Volume 16, Issue 9 p. 1667-1679
ARTICLE
Open Access

Anti-inflammatory effects of dexamethasone in COVID-19 patients: Translational population PK/PD modeling and simulation

Artur Świerczek

Corresponding Author

Artur Świerczek

Department of Pharmacokinetics and Physical Pharmacy, Faculty of Pharmacy, Jagiellonian University Medical College, Kraków, Poland

Correspondence

Artur Świerczek, Department of Pharmacokinetics and Physical Pharmacy, Faculty of Pharmacy, Jagiellonian University Medical College, 30-688 Kraków, Poland.

Email: [email protected]

Search for more papers by this author
William J. Jusko

William J. Jusko

Department of Pharmaceutical Sciences, School of Pharmacy and Pharmaceutical Sciences, State University of New York at Buffalo, Buffalo, New York, USA

Search for more papers by this author
First published: 29 June 2023

Abstract

Dexamethasone (DEX) given at a dose of 6 mg once-daily for 10 days is a recommended dosing regimen in patients with coronavirus disease 2019 (COVID-19) requiring oxygen therapy. We developed a population pharmacokinetic and pharmacodynamic (PopPK/PD) model of DEX anti-inflammatory effects in COVID-19 and provide simulations comparing the expected efficacy of four dosing regimens of DEX. Nonlinear mixed-effects modeling and simulations were performed using Monolix Suite version 2021R1 (Lixoft, France). Published data for DEX PK in patients with COVID-19 exhibited moderate variability with a clearance of about half that in healthy adults. No accumulation of the drug was expected even with daily oral doses of 12 mg. Indirect effect models of DEX inhibition of TNFα, IL-6, and CRP plasma concentrations were enacted and simulations performed for DEX given at 1.5, 3, 6, and 12 mg daily for 10 days. The numbers of individuals that achieved specified reductions in inflammatory biomarkers were compared among the treatment groups. The simulations indicate the need for 6 or 12 mg daily doses of DEX for 10 days for simultaneous reductions in TNFα, IL-6, and CRP. Possibly beneficial is DEX given at a dose of 12 mg compared to 6 mg. The PopPK/PD model may be useful in the assessment of other anti-inflammatory compounds as well as drug combinations in the treatment of cytokine storms.

Study Highlights

  • WHAT IS THE CURRENT KNOWLEDGE ON THE TOPIC?

Dexamethasone (DEX) given orally at a dose of 6 mg once-daily for 10 days is a recommended and effective dosing regimen in patients with coronavirus disease 2019 (COVID-19) requiring oxygen therapy. TLR4 signaling is involved in the development of COVID-19 and increased plasma concentrations of TNFα, IL-6, and CRP are strongly correlated with poor outcomes in patients with the COVID-19 cytokine storm.

  • WHAT QUESTION DID THIS STUDY ADDRESS?

Is there sufficient knowledge regarding the pharmacokinetics (PKs) of DEX and its mechanisms of action in suppression of cytokines to lend insight into current dosing regimens in treatment of the cytokine storm in COVID-19?

  • WHAT DOES THIS STUDY ADD TO OUR KNOWLEDGE?

Our parsimonious and mechanism-based PK/pharmacodynamic (PD) model describes inter-relations among DEX and relevant biomarkers of inflammation in a population of patients with COVID-19. Despite the lower DEX clearance in patients with COVID-19 no substantial accumulation of DEX is expected for 6 or 12 mg daily doses for 10 days. Current DEX regimens are not fully effective and might be improved with joint administration of anti-IL-6 therapeutics.

  • HOW MIGHT THIS CHANGE CLINICAL PHARMACOLOGY OR TRANSLATIONAL SCIENCE?

Our PK/PD model may be used in the present form to assess dosing regimens of DEX and other GCs in the treatment of cytokine storms, may facilitate translation of the results of preclinical studies, and could allow for the assessment and dose optimization of other anti-cytokine and anti-inflammatory drugs given alone or in combination.

INTRODUCTION

Cytokine release syndrome (CRS), also known as a cytokine storm, has been observed in patients with severe acute respiratory syndrome and Middle East respiratory syndrome (i.e., diseases caused by coronaviruses as well as in patients with coronavirus disease 2019 [COVID-19]).1 In COVID-19, CRS has been shown to participate in the development of acute respiratory distress syndrome (ARDS), a condition often leading to respiratory failure being the most common cause of death of severely ill patients with COVID-19.2 CRS in COVID-19 is associated with increased production and release of interleukin (IL)-1β, IL-6, IL-4, and IL-10, and tumor necrosis factor (TNF)α.3 In addition, increased concentrations of CRP and D-dimer levels were observed. Increases in TNFα, IL-6, and CRP plasma concentrations were strongly correlated with poor outcomes in patients with COVID-19.4

