Real‐world population pharmacokinetics of tezacaftor‐ivacaftor in children with cystic fibrosis: The SYM‐CF study

Abstract

Aims

The clinical effectiveness of tezacaftor‐ivacaftor in children with cystic fibrosis (cwCF) varies; some patients respond while others do not or have adverse effects. The pharmacokinetics (PK) of tezacaftor‐ivacaftor are inadequately published, especially in children. Knowledge of the PK in this cohort in relation to clinical outcomes may give further insight into the drug's exposure–response relationship and its associated interindividual variability. The aim of this study was to assess the real‐world PK of tezacaftor‐ivacaftor in cwCF.

Methods

A prospective, observational PK study was performed in cwCF using tezacaftor‐ivacaftor. PK samples were obtained by dried blood spots at home and during routine outpatient hospital visits. Population PK (popPK) models were created using nonlinear mixed‐effects modelling. Due to data scarcity, prior information from adolescent/adult PK models was required.

Results

The study involved 21 children (age 6–17 years, weight 24–70 kg). Novel popPK models were created for tezacaftor‐ivacaftor and its main metabolites. Variability in PK was explained by variation in body weight. The area under the curve of tezacaftor‐ivacaftor varied significantly within and across age groups, which corresponded to the reported area under the curve in the product information. Maximum concentration and elimination half‐lives closely matched adult reported values.

Conclusions

This is the first study to investigate the popPK of tezacaftor‐ivacaftor in cwCF. The established models can be used for more personalized dosing in children experiencing suboptimal efficacy, adverse effects, drug–drug interactions, or where adherence is a concern.

What is already known about this subject

• The clinical efficacy and safety of tezacaftor‐ivacaftor in children with cystic fibrosis is highly variable in clinical practice.

• Tezacaftor‐ivacaftor has 2 fixed dosing regimens in children: ≥30 kg receive the adult dose, and <30 kg receive half of the adult dose.

• A limited number of pharmacokinetic studies has been performed in children; data are sparsely published. No population pharmacokinetic data of tezacaftor‐ivacaftor in children have been published so far.

What this study adds

• The first population pharmacokinetic models with real‐world data were developed for tezacaftor‐ivacaftor and its main metabolites using prior information from adolescents/adults.

• Area under the curve of tezacaftor‐ivacaftor varied greatly within and across age groups.

• The developed population pharmacokinetic models for tezacaftor‐ivacaftor can be used in future studies evaluating the exposure–response relationship and its variability as a basis for more personalized dosing.

1. INTRODUCTION

Cystic fibrosis (CF) is an autosomal genetic disease, characterized by a mutation in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. A mutation in this gene results in thick and sticky mucus, which impairs various organ functions including the lungs and pancreas. Until the early 2010s, therapy concentrated on airway clearance to remove mucus, pancreatic enzyme supplementation to aid digestion and antibiotics to treat lung infections. The therapeutic management of people with CF (pwCF) is rapidly changing due to the emergence of highly effective modulator therapy. CFTR modulators are the first treatments for CF to address the fundamental cause of the disease. ¹ Tezacaftor‐ivacaftor was the second available combination of a CFTR corrector and a potentiator, respectively. In the Netherlands, tezacaftor‐ivacaftor was approved in 2020 for children and adults from 12 years. In 2021 the approval was extended to children aged 6–11 years. Tezacaftor‐ivacaftor is registered for pwCF with a homozygous F508del‐mutation or a heterozygous F508del‐mutation, in combination with 1 of the 14 residual function mutations (P67L, R117C, L206W, R352Q, A455E, D579G, 711 + 3A → G, S945L, S977F, R1070W, D1152H, 2789 + 5G → A, 3272‐26A → G and 3849 + 10kbC → T), which account for ~60% of the pwCF in the Netherlands. ²

Eligible children with CF (cwCF) start modulator treatment as soon as it becomes available for their age, and treatment is required indefinitely. At the moment of tezacaftor‐ivacaftor introduction, children weighing ≥30 kg receive the adult dose, while those weighing <30 kg receive half the dose. ³ In real‐life therapy response varies greatly, with some cwCF responding and some experiencing side effects. The pharmacokinetics (PK) appear to exert significant interindividual variability (IIV), raising the possibility that specific patient groups are receiving dosages that are either too high or too low. Aside from that, tezacaftor‐ivacaftor is susceptible to drug–drug interactions, since both components are substantially metabolized by cytochrome P450 3A4. This can lead to drug–drug interactions with for example –azole antifungals or rifampicin, which are commonly used by pwCF to treat infections. ³

A limited number of PK studies has been performed in children. The only population PK (popPK) results available are those published in the registration documents, where data from adolescents (>12 years) and adults are presented together. ⁴ In a phase 3 study conducted by the registration holder, area under the curve (AUC) of tezacaftor‐ivacaftor has been evaluated in children aged 6–11 years, but data are not presented in the publication. ⁵ Furthermore, there is currently a lack of independent studies evaluating the PK of tezacaftor‐ivacaftor in children. Therefore, it is crucial to investigate the PK of tezacaftor‐ivacaftor in children aged 6–17 years in real‐world clinical settings. Even though most children have since switched to the newest triple therapy elexacaftor‐tezacaftor‐ivacaftor (ETI), this is still relevant since tezacaftor‐ivacaftor are 2 of the 3 components in ETI. Gaining a deeper understanding of PK could improve knowledge of the exposure‐response relationship and its associated IIV. This information may lead to better insights into drug efficacy and side effects and support the development of personalized dosing regimens.

