11β-Hydroxysteroid dehydrogenase type 1 inhibition in idiopathic intracranial hypertension: a double-blind randomized controlled trial

Abstract Treatment options for idiopathic intracranial hypertension are limited. The enzyme 11β-hydroxysteroid dehydrogenase type 1 has been implicated in regulating cerebrospinal fluid secretion, and its activity is associated with alterations in intracranial pressure in idiopathic intracranial hypertension. We assessed therapeutic efficacy, safety and tolerability and investigated indicators of in vivo efficacy of the 11β-hydroxysteroid dehydrogenase type 1 inhibitor AZD4017 compared with placebo in idiopathic intracranial hypertension. A multicenter, UK, 16-week phase II randomized, double-blind, placebo-controlled trial of 12-week treatment with AZD4017 or placebo was conducted. Women aged 18–55 years with active idiopathic intracranial hypertension (>25 cmH2O lumbar puncture opening pressure and active papilledema) were included. Participants received 400 mg of oral AZD4017 twice daily compared with matching placebo over 12 weeks. The outcome measures were initial efficacy, safety and tolerability. The primary clinical outcome was lumbar puncture opening pressure at 12 weeks analysed by intention-to-treat. Secondary clinical outcomes were symptoms, visual function, papilledema, headache and anthropometric measures. In vivo efficacy was evaluated in the central nervous system and systemically. A total of 31 subjects [mean age 31.2 (SD = 6.9) years and body mass index 39.2 (SD = 12.6) kg/m2] were randomized to AZD4017 (n = 17) or placebo (n = 14). At 12 weeks, lumbar puncture pressure was lower in the AZD4017 group (29.7 cmH2O) compared with placebo (31.3 cmH2O), but the difference between groups was not statistically significant (mean difference: −2.8, 95% confidence interval: −7.1 to 1.5; P = 0.2). An exploratory analysis assessing mean change in lumbar puncture pressure within each group found a significant decrease in the AZD4017 group [mean change: −4.3 cmH2O (SD = 5.7); P = 0.009] but not in the placebo group [mean change: −0.3 cmH2O (SD = 5.9); P = 0.8]. AZD4017 was safe, with no withdrawals related to adverse effects. Nine transient drug-related adverse events were reported. One serious adverse event occurred in the placebo group (deterioration requiring shunt surgery). In vivo biomarkers of 11β-hydroxysteroid dehydrogenase type 1 activity (urinary glucocorticoid metabolites, hepatic prednisolone generation, serum and cerebrospinal fluid cortisol:cortisone ratios) demonstrated significant enzyme inhibition with the reduction in serum cortisol:cortisone ratio correlating significantly with reduction in lumbar puncture pressure (P = 0.005, R = 0.70). This is the first phase II randomized controlled trial in idiopathic intracranial hypertension evaluating a novel therapeutic target. AZD4017 was safe and well tolerated and inhibited 11β-hydroxysteroid dehydrogenase type 1 activity in vivo. Reduction in serum cortisol:cortisone correlated with decreased intracranial pressure. Possible clinical benefits were noted in this small cohort. A longer, larger study would now be of interest.


Introduction
Idiopathic intracranial hypertension (IIH) is a debilitating condition characterized by raised intracranial pressure (ICP), papilledema, with the risk of permanent visual loss (Mollan et al., 2018b) and chronic headaches, which reduce the quality of life (Mulla et al., 2015). IIH predominately affects obese women between the ages of 25 and 36 years with a distinct androgen excess signature recently identified (Daniels et al., 2007;Markey et al., 2016;O'Reilly et al., 2019). Incidence is increasing in line with escalating worldwide obesity rates (Mollan et al., 2018a).
We have previously demonstrated that the enzyme 11bhydroxysteroid dehydrogenase type 1 (11b-HSD1) is expressed and active in the choroid plexus to amplify cortisol availability and acts to regulate cerebrospinal fluid (CSF) production (Sinclair et al., 2007(Sinclair et al., , 2010b; Gathercole et al., 2013). In patients with IIH, resolution of disease (reduced ICP, improvements in papilledema and headaches) was associated with reduced 11b-HSD1 activity (Sinclair et al., 2010a, b), with a study suggesting that inhibition of 11b-HSD1 with a non-selective inhibitor lowered intraocular pressure. Importantly, 11b-HSD1 expression and activity are dysregulated in obesity (Wake and Walker, 2004;Sandeep et al., 2005).
Selective inhibitors of 11b-HSD1 have been developed as treatments for obesity, hepatic steatosis, metabolic syndrome and type 2 diabetes (Boyle, 2008;Stefan et al., 2014). Based on these data, 11b-HSD1 could represent a therapeutic target for lowering CSF pressure. AZD4017 is a highly selective, fully reversible, competitive 11b-HSD1 inhibitor. It has been tested over short time intervals in healthy males (9 days) and abdominally obese subjects (10 days) and found to be safe and tolerable (AstraZeneca, 2000a, b, c, d, e). The ability of AZD4017 to penetrate the blood-brain barrier is not established; however, the choroid plexus lies outside the blood-brain barrier and consequently can be targeted directly following oral administration (Davson, 1966;Eftekhari et al., 2015).
We hypothesized that the inhibition of 11b-HSD1 could be therapeutically beneficial in IIH. To test this theory, we conducted a multicenter, phase II doubleblind, placebo-controlled RCT in IIH using the selective 11b-HSD1 inhibitor AZD4017, aiming to assess therapeutic efficacy, safety and tolerability, and investigated in vivo systemic and central nervous system efficacy.