Toll like receptor (TLR)4 is a pattern recognition receptor that belongs to the innate immune system. It recognizes multiple pathogen-associated molecular patterns, such as bacteria or viruses.5 In addition, it recognizes certain damage-associated molecular patterns released from dying cells during tissue injury or infection.6 The canonical pathway of TLR4 signaling, triggered upon ligand binding to the receptor, culminates in the activation of the transcription factor NF-κB that induces the expression of pro-inflammatory cytokines and chemokines, especially TNF-α, IL-1β, and IL-6.5 TLR4 signaling was shown to be closely associated with the development and progression of CRS in COVID-19. It was demonstrated that TLR4 has the strongest protein–protein interaction with the spike (S) glycoprotein, which is present on the surface of severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2) virions, compared to other TLR family members. The S protein binding to TLR4 results in the activation of transcription factors, such as NF-κB, interferon (IFN) regulatory factor, and activator protein 1 (AP-1) leading to upregulation of pro-inflammatory cytokines and IFNs.7 The S glycoprotein of SARS-CoV-2 activates TLR4 on monocytes and macrophages thereby inducing the production of pro-inflammatory cytokines through TLR4-mediated signaling pathways, namely the NF-κB and the mitogen-activated protein kinase pathways.8 Moreover, in situ TLR4 activation in the alveolar macrophages leads to intense local inflammation resulting in accumulation of inflammatory factors that ultimately cause disruption of respiratory gas exchange, breathing problems, and even death.9 The archetypal agonist for TLR4 is the gram-negative bacterial lipopolysaccharide (LPS).5 Therefore, LPS-induced endotoxemia is a widely used animal and human model that allows effects of anti-inflammatory compounds to be assessed in vivo. Moreover, LPS given at high doses in rodents serves as an animal model of ARDS associated with excessive production of inflammatory mediators.10

Dexamethasone (DEX) is a synthetic glucocorticoid (GC) that has a relatively high affinity for GC receptors that results in relatively low half-maximal inhibitory concentration (IC50) and half-maximal effective concentration values for many effects.11, 12 DEX is commonly used in the treatment of many inflammatory and autoimmune conditions and, based on the results of the RECOVERY clinical trial (NCT04381936), this drug given at a dose of 6 mg daily for up to 10 days was shown to be effective in reducing mortality of severely ill patients with COVID-19 requiring oxygen supplementation or mechanical ventilation but not in patients with mild symptoms who did not require oxygen supply.13 In a separate clinical trial that compared the effects of 6 mg versus 12 mg a day of DEX in patients with COVID-19 with hypoxemia, there was no significant difference reported in the number of days alive without life support between these treatment groups; however, the study might have had not enough statistical power to detect the difference in favor of the higher dose.14, 15 On the other hand, it has been observed that high-dose DEX therapy in patients with COVID-19 may be harmful, indicating that decreasing the DEX dosage could be a possible approach to improving treatment outcomes.16

The mechanism of action of DEX in the treatment of COVID-19 may be attributed to its potent anti-inflammatory activity leading to the inhibition of transcription factors, including NF-κB and AP-1, resulting in decreased expression and production of inflammatory mediators, including TNFα and IL-6.17 Indeed, decreased IL-6 concentrations were observed in patients with COVID-19 receiving GC including DEX and lower TNFα and IL-6 levels are positive prognostic factors of greater survival from COVID-19.18, 19 DEX is mainly metabolized in humans by the cytochrome P450 (CYP)3A4 enzyme.20 Increased concentrations of IL-6 and TNFα inhibit the expression of CYP enzymes including CYP3A4.21 These alterations may potentially influence the pharmacokinetics (PKs) of drugs such as DEX.22

This study develops a population PK (PopPK)/pharmacodynamic (PD) model of DEX effects beginning with analysis of published DEX PK data in patients with COVID-19 and assesses the expected suppression of CRS inflammatory biomarkers TNFα, IL-6, and CRP in patients with COVID-19 for dosing regimens of 1.5, 3, 6, and 12 mg daily. Treatment end points were numbers of subjects that exhibited suitable degrees of biomarker suppression over 10 days of DEX dosing.

METHODS

Source of PK data

Blood PK data of DEX in hospitalized patients with COVID-19 were provided in the supplementary materials of Abouir et al.23 based on a clinical trial (ClinicalTrials.gov Identifier: NCT04996784). The trial was a prospective, observational, exploratory PK study conducted at Geneva University Hospitals in Switzerland. They included 30 hospitalized patients with COVID-19 that were both male and female patients at an average age of 63.5 years and mean body weight of 81 kg. Further information regarding the demographic characteristics of the clinical trial's participants can be found in the publication by Abouir et al.23 Patients received a single oral dose of 6 mg DEX (Dexamethasone Galepharm tablets) under fasting conditions. Capillary blood samples were obtained before (time 0) and 0.5, 1, 2, 4, 6, and 8 h after dosing. The DEX blood concentrations were quantified using a liquid-chromatography tandem mass spectrometry method validated according to guidelines of the European Medicines Agency. In the present analysis, three out of 30 concentration-time profiles of DEX were excluded due to high (>50 ng/mL) concentrations of DEX measured at time 0 indicating probable previous drug exposure.