Paediatric trials are often limited by the number of PK samples that can be collected, and traditional methods for PK analysis are not suitable. PopPK modelling methods can offer a solution in this case. However, because of the limited amount of data available, it is often not possible to precisely estimate all PK parameters of the model. Combining popPK methods with prior information from previous or adult models can improve the precision of the estimated PK parameters. ⁶ The objective of this study is to describe the popPK of tezacaftor‐ivacaftor and its main metabolites (tezacaftor‐M1, ivacaftor‐M1 and ivacaftor‐M6) in cwCF using prior information. Secondary goals are to assess AUC, maximum concentration (C_max) and the elimination half‐life (t _1/2) of tezacaftor and ivacaftor.

2. METHODS

2.1. Study design

Real‐word data were prospectively collected in a multicentre, observational PK study in a cohort of 21 cwCF using tezacaftor‐ivacaftor as chronic CF treatment. Between May 2021 and August 2022 cwCF were enrolled from 3 Dutch hospitals. Main inclusion criteria were children aged 6–17 years with at least 1 F508del‐mutation, and the use of tezacaftor‐ivacaftor according to regular care protocols: with tezacaftor‐ivacaftor dosages of 100–300 mg daily in children of ≥12 years and ≥6 years weighing ≥30 kg, and 50–150 mg in 6–11 years weighing <30 kg. Main exclusion criteria were history of poor compliance deemed by the physician, and concomitant use of drugs with inhibitory or inducing effect on the CYP3A4 enzyme metabolism during 14 days before blood collection. The study was approved by the Institutional Review Board (METC Amsterdam UMC, ABR NL75811.018.21). Informed consent was obtained from all participants and their parents or legal guardians prior to any study‐related procedures. Patients were included after a minimum of 2 weeks treatment with tezacaftor‐ivacaftor, in order to reach steady‐state concentrations. Patients were followed until they switched to elexacaftor‐tezacaftor‐ivacaftor. To facilitate PK sampling, patients and their parents were trained to perform dried blood spot (DBS) sampling. At 1 time point during the study, DBS samples were taken at home at t = 0, 4 and 8 h after administration of tezacaftor‐ivacaftor by the participant/parent(s). During every regular visit at the outpatient clinic (~every 3 months), a single sample was taken by venous or DBS sampling at a random time point after administration of tezacaftor‐ivacaftor. No additional venipunctures were performed for PK analysis. Tezacaftor‐ivacaftor administration and sampling times were recorded, as well as adherence (self‐reported and physicians assessment) and the (fat) food with which tezacaftor‐ivacaftor was taken. Clinical data as part of regular care were collected including patient demographics, CF‐related comorbidities, comedication and liver function tests (alanine aminotransferase, aspartate aminotransferase, bilirubin).

2.2. Sample preparation and analysis

DBS samples were stored at −20°C (after drying for a minimum of 30 min at room temperature) until analysis. DBS samples collected at home were sent to the hospital by mail in a sealed plastic bag and envelope. Venous blood samples were centrifuged and the plasma was stored at −80°C until analysis. The quantification of concentrations of tezacaftor, tezacaftor‐M1, ivacaftor, ivacaftor‐M1 and ivacaftor‐M6 in plasma and DBS was performed using liquid chromatography–tandem mass spectrometry, using the method detailed in our previous publications. ⁷ , ⁸ For all compounds the respective lower and upper limits of quantification (LLOQ/ULOQ) were 0.01 mg/L and 10 mg/L in plasma and DBS, except for tezacaftor‐M1 (LLOQ 0.025 mg/L and ULOQ 12.5 mg/L). DBS concentrations were converted to their estimated plasma concentrations by the Passing–Bablok regression equation as described in our previous article. ⁸

2.3. Population PK analysis using the PRIOR subroutine in NONMEM

In literature the PK of tezacaftor‐ivacaftor has been described by 2‐compartment models (Figures 1 and 2). ⁴ , ⁹ In the present study, the available sparse data did not allow a precise estimation of all model PK parameters. Instead of fixing parameters to known values from the literature, the PRIOR subroutine was used in the nonlinear mixed‐effects modelling (NONMEM) software (v7.5.1 ICON Development Solutions, Dublin, Ireland). ⁶ For development of the popPK models for tezacaftor‐ivacaftor and their main metabolites, popPK information was derived from the registration document of tezacaftor‐ivacaftor (Symdeko). ⁴ In this document popPK parameters for tezacaftor, tezacaftor‐M1 and ivacaftor models were described; no information was available for ivacaftor‐M1/M6.

FIGURE 1.

