Abstract
Over the past 10 years, osilodrostat has become one of the most commonly used steroidogenesis inhibitors in patients with Cushing’s syndrome. The starting dose is usually determined based on the product characteristics, the prescriber’s experience, and cortisol levels. However, no study has attempted to determine whether there was a dose–response relationship between osilodrostat and cortisol reduction. In this study, we developed a preliminary kinetic–pharmacodynamic model to tailor osilodrostat in patients with Adrenocorticotropin hormone (ACTH)-dependent Cushing’s syndrome. We first analyzed the decrease in cortisol 48 hours after initiation or dose change of osilodrostat in 18 patients. Simulations were then performed for different doses of osilodrostat to evaluate the variation in cortisol concentrations. Our results report the first dose–response relationship between osilodrostat dose and cortisol levels, which should be helpful in identifying the optimal dosing regimen in patients with Cushing’s syndrome and in individualizing treatment to approximate a nychthemeral rhythm.
Significance
The current preliminary study is a first step in trying to better understand the effect of osilodrostat on cortisol, which should help determine the optimal dose for each patient.
Introduction
Cushing’s syndrome is a rare condition in which increased cortisol levels lead to a wide range of comorbidities and increased mortality. Surgery is usually regarded as the first-line and most effective treatment.1 In some cases, cortisol-lowering drugs are necessary, mainly after failed surgery.2,3 Among several steroidogenesis inhibitors such as ketoconazole and metyrapone,4,5 osilodrostat, which acts through inhibition of 11β-hydroxylase, is now being considered an effective drug in controlling cortisol hypersecretion. Initially designed as a CYP11B2 inhibitor, the study by Ménard et al.6 involving both animal models and healthy human subjects showed that osilodrostat reduced cortisol levels from a dose of 1 mg/day, while lower doses exerted an anti-aldosterone effect. Since then, several clinical trials and retrospective studies emphasized its efficacy in all etiologies of Cushing’s syndrome.7-9 While the usual recommended starting dose is 2 mg twice a day, precise studies on the short-term effect of osilodrostat on plasma cortisol are lacking. These data could, however, be of interest to tailor the treatment. Moreover, baseline urinary free cortisol (UFC) level is not able to predict response to osilodrostat.10 Taking advantage of serial cortisol measurements performed in inpatient clinics in our center at the time osilodrostat became available, we developed a pharmacokinetic (PK)/pharmacodynamic model of plasma cortisol variation as a function of osilodrostat dose in patients with Adrenocorticotropin-hormone (ACTH)-dependent Cushing’s syndrome.
Patients and methods
Clinical data and hormonal measurements
We retrospectively included patients with ACTH-dependent Cushing’s syndrome, who had serial measurements of plasma cortisol (every 4 hours for 24 hours) before and after the first osilodrostat dose between 2019 and 2024. These measurements were part of our standard of care approach when osilodrostat became available in our tertiary expert center as a thorough evaluation of the efficacy and tolerance of a new drug. The initial dose ranged from 2 to 15 mg/day, depending on the severity of hypercortisolism. Subsequently, osilodrostat dose was gradually adjusted based on the successive cortisol measurements described above. Sex, age at diagnosis, and etiologies were recorded, as well as plasma cortisol measurements 48 hours after the initiation or any change in the osilodrostat dose and time elapsed since change of dose and last administration were recorded. All plasma cortisol measurements were performed with the same Elecsys II Cortisol, Cobas (Roche Diagnostics) assay in the hormonal laboratory of our center; cross-reactivity with 11-deoxycortisol is 4.9%. According to our institutional policy, this retrospective study did not require specific signed informed consent from patients as the data collected were anonymized. It was thus approved by the Ethics Committee of Assistance Publique—Hopitaux de Marseille (RGPD PADS reference RUXXX2). The current study complies with the Declaration of Helsinki.