PopPK modeling

One- and two-compartment PK models with first- and zero-order absorption with or without lag time (tlag), and linear or nonlinear elimination were tested. For most PK parameters, interindividual variability (IIV) was assumed to be log-normally distributed, whereas tlag was assumed to be normally distributed. Additive, proportional, and combined error models were tested to account for a random unexplained variability. The final structural model for DEX PKs is described by:
d A d t = k a · A ; A 0 = D · F (1)
V c · d C c d t = k a · A + C p C c · CL d CL · C c ; C c 0 = 0 (2)
V p · d C p d t = C c C p · CL d ; C p 0 = 0 (3)
where CL is the elimination clearance; D is the oral dose, F is the oral bioavailability; ka is the first-order absorption rate constant; CLd is the intercompartmental clearance; Vc and Vp are the volumes of distribution of the central and peripheral compartments; and C c and C p are the drug concentrations in the blood and peripheral compartments. The symbol definitions are also listed in Table 1.
TABLE 1. PopPK model parameter estimates of DEX in patients with COVID-19.
Parameter (unit) Definition Fixed effects (RSE%) SD of random effects (RSE%)
tlag (h) Absorption delay time 0.29 (20.5) 0.16 (33.9)
ka (h−1) Absorption rate constant 2.24 (42.2) 1.69 (16.3)
CL/F (L/h) Apparent systemic clearance 6.27 (11.7) 0.55 (15.8)
Vc/F (L) Apparent central volume of distribution 20.83 (25.9) 0.62 (23.3)
CLd/F (L/h) Apparent distributional clearance 19.72 (48.6) 0.67 (57.7)
Vp/F (L) Apparent peripheral volume of distribution 13.02 (26.7) 0.4 (52.1)
Residual variability
a (ng/mL) Additive error 14.76 (8.42)
  • Abbreviations: COVID-19, coronavirus disease 2019; DEX, dexamethasone; PopPK, population pharmacokinetics.

PD model of DEX anti-inflammatory effects in patients with COVID-19

A PD model of DEX effects on the inflammatory mediators, TNFα, IL-6, and CRP in patients with COVID-19 was developed, as shown in Figure 1, utilizing related published modeling efforts, our own in vivo preclinical studies, and literature data. Elevated TNFα, IL-6, and CRP plasma concentrations occur in the severe course of COVID-1924, 25 and are correlated with disease severity and mortality.18, 19, 26-28 TLR4 signaling is closely associated with the pathogenesis of CRS in patients with COVID-19 and a natural agonist of TLR4 is LPS. Our PopPK/PD model assumes that DEX inhibits the production of TNFα and IL-6 in COVID-19 by the same mechanism as in LPS-induced inflammation, namely by inhibiting expression and production of inflammatory mediators.29 It is assumed that CRP production is exclusively stimulated by IL-6 per Liu et al.30 A constant increased production of TNFα and IL-6 was assumed as described by zero-order rate constants with degradation being first-order leading to elevated steady-state concentrations of these cytokines in COVID-19. This assumption is in line with longitudinal monitoring indicating relatively constant concentrations of these cytokines in hospitalized patients with COVID-19.18, 31 The PD part of the model is described by:
dTNF d t = k inTNF · 1 I maxDEX TNF · C p IC 50 DEX TNF + C p k outTNF · TNF ; TNF 0 = R 0 TNF (4)
dIL 6 d t = k inIL 6 · 1 I maxDEX IL 6 · C p IC 50 DEX IL 6 + C p k outIL 6 · IL 6 ; IL 6 0 = R 0 IL 6 (5)
dCRP d t = k inCRP · 1 + S IL 6 onCRP · IL 6 k outCRP · CRP ; CRP 0 = R 0 CRP (6)
where k inTNF , k inIL 6 , and k inCRP are zero-order production rate constants and k outTNF , k outIL 6 , and k outCRP are first-order disposition constants of the indicated mediators. The I maxDEX TNF and I maxDEX IL 6 reflect maximum inhibition of production by DEX and IC 50 DEX TNF and IC 50 DEX IL 6 are the concentrations resulting in 50% of the maximum inhibition of cytokine production. The S IL 6 onCRP is the stimulatory coefficient of IL-6-induced CRP production. The symbol definitions are also listed in Table 2.
Details are in the caption following the image
Schematic of the proposed population pharmacokinetic/pharmacodynamic model of dexamethasone anti-inflammatory effects on the indicated biomarkers in patients with coronavirus disease 2019. Symbols are described in the text and in Tables 1 and 2. CL, clearance; IC50, half-maximal inhibitory concentration; Imax, maximum inhibition of production by DEX; kin, zero-order production rate constants; kout, first-order disposition constants.
TABLE 2. Pharmacodynamic model parameter values used to perform simulations of DEX anti-inflammatory effects in patients with COVID-19.
Parameter (units) Definition Fixed effects (reference) SD of random effects
kinTNF (pg mL−1 h−1) Production rate constant of TNFα 3.890a 0b
koutTNF (h−1) Degradation rate constant of TNFα 0.35730 0.1071c
IC50DEX(TNF) (ng/mL) DEX plasma concentration resulting in 50% of ImaxDEX(TNF) 2.2529 0.45b
I maxDEX(TNF) Maximum inhibition of TNFα production by DEX 0.80529 0.166b
IC50DEX(IL6) (ng/mL) DEX plasma concentration resulting in 50% of ImaxDEX(IL-6) 3.3529 0.7b
I maxDEX(IL6) Maximum inhibition of IL-6 production by DEX 0.8329 0.161b
kinIL6 (pg mL−1 h−1) Production rate constant of IL-6 18.21a 0b
SIL6onCRP (mL/pg) Stimulatory coefficient of IL-6-induced CRP production 19.230 0b
koutIL6 (h−1) Degradation rate constant of IL-6 0.30530 0.183c
kinCRP (mg L−1 h−1) Production rate constant of CRP 0.0016a 0b
koutCRP (h−1) Degradation rate constant of CRP 0.018330 0.0037c
R0TNF (pg/mL) Baseline plasma TNFα concentration 10.919 0b
R0IL6 (pg/mL) Baseline plasma IL-6 concentration 59.719 0b
R0CRP (mg/L) Baseline plasma CRP concentration 100.519 0b
  • Abbreviations: COVID-19, coronavirus disease 2019; DEX, dexamethasone.
  • a Secondary parameter.
  • b Set based on the assumptions.
  • c Based on the trial-end-error optimization.