Schematic illustration of the pharmacokinetic model of tezacaftor and its main metabolite tezacaftor‐M1. D1, zero‐order absorption into the gut compartment; KA, absorption rate constant; V_{c TEZ}/F, apparent tezacaftor central volume of distribution; V_{p TEZ}/F, apparent tezacaftor peripheral volume of distribution; Q/F, apparent intercompartmental clearance; CL_TEZ/F, apparent tezacaftor clearance; V_{c M1}/(F*f_m), apparent tezacaftor‐M1 central volume of distribution of the fraction metabolized; V_{p M1}/(F*f_m), apparent tezacaftor‐M1 peripheral volume of distribution of the fraction metabolized; CL_M1/(F*f_m), apparent tezacaftor‐M1 clearance of the fraction metabolized; Q_m/f_m, intercompartmental clearance of the fraction metabolized.

FIGURE 2.

Schematic illustration of the pharmacokinetic model of ivacaftor and its main metabolites ivacaftor‐M1 and ivacaftor‐M6. D1, zero‐order absorption into the gut compartment; KA, absorption rate constant; V_{c IVA}/F, apparent ivacaftor central volume of distribution; V_{p IVA}/F, apparent ivacaftor peripheral volume of distribution; Q/F, apparent intercompartmental clearance; CL_IVA/F, apparent ivacaftor clearance; V_{c M1/6}/(F*f_m), apparent ivacaftor‐M1/6 central volume of distribution of the fraction metabolized; CL_M1/6/(F*f_m), apparent ivacaftor‐M1/6 clearance of the fraction metabolized; Q_m/f_m, intercompartmental clearance of the fraction metabolized.

The PK of tezacaftor(−M1) was described by a 2‐compartment model for both the parent compound and the metabolite, with zero‐order absorption into the depot compartment, followed by first‐order absorption to the central compartment and first‐order elimination of both the parent and metabolite (Figure 1). ⁴ The PK of ivacaftor was described by a 2‐compartment model for the parent compound with zero‐order absorption into the depot compartment, followed by first‐order absorption to the central compartment. Elimination and conversion of the parent to the metabolites was described with first‐order rate constants. ⁴ As there were no models available for ivacaftor's metabolites, they were described by a 1‐compartment model with first‐order absorption and elimination, and linked to the central compartment of ivacaftor (Figure 2). ⁹ , ¹⁰

These models were used as a starting point for the popPK analysis of the current study. Concentrations of the metabolites were adjusted to their parent equivalents using the molecular weight. Since there is no intravenous formulation available, PK parameters of the parent compounds were estimated as apparent clearance (CL/F), apparent intercompartmental clearance (Q/F) and apparent volume of distribution (V/F). For tezacaftor the fraction parent drug metabolized into the metabolite was fixed to 1 for f _m M1. For ivacaftor the fraction parent drug metabolized into the metabolites were fixed to 22 and 43% for f_mM1 and f_mM6, respectively. ⁹ PK parameters of the metabolites were estimated as apparent clearance of the fraction metabolized (CL/[F*f_m]) and apparent volume of distribution of the fraction metabolized (V/[F*f_m]).

As there was a wide range in body weight (BW), allometric scaling for BW was applied on both parent (CL/F, V/F) and metabolite CL/(F*f_m), V/(F*f_m) PK parameters. The latter are shown in Equations (1) and (2), where θ is the typical estimate for the metabolite CL and, V normalized for a BW of 70 kg.

$$\frac{\mathit{CL}}{\left(F*f_m\right)}={\theta }_{\mathit{CL}}*{\left(\frac{\mathit{BW}}{70}\right)}^{0.75}$$ (1)

$$\frac{V}{\left(F*f_m\right)}={\theta }_V*{\left(\frac{\mathit{BW}}{70}\right)}^1$$ (2)

IIV was assessed on CL Equation (3). Where η_CL is a normally distributed random variable with a mean of zero and an estimated variance of ω², indicating the IIV in CL.

$$\mathit{CL}={\theta }_{\mathit{CL}}*e^{{\eta }_{\mathit{CL}}}$$ (3)

Residual variability was assessed by testing proportional, additive and combined error models to describe differences between individual predictions and observations Equation (4). Where θ is the typical estimate for the proportional or additive error, IPRED is the individual predicted concentration and Y is the modelled value of the observed variable.

$$Y=\mathit{\text{IPRED}}\left(1+{\theta }_{\mathit{\text{prop}}}\right)+{\theta }_{\mathit{add}}$$ (4)

For ivacaftor‐M1 and M6 separate proportional error models for plasma and DBS samples were implemented.

To define the weight of the prior value of the PK parameters, predefined residual standard error (RSE) values were used. Three types of prior weight were predefined and based on numbers described in literature: informative (10% RSE), moderately informative (30% RSE) and weakly informative/vague (10⁵). ⁶ The first step was to assign informative priors on all parameters, except for CL, IIV and the residual error. No priors were assigned to CL, as data were thought to be rich enough to estimate this parameter without a prior. Also, as described above IIV was only applied on CL and no priors were used to estimate the IIV on other PK parameters. As well as the residual error, which was also estimated without priors on basis of the available data.

