Relative bioavailability of delamanid 50 mg tablets dispersed in water in healthy adult volunteers

Abstract

Aim

Delamanid is a novel drug for the treatment of drug‐resistant tuberculosis, manufactured as 50‐mg solid and 25‐mg dispersible tablets. We evaluated the effects of dispersing the 50‐mg tablet, focusing on the relative bioavailability.

Methods

Delamanid, 50‐mg tablets administered dispersed vs swallowed whole, was investigated in a phase I, four‐period, crossover study. Two of three dose strengths of delamanid (25, 50 or 100 mg) were given to healthy adult participants, in both whole and dispersed forms, with a 7‐day washout period. Blood samples were collected over 168 h after each dose. Delamanid and its metabolite DM‐6705 were analysed with a validated liquid chromatography tandem mass spectrometry assay. The pharmacokinetics of both analytes were analysed using nonlinear mixed‐effect modelling. Palatability and acceptability were determined using a standardized questionnaire.

Results

Twenty‐four participants completed the study. The bioavailability of dispersed tablets was estimated to be 107% of whole tablets, with a 90% confidence interval of 99.7‐114%, fulfilling bioequivalence criteria. The two formulations were not significantly different regarding either bioavailability or its variability. Bioavailability increased at lower doses, by 34% (26‐42%) at 50 mg and by 74% (64‐86%) at 25 mg, relative to 100 mg. The majority of participants (93%) found the dispersed formulation acceptable in palatability across all delamanid doses.

Conclusions

Dispersed 50‐mg delamanid tablets have similar bioavailability to tablets swallowed whole in adult volunteers. This can be an option for children and other patients who cannot swallow whole tablets, improving access to treatment.

What is already know about this subject?

• Delamanid is available as 50‐mg solid tablets for adults and 25‐mg dispersible tablets for children.

• The latter is not yet widely available in some settings.

• Manipulating the adult formulation by dispersing could be an option for children without access to 25‐mg dispersible tablets, but there is a risk of altering the bioavailability.

What this study adds?

• We evaluated the relative bioavailability of delamanid 50‐mg tablets administered dispersed in water compared to swallowed whole.

• The results showed that delamanid dispersed and swallowed whole are bioequivalent according to the 80%–125% criteria, and dispersed delamanid has favourable palatability, supporting the use in children and other populations who have difficulty in swallowing.

1. INTRODUCTION

Delamanid (OPC‐67683, Deltyba) is a nitro‐dihydro‐imidazooxazole antibiotic with antimycobacterial activity derived from blocking the production of mycolic acids of Mycobacterium tuberculosis. ¹ Based on the phase II trial data, delamanid as 50‐mg tablets received conditional regulatory approval from the European Medicines Agency (EMA) for the treatment of multidrug‐resistant TB in 2014. ² , ³ In the same year, the World Health Organization (WHO) issued guidance for delamanid's use in adults under limited conditions and later, in the guidelines for treatment of rifampicin‐resistant (RR) TB, delamanid was classified as Group C medication. ⁴ , ⁵ For adults, delamanid is currently recommended to be given for 6 months as 100 mg twice daily taken together with food. Child‐friendly 25‐mg dispersible delamanid tablets were approved by the EMA in 2021 for use in children down to 10 kg based on paediatric clinical trials. ⁶ However, access to the dispersible formulation is limited currently, and it may be several years before it is widely available in programmatic settings.

The pharmacokinetics (PK) of delamanid 50‐mg tablets have been described. ³ , ⁶ , ⁷ Peak plasma concentration is observed after about 4 h. The oral bioavailability of delamanid is highly affected by food due to its high lipophilicity, increasing about 2.7‐fold with a standard meal and >4‐fold with high‐fat meal. The plasma exposures exhibit nonlinearity with dose, increasing less than proportionally with increasing doses. Delamanid is metabolized primarily through albumin to metabolite DM‐6705 and thereafter further metabolized modestly by CYP3A4. Both delamanid and DM‐6705 have high fractions of protein binding (>99%) and long half‐lives (30‐38 and 159‐600 h, respectively). It is important to monitor the concentration of DM‐6705 in plasma concurrently considering its strong correlation with the prolongation of QT interval. ² , ³