Verification of the model

To assess the predictive performance of the proposed PopPK/PD model, we plotted the observed plasma CRP concentration-time data obtained from patients with COVID-19 treated with DEX against an output distribution plot of 1000 model simulated plasma CRP concentration-time profiles in subjects with COVID-19 after i.v. DEX doses of 6 mg daily for 10 days. Mean and median plasma CRP concentrations monitored at several timepoints in patients with COVID-19 treated with these DEX doses were found in Kooistra et al.32 and in Zacharias et al.33 The dosing regimen of DEX in Zacharias et al.33 was not specified and we assumed the same regimen that is recommended for the treatment of COVID-19. The simulation was performed using values of PK and PD parameters and their IIV by sampling, with replacement, individual values from their distributions. Typical PK parameter values and their IIV were obtained in the current PopPK analysis, whereas typical PD parameters and their IIVs were based on the results of previous clinical trials, our preclinical studies, and assumptions. In order to capture the observed variability in baseline plasma TNFα, IL-6, and CRP concentrations in patients with COVID-19,19 we allowed IIV on the degradation rate constants of these biomarkers. This implies that IIV in baseline biomarker concentrations and in their production and degradation rate constants was lumped into IIV in kout values and it was assumed to be log-normally distributed. For koutTNF, IIV expressed as SD of random effect was set at 0.1071, for koutIL6 at 0.183, and for koutCRP at 0.0037. These values were selected based on a trial-and-error optimization performed by changing the values of SD of random effects of these parameters, simulating 1000 individuals, and comparing ranges of the simulated baseline biomarker concentrations with those from patients with COVID-19.19 In addition, IIV was applied on the parameters related to the DEX inhibitory activity against cytokines (i.e., IC 50 DEX TNF , I maxDEX TNF , IC 50 DEX IL 6 , and I maxDEX IL 6 ) and set at 20% of the typical value. Also assumed was IIV in IC 50 DEX TNF and IC 50 DEX IL 6 to be log-normally distributed and logit-normal IIV distribution for I maxDEX TNF and I maxDEX IL 6 . The baseline plasma TNFα, IL-6, and CRP concentrations ( R 0 TNF , R 0 IL 6 , and R 0 CRP ) were set at their median values in non-survivors from COVID-19 that were 10.9, 59.7 pg/mL, and 100.5 mg/L based on the data from Tongji Hospital in Wuhan, China.19 It was assumed that the blood to plasma ratio of DEX in humans equals 1 and it is not changed in patients with COVID-19.34