The next step was to change 1 parameter at the time to a vague prior, and assess whether this parameter could be estimated on basis of the available data—with weakly prior information. If estimation was not possible, the following step was to change this parameter to a moderate informative prior. If estimation was still not possible, the parameter was set to the informative prior again. These steps were repeated for all parameters in order to obtain a stable structural model, which was defined by being able to estimate the model parameters with a value within an expected range with a maximum RSE of 30%.

Following the development of the structural model, a covariate analysis was conducted to determine whether covariates could explain IIV. A covariate search can only be applied on parameters without a prior, in this case CL. ⁶ The relationship between CL and covariates was evaluated by visual inspection of scatter and box plots (continuous and categorical variables, respectively) of empirical Bayesian etas of CL vs. the following covariates: sex, adherence, mutation, exocrine pancreatic insufficiency, CF related diabetes, distal intestinal obstructive syndrome, age, alanine aminotransferase, aspartate aminotransferase, bilirubin. Only covariates that visually showed potential effects were included in the statistical procedure. The covariates assessed included age, adherence and CF mutation. Stepwise forward inclusion was used in the covariate analysis. A reduction in objective function value ≥3.81 (P = .05) was considered statistically significant. Dichotomous and continuous covariates were included in the model Equation (5) and (6). In this context, θ_i represents the individual model predicted PK parameter for an individual with covariate value cov_i. θ_pop is the population estimate for that parameter, cov_m is the median covariate value and θ_cov denotes the covariate effect.

$${\theta }_i={\theta }_{\mathit{pop}}*{{\theta }_{\mathit{cov}}}^{{\mathit{cov}}_i}$$ (5)

$${\theta }_i={\theta }_{\mathit{pop}}*{\left(\frac{{\mathit{cov}}_i}{{\mathit{cov}}_m}\right)}^{{\theta }_{\mathit{cov}}}$$ (6)

The first‐order conditional estimation with interaction (FOCE‐I) in NONMEM was used for all runs.

The model fits were based on visual inspection of goodness‐of‐fit plots, objective function value, parameter precision and shrinkage values. The robustness of the parameter estimates and the validity of the models were evaluated with a bootstrap analysis (n = 1000) and a visual predictive check, respectively. Data handling, visualization and descriptive statistics were performed using Pirana (v2.9.4 Certara, Radnor, PA, USA), R Studio version 4.3.1, with the packages xpose4, lattice and ggplot2 for data visualization, and GraphPad Prism 9.0 (GraphPad Software, Boston, MA, USA).

2.4. AUC, C_max and elimination half‐life of tezacaftor‐ivacaftor real‐world compared to reported values

AUC_0‐24h for tezacaftor(−M1) and AUC_0‐12h for ivacaftor(−M1 and –M6), C_max and terminal elimination half‐life (t _1/2,β) were assessed during every occasion by Bayesian analysis. The obtained values were compared to their reported values as described in the product information. ³

3. RESULTS

3.1. Population and data characteristics

In total 21 patients were included in the study, contributing to a total of 97 PK samples (13 plasma samples and 84 dB samples). Results from 3 (3%) samples were excluded due to incorrect DBS sampling, ¹ and missing dosing information. ² Ivacaftor‐M6 concentration in 3 samples (3%) was below the LLOQ and excluded. No samples were above the ULOQ.

Patient demographics and baseline characteristics are summarized in Table 1. All patients used tezacaftor‐ivacaftor according to the dose recommendation as stated in the product information. ³

TABLE 1.

Patient demographics and baseline characteristics.

Demographic	Value
Patients, n	21
Female, n (%)	11 (52)
Age (years), median (range)	12 (6–17)
Weight (kg), median (range)	43.5 (23.6–69.8)
Height (cm), median (range)	153 (122–191)
Mutation homozygous F508del, n (%)	16 (76)
Mutation heterozygous F508del, n (%)	5 (24)
‐ Other mutations, n (%)
○ 3849 + 10kbC → T	1 (5)
○ A455E	4 (19)
Children per age/dosing group, n (%)
‐ 6–11 years, <30 kg	3 (14)
‐ 6–11 years, ≥30 kg	7 (33)
‐ 12–17 years	11 (52)
Comorbidities, n (%)
‐ CF‐related diabetes	0 (0)
‐ Distal intestinal obstruction syndrome	4 (19)
‐ Exocrine pancreatic insufficiency	20 (95)
Laboratory parameters, median (range)
‐ ALAT (U/L)	22 (13–39)
‐ ASAT (U/L)	27 (18–50)
‐ Bilirubin (μmol/L)	3 (1–12)
Samples for PK analysis, n (%)	97
‐ Plasma samples	13 (13)
‐ DBS samples	84 (87)
Samples per patient, median (range)	5 (2–7)

Note: Variables are presented as numbers (%) or median (range min–max).

Abbreviations: ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; PK, pharmacokinetic; DBS, dried blood spot.