Pharmacokinetics and statistical analysis
The pharmacodynamic parameters of osilodrostat on cortisol concentrations were analyzed using a kinetic–pharmacodynamic (PD) model in the software Nonlinear Mixed Effects Modeling version 7.4 (NONMEM Icon Development Solutions, Ellicott City, MD, United States). PK analysis from a previously published study6 was used to predict plasma concentration in our patients. The PK parameters were described in the article, and mean concentration values were obtained by digitizing the graph of osilodrostat vs time using the software WebPlotDigitizer version 4.2.11 With these data, a one-compartment population PK model was used to predict osilodrostat concentrations for different dosing regimens. Direct and indirect relationship between osilodrostat-predicted concentration and variation of cortisol concentrations were evaluated to consider a delay. The variation of cortisol concentrations was calculated with reference to a session without treatment. Several functions were tested to describe the relationship such as linear and sigmoidal. Model selection and evaluation were done by the likelihood ratio test (objective function), goodness-of-fit plots (observed vs predicted variation of cortisol concentrations, observed vs individual predictions, normalized prediction distribution errors vs time and variation of cortisol predictions), bootstrap, and visual predictive checks. Graphical analysis was performed with the R software version 4.4.012 using the ggplot2 package.13 Simulations were performed for different doses of osilodrostat to evaluate the variation on cortisol concentrations using the package rxode2.14
Results
Of the patients who were prescribed osilodrostat at least once between 2019 and 2024, 18 were presenting ACTH-dependent Cushing’s syndrome, 12 women (66.6%) and 6 men (33.3%). Mean age was 53.2 ± 15 years. The cause of Cushing’s syndrome was Cushing’s disease in 16 patients (88.9%), ectopic ACTH secretion in 1 patient (5.6%), and ACTH-dependent hypercortisolism of uncertain diagnosis in 1 patient (5.6%). Clinical characteristics are presented in Table 1. It should be noted that none of the patients included were Asian.
Table 1.
Clinical characteristics of patients with all included patients and differentiated according to gender.
| All patientsa | Women | Men | |
|---|---|---|---|
| Age at diagnosis | 53.2 ± 15 | 54 ± 17.2 | 51.5 ± 10.5 |
| Weight | 81.7 ± 13.7 | 79.5 ± 12.7 | 86.2 ± 15.6 |
| % of CD | 88.9 | 83.3 | 100 |
| ULN of 24 hour UFC | 4.4 ± 8.3 | 5.5 ± 10.3 | 2.5 ± 1.8 |
| Osilodrostat starting dose | 3.3 ± 2.2 | 3.7 ± 2.4 | 2.5 ± 1.4 |
| Cortisol before osilodrostat intake | 422.9 ± 159.2 | 414.7 ± 176.6 | 439.4 ± 130.7 |
| Cortisol 4 hour after osilodrostat | 404 ± 165.6 | 408.2 ± 200.1 | 395.5 ± 70.8 |
Abbreviations: CD, Cushing’s disease; ULN, upper limit range; UFC, urinary free cortisol.
aOf note, none of the included patients were Asian.
In their article, Ménard et al.6 showed that the dose–exposure relationship was not strictly proportional. A one-compartment model was enhanced by increasing the relative bioavailability with the dose and was estimated that the dose resulting in a 50% increase in bioavailability was 1.06 mg. The PK parameters derived from Ménard et al.6 were fixed and used to predict osilodrostat concentration in our patients. A direct relationship between the predicted osilodrostat concentrations and variation of cortisol concentrations (%) gave a better fit than an indirect model. The drug effect was modeled with the following sigmoidal function (Eq. 1);
(1)
where Imax is the maximal inhibition and IC50 is the apparent half-maximal inhibitory concentration.
The estimated PD parameters were IC50 and Imax. Their values as well as the relative standard errors (RSE%) and the corresponding bootstrap IC50 are shown in Table 2. Final parameters were used to simulate n = 500 profiles following a single dose of osilodrostat.
Table 2.