Simulations and analysis of end points

Simulations were performed for DEX anti-inflammatory effects following once-daily oral doses for 10 days for each of four doses (i.e., 1.5, 3, 6, or 12 mg in patients with COVID-19. Utilizing our PK/PD model to run population simulations at these four dose levels, our goal was to assess the effects of halving or quartering the dose, as well as doubling it, relative to the recommended DEX dose, on the treatment efficacy in terms of the drug's ability to reduce cytokine concentrations. There were 1000 simulations of blood DEX, plasma TNFα, plasma IL-6, and plasma CRP concentration-time profiles for each dosing regimen. The PK and PD parameters for each individual were sampled with replacement from their distributions. A different set of individuals was simulated for each regimen. Whereas the baseline plasma TNFα, IL-6, and CRP concentrations ( R 0 TNF , R 0 IL 6 , and R 0 CRP ) were set at values in non-survivors from COVID-19, it may be assumed they remain relevant during DEX therapy. The PD parameter values, sources, and their IIV variability that were used to perform simulations are presented in Table 2.

Several treatment end points were specified. We assumed that a successful treatment leads to the reduction in TNFα, IL-6, and CRP plasma concentrations to below cutoff levels that were set based on the results of a clinical trial in patients with COVID-19.19 The biomarker concentrations in non-survivors from COVID-19 ranged from 7.7 to 15.9 pg/mL (median: 10.9 pg/mL) for TNFα, 23.6–137.4 pg/mL (median: 59.7 pg/mL) for IL-6, and 62.4–161.2 mg/L (median: 100.5 mg/L) for CRP. On the other hand, the biomarker concentrations in the survivors from COVID-19 were significantly lower and ranged 6.1–9.7 pg/mL (median 7.8 pg/mL) for TNFα, 2.7–22.8 pg/mL (median 7.9 pg/mL) for IL-6, and 2.4–36.7 mg/L (median 10.7 mg/L) for CRP. Therefore, it was designated that reductions of biomarker concentrations below the latter values, 7.7 pg/mL for TNFα, 22.8 pg/mL for IL-6, and 36.7 mg/L for CRP, is a desired treatment end point. In addition, simultaneous reduction of all three biomarkers was also considered. For TNFα, the cut-off concentration was set at 7.7 pg/mL because this value is lower than the highest TNFα concentration observed in the survivors from COVID-19 (9.7 pg/mL). The assessment of DEX efficacy at different dosing regimens was based on the statistical comparison of the numbers of individuals that achieved the specified treatment end points in each dosing group. The group treated with DEX given orally at 6 mg daily for 10 days, was a reference group for these comparisons. The assessments of the treatment end points were performed at day 5 of the treatment (after the fifth dose) and at day 10 of the treatment (after the last dose).

Computation and statistical analysis

PopPK/PD model building, evaluation, and PopPK parameter estimation were performed in the Monolix module of Monolix Suite version 2021R1 (Lixoft, France) using the Stochastic Approximation Expectation–Maximization algorithm. The final PopPK model was selected based on goodness-of-fit diagnostic criteria, such as population and individual weighted residuals (PWRES and IWRES) versus time and PWRES and IWRES versus observed data plots, observed versus predicted plots, Akaike Information Criteria, precision of parameter estimates expressed as relative standard errors (RSE)% and parsimony. A visual predictive check (VPC) was performed by simulating 1000 observations at each timepoint and the 90% prediction intervals of 10th, 50th, and 90th simulated percentiles were plotted against empirical 10th, 50th, and 90th percentiles. Simulations of DEX effects for the tested dosing regimens were performed in the Simulx module of Monolix Suite 2021R1 by sampling with replacement PK and PD parameters from their distributions. Statistical comparison of the numbers of individuals that achieved treatment end points for the treatment groups was performed in the Simulx module. The statistical test used was a two-sided Fisher's exact test with the p value set at 0.05. The odds ratio ≠1 represented the alternative hypothesis.

RESULTS

PopPK modeling

The total were 167 DEX blood concentrations obtained from 27 hospitalized patients with COVID-19 after single oral administration of 6 mg doses. As it can be seen in Figure 2, a two-compartment model with first-order absorption well captured the observed DEX blood concentration-time data. No substantial systematic under- or over-predictions (Figure S1) were observed. The VPC plot (Figure S2) shows that the model captured the variability of data reasonably well with most parts of the medians and 10th and 90th percentile curves for the observed data contained within the corresponding model-predicted confidence regions. The estimated PopPK model parameters with corresponding RSE% values are listed in Table 1. Although the blood samples were collected for only 8 h after DEX dosing, the typical PK parameter values and their IIV variability were estimated with satisfactory precision reflected in relatively low values of RSE% for both fixed effects and SD of random effects. The absorption rate constant was appreciably higher compared to the typical value observed in healthy Indian woman that was around 0.94 h−1 after 6 mg doses of DEX phosphate tablets.35 Of course, the F is not known. The apparent weight-normalized total apparent clearance (CL/F) value (considering mean body weight of 81 kg) in patients with COVID-19 of 0.077 L h−1 kg−1 was around two-fold lower compared to the CL/F values in healthy subjects that ranged from 0.079 to 0.300 L h−1 kg−1, but typically around 0.16 L h−1 kg−1.36 The IIV in tlag and ka values was relatively high and comparable to healthy women,35 whereas IIV in the values of remaining PK parameters was moderately low and estimated with acceptable precision.