3.2. PopPK analysis

3.2.1. Tezacaftor and tezacaftor‐M1

In Table 2 the estimated PK parameters of the final models of tezacaftor and its M1‐metabolite are shown. The most stable model was created with a vague prior on V_{c tez}, with moderate informative priors on V_{p tez}, Q_tez, D1 and Q_tez‐M1, and with informative priors on KA, V_{c tez‐M1} and V_{p tez‐M1}. No prior information was applied on CL and its IIV, as the data were rich enough to estimate these parameters. No correlations of the covariates age, CF mutation or adherence were observed on CL.

TABLE 2.

Pharmacokinetic parameter estimates of tezacaftor, tezacaftor‐M1, ivacaftor, ivacaftor‐M1 and ivacaftor‐M6 of the final population pharmacokinetic models.

Parameters	Tezacaftor			Tezacaftor‐M1			Ivacaftor			Ivacaftor‐M1		Ivacaftor‐M6
	Estimates Value (RSE) [Shr]	Bootstrap Median (95% CI)	Prior type b	Estimates Value (RSE) [Shr]	Bootstrap Median (95% CI)	Prior type b	Estimates Value (RSE) [Shr]	Bootstrap Median (95% CI)	Prior type b	Estimates Value (RSE) [Shr]	Bootstrap Median (95% CI)	Estimates Value (RSE) [Shr]	Bootstrap Median (95% CI)
CL a (L/h/70 kg)	1.95 (6)	1.93 (1.71–2.24)	‐	1.01 (6)	1.00 (0.905–1.13)	‐	15.9 (9)	15.8 (13.3–19.1)	‐	2.10 (10)	2.10 (1.74–2.61)	12.2 (16)	12.2 (8.81–17.3)
Vc a (L/70 kg)	38.4 (7)	38.5 (33.6–45.6)	V	4.86 (28)	4.89 (4.53–5.50)	I	178 (20)	176 (111–342)	V	0.1 * Viva	0.1 * Viva	0.1 * Viva	0.1 * Viva
Q a (L/h/70 kg)	0.19 (29)	0.190 (0.181–0.203)	M	3.70 (19)	3.67 (2.83–4.25)	M	13.2 (25)	13.0 (11.5–15.4)	M	‐	‐	‐	‐
Vp a (L/70 kg)	36.4 (29)	36.4 (36.4–36.4)	M	37.5 (10)	37.5 (37.0–38.5)	I	106 (26)	104.5 (97.7–119)	M	‐	‐	‐	‐
Ka (h−1)	2.95 (10)	2.94 (2.88–3.00)	I	‐	‐	‐	0.506 (10)	0.505 (0.482–0.523)	I	‐	‐	‐	‐
D1 (h)	1.06 (28)	1.08 (0.885–1.34)	M	‐	‐	‐	2.59 (10)	2.59 (2.41–2.77)	I	‐	‐	‐	‐
Inter‐individual variability
CL (CV%)	26 (19) [6]	24 (8–39))	‐	24 (19) [5]	22 (11–34)	‐	40 (37) [5]	37 (13–57)	‐	44 (38) [3]	43 (17–63)	76 (37) [3]	72 (43–114)
Residual variability
Prop. Error	0.26 (9)	0.26 (0.21–0.30)	‐	0.20 (9)	0.20 (0.16–0.24)	‐	0.34 (10)	0.33 (0.26–0.41)	‐	‐	‐	‐	‐
Prop. Error Plasma	‐	‐	‐	‐	‐	‐	‐	‐	‐	0.37 (24)	0.36 (0.18–0.52)	0.98 (25)	0.94 (0.56–1.38)
Prop. Error DBS	‐	‐	‐	‐	‐	‐	‐	‐	‐	0.36 (10)	0.35 (0.29–0.42)	0.50 (10)	0.50 (0.36–0.62)

Abbreviations: CI, confidence interval; CL, clearance; DBS, dried blood spot; CV%, coefficient of variation; D₁, zero‐order absorption into the depot compartment; f_m, fraction metabolized into metabolite; K_a, absorption rate constant; prop. Error, proportional error; Q, intercompartmental clearance; RSE, relative standard error in %; Shr., shrinkage in %; V_c, central volume of distribution; V_p, peripheral volume of distribution.

^a Apparent CL (CL/F), Q (Q/F) and V_c/p (V_c/p/F) were described for the parent compounds, and apparent CL (CL/[F * f_m]), Q (Q/[F * f_m]) and V_c/p (V_c/p/[F * f_m]) for the metabolites. In the final models CL, Q and V_c/p were allometrically scaled to weight (kg): CL = θ_CL * (weight/70)^0.75, Q = θ_Q * (weight/70)^0.75 and V_c/p = θ_V * (weight/70)¹ for both the parent compounds and the metabolites if applicable.

^b Prior types: V, vague (10⁵); M, moderately informative (RSE 30%); I, informative (RSE 10%).