There is an urgent need to accelerate access to novel treatments for children with drug‐resistant TB globally. ⁸ Delamanid remains crucial in the treatment for children with RR TB where therapeutic options are limited. Given the difficulty of accessing the delamanid paediatric formulation in some settings, it could be an option to manipulate the 50‐mg tablets, such as by dispersing or crushing, for children and people who cannot swallow whole tablets. However, off‐label use by these physical manipulations introduces potential challenges, such as inaccurate dosing, impact on bioavailability and issues with stability, as reported by the EMA. ⁹ The WHO raised concern for crushing or dispersing delamanid 50‐mg tablets in its 2019 drug‐resistant TB guidelines and suggested that formulation manipulation should be avoided until the effects on bioavailability are known. ¹⁰

In this study, we evaluated the PKs and relative bioavailability of the delamanid 50‐mg tablet formulation as dispersed in water compared to swallowed whole. We also compared the difference between the two methods of administration in rate of absorption, as well as safety, palatability and acceptability.

2. METHODS

2.1. Study design

The relative bioavailability of delamanid 50‐mg tablets when administered in dispersed formulation vs swallowed whole under fed conditions was investigated in a phase I single‐dose, open‐label, randomized, four‐period, crossover study at a single centre in Cape Town, South Africa (TASK, Brooklyn). The study was approved by the Stellenbosch University Health Research Ethics Committee (reference number M20/02/005) and the WHO Ethical Review Committee prior to initiation and conducted in accordance with the International Council for Harmonization and South African guideline for Good Clinical Practice (E6). All participants provided written informed consent prior to the conduct of study procedures. The trial is registered with the Pan African Clinical Trials Registry (PACTR202007578774941).

Healthy adult participants were randomly assigned 1:1:1 to separately receive two of three dose strengths of delamanid after a meal; both whole and dispersed tablets were given for each dose strength. The two dose strengths administered to each participant were randomly selected from 25 (½ × 50 mg), 50 and 100 mg (2 × 50 mg). Before each dose, participants were fasted for at least 10 h overnight, and thereafter they received a standard breakfast with about 670 kcal, of which 33% was fat. Participants received a dose within 30 (±5) min after the meal either first as the dispersed formulation and second as the whole tablet, or vice versa. In total, four dosing occasions with combinations of two doses and two formulations were assigned to each participant at days 1, 8, 15 and 22, with 7 days between each occasion as washout period.

The two dose strengths of delamanid received and the order of the formulation in which each was first administered were determined by the randomisation sequence. The study pharmacist was responsible for allocating the predetermined randomisation sequence to eligible participants. The study was conducted in an unblinded fashion, and only the laboratory staff performing the PK analysis were blinded to delamanid dose and formulation administered.

The planned sample size, 24 participants, which contributed 16 participants per dose and formulation (according to the 1:1:1 dosing allocation mentioned above), and the PK sampling schedule (further described in Section 2.3), was determined based on clinical trial simulations in which a population PK model built on a paediatric clinical trial (Trial 232, ClinicalTrials.gov identifier: NCT01856634) was used. The simulation model was a prefinal version of the later published model adjusted with allometric scaling factors. ¹¹ The simulations showed that the study design provided a power of 84% to detect a 20% decrease (regarded as potentially clinically relevant and bioinequivalence) in bioavailability with dispersing, and a power of 77% to detect formal bioequivalence, that is, within 80‐125% of the difference in bioavailability. ¹²

2.2. Delamanid preparation and administration

The dispersed formulation was prepared by the study pharmacists (TASK Pharmacy, Bellville) by allowing the tablet(s) to disperse over 5‐6 min in 5 mL of water, after which 5‐15 mL of sugar syrup was added. For each dispersed delamanid formulation, a volume of 10 mL was administered, followed by a rinse step with another 10 mL of distilled water. To prepare the 25‐mg tablet to be received as “whole tablet”, a 50‐mg tablet was first weighed then divided along the middle using a pill cutter. Each of the two halves was then weighed, and any half tablet within 2.5 mg of the expected half weight of the whole tablet was considered acceptable for administration. See the detailed methods for preparing doses in Data S1.