Details are in the caption following the image
Individual fits of blood DEX concentration versus time profiles following single oral doses of 6 mg DEX in 27 hospitalized patients with coronavirus disease 2019. Dots are measured DEX concentrations obtained from Abouir et al.23 and the red lines are population pharmacokinetic model-fitted profiles. The number above each plot indicates a patient ID number. DEX, dexamethasone.

Verification of the model

We could not find published data for longitudinal monitoring of IL-6 or TNFα in patients with COVID-19 treated with DEX, probably because these measurements require more expensive enzyme-linked immunosorbent assay or multiplex assays. On the other hand, plasma CRP can be easily monitored and is a commonly used, sensitive, but not specific, biomarker of inflammation. Indeed, CRP concentrations were assessed in patients with COVID-19 including those treated with DEX,32, 33 and plasma CRP concentrations have a prognostic value in predicting COVID-19 severity and mortality.37 The PD parameters used to perform simulations are listed in Table 2. Figure 3 presents observed and model-predicted plasma CRP concentration-time profiles in patients with COVID-19 treated with DEX at a dose of 6 mg for 10 days. As it can be seen in Figure 3, most of the observed plasma CRP concentrations were within the model-predicted percentiles. The decreasing trend in plasma CRP concentration-time profiles upon serial DEX dosing was accurately captured. The moderate underprediction of CRP concentrations between the sixth and 11th days may be partially explained by different baselines (median values) in patients from Zacharias et al.33 that was ~109.0 mg/L (interquartile range: 65.5–205.5) compared to 100.5 mg/L (62.4–161.2) from Liu et al.19

Details are in the caption following the image
Plasma CRP concentration-time profiles in COVID-19 patients treated with DEX at the recommended 6 mg daily dosing regimen. The black circles are the median plasma CRP concentrations from Zacharias et al.,33 black triangles are the mean plasma CRP concentrations from Kooistra et al.,32 and a black square is median baseline CRP plasma concentration in patients with COVID-19 from Liu et al.19 The black solid line represents the median of 1000 simulated concentration-time profiles of plasma CRP in patients with COVID-19 after DEX i.v. doses of 6 mg daily for 10 days (each arrow indicates DEX administration). The blue lines depict the 5th and 95th percentiles of predictions. COVID-19, coronavirus disease 2019; DEX, dexamethasone.

Simulations and analysis of end points

Simulations of 1000 concentration-time profiles for DEX and each biomarker were performed for patients with COVID-19 receiving 1.5 to 12 mg DEX orally once daily for 10 days. The simulations take into account IIV in PKs of DEX, in PD parameters representing turnover of the inflammatory mediators, and in parameters related to the inhibitory activity of DEX on the biomarkers. The PopPK/PD parameters that were used to perform simulations are listed in Tables 1 and 2. The simulated concentration-time profiles of DEX, TNFα, IL-6, and CRP at the four tested dosing regimens are presented in Figure 4. According to the simulations, there is very little accumulation of DEX after multiple oral dosing even at the highest dose of 12 mg despite the reduced CL/F of this drug. Interestingly, the simulations accurately reflect the trajectory of serum IL-6 concentrations in DEX-treated patients with COVID-19 shown previously.18 However, we were unable to resolve data points or the dosing regimens for patients from that study. As depicted in Figure 5, the analysis of end points indicates that for DEX dosing regimens of 1.5, 3, 6, and 12 mg, the success rates for inhibition of TNFα were 55%, 65.9%, 74.8%, and 81.2% at both the fifth and 10th days of treatment. For IL-6, the success rates were 13.0%, 26.5%, 38.9%, and 53.1%, for DEX doses of 1.5, 3, 6, and 12 mg. The same percentages reflect both the fifth and 10th days of treatments. For CRP, the corresponding success rates were 5.5, 13.5, 21.6, and 34.5 at the fifth day and 20.1%, 37.3%, 53.5%, and 67.4% of the simulated subjects at the 10th day. Thus, for all biomarkers, the use of 6 mg doses of DEX are superior to the 1.5 and 3 mg doses. At the DEX dose of 6 mg, there was a 2.5-fold increase in the number of patients that achieved the CRP treatment end point at day 10 compared to day 5 of the regimen. For the cumulative end point, this was 1.9-fold. The anti-inflammatory activity of DEX against CRP manifests later in time due to the relatively long half-life of this mediator, whereas TNFα and IL-6 have more rapid turnover. The full dosing regimen of 10 doses of DEX every 24 h is required to achieve a sufficient reduction of plasma CRP in COVID-19.