Pharmacodynamic parameters of osilodrostat’s effects on the variation of cortisol concentrations.
| Parameters | Unit | Estimation | RSE% | Bootstrap | |
|---|---|---|---|---|---|
| 0.025 | 0.975 | ||||
| KA (fixed)a | 1/hour | 4.03 | — | — | — |
| CL/F (fixed)a | L/hour | 18.3 | — | — | — |
| V/F (fixed)a | L | 125 | — | — | |
| Imax | % | 44.5 | 18.7 | 12.51 | 90.9 |
| IC50 | mg/L | 0.011 | 37.4 | 0.0001 | 0.10 |
| Interindividual variability (ω) | |||||
| Imax | 0.40 | 30.9 | 0.003 | 1.86 | |
| IC50 | 3.78 | 41.0 | 0.003 | 9.22 | |
| Residual unexplained variability (σ) | |||||
| Additive | % | 23.8 | 12.2 | 18.2 | 29.9 |
Abbreviations: CL/F, apparent clearance; IC50, osilodrostat concentration associated with half the maximal inhibition of the cortisol variation; Imax, maximum inhibitory effect of osilodrostat on the variation of cortisol; KA, first-order absorption rate constant; RSE, relative standard error; V/F, apparent volume of distribution.
aAdapted from Ménard et al.6
The effects on plasma cortisol variation are depicted in Figure 1. Cortisol concentration declines during the first hour after taking osilodrostat, from 24% for a 1 mg dose to over 42% for a 20 mg dose. Thereafter, from the first hour onward, cortisol increases progressively, with loss of treatment efficacy occurring around the 10th-15th hour for 1 and 2 mg, while for doses above 5 mg, a moderate effect persists over the following hours. Figure 2 shows the variation in cortisol concentration for a 2 mg dose, with median decrease in cortisol variation of 31%, ranging from 0% to 67.5%, with, as mentioned above, a maximum effect 1 hour after osilodrostat intake, and a progressive increase in cortisol levels, mainly during the 12 hours following treatment. The same analysis for 10 mg revealed a median reduction in cortisol of 38%, ranging from 5% to 80%. Figure 3 describes the relationship between osilodrostat concentration and cortisol variation, showing that the maximum effect corresponds to the maximum concentration and that a decrease in osilodrostat concentration results in an increase in cortisol level.
Relationship between time since last administration of osilodrostat and cortisol concentrations.
Visual predictive variation on cortisol concentrations following 2 or 10 mg osilodrostat administration.
Relation between osilodrostat concentration and cortisol variation.
Discussion
To the best of our knowledge, this is the first study that attempts to define a dose/efficacy relationship between osilodrostat dose and the variation of plasma cortisol. First, our results suggest that the effect of osilodrostat appears immediately after the peak of concentration, 1 hour after treatment intake, which highlights the parallel evolution of osilodrostat and cortisol concentrations. This is unusual, as typically effect peak takes few hours, following concentration peak.15 The relationship between osilodrostat concentration and the effect on cortisol is not linear but sigmoidal with a rapid increase in concentrations producing a rapid significant effect, leading to a maximal effect. Because elimination is a slower process than absorption, the effect’s decline will also be slower: this means that efficiency remains stable during the first 5 hours, with a further progressive increase of cortisol and a loss of efficiency around 10-15 hours after intake. This confirms the need for two intakes per day, with one early in the morning and the other 12 hours later in the evening. In addition, even if our simulation suggests a wide interindividual variability, we were able to determine the impact of different doses of osilodrostat on the percent decrease in plasma cortisol levels. For instance, 20 mg osilodrostat leads to an estimated 42% decrease in cortisol concentration. Interestingly, Ferrari et al.16 recently showed that patients controlled with two doses of osilodrostat for at least 1 month had the same efficacy with a single intake (combing both doses) at 4 or 7 Pm. This is quite surprising and will need to be evaluated in future studies: our preliminary model could give more precise information on this point.