3.2.2. Ivacaftor, ivacaftor‐M1 and ivacaftor‐M6

In Table 2 the estimated PK parameters of the final models of ivacaftor (−M1/6) are shown. Values for V_iva‐M1 and V_iva‐M6 were fixed at 0.1*V_iva, as we were unable to estimate them and no popPK models have been described for the metabolites in literature. A factor of 0.1 was chosen because basic lipophilic drugs such as ivacaftor often have a large V (>100 L), whereas their more polar and acidic metabolites have volumes closer to 10 of 20 L. ¹¹ For ivacaftor‐M6 the proportional error (RSE) in plasma samples was larger than DBS samples with values of 0.98 (225%) and 0.50 (10%), respectively. The most stable model was created with a vague prior on V_{c iva}, with moderate informative priors on V_{p iva} and Q_iva, and with informative priors on KA and D1. No prior information was applied on CL and its IIV, as the data were rich enough to estimate these parameters. No correlations of the covariates age, CF mutation or adherence were observed on CL.

3.2.3. Model evaluation

RSE values of estimated parameters were generally low both for the typical PK parameters (≤29%) and the random effects (IIV on CL ≤ 38%). Goodness‐of‐fit plots (Figure A1–3) and visual predictive check plots (Figure 3) demonstrate that the developed models adequately describe the observations. The robustness of the models was evaluated by a bootstrap analysis; its results are presented in Table 2.

FIGURE 3.

Prediction corrected visual predictive checks (VPC) of the final models. The open circles represent the prediction correct concentrations of tezacaftor, tezacaftor‐M1, ivacaftor, ivacaftor‐M1 and ivacaftor‐M6. The solid black line represents the observed median and the dashed black lines represent the 5th and 95th percentiles of the observed prediction‐corrected data. The blue areas represent the 80% confidence interval of the model‐predicted 5th and 95th percentiles. The orange area represents the 80% confidence interval of the model‐predicted median. For tezacaftor, tezacaftor‐M1, ivacaftor and ivacaftor‐M6, the solid black line slightly rises above the orange shaded area at the end of the dosing interval, indicating a minor underestimation of the observed 50th percentile. For ivacaftor‐M1, the dashed black line slightly rises above the blue shaded area at the beginning of the dosing interval, indicating a minor underestimation of the observed 5th percentile. For ivacaftor‐M6, the dashed black line slightly rises above the blue shaded area in the middle of the dosing interval, indicating a minor underestimation of the observed 95th percentile.

3.3. AUC, C_max and half‐life of tezacaftor‐ivacaftor real‐world compared to reported values

In Table 3 the average AUC values per age and dosing group are shown and compared with the corresponding reported value in the product information. ³ The variability is large within age groups, as demonstrated by the large standard deviation with corresponding coefficients of variation (CV) between 16 and 88%. The AUC differs per age group, as children aged 6–11 years and ≥30 kg tend to have a higher AUC value than children from other age groups. Difference in average AUC vs. reported values in the product information was less than ±17%, except for tezacaftor in the age 12–17 years group and ivacaftor in the 6–11 years, ≥30 kg groups with −32 and 48% difference, respectively.

TABLE 3.

AUC estimates by the final pharmacokinetic models vs. reported AUC in product information.

		Dose a (mg/day)	AUC b (mg*h/L)	Reported AUC c (mg*h/L) 3
			Mean (SD)	Mean (SD)
Tezacaftor	6–11 years < 30 kg	50	53.2 (12.9)	58.9 (17.5)
	6–11 years ≥30 kg	100	91.7 (25.5)	107 (30.1)
	12–17 years	100	66.2 (11.4)	97.1 (35.8)
	Adult	100		85.9 (28.0)
Tezacaftor‐M1	6–11 years <30 kg	(50)	107 (30.4)	126 (30.0)
	6–11 years ≥30 kg	(100)	192 (47.3)	193 (45.8)
	12–17 years	(100)	124 (21.8)	146 (35.7)
	Adult	(100)		126 (34.9)
Ivacaftor	6–11 years <30 kg	150 in 2 doses	7.51 (2.34)	7.1 (1.95)
	6–11 years ≥30 kg	300 in 2 doses	17.5 (7.29)	11.8 (3.89)
	12–17 years	300 in 2 doses	13.3 (3.39)	11.4 (5.5)
	Adult	300 in 2 doses		11.4 (4.14)
Ivacaftor‐M1	6–11 years <30 kg	(150 in 2 doses)	14.9 (9.32)
	6–11 years ≥30 kg	(300 in 2 doses)	34.2 (16.8)
	12–17 years	(300 in 2 doses)	21.1 (3.46)
	Adult	(300 in 2 doses)
Ivacaftor‐M6	6–11 years <30 kg	(150 in 2 doses)	4.78 (2.20)
	6–11 years ≥30 kg	(300 in 2 doses)	15.8 (13.9)
	12–17 years	(300 in 2 doses)	7.97 (4.81)
	Adult	(300 in 2 doses)

Abbreviations: AUC, area under the curve; SD, standard deviation; y, years.

^aTezacaftor daily dose is administered in 1 dose, ivacaftor daily dose is given in 2 equal doses.

^b AUC_{0‐24h ss} for elexacaftor, elexacaftor‐M23, tezacaftor and tezacaftor‐M1. AUC_{0‐12h ss} for ivacaftor, ivacaftor‐M1 and ivacaftor‐M6.