Study conduct
The study was conducted from March 2014 to December 2016 in three UK hospitals (Fig. 1). The National Research Ethics Committee York and Humber-Leeds West gave ethical approval (13/YH/0366). In vitro subcutaneous and omental adipose explants were collected from a separate IIH population undergoing bariatric surgery (National Research Ethics Committee Black Country 14/WM/0011). All patients provided written informed consent in accordance with the declaration of Helsinki. Detailed clinical trial methodology has been published (Markey et al., 2017).

Study population
Women (18-55 years) were eligible if they had a clinical diagnosis of active IIH meeting the updated, modified Dandy criteria (ICP > 25 cmH 2 O and active papilledema) and normal brain imaging (including magnetic resonance venography or CT with venography) at recruitment (for detailed eligibility criteria see Supplementary Table 1; Friedman et al., 2013;Markey et al., 2017).

Study design
This study was a 16-week phase II, UK, multicenter, double-blind, placebo-controlled RCT with a 12-week dosing duration and 4-week follow-up of drug.

Randomization and blinding
Participants were allocated to either the study drug or placebo using a trial number randomly allocated by phone, using block-of-6 randomization. Randomization was performed by an independent manufacturer (Almac) (Markey et al., 2017). Participants and investigators were masked to treatment allocation during the trial.

Intervention
Participants received 400 mg of an oral selective 11b-HSD1 inhibitor, AZD4017, twice daily for 12 weeks, compared with a matched placebo. Trial dosing was added to existing therapy for IIH; other drugs were maintained at a fixed dose throughout the study. Patients with IIH are typically informed about the importance of weight loss to treat their IIH as part of routine standard of care at diagnosis; however, they were not treated with an additional weight management intervention during this trial.

Clinical assessments
The primary outcome for clinical efficacy was the difference in ICP between AZD4017 and placebo, as measured by lumbar puncture at 12 weeks. Secondary outcomes included IIH symptoms, visual function [visual acuity measured using log of the minimum angle of resolution, perimetric mean deviation (PMD) using Humphrey 24-2 central threshold automated perimetry and contrast sensitivity assessed by MARs charts; Mars Perceptrix, USA], papilledema, headache-associated disability and anthropometric measures. Papilledema was evaluated using spectral domain optical coherence tomography (Spectralis, Heidelberg Engineering) to quantify the peripapillary retinal nerve fibre layer (RNFL) average and maximal values. Papilledema was graded from fundal photographs by three masked neuroophthalmologists using the Frisen classification (0 denotes no papilledema to grade 5 severest papilledema) (Frisen, 1982). Headache was evaluated through the headache impact test-6 disability questionnaire, headache severity (verbal rating scores 0-10), frequency (days per month), duration and analgesic use (days per month) (Bayliss et al., 2003). Pill counting at each visit determined drug compliance.
In the original grant application and early versions of the protocol, the primary outcome measure was stated as the change in ICP between baseline and 12 weeks. Following adoption of the study by the Birmingham Clinical Trials Unit, the primary outcome was changed to ICP at 12 weeks, with adjustment for baseline ICP in the analysis. This change was made blind to any data analysis.
Stopping criteria for the trial would occur if more than one participant met Hy's law criteria or 10% or more had bilirubin levels over two times the upper limit of normal or alanine transaminase and/or aspartate transaminase over five times the upper limit of normal for 7 days, and no other cause for the liver dysfunction could be identified (Markey et al., 2017).