Details are in the caption following the image
Simulated concentration-time profiles of blood DEX and plasma TNFα, IL-6, and CRP after oral DEX doses of (a) 1.5 mg, (b) 3 mg, (c) 6 mg, and (d) 12 mg once daily for 10 days in patients with coronavirus disease 2019. The solid black lines depict the median of simulated concentrations, shaded blue areas represent percentiles of predicted DEX blood concentrations and plasma TNFα, IL-6, and CRP for each dosing regimen of DEX. The red lines depict the cutoff plasma concentrations that are 7.7 pg/mL for TNFα, 22.8 pg/mL for IL-6, and 36.7 mg/L for CRP. DEX, dexamethasone.
Details are in the caption following the image
Numbers of simulated subjects that achieved the treatment end points for the indicated dosing regimens of DEX when assessed: (a) after the fifth dose (at the 5th day) and (b) after the last dose of DEX (at the 10th day). The treatment end points are drops in plasma concentrations below: 7.7 pg/mL for TNFα, 22.8 pg/mL for IL-6, 36.7 mg/L for CRP, and simultaneous drop below all three cutoffs for cumulative. DEX, dexamethasone.

There is a statistically significant difference in the numbers of simulated subjects that achieved the treatment end points between each group given 1.5, 3, and 12 mg doses of DEX compared to the 6 mg reference group. Odds ratios were calculated as ratios between chances of a successful treatment in a tested group versus the 6 mg reference group. Odds ratios of greater than 1 indicates a greater chance, odds ratios of less than 1 a lower chance, and an odds ratio = 1 indicates that a chance of attaining a successful treatment is equal in both groups. The odds ratios of successful reduction in plasma CRP concentration were 0.22, 0.52, and 1.8 for DEX given at doses of 1.5, 3, and 12 compared to 6 mg DEX. Corresponding odds ratios for IL-6 were 0.24, 0.57, and 1.78, for TNFα were 0.41, 0.65, and 1.46, and for the cumulative end point were 0.23, 0.57, and 1.78 (Table S1). All differences in odds ratios were statistically significant when 1000 simulations were performed for each dosing regimen, but not between 6 and 12 mg if only 100 individuals were simulated (Table S2).

DISCUSSION

The PKs of DEX among hospitalized patients with COVID-19 was assessed without evaluation of potential influences of covariates, such as weight or sex or disease severity. The source publication lacked this information for individual patients.23 However, our approach was sufficient to allow simulations of DEX concentration-time profiles in patients with COVID-19. The apparent weight-normalized CL/F was around two-fold lower compared to previous studies in healthy subjects.36 The metabolism of DEX is through CYP3A4-mediated hydroxylation.21 The pro-inflammatory cytokines, including IL-6 and TNFα strongly decrease mRNA expression and enzyme activity of CYP enzymes including CYP3A4 in human hepatocytes.22, 23, 38 However, our simulations (Figure 4) show no substantial risk of DEX accumulation, even at the highest dose of 12 mg daily for 10 days. Moreover, an increase in DEX dose may be required in obese patients, because a decrease in DEX area under the curve of 4% per body mass index (BMI) unit was observed in patients with COVID-19.23 Our PopPK/PD model could be used to assess the potential impact of covariates such as weight or BMI on the PKs and PDs of DEX.

We previously proposed a PK/PD model of DEX effects in a rat model of CRS related to the activation of TLR4 signaling.29 Due to the engagement of TLR4 in the pathogenesis of COVID-19, the model of TLR4 activation by LPS mimics some pathophysiological features of COVID-19-related CRS. Using that model, we previously estimated the IC50 and Imax values of DEX against IL-6 and TNFα and used them in our present model. We assumed that CRP is exclusively produced in response to the presence of IL-6 as supported by the literature39 and previous modeling efforts.30 The turnover parameters of TNFα, IL-6, and CRP used currently were set at the values observed in human LPS-induced inflammation.30 Our mathematical model is parsimonious describing inter-relations among relevant mediators of inflammation for DEX in a population of patients with COVID-19 with CRS. Our model translates preclinical data that, in combination with the results of clinical trials and previous modeling efforts, was useful to evaluate the effects of different dosing regimens of DEX for treatment of CRS at the population as well as at the individual level.