2.3. Study participants

Male and female volunteers between 18 and 55 years, and 40 to 90 kg (inclusive) were recruited. Exclusion criteria included body mass index over 30 kg/m², active TB or recent contact with an infectious TB case, human immunodeficiency virus infection, abnormal cardiac arrhythmia, other medical conditions that were judged to jeopardize the safety of participants or affect the study validity, and the use of prohibited medications, such as cytochrome P450 3A4 (CYP3A4) inhibitors or inducers, or drugs used to treat TB or known to prolong QT interval. Female participants were excluded if pregnant or breastfeeding. Data were deidentified before analysis.

2.4. Pharmacokinetic sample collection

Clinical trial simulations were utilized to optimize the timing of blood samples for PK analysis in the study, with the aim of reducing the number of samples and shortening the washout period. Sampling schedules from 2 to 4 days, with or without an additional sample at day 7, were considered. The selected schedule reached power over 80% to detect a 20% change in bioavailability with dispersing, with a controlled type I error. The final schedule of sample collection was predose, 1, 2, 4, 6, 8, 24 and 48 h after each dose and 168 h after the fourth dose.

K2‐EDTA tubes were used to collect 2 mL of blood at each timepoint, which was placed in an ice‐bath immediately after collection and protected from light during processing. Samples were centrifuged at 1500 g for 10 min at 4°C, followed by transfer of 450 μL of plasma into each of two amber cryovials (primary and back‐up) for storage at −80°C within 1 h of sample collection. Delamanid and its metabolite DM‐6705 were analysed with a validated liquid chromatography tandem mass spectrometry assay developed at the Division of Clinical Pharmacology, University of Cape Town (see Data S1 for detailed assay methods).

2.5. Pharmacokinetic data analysis

The PK data were analysed using nonlinear mixed‐effects modelling. The data below the limit of quantification of 1 ng/mL were excluded from the analysis. The model used for the clinical trial simulations was used as a starting point for the analysis. ¹¹ In this model, both delamanid and DM‐6705 were described by two‐compartment models with first‐order elimination, with the delamanid clearance being the input to DM‐6705. Delayed absorption of delamanid was described by a transit compartment, parameterized by mean absorption time. ¹³ Allometric scaling with weight was applied to clearance, volume distributions (central and peripheral compartment) and inter‐compartment clearance of both compounds. Bioavailability increased at doses lower than 50 mg. The fraction metabolized to DM‐6705 was fixed to 100% but allowed to have interindividual variability (IIV). IIV was also included for clearance and volume distribution (central compartment) of delamanid and clearance of DM‐6705. Variability between occasions (defined as each new dosing event, called interoccasion variability [IOV]) was included in mean absorption time and bioavailability. The IIV and IOV followed log‐normal distributions around the typical PK parameters. The unexplained residual variability was modelled as proportional.

The adequacy of the starting model was assessed and the parameter values were re‐estimated. Effects of dose and formulation on bioavailability and absorption parameters, and their variabilities, and dose dependency on bioavailability were tested at a significance level of 0.05 using the likelihood ratio test. The updated model was evaluated through visual predictive checks (VPC, n = 1000 simulations). Individual secondary PK parameters of the observations, that is, areas under the concentration‐time curve (AUC) from time 0 to 48 h, maximum concentrations (C _max) and time to C _max (T _max), were derived using noncompartment analysis (NCA). Posterior predictive checks based on NCA were also used to evaluate the performance of the final model. The secondary PK parameter of interest of delamanid, AUC_0‐48, was calculated from 200 model‐simulated datasets with the same study design; the distribution of simulated AUC_0‐48 was then compared with the AUC_0‐48 derived from original dataset. Both AUC calculation and predictive checks were performed using the ncappc tool. ¹⁴ The method to calculate AUC was linearup‐logdown, that is, linear interpolation for the absorption phase and log interpolation for the elimination phase. For all the estimated parameters in the final model, the 90% confidence intervals (CIs) were determined with the sampling importance resampling (SIR) method. ¹⁵