Glucocorticoid and AZD4017 blood and cerebrospinal fluid levels
Samples were collected and stored at À80 C. Cortisol and cortisone levels in serum and CSF were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS) at the University of Birmingham, as previously described (Hassan-Smith et al., 2015;Long et al., 2016). Plasma and CSF AZD4017 levels were quantified by an external laboratory (Alderley Analytical, Knutsford, UK).

In vivo hepatic 11-hydroxysteroid dehydrogenase activity
Inhibition of hepatic 11b-HSD1 activity was informed by measuring first-pass metabolism of 10 mg of oral prednisone to prednisolone. Serum prednisone and prednisolone were measured every 20 minutes over 4 hours using LC-MS/MS (Richards et al., 2012;Hassan-Smith et al., 2015).

Statistical analysis
Analysis of the clinical data was based on the full analysis set according to the statistical analysis plan (Supplementary material). Analysis was conducted using intention-to-treat with data from all available randomized participants used. The primary comparison was between AZD4017 versus placebo at 12 weeks. The majority of data was continuous, so groups were compared using linear regression models with baseline measurements included as a covariate in the model. IIH symptom data were binary and were analysed using log-binomial models with baseline symptom included as a covariate in the model. The primary analysis of visual data included data from both eyes, using a linear mixed model with participant included as a random effect. We also analysed data from the most affected eye at baseline as defined by PMD (Friedman et al., 2014). Statistical significance was set at P < 0.05, with no adjustment for multiple comparisons made. Clinical data were analysed using SAS (version 9.4) and STATA (version 14).
Analysis of laboratory data was performed using SPSS (version 24; IBM, New York, NY, USA). All laboratory data were continuous. The primary comparison between groups used an unpaired t-test for normally distributed data (Mann-Whitney U test for non-parametric data). For within-group comparisons (e.g. comparing baseline with 12-week data in one group), either the paired t-test or Wilcoxon signed-rank test was used for parametric or non-parametric data, respectively. We reported mean and standard deviation for parametric data (medians and ranges for non-parametric data).

Sample size
To detect a difference between groups of 14% in ICP (assuming a standard deviation of 10% for ICP) with 90% power and two-sided alpha ¼ 0.05, required 12 participants per group. Allowing for 20% drop out, we aimed to recruit 30 participants.

Data availability
The trial is registered at Clinicaltrials.gov NCT02017444; European Clinical Trials Database (EudraCT Number: 2013-003643-31). The data that support the findings of this study are available from the corresponding author, upon reasonable request.

Results
A total of 31 participants were recruited: 17 participants were randomized to AZD4017 and 14 participants were randomized to placebo (Fig. 1B). Baseline characteristics represent the cohort of patients with IIH with active disease recruited (Table 1). Baseline characteristics were not significantly different between trials arms, although mean deviation differed between groups (AZD4017: À3.4 db versus placebo: 6.1 db; P ¼ 0.077). Acetazolamide was continued at a stable dose in 32% of participants (balanced between the trial arms; Table 1) and no other pharmacological IIH treatments were taken by the trial cohort.
Data from both eyes were also analysed but yielded equivalent results to that of the worst eye.
All headache outcomes were not statistically significantly different between AZD4017 and placebo at Weeks 12 or 16 (Supplementary Table 4). There were also no statistically significant differences in any of the anthropometric outcomes (body mass index, waist:hip ratio). Specifically, the mean difference in body mass index at 12 weeks between arms was 0.4 kg/m 2 (95% CI: À0.6 to 1.4). Both trial arms saw a minimal increase in weight: AZD4017 group increased by 1.21 kg (95% CI: À0.47 to 2.89) and the placebo group increased by 0.04 kg (95% CI: À1.88 to 1.96). The body mass index change also saw a small increase in both groups: AZD4017 group by 0.6 kg/m 2 (95% CI: À0.2 to 1.3) and placebo group by 0.2 kg/m 2 (95% CI: À0.5 to 0.8).

Safety and tolerability
Study medication was well tolerated with participants in both arms taking on average 98% of the total 168  No differences were noted between treatment groups for the safety blood tests (Supplementary Table 6). As expected, there was a rise in the hypothalamic pituitary adrenal stimulatory hormone, adrenocorticotropic hormone, over 12 weeks in the AZD4017 group (mean difference at 12 weeks: 12.36 ng/l, 95% CI: À0.03 to 24.74). There was no difference in serum cortisol, testosterone or androstenedione, although serum dehydroepiandrosterone sulphate, a marker of adrenal androgen production, was higher at 12 weeks in the AZD4017 group (mean difference at 12 weeks: 5.44 nmol/l, 95% CI: 1.09-9.79); levels returned to normal 4 weeks after treatment cessation (Week 16).