^c For ivacaftor‐M1 and ivacaftor‐M6 no AUC are reported in the product information of Symkevi. Exposures in ≥ 30 kg to < 40 kg weight range are predictions derived from the population PK model. (3).

In Table 4 mean (standard deviation) C_max and half‐lives per age and dosing group are presented for tezacaftor‐ivacaftor in relation to their reported adolescent/adult values in the product information and registration document. ³ , ⁴ Children aged 6–11 years with a body weight ≥30 kg tend to have a higher mean C_max and, in the 6–11 years, <30 kg group, the mean half‐lives tend to be shorter. Overall the pooled mean C_max values seems slightly lower for tezacaftor in our paediatric data, compared to the reported adult values. ³ For tezacaftor the half‐lives in children tend to be shorter compared to the reported adult values. ³ For ivacaftor, C_max values are overall comparable and half‐lives tend to be higher than the reported values in adults, except for the 6–11 years, <30 kg group. ³

TABLE 4.

C_max and half‐lives estimates by the final pharmacokinetic models vs. reported in the product information.

		Dose* (mg/day)	Cmax (mg/L) Mean (SD)	Reported Cmax a (mg/L) 3 Mean (SD)	T½ (h) Mean (SD)	Reported T½ a (h) 3 Mean (SD)
Tezacaftor	6–11 years <30 kg	50	4.01 (0.660)	6.52 (1.83)	116 (1.10)	156 (52.7)
	6–11 years ≥30 kg	100	6.61 (0.910)		124 (9.79)
	12–17 years	100	4.39 (1.12)		136 (7.14)
	Pooled		4.84 (1.40)		132 (9.87)
Ivacaftor	6–11 years <30 kg	150 in 2 doses	0.840 (0.208)	1.28 (0.440)	9.78 (1.88)	9.3 (1.7)
	6–11 years ≥30 kg	300 in 2 doses	1.80 (0.600)		14.0 (6.18)
	12–17 years	300 in 2 doses	1.34 (0.308)		14.9 (2.71)
	Pooled		1.40 (0.449)		14.4 (3.85)

Abbreviations: C_max, maximum concentration; T½, half‐life; SD, standard deviation; y, years.

^a C_max and T½: mean (SD) reported values as described in the product information (3) are PK parameters of tezacaftor and ivacaftor at steady‐state in pooled data from adolescent(>12 years)/adult patients with CF receiving 100–300 mg tezacaftor‐ivacaftor daily.

4. DISCUSSION

In this study, popPK models for tezacaftor‐ivacaftor and its main metabolites were successfully developed in children with real‐world data using prior information from adolescent/adult models. Substantial variability in AUC was observed both within and across age and dosing groups. In general, AUC corresponded well with reported values in the product information, except for tezacaftor in the 12–17 years group and ivacaftor in the 6–11 years, ≥30 kg group. C_max and half‐lives also corresponded closely with reported values, although in children aged 6–11y and ≥30 kg, C_max tended to be higher and half‐lives tended to be shorter in children aged 6–11 years and <30 kg. ³

Although sparse data were available in this study, the prior subroutine was essential to successfully develop full popPK models, that also described the absorption phase and the distribution to peripheral compartments. In a sensitivity analysis, we found that it was not possible to obtain these extended models without prior information, emphasizing the importance of prior information in popPK modelling when data are limited, as is frequently the case in paediatric studies. When data are sparse, there are 2 methods to stabilize difficult‐to‐estimate parameters: (i) fix the parameters to a previous value reported in literature; (ii) inform them from previous studies. In a sensitivity analysis, we found that the latter strategy minimized bias in situations where the parameters differed somewhat between the preceding population and the population from which the sparse data were taken (Table A1). ⁶

The key covariate of relevance for this analysis was weight as predictor of CL/F(*f_m), Q/F(*f_m) and V/F(*f_m), and was predefined. This is fairly typical in paediatric studies due to the large weight range, partially explaining the IIV in PK parameters. No other covariates explaining IIV were identified, probably due to the study being underpowered for this type of analysis. The proportional residual variability for ivacaftor‐M6 in plasma (0.98) raises concerns about the reliability of individual predictions. However, the vast majority (87%) of the samples were DBS samples, which showed a lower proportional error (0.50). We do not have an explanation for the relatively high residual variability. However, as ivacaftor‐M6 has only 1/50th of ivacaftor's activity this limitation was accepted. Interestingly, despite the real‐world context, AUC variability in our studies (CV% = 16–88) was in close agreement with the reported values in the product information. ³ A notable contribution of this study is the first real‐world AUC data for children in the 30–40 kg weight class, as the registration studies relied on model‐based predictions in this weight group due to other weight‐dose group categorizations. ³ , ⁵ Additionally, our findings of elevated AUC and C_max in children aged 6–11 years and ≥30 kg raise questions about whether the tezacaftor‐ivacaftor dose is appropriate for all children within this subgroup, as they receive the adult dose. Lowering the dose could reduce the risk of overexposure and possible development of side effects, as well as saving costs.