The 90% CI of the formulation effect (dispersed vs whole) on bioavailability was additionally determined with log‐likelihood profiling (LLP), which is a function available in the PsN toolkit. ¹⁶ To further evaluate the formulation effect, simulations for virtual bioequivalence were performed. One thousand datasets were simulated using the final model with parameter uncertainty, including 500 simulated participants in each formulation group (dispersed vs whole) at each dose level (25, 50 and 100 mg). Participants received single doses in parallel; the washout period was not included in the simulations. Simulated body weight was assumed to be normally distributed with the same mean, standard deviation, maximum and minimum as in the study. The uncertainty of the model parameters was accounted for by sampling 1000 parameter sets from the SIR distribution and updating the model with these samples to simulate corresponding 1000 datasets. In each dataset, geometric mean ratios of model‐derived, log‐transformed C _max, AUC_0‐48 and AUC_0‐168 were summarized, respectively, by comparing the two formulation groups. The final CI of the geometric mean ratios was computed as the 90% CI of the 1000 ratios from all the datasets. The dispersed formulation was judged as bioequivalent to the whole tablets if the 90% CI of the formulation effect and geometric mean ratio of PK metrics fell within 80‐125%, which indicated a nonsignificant (level of .05) change in the bioavailability of the dispersed formulation relative to the whole tablets. ¹⁷

The modelling and evaluation procedures were conducted with NONMEM 7.5, PsN, R, xpose4 and pirana workbench. ¹⁸ , ¹⁹ , ²⁰ , ²¹ The modelling and data processing were performed at the Uppsala Multidisciplinary Center for Advanced Computational Science (UPPMAX).

2.6. Palatability assessment

A standardized questionnaire with a Likert‐type scale was used to determine participants' experience of palatability and acceptability of both delamanid formulations, dispersed and whole tablets. The questionnaires were completed within 1 h after the delamanid dose on each of the four dosing days. The questionnaire considered various properties of each dose and formulation, including visual appearance, smell, taste, texture and volume, as well as each participant's perception of how a child may feel about these properties. We reported the results dichotomized into acceptable (neutral, like, like very much) vs unacceptable (dislike, dislike very much) using summary statistics.

2.7. Safety monitoring

For each participant, we performed a physical examination, 12‐lead ECG and assessment of vital signs at the screening visit prior to each dose and at the follow‐up visits 7 and 14 days after the last dose of delamanid administered. We also collected blood and urine samples at screening, once during the dosing period and at the follow‐up visits to detect any clinically significant changes from baseline. Participants were monitored for the development of adverse events from the first dose of study treatment until the final follow‐up visit. The attribution of events to delamanid was assessed by study investigators, and the severity was graded according to the Common Terminology Criteria for Adverse Events version 5.0.

3. RESULTS

3.1. Study demographics

A total of 21 female and five male participants were enrolled, of which two withdrew before completion of the study and were replaced with new participants (see participant disposition in Supporting Information Table S1). Data from the two early‐withdrawn participants were kept in the analysis. Detailed demographics are summarized in Table 1 (demographics summary per dose and formulation are available in Supporting Information Table S2).

TABLE 1.

Demographics of all participants (n = 24 + 2 withdrawals).

Characteristics	Number (percentage) or median (minimum‐maximum)
Gender
Male	5 (19.2)
Female	21 (80.8)
Race
Coloured a	14 (53.8)
Black	8 (30.8)
Caucasian	4 (15.4)
Age (years)	23.8 (19.0‐53.5)
Weight (kg)	61.0 (43.4‐81.3)
Height (m)	1.61 (1.52‐1.83)
BMI (kg/m2)	23.7 (16.7‐29.8)

Abbreviation: BMI, body mass index.

^a Coloured race here refers to South African Coloured, a multiracial ethnic group native to Southern Africa.

3.2. Pharmacokinetic data analysis

After removing 141 observations below the limit of quantification, 763 and 714 observations of delamanid and DM‐6705 were included in the analysis. The structural model remained unchanged for both delamanid and DM‐6705, except that the number of transit compartments was increased to four.

Neither bioavailability nor the variability in bioavailability showed a significant difference between the two methods of administration. The bioavailability of the dispersed tablets was estimated to be 107% compared to whole tablets, with a 90% CI of 99.7‐114% using LLP, fulfilling the standard 80‐125% criterion for bioequivalence. Similar CIs were obtained for the formulation effect using SIR (100‐113%). The simulation‐based evaluation is shown in Figure 1, where all the secondary PK metrics at all dose levels fulfil the bioequivalence criterion.

