, USA), with a matching retention time and identical mass transitions (MRMs) required for positive identification (Supporting Information Table S1 and Figure S1). Steroids were quantified relative to a calibration series ranging from 10-5000 ng/ mL prepared in synthetic urine, including a blank. Steroid concentrations were calculated relative to an assigned internal standard.
Validation of this method is described in the supplementary section (Supporting Information Table S2). Measurements of urinary cortisol and cortisone for supplementary analysis followed similar principles, and full details for this methodology are provided in the supplementary materials (Supporting Information Appendix S1, Table S1 and Figure S2).

| Statistical analysis
Data for continuous variables were not normally distributed. Mann-Whitney U test, Kruskal-Wallis test followed by Dunn's multiple comparisons test and Spearman's correlation test were used as appropriate. For multiple linear regression analysis, outcome variables were log-transformed to comply with the requisite statistical assumptions. For binary logistic regression analysis, urinary (THF+5αTHF)/ THE ratio, eGFR and CRP did not comply with the linearity assumption of independent variables and log odds. These were therefore included as log-transformed variables (CRP) or as stratified variables

| RE SULTS
The clinical and laboratory characteristics of the study participants are summarized in Table 1 (for break-down by aetiology of renal disease see Supporting Information Table S3).

| Renal diagnosis and function
Previous studies have reported increasing GC activation by 11β-HSD enzymes with declining renal function, 4,6 which we tested in our larger cohort. Urine (THF+5αTHF)/THE was elevated in renal disease regardless of aetiology compared to healthy volunteers, except for polycystic renal disease ( Figure 1A). Among cases with renal disease, urine (THF+5αTHF)/THE was lowest in polycystic renal disease and highest in diabetic nephropathy. As in previous reports, we found a negative but weak correlation of urine (THF+5αTHF)/ THE with eGFR (ρ = −0.116, P = 0.032; Figure 1B). There was no association with urinary albumin-creatinine ratio (ACR; ρ = 0.014, P = 0.796; Figure 1C). These results consolidate previous observations that people with renal impairment have higher GC activation by 11β-HSD.

| Inflammation
Chronic kidney disease is known to entail a pro-inflammatory state. 9 This was apparent in our cohort in the negative correlation for eGFR with Creactive protein (CRP; ρ = −0.281, P < 0.001; Figure 2A). Importantly, we identified a significant positive correlation between CRP and urine (THF+5αTHF)/THE (ρ = 0.208, P < 0.001; Figure 2B). This is reflected in the earlier observation of relatively low urine (THF+5αTHF)/THE in patients with polycystic kidney disease, as this subgroup also exhibited low CRP (Supporting Information Table S3). We constructed a multivariable linear regression model to explore independent determinants of urine (THF+5αTHF)/THE. Covariates included demographic factors, markers of renal function (eGFR, urine ACR), inflammation (CRP) and glycaemic control (HbA1c) ( Table 2). Inflammation remained independently associated with urine (THF+5αTHF)/THE. However, there was no independent relationship between eGFR and urine (THF+5αTHF)/ THE in the multiple linear regression model. This suggests that GC activation by 11β-HSD is more closely associated with inflammation than with renal function in this CKD cohort.

| Diabetes mellitus
High 11β-HSD1 activity leads to metabolic dysfunction and impaired insulin signalling, but this has not yet been tested in human CKD.
Having identified that glucocorticoid activation by 11β-HSD is elevated in CKD, we examined the association of urine (THF+5αTHF)/THE with prevalent diabetes. Participants with known diabetes mellitus had higher urine (THF+5αTHF)/THE (median [interquartile range (IQR)]:  (Table 3; urine (THF+5αTHF)/THE and eGFR were categorized for this analysis to comply with the requisite statistical assumptions as detailed in the methods section). The association of high urine (THF+5αTHF)/THE with diabetes was independent of these covariates. The association also persisted after the inclusion of body mass index and family history of diabetes as additional covariates. This supports a link between glucocorticoid activation by 11β-HSD and diabetes mellitus in our CKD cohort.

| HbA1c and poor glycaemic control
To further corroborate the link between elevated GC activation by 11β-HSD with impaired glucose metabolism in CKD, we assessed the relationship of GC activation by 11β-HSD with HbA1c. HbA1c was positively correlated with urine (THF+5αTHF)/THE (ρ = 0.144, P = 0.008; Figure 3B). HbA1c also exhibited a negative association with eGFR (ρ = −0.127, P = 0.020) and a positive association with CRP (ρ = 0.337, P < 0.001) in our cohort. We controlled for these covariates and demographic factors in a multiple linear regression analysis. Age, South Asian ethnicity and CRP remained associated with HbA1c (P < 0.001, P < 0.001 and P = 0.001, respectively; Supporting Information Table   S5), but urine (THF+5αTHF)/THE did not show an independent association in the full cohort (P = 0.659; Supporting Information Table S5).
We considered that the association of urine (THF+5αTHF)/THE with  Table S5). Taken together, these results indicate an association of GC activation by 11β-HSD and disturbances in glucose metabolism in the context of renal impairment.