Achieving the treatment end points for reduction of a biomarker plasma concentrations below our designated cutoff levels is 74.8% for TNFα and only 38.9% for IL-6 for DEX given at an oral dose of 6 mg. Reducing the DEX dose from 6 to 3 mg is less effective, whereas the 12 mg dose may be beneficial. The analysis of end points suggests that the DEX dosing regimen of 12 mg may be superior to 6 mg, but a larger sample size (n = 1000 vs. n = 100) was required to increase statistical power and identify this difference (Tables S1 and S2). This may be due to moderate IIV in PKs and PDs of DEX in patients with COVID-19 leading to relatively high variability in simulated concentration-time profiles of plasma CRP, IL-6, and TNFα in the population. However, the higher dose may be accompanied by greater adverse effects of DEX.16 Preliminary statistical analysis of the results of the COVID STEROID 2 clinical trial (ClinicalTrials.gov Identifier: NCT04509973) comparing outcomes of DEX given at doses of 12 and 6 mg did not reveal any statistically significant difference in the number of days alive without life support in patients with COVID-19.14 However, re-assessment of these results using a preplanned, secondary Bayesian analysis revealed high probabilities of benefit and low probabilities of harm with DEX 12 versus 6 mg daily in patients with COVID-19 and severe hypoxemia.15 In light of our findings, the COVID STEROID 2 clinical trial might have been underpowered to identify a significant difference in outcomes between dosing groups using a traditional p value-based approach.

The incomplete expected efficacy of all doses of DEX indicates that a combinational therapy with an IL-6 inhibitor and DEX may be superior compared to DEX monotherapy because it may increase a probability of achieving the cumulative treatment end point. IL-6 is one of the key mediators of inflammation in COVID-19 and concentrations of this cytokine are highly correlated with disease severity, progression, and mortality among SARS-CoV-2 infected patients.19 Anti-IL-6 treatments were successfully applied in patients with COVID-19.40 Thus combinational therapy of GC with anti-IL-6 drugs constitute an alternative and more efficient treatment in hospitalized patients with COVID-19 compared to high doses (6 or 12 mg daily) of DEX alone. Indeed, some recent reports indicate that combinational therapies with GC and anti-cytokine biological drugs, such as tocilizumab, an anti-IL-6 receptor monoclonal antibody, are superior compared to anti-inflammatory treatments involving GC given alone.41 The addition of the PK and a mechanistic role of anti-IL-6 treatments could be readily added to our model to assess potential synergistic interactions on pro-inflammatory cytokines.

So far, several PK/PD and quantitative systems pharmacology (QSP) models of COVID-19 were published describing the dynamics of viral load and accounting for disease progression.42, 43 In these models, the effects of immune responses were lumped into the viral dynamic parameters, such as rate constants of infection. A more complex model accounting for inter-relations among infected cells, CD8+ T cells, and innate immune response was proposed recently.44 Complex QSP models of COVID-19 incorporating viral dynamics, innate and adaptive immune responses, accounting for covariates, such as age or comorbidities may further guide rational optimization of treatments with antiviral, anti-inflammatory, and immunomodulatory drugs as well as help in the development of vaccines.45, 46 However, one of the issues of complex QSP models is that robust data related to spatio-temporal dynamics of immune responses are required that are often very difficult to obtain.47 Our model for DEX effects on CRS is complementary to these QSP models. However, the proposed model cannot be directly translated to immunocompromised patients. People who suffer from diseases that impair immune response, such as AIDS, but also patients treated with anticancer drugs or other immunosuppressives may have different baseline cytokine concentrations and may respond differently to DEX treatment compared to subjects with fully functional immune systems. In immunocompromised patients with COVID-19, CRS may not be the main cause of death but an uncontrolled viral load may mainly contribute to poor outcomes. In such cases, more complex models of COVID-19 progression involving viral load, innate, and adaptive immune responses may offer better prediction for COVID-19 outcomes. Another limitation of this study is that PK and PD data and parameters were obtained from various publications, including preclinical studies and clinical trials that may increase the uncertainty in the predictions of the model.

In summary, DEX PK was well-characterized in patients with COVID-19 showing a low CL/F, but no concerns for unexpected accumulation at doses of 6 or 12 mg daily for 10 days. Our PD model premised on our previous work and literature data allowed simulations of DEX effects on three biomarkers in patients with COVID-19. Daily DEX doses of 6 and preferably 12 mg were found to be superior to 1.5 or 3 mg doses and are required to obtain sufficient reduction in CRP plasma concentrations in the majority of the population. Our mathematical model may be used to guide dosing selection of DEX and other GC as well as assess the efficacy of other anti-inflammatory and anti-cytokine treatments and their combinations in the CRS.

AUTHOR CONTRIBUTIONS

A.Ś., and W.J.J. designed the research. A.Ś., and W.J.J. wrote the manuscript. A.Ś. performed the research. A.Ś. analyzed the data.

FUNDING INFORMATION

The work was supported by the National Institute of General Medical Sciences (NIH) grant R35 GM131800 and statutory activity of Jagiellonian University N42/DBS/000259.

CONFLICT OF INTEREST STATEMENT

The authors declared no competing interests for this work.