FIGURE 1.

Bioequivalence test ratio using simulation‐based evaluation, summarized for C _max, AUC_0‐48 and AUC_0‐168 at dose levels of 25, 50 and 100 mg. Dashed vertical lines represent the standard bioequivalence limit of 80‐125%, with a reference line (solid) at 100%.

A formulation effect (dispersed vs whole) was included in the mean transit time (MTT), resulting in mean absorption times (MTT + 1/[absorption rate, KA]) of typically 2.64 (90% CI 2.29‐3.04) h for whole tablets and 1.93 (90% CI 1.56‐2.40) h for dispersed tablets. The variability in MTT was not found to be significantly different between the formulations. Figure 2A illustrates the typical concentration curves of delamanid and DM‐6705, showing similar profiles with different formulations across the doses, except the minor difference in the absorption phase. The nonlinear relationship between bioavailability and dose was confirmed and was the same for both administration methods. Relative to 100 mg, the bioavailability at 50 mg was 34% (90% CI 26‐ 42%) higher, while at 25 mg it was 74% (90% CI 64‐86%) higher, as shown in Figure 2B.

FIGURE 2.

(A) Typical concentration vs time profiles after dose to 24 h at each dose level for delamanid and DM‐6705, simulated from the final PK model. (B) Change in bioavailability of delamanid with dose ranged from 25 to 100 mg relative to 100 mg, with 90% confidence intervals based on uncertainty in parameter estimates.

The NONMEM code of the final model is available in Data S2. The final estimates of the PK parameters with their precisions are summarized in Table 2. The VPCs shown in Figures 3 and 4 demonstrate that the final model fits the data well for both delamanid and DM‐6705. Ncappc in Figure 5 showed general agreement between model‐predicted and observed AUC_0‐last of delamanid, except for the underprediction shown at the group of delamanid dispersed 50 mg (the observed median slightly outside 95% model prediction interval). The NCA‐derived T _max, C _max and AUC_0‐48 of the observed data are summarized in Table 3 for illustration. However, the carryover effect in the NCA summary is not adjusted for here, thus any conclusion should be drawn carefully.

TABLE 2.

Parameter estimates of the final population pharmacokinetic model in healthy participants.

Parameter	Estimate a	90% confidence interval
Typical parameter (unit) b
KA (h−1)	0.812	0.681‐0.959
MTT (h)	1.41	1.25‐1.57
Dispersing effect ( ${\theta }_1$): ${\mathrm{MTT}}_{\text{disperse}}=\mathrm{MTT}\times {\theta }_1$	0.494	0.413‐0.590
F	1 (fixed)
Dose effect: $F_{\text{Dose}}=F\times {\theta }_2^{\left[\text{if}50\mathrm{mg}\right]}\times {\theta }_3^{\left[\text{if}25\mathrm{mg}\right]}$
${\theta }_2$	1.34	1.26‐1.42
${\theta }_3$	1.74	1.64‐1.86
CLDLM/F (L/h)	20.3	19.1‐21.7
VdDLM,cent/F (L)	258	223‐297
VdDLM,peri/F (L)	610	566‐660
Q DLM/F (L/h)	34.3	30.3‐38.5
CLM1/F/f m (L/h)	90.9	83.2‐98.1
VdM1,cent/F/f m (L)	532	405‐666
VdM1,peri/F/f m (L)	20 800	19 500‐22 100
Q M1/F/f m (L/h)	1800	1700‐1910
Interindividual/occasion variability c
IIV on CLDLM	11.3%	8.63‐13.5
IIV on VdDLM,cent	41.1%	35.2‐47.1
IIV on CLM1	22.0%	17.3‐27.8
IIV on f m	13.8%	10.2‐17.5
IOV on F	15.5%	13.2‐17.8
IOV on KA	49.8%	41.0‐60.0
IOV on MTT	52.6%	45.5‐60.3
Residual unexplained variability c
Proportional error in DLM	24.9%	23.7‐26.2
Proportional error in M1	9.70%	9.29‐10.2

Abbreviations: CL, clearance; DLM, delamanid; F, bioavailability; f _m, fraction metabolized; IIV, inter‐individual variability; IOV, inter‐occasion variability; KA, absorption rate; M1, metabolite DM‐6705; MTT, mean transit time; Q, inter‐compartment clearance; Vd, volume distribution.