In vivo assessments
Blood and cerebrospinal fluid levels of AZD4017 and glucocorticoids AZD4017 concentrations were detected in the serum after 1 week of treatment and sustained at Week 12 (n ¼ 6). The presence of AZD4017 in the CSF was 0.5% that of the serum (Supplementary Table 7). No AZD4017 was detected in the placebo group at any time point.
Serum and CSF cortisol and cortisone were examined in the placebo and AZD4017 groups at baseline and at 12 weeks. There was no difference in the serum cortisol:cortisone ratio at baseline; however, the ratio fell significantly in the AZD4017 group between baseline and 12 weeks (P ¼ 0.0083), while it did not in the placebo group, and was significantly lower in the AZD4017 group than in the placebo group at 12 weeks (P ¼ 0.0125), indicating inhibition of 11b-HSD1 activity (Fig. 3A). Similarly, the CSF cortisol:cortisone ratio did not differ between arms at baseline; however, at Week 12, there was a significant decrease in the CSF cortisol:cortisone in the AZD4017 group compared with placebo (P ¼ 0.002) and the AZD4017 group between baseline and 12 weeks (P ¼ 0.03) (Fig. 3B), implying that the systemic inhibition of 11b-HSD1 activity can regulate CSF glucocorticoid exposure. Importantly, in the AZD4017 group, changes between baseline and 12 weeks (n ¼ 15 with paired data) in both the serum cortisol and the cortisol:cortisone ratio significantly correlated with change in lumbar puncture pressure (R ¼ 0.65, P ¼ 0.01 and R ¼ 0.70, P ¼ 0.005, respectively), while, as expected, changes in the cortisone levels did not (Fig. 3C-E). There was no correlation between body mass index and the serum cortisol or cortisol:cortisone ratio. In the small subgroup (n ¼ 6) in which changes in CSF glucocorticoids were measured, we did not identify a significant correlation with changes in lumbar puncture pressure.

Hepatic 11-hydroxysteroid dehydrogenase activity
The placebo group had robust capacity to generate prednisolone following oral prednisone at both baseline and after 12 weeks. The baseline prednisolone generation curve for the AZD4017 group was indistinguishable from the placebo curve; however, at 12 weeks, the AZD4017 group was essentially unable to generate prednisolone ( Fig. 3D and E), indicating effective inhibition of hepatic 11b-HSD1 activity. Area under the curve analysis of the mean time points at 12 weeks showed significantly impaired prednisolone generating capacity for AZD4017 versus placebo (228 6 99 versus 1738 6 142; P < 0.0001), an 85.9% reduction (P < 0.0001) in overall prednisolone generating capacity after 12 weeks ( Fig. 3D and E). There was no correlation between the change in the area under the curve for prednisolone and ICP (R ¼ 0.1; Adipose 11-hydroxysteroid dehydrogenase activity While AZD4017 effectively inhibited hepatic 11b-HSD1, we were unable to show impaired capacity to generate cortisol from cortisone in explanted subcutaneous adipose tissue biopsies. At baseline and following 12 weeks of oral AZD4017 (n ¼ 5), there was no significant change in total cortisol versus placebo (9.0 6 5.6 versus 12.4 6 4.9 nmol; P ¼ 0.3) or percentage conversion of cortisone to cortisol (23 6 14 versus 27 6 18%; P > 0.99) and no change in those treated with placebo (n ¼ 6; Fig. 3F). However, AZD4017 was able to significantly inhibit 11b-HSD1 activity when added to ex vivo adipose explants from subcutaneous and omental depots. The 20 nM AZD4017 significantly impaired the conversion of cortisone to cortisol (>70% versus control), and 200 nM onwards was sufficient to effectively block cortisol generation, particularly in the subcutaneous depot ( Fig. 3G and H).