This study benefits from several strengths, including the use of real‐world data in children aged 6–17 years that allowed the analysis of AUC in the 30–40 kg weight class, where limited data previously existed. Also, real‐world PK data are relevant to account for the difference between controlled clinical trials and the complexity of drug use in diverse, everyday settings. Next, the selective use of informative prior information from adolescents/adults was used to support portions of the model that were not well defined from the currently available paediatric data (Ka, D1, V_p, Q) and allowed for the estimation of the remaining parameters (CL, V). Limited sampling strategies were applied to reduce the burden of PK sampling in this age group. As well as the use of DBS sampling (at home), which served as a feasible PK sampling method and patients experienced this as less invasive than a venipuncture. ⁸ Especially in the era of changing CF care this method is preferable, as pwCF have better outcomes and will probably visit the hospital less frequent. ¹² Furthermore, intake with fatty food was registered, which allowed for additional control over factors influencing drug absorption as fatty food increases the absorption of tezacaftor‐ivacaftor. ³ Concomitant fatty food intake was not further investigated in the covariate analysis, as the absorption parameters were estimated with prior information and covariates can only be applied on parameters estimated without priors. ⁶

Despite this, the study has some limitations. The sample size was small, due to faster access to ETI than expected, which limited the recruitment as some children did not start with tezacaftor‐ivacaftor awaiting ETI. This also immediately implies that tezacaftor‐ivacaftor is hardly used, since the introduction of ETI. However, the findings in this study are still useful because elexacaftor had little effect on the PK of tezacaftor‐ivacaftor, as indicated in the registration report. ⁴ The small sample size, while appropriate for popPK modelling, may limit the generalizability of the findings. A larger sample size could improve the estimation of parameter estimates by the model and better characterization of IIV. Furthermore, due to the short study duration, only 1 AUC curve was acquired for certain patients who used home‐based DBS collections. This also resulted in incomplete covariate data in those patients (e.g. liver‐enzyme measurements), as they did not visit the hospital during the study visit. This could have resulted in an underpowered covariate analysis, prohibiting a thorough assessment of the IIV in PK of tezacaftor‐ivacaftor. This also made it impossible to assess the interoccasion variability. Another limitation of this analysis is the absence of efficacy and safety data, which impairs direct evaluation of the relationship between drug exposure and clinical outcomes such as suboptimal efficacy or adverse events. Also, the PK data were collected using a sparse sampling design, which may limit the ability to describe drug absorption and distribution. Therefore, prior knowledge was used in order to enhance parameter estimation. Last, self‐reported adherence could have influenced the results of the study, as some dosing records may have been subject to inaccuracy; however, this reflects a real‐world situation.

To conclude, this study is the first to describe the popPK of tezacaftor‐ivacaftor and its main metabolites in cwCF based on real‐world data. The selective use of prior information from adolescent/adult models enabled the development of stable and robust models. The popPK models developed in this study could be used as a basis for more personalized medicine. Future applications of such TDM models could enhance dose optimization, particularly for children experiencing suboptimal efficacy, adverse effects, drug–drug interactions or where adherence is a concern.

AUTHOR CONTRIBUTIONS

Steffie E. M. Vonk: Conceptualization; data curation; investigation; methodology; project administration; formal analysis; writing—original draft; writing—review and editing. Suzanne W. J. Terheggen‐Lagro: Conceptualization; supervision; writing—review and editing. Eric G. Haarman: Conceptualization; supervision; writing—review and editing. Hettie M. Janssens: Writing—review and editing. Anke‐Hilse Maitland–van der Zee: Conceptualization; supervision; writing—review and editing. E. Marleen Kemper: Conceptualization; funding acquisition; supervision; writing—review and editing. Ron A. A. Mathôt: Conceptualization; methodology; formal analysis; supervision; writing—review and editing.

CONFLICT OF INTEREST STATEMENT

The authors have no conflict of interest to declare.

Supporting information

FIGURE A1. Goodness of fit plots of the final population pharmacokinetic models of tezacaftor and its metabolite tezacaftor‐M1. The dashed lines represent the line of identity. CWRES; conditional weighted residuals.

FIGURE A2. Goodness of fit plots of the final population pharmacokinetic models of ivacaftor and its metabolite ivacaftor‐M1. The dashed lines represent the line of identity. CWRES; conditional weighted residuals.

FIGURE A3. Goodness of fit plots of the final population pharmacokinetic models of ivacaftors'’ metabolite ivacaftor‐M6. The dashed lines represent the line of identity. CWRES; conditional weighted residuals.

TABLE A1. Sensitivity analysis to assess the influence of using the $PRIOR function instead of fixing the parameters for tezacaftor, tezacaftor‐M1, and ivacaftor.

Vonk SEM, Terheggen‐Lagro SWJ, Haarman EG, et al. Real‐world population pharmacokinetics of tezacaftor‐ivacaftor in children with cystic fibrosis: The SYM‐CF study. Br J Clin Pharmacol. 2025;91(10):2969‐2978. doi: 10.1002/bcp.70131

DATA AVAILABILITY STATEMENT

The authors confirm that the data supporting the findings of this study are available within the article and its appendices. Raw data that support the findings of this study are available from the corresponding author, upon reasonable request.