^a Estimates of typical parameters are relative to a healthy participant weighing 60 kg who received 100‐mg whole tablets.

^b An allometric scaling factor is added on CL, Vd and Q of both two analytes as $P_i=P\times {\left(\frac{W{T}_i}{60}\right)}^x$, where P _i is individual typical parameter; $x$ is 0.75 for CL and Q and 1 for Vd.

^c Estimates of variabilities are reported as %coefficients of variations (CV%, the square root of estimated variances × 100%).

FIGURE 3.

Visual predictive check of the final model on the (log‐scaled) observed delamanid concentrations (open circles) from the study by administration methods and doses from time after dose to 60 h. The solid and dashed black lines, respectively, represent the 50th and 5th/95th percentiles of the observations in the data set, with corresponding red and blue areas representing the 95% model‐predicted confidence intervals.

FIGURE 4.

Visual predictive check of the final model on the (log‐scaled) observed metabolite DM‐6705 concentrations (open circles) from the study by administration methods and doses from time after dose to 60 h. The solid and dashed black lines, respectively, represent the 50th and 5th/95th percentiles of the observations in the data set, with corresponding red and blue areas representing the 95% model‐predicted confidence intervals.

FIGURE 5.

Histogram distribution of simulated AUC_0‐last of delamanid from 200 model‐based simulations, compared with the observed AUC represented by red line. Grey lines indicate the 2.5th, 97.5th (dashed) and 50th (solid) percentiles of the simulated distribution.

TABLE 3.

Secondary PK metrics summary using noncompartmental analysis (carryover not adjusted for).

	Whole	Disp	Whole	Disp	Whole	Disp
	25 mg (n = 16)	25 mg (n = 17)	50 mg (n = 16)	50 mg (n = 17)	100 mg (n = 16)	100 mg (n = 16)
Delamanid
T max (h)	3.81 (1.77, 5.87)	3.88 (1.77, 5.93)	3.86 (2.10, 5.90)	3.87 (1.80, 5.88)	3.82 (1.82, 5.93)	3.97 (3.75, 6.07)
C max (ng/mL)	136 (86.4, 193)	106 (66.9, 169)	192 (119, 290)	208 (106, 241)	233 (112, 424)	224 (126, 442)
AUC0‐48 (ng·h/mL)	1522 (1076, 2065)	1468 (870, 2097)	2346 (1525, 3710)	2602 (1508, 3683)	3488 (1994, 4746)	3809 (2312, 4789)
DM‐6705
T max (h)	5.84 (3.77, 48.4)	7.88 (3.8, 48.3)	5.87 (2.10, 47.1)	5.88 (3.75, 47.9)	7.82 (3.77, 48.7)	23.4 b (3.75, 48.3)
C max (ng/mL)	2.5 (1.2, 5.3)	2.8 (1.2, 5.3)	3.3 (1.3, 7.1)	2.8 (1.9, 9.5)	4.5 (3.2, 6.7)	5.2 (2.6, 8.1)
AUC0‐48 (ng·h/mL)	105 (50, 222)	106 a (50, 221)	136 (55, 312)	122 (75, 359)	184 (113, 300)	187 (98, 359)

Note: PK metrics are summarized as median (minimum, maximum).

Abbreviations: Disp, delamanid dispersed; Whole, delamanid whole tablet(s).

^a The AUC minimum of 50 was reported after excluding an incomplete PK profile (<5 h).

^b Seven participants had T _max above 40 h; seven were less than 10 h; two were around 24 h, contributing to the median of 23.4 h.