Discussion
We report the first phase II RCT assessing an 11b-HSD1 inhibitor AZD4017 for the treatment of IIH. We have shown some possible clinical benefit for AZD4017 and have also shown that it was well tolerated and safe. We found evidence for effective in vivo 11b-HSD1 inhibition.
12 weeks of treatment. ICP was the primary clinical outcome measure, representing the hallmark of the disease driving clinical sequelae. At 12 weeks, although ICP was lower in the AZD4017 group compared with placebo, the difference between groups was not statistically significant. Exploratory analyses of the mean change within groups found a significant improvement in ICP in the AZD4017 group between baseline and 12 weeks but not in the placebo group. In support of our hypothesis, among the AZD4017 group, the change in serum cortisol and cortisol:cortisone ratio over the treatment period, a marker of 11b-HSD1 inhibition, correlated significantly with reduction in ICP. Of note, a minimal clinically important change in ICP in IIH has not been determined in IIH and establishing one would be useful for future trials. In addition, previous trials have noted that ICP reduction below the cut-off of 25 cmH 2 O is not universally required to translate into resolution of IIH clinical features (Sinclair et al., 2010a).
The visual field perimetric assessment is another clinically meaningful measure and has been selected as the primary outcome measures in previous IIH trials. We found no difference between groups in PMD at 12 weeks; however, there was significant improvement over time in the AZD4017 arm but not in the placebo arm. This may reflect the pragmatic recruitment of all degrees of PMD at enrolment (including those with severe visual loss with limited capacity to improve), while other trials have restricted enrolment to a selected cohort (e.g. À2 to À5 dB) (NORDIC Idiopathic Intracranial Hypertension Study Group Writing Committee et al., 2014). In addition, in those with a PMD near normal at baseline (despite papilloedema), there may be a floor effect, where no further improvement is possible in the PMD in these individuals. In addition, this small trial was not powered to determine significance in the secondary outcome measures.
Headache is a key disabling feature in IIH (Mulla et al., 2015). We did not detect differences between the groups in any of the headache assessments at 12 weeks, although data from the patient-completed headache impact test-6 favoured the AZD4017 group. Evaluating the effect of AZD4017 on headache measures over a longer treatment duration would be of interest.
Previous trials showed that 11b-HSD1 inhibition leads to adaptive changes in hypothalamic pituitary adrenal stimulatory hormone adrenocorticotropic hormone and the adrenal androgen precursor dehydroepiandrosterone sulphate. Our data support these findings, but with no significant change in downstream effector hormones (cortisol and testosterone).
In vivo evaluation of our patients demonstrated that AZD4017 was a highly effective systemic and hepatic 11b-HSD1 inhibitor, in line with previous studies using 11b-HSD1 inhibitors in humans (Courtney et al., 2008;Schwab et al., 2017). Systemic efficacy may modify metabolic aspects of IIH with indirect benefits on ICP (Hornby et al., 2018).
While AZD4017 effectively inhibited 11b-HSD1 when applied to subcutaneous and omental adipose tissue explants, we were unable to prove inhibition in vivo and propose that with this experimental design, 11b-HSD1 activity recovers over the assay period once removed from AZD4017, a reversible competitive inhibitor.
Blood-brain barrier AZD4017 penetrance was low, with levels in the CSF 0.5% those of plasma levels, but was associated with reduced CSF cortisol:cortisone ratio suggesting that 11b-HSD1 may contribute to cortisol availability in the CSF.

Limitations
We were unable to directly evaluate 11b-HSD1 inhibition at the choroid plexus, the tissue responsible for CSF secretion; hence, we cannot be certain of inhibition by 11B-HSD1 at the target tissue. We have evaluated the efficacy of other IIH drugs using rodent ICP monitoring models (Botfield et al., 2017;Scotton et al., 2019), but AZD4017 is only effective in humans and primates, thus limiting our ability to evaluate its action in rodent models. The trial duration was likely too short, with insufficient time to detect clinical efficacy. A duration of 12 weeks was chosen for the evaluation of safety and tolerability and represented the longest duration of dosing to date with AZD4017. This may not have been sufficient for the meaningful evaluation of clinical outcomes with other IIH RCTs evaluating drugs over a 6-month period (Committee et al., 2014). The enrolment criteria for the study were deliberately broad allowing inclusion of a spectrum of patients with IIH with active disease and ensuring the generalizability of results; however, this did not allow evaluation in disease subgroups such as those with mild visual loss versus those with severe irreversible visual loss. Finally, the sample size (31 participants) is small, which may have reduced our power and limited meaningful evaluation of clinical measures and the trial was not designed to establish significant changes in the secondary clinical outcome measures.

Conclusion
This is the first phase II study evaluating the novel pharmacological therapy AZD4017 in IIH. We demonstrate safety, tolerability and provide strong in vivo evidence for effective 11b-HSD1 inhibition. There was a significant reduction in ICP in the AZD4017 and not the placebo group over the treatment duration (exploratory within-group analysis) and reduction in ICP significantly correlated with reduction in serum cortisol:cortisone ratio; however, the primary analysis evaluating the difference between groups at 12 weeks did not reach statistical significance. The data suggest that 11b-HSD1 inhibition may have utility for reducing the effects and consequences of raised ICP in patients with IIH. Further evaluation of these therapeutic strategies in this disabling disease, for which few useful medical options exist, would be worthwhile.

Supplementary material
Supplementary material is available at Brain Communications online.