| Urinary cortisol-cortisone ratio
A supplementary analysis of urinary cortisol-cortisone ratio identified no significant association with eGFR, CRP, prevalent diabetes mellitus or HbA1c (Supporting Information Figure S3). Previous studies did not control for inflammation, which is concomitant with renal impairment. Inflammation is a potent inducer of 11β-HSD1 expression and activity, with significant increases in cortisol activation by 11β-HSD1 seen in several inflammatory diseases. 11,12,14,34,36,37 Whilst this study cannot accurately discriminate between the 11β-HSD types 1 and 2

| D ISCUSS I ON
contributions to systemic GC metabolism, it is unlikely that our findings are solely attributable to loss of renal cortisol inactivation by 11β-HSD2. The inflammatory cytokines TNFα and IL-1β, which accumulate in uraemia, are potent inducers of 11β-HSD1. 9,11 Furthermore, studies in human hepatocytes and animal models of uraemia report marked increases in 11β-HSD1 activity. 7,8 Our results are therefore consistent with a growing body of evidence that increased inflammatory induction of 11β-HSD1 contributes considerably to abnormal GC metabolism in patients with CKD.
In this study, we demonstrate that diabetes, and HbA1c among patients with diabetes, is associated with elevated urine Urinary ACR is displayed on a log scale. Spearman's test was used to assess correlations. ACR, albumin-creatinine ratio; eGFR, estimated glomerular filtration rate; (THF+5αTHF)/THE, tetrahydrocortisol+5α-tetrahydrocortisol/tetrahydrocortisone control. 18 Our data therefore remain consistent with a role for 11β-HSD1 in insulin resistance in human CKD.
Inflammation is a known driver for insulin resistance in CKD. This is evident by CRP showing a negative correlation with eGFR and an independent positive correlation with HbA1c in this study.
Pro-inflammatory cytokines TNFα and IL-1 accumulate in CKD in serum, as well as in insulin target tissues, where they act on IRS-1 and PKB/Akt to impede post-receptor insulin signal transduction. 15,16 In the same tissues, in vitro experiments have demonstrated the capability of TNFα and IL-1β to upregulate 11β-HSD1, which in turn also causes insulin resistance through action on IRS-1 and PKB/Akt. 11,25 Existing biochemical data therefore illustrate a coherent pathway through which inflammation-mediated upregulation of 11β-HSD1 leads to insulin resistance. Importantly, this opens 11β-HSD1 inhibition as a therapeutic target in CKD to treat insulin resistance-and possibly other disturbances like dyslipidaemia-for which a proof of concept experiment in rodents has already been successful. 7 Urinary cortisol-cortisone ratio exhibited a moderate correlation with renal 11β-HSD2 expression in a previous study by Quinkler et al 6 . An association between renal function and cortisol-cortisone ratios in 24 hour urine samples has been described in some but not all studies. 4,6,40 In this study, we did not identify any significant correlations between urinary cortisol-cortisone ratios with eGFR, nor with CRP or Hb1Ac. These divergent results may reflect significant heterogeneity between study populations, F I G U R E 2 Correlation of inflammation with renal function and urine (THF+5αTHF)/THE. A, Inflammation, measured by C-reactive protein, is negatively correlated with eGFR and (B) positively correlated with urine (THF+5αTHF)/THE ratio. C-reactive protein is displayed on a log scale. Spearman's test was used to assess correlations. eGFR, estimated glomerular filtration rate; (THF+5αTHF)/THE, tetrahydrocortisol+5α-tetrahydrocortisol/tetrahydrocortisone TA B L E 2 Independent determinants of 11β-HSD-mediated glucocorticoid activation (n = 337) or that spot urines lack sufficient sensitivity to detect these changes.
Several strengths and limitations are worth noting about this study.
The observational design produced results that demonstrate associations between shifts in glucocorticoid metabolism, inflammation and impaired glucose metabolism, but cannot confirm causal relationships.
By choosing this design, it was nevertheless possible to test predictions from pre-clinical models, which link inflammation with raised 11β-HSD1 activity and insulin resistance, in the clinical setting and thereby providing new evidence for the significance of this pathway in human disease.
The study cohort was large and diverse, which permitted control for potential confounders, strengthens the validity of the results and enhances their relevance for people with CKD. Measuring urine metabolite ratios by liquid chromatography-mass spectrometry as indicator of 11β-HSD activity, whilst offering high convenience for a large study population, limits conclusions about 11β-HSD type-or tissue-specific alterations.
Alterations in 11β-HSD activity lead to much more pronounced shifts of cortisol-cortisone equilibrium in local tissues than in circulation, 7  renal impairment. We identified inflammation as a major determinant of GC activation by 11β-HSD and as a confounder in the correlation of GC activation with renal impairment, thereby expanding on knowledge from previous studies. Higher GC activation by 11β-HSD was associated with disturbances in glucose homoeostasis, offering novel evidence for its clinical significance in patients with CKD. Our data agree with studies from preclinical and non-renal settings that inflammation-induced upregulation of 11β-HSD1 can contribute to metabolic pathology. This study therefore adds credibility to the potential of 11β-HSD1 inhibitors to reduce insulin resistance in populations with pro-inflammatory conditions and improve their cardiovascular risk.

D I SCLOS U R E
All authors declare that they have no relevant financial interests.