Cushing’s syndrome is also characterized by a loss of circadian rhythm leading to increased comorbidities such as diabetes, hypertension, and cardiovascular disease.17,18 This is why 24 hour UFC can only be considered an imperfect marker of glucocorticoid overexposure even though it is an easy-to-use marker, as exemplified by its use in all the clinical trials performed on cortisol-lowering drugs.7,8,10,19 Predicting the efficacy of osilodrostat on plasma cortisol might be helpful to tailor the treatment as a titrating approach. Of note, some studies suggested that there might be an inpatient variability of cortisol secretion in Cushing’s syndrome,20 and this might account for a bias in our results. However, none of our patients had cyclical Cushing’s syndrome. Moreover, 12 patients in our cohort had at least two cortisol cycles (every 4 hours during the day) before starting treatment. A comparison of these two cycles using Student’s t-test showed no significant difference (P = .7), indicating no obvious spontaneous variability. Our preliminary report gives interesting insights into the maximal efficacy expected for a single dose of osilodrostat, thus defining the initial dosage needed to rapidly control hypercortisolism, as opposed to the dose currently recommended by the manufacturer (2 mg twice daily). Thus, our results could help define an optimal dose in the morning, but also in the evening, with the aim of re-establishing a circadian profile. This will, however, have to be confirmed on an interventional study focusing on comorbidities, quality of life and their potential improvements while using this PK model.
The main limitation of this proof-of-concept study is the large CI. This may be due to the relatively low number of patients and the fact that cortisol was measured every 4 hours instead of every hour, but also to the large variability in efficacy between subjects. Due to the number of patients included in the analysis, it was not possible to investigate further if a covariate, such as the gender, may explain these differences between individuals. It is important to highlight that although our model predicts cortisol levels 1 hour post intake as the most reliable predictor of future efficacy, cortisol measurements were taken every 4 hours. Thus, this finding should be confirmed in prospective studies with more frequent cortisol measurements, particularly 1 hour after osilodrostat administration. While the kinetic–pharmacodynamic approach used in this study can present with some inherent limitations, this type of approach is regularly used to define the modalities of use for a medication in a new indication. A nonlinear mixed-effects modeling allows the use of data from the routine clinical follow-up of patients. This method is thus effective and particularly well-suited for sparse data. Finally, a larger study could include closer measurements of cortisol. Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) is the best method for avoiding cross-reactivity with steroid precursors and could be used for these measurements. However, we used the Elecsys Cortisol II Immunoassay, which shows <5% cross-reactivity with 11-deoxycortisol; thus, our results are credible.
In conclusion, we designed a kinetic–pharmacodynamic model to adapt osilodrostat in patients with ACTH-dependent Cushing’s syndrome. Our model shows that cortisol level 1 hour after treatment is the best indicator of future efficacy. Moreover, depending on the initial cortisol level and the goal to be achieved, different doses should be prescribed. Despite wide inter-patient variability, we believe our model provides insight into the minimal dose necessary to decrease cortisol levels and the maximal efficacy expected for a given dose. Thus, it should help physicians tailor the treatment to reach maximal efficacy in the shortest possible time. The next step will be to analyze whether this percent decrease remains stable on a long-term basis or becomes more important with time, as suggested by some clinical cases showing delayed adrenal insufficiency on stable doses of osilodrostat.21
Authors’ contributions
Cecilia Piazzola (Conceptualization [equal], Formal analysis [equal], Writing—original draft [equal]), Frederic Castinetti (Conceptualization [equal], Formal analysis [equal], Writing—review & editing [equal]), Katharina von Fabeck (Conceptualization [equal], Writing—review & editing [equal]), and Nicolas Simon (Conceptualization [equal], Methodology [equal], Supervision [equal], Validation [equal], Writing—original draft [equal], Writing—review & editing [equal])
Funding
This work received an unrestricted educational grant from Recordati Rare Diseases.
To see the references and the original article, please go here: https://academic.oup.com/ejendo/article/193/4/K11/8255719?login=false
Filed under: adrenal crisis, Clinical trials, Cushing's, pituitary, Treatments | Tagged: adrenal insufficiency, cortisol, Cushing's, Osilodrostat | Leave a comment »
CushieBlog 