3.3. Palatability assessment

Table 4 summarizes participant responses on the acceptability of the formulations. Overall, 95% of participants found the combined properties of both formulations and all dose strengths of delamanid acceptable, with similar overall acceptability between the different dose strengths for the dispersed formulation groups. Based on the participants' responses on how they anticipate a child would perceive the taste, volume and overall acceptability, 100%, 82% and 87% of participants found the dispersed formulations of 25, 50 and 100 mg, respectively, to be acceptable.

TABLE 4.

Acceptability by delamanid dose and formulation.

	All participants	Disp	Disp	Disp	Whole	Whole	Whole
		25 mg	50 mg	100 mg	25 mg	50 mg	100 mg
	n = 98 (%)	n = 17 (%)	n = 17 (%)	n = 16 (%)	n = 16 (%)	n = 16 (%)	n = 16 (%)
Appearance	87 (88.8)	15 (88.2)	16 (94.1)	14 (87.5)	16 (100.0)	11 (68.8)	15 (93.8)
Smell	92 (93.9)	16 (94.1)	16 (94.1)	14 (87.5)	16 (100.0)	15 (93.8)	15 (93.8)
Taste	90 (91.8)	15 (88.2)	17 (100.0)	15 (93.8)	15 (93.8)	13 (81.3)	15 (93.8)
Texture	90 (91.8)	17 (100.0)	15 (88.2)	13 (81.3)	15 (93.8)	16 (100.0)	14 (87.5)
Amount	90 (91.8)	15 (88.2)	17 (100.0)	15 (93.8)	16 (100.0)	12 (75.0)	15 (93.8)
Overall	93 (94.9)	17 (100.0)	16 (94.1)	15 (93.8)	16 (100.0)	13 (81.3)	16 (100.0)
Child‐Taste	83 (84.7)	17 (100.0)	17 (100.0)	16 (100.0)	11 (68.8)	9 (56.3)	13 (81.3)
Child‐Amount	75 (76.5)	17 (100.0)	16 (94.1)	14 (87.5)	12 (75.0)	6 (37.5)	10 (62.5)
Child‐Overall	78 (79.6)	17 (100.0)	14 (82.4)	14 (87.5)	13 (81.3)	9 (56.3)	11 (68.8)

Note: For Likeability and Acceptability, only Likeable and Acceptable values are displayed in this table.

Abbreviations: (%), percentage of the total number of participants per group; Acceptable, neutral, like, like very much; Child‐, if a child was required to take this formulation, how the participant thought the child would feel about the formulation; Disp, delamanid dispersed; Unacceptable, dislike, dislike very much; Whole, delamanid whole tablet(s).

3.4. Safety monitoring

A total of 30 adverse events were reported (see Supporting Information Table S3), with no clear correlation with dose or formulation of delamanid administered. The most commonly reported adverse event of pruritis was reported with similar frequencies between all groups of delamanid dose and formulation. All adverse events were graded as either mild (77%) or moderate, with no severe, serious or fatal adverse events occurring. Only five events were considered to be possibly related to delamanid. One participant, who received the dispersed formulation with dose 50 mg, experienced an adverse event (gastroenteritis) leading to study withdrawal. No participants experienced any clinically significant changes in ECG parameters, vital signs, safety laboratory tests or physical examinations over the course of the study.

4. DISCUSSION

In this study, we compared among healthy participants two methods of administration of delamanid, dispersing in water before administration vs. swallowing whole, at several dose levels, for differences in bioavailability and secondarily palatability and acceptability. Using a population PK modelling approach, we established that the bioavailability when dispersing a 50‐mg tablet of delamanid was unaltered compared to swallowing whole or halved tablets. Dispersing the tablets following the method here was highly acceptable to participants.

A difference in mean absorption time was found in the two preparations of delamanid, with typically 27% faster mean absorption time for dispersed compared to whole tablets. This is in contrast to the reported effect of delamanid dispersible tablets (50% slower) compared to whole tablets. ¹¹ However, this finding was considered not clinically relevant, especially as the maximum concentration (Figure 2A) remained similar between the formulations. The earlier observed negative correlation between dose and bioavailability was confirmed in this study. ¹¹ , ²² , ²³ This effect could be due to limited solubility of delamanid with increasing dose, leading to limited absorption. Larger variability in absorption parameters was informed by the developed model. Evaluating study doses ranging from 25 to 100 mg, which covers the dose of paediatric formulation to the recommended adult dose, provides more direct comparison and suggests the interchangeability between the methods across these dose levels.

The dispersed formulations of delamanid were considered acceptable by the majority of participants, including participants' expectation of how a child may perceive this formulation across the three dose strengths. In our study, the dispersed formulations were prepared with water and sugar syrup, which is expected to improve palatability. Although sugar syrup is simple to prepare, it may not always be accessible in the field. Delamanid is not known to be a particularly bitter compound and would not be expected to have poor palatability when dispersed just in water, but this was not assessed in our study. As expected for this study, no significant safety concerns were raised.

Although the child‐friendly 25‐mg dispersible tablet formulation of delamanid has recently become commercially available and active scale‐up in access is underway, it will take some time before this paediatric formulation is widely available. ²⁴ , ²⁵ Even though the WHO has recommended delamanid for children 3 to <6 years of age since 2019, limited access to the 25‐mg dispersible tablet and concerns about the impact of manipulating the 50‐mg tablet have contributed to few children actually taking this much‐needed drug. ⁵ , ¹⁰ This study provides important evidence that manipulating the 50‐mg tablet by dispersing for children or other patients who cannot swallow whole tablets does not affect its bioavailability and can be safely done in settings where the 25‐mg dispersible tablet is not available. This is very timely as the WHO has now (March 2022) recommended that delamanid be used in children of all ages diagnosed with RR TB. ¹⁰

The study was designed and analysed using a modelling approach, considering the long half‐lives of both delamanid and DM‐6705. For compounds with a long half‐life, it may be challenging to evaluate in crossover studies using conventional methods due to a limited washout period causing carryover effects. With model‐based analysis, it was not required to confirm negligible concentrations prior to each administration, as the model quantitatively accounted for the carryover effects. Clinical trial simulations were used in this study to ensure the desired study power and quality of the data, while reducing the sampling period and shortening the washout period. Studies demonstrated that model‐based bioequivalence evaluation has relatively higher power than traditional NCA‐based methods, especially for long half‐life drugs such as delamanid and DM‐6705, and SIR was also shown to have better performance to control type I error and increase study power. ²⁶ , ²⁷ , ²⁸

The data below the quantification limit (BQL) were directly removed from the analysis. There is concern that removing BQL data unduly could cause biased estimates, for instance the remaining data points at low concentration levels could be mistreated as the real lowest levels (biased upwards). ²⁹ In this analysis, evaluation was made and we found that the model‐predicted concentrations for DM‐6705 at time points of observed BQL were well below limit, reflecting the observation data without bias (91 out of 95, ie, 96% were predicted below limit). A bias is not expected in the estimation of delamanid PKs as only 1% of data was removed.

However, some study limitations may be noted. The study was conducted in healthy volunteers with a controlled predose meal. The absorption behaviour in children could be different from adults. ³⁰ The meal type and effect in paediatric patients could also be unpredictable as delamanid exposure is highly affected by food intake. Splitting 50‐mg tablets into 25 mg does in the study did not introduce influential error and was well controlled within acceptable limit under rigorous operation. It should be noted that the sensitivity of operation error was not evaluated and in reality the error could be larger without careful operations.

Overall, based on evidence from this study, we conclude that dispersing 50‐mg delamanid tablets results in the same bioavailability as whole tablets and can be used in children or other patients who cannot swallow whole tablets, therefore improving access to treatment.

COMPETING INTERESTS

No conflicts of interest to declare.

Supporting information

DATA S1 Dose preparation and assays

SUPPORTING INFORMATION TABLE S1. Participant disposition

SUPPORTING INFORMATION TABLE S2. Participant demographics by delamanid dose and formulation

DATA S2. NONMEM model code

SUPPORTING INFORMATION TABLE S3. Summary of adverse events by delamanid dose and formulation

Zou Y, de Jager V, Hesseling AC, et al. Relative bioavailability of delamanid 50 mg tablets dispersed in water in healthy adult volunteers. Br J Clin Pharmacol. 2025;91(4):957‐967. doi: 10.1111/bcp.15672

DATA AVAILABILITY STATEMENT

The data of this study are not openly available due to sensitivity issues.