Low-volume LC – MS/MS method for the pharmacokinetic investigation of carvedilol, enalapril and their metabolites in whole blood and plasma: Application to a paediatric clinical trial

Evidence-based pharmacotherapy with carvedilol and enalapril in children suffering from heart failure is insufficient owing to limited pharmacokinetic data. Although a few data sets regarding enalapril, its metabolite enalaprilat and carvedilol in children have been published, pharmacokinetic data on carvedilol metabolites are missing. However, for both drug substances, their active metabolites contribute substantially to drug efficacy. As data can hardly be derived from adults owing to the unknown impacts of enzymatic maturation and ontogeny during childhood, customised assays are important to facilitate paediatric evidence-based pharmacotherapy. Considering ethical paediatric constraints, a low-volume liquid chromatography coupled to mass spectrometry (LC – MS/MS) assay was developed using whole blood or plasma for the quantification of enalapril, enalaprilat, carvedilol, O-desmethyl carvedilol, 4- and 5-hydroxyphenyl carvedilol as well as 3- and 8-hydroxy carvedilol. To facilitate broader applications in adults, the elderly and children, a wide calibration range — between 0.024/0.049 and 50.000 ng/ml — was achieved with good linearity ( r 0.995 all analytes). international bioanalytical guidelines, accuracy, precision, sensitivity and internal standard normalised matrix effects further successfully validated with the exception of those for 3-hydroxy carvedilol, which was therefore assessed semi-quantitatively. Distinct haematocrits did not impact matrix effects or recoveries when analysing whole blood. Blood-to-plasma ratios were determined for all analytes to form the basis for pharmacokinetic model-ling. Finally, incurred sample reanalysis of paediatric samples confirmed the reproducibility of the developed low-volume LC – MS/MS method during study sample analysis. The assay facilitates the reliable generation of important data and contrib-utes towards a safe drug therapy in children.

hepatic carboxylesterase 1 and is converted into the active enalaprilat. 13 Enalaprilat inhibits the ACE and thus the conversion of Angiotensin I into Angiotensin II, which results in vasodilation, decreased preload, natriuresis and prevention of heart remodelling. 14 Carvedilol nonselectively inhibits α 1 -and β-receptors, resulting in peripheral vasodilation and negative chronotropic and inotropic effects. Carvedilol is applied as a racemate and enantioselectively metabolised through aromatic ring hydroxylation by cytochrome P enzymes and glucuronidation. 15 Whereas a pharmacokinetic model for the enantioselective disposition of carvedilol in children has been developed, 16 no data are available on the carvedilol metabolites. Five positional isomers are reported, which include next to the inactive 1-hydroxy carvedilol, 3-hydroxy carvedilol and 8-hydroxy carvedilol, the active 4-hydroxyphenyl carvedilol and 5-hydroxyphenyl carvedilol as well as the further active metabolite O-desmethyl carvedilol. 15,17 The metabolite 4-hydroxyphenyl carvedilol shows a 13-fold increase in β-receptor blocking activity compared with carvedilol and is present in quantities of about 10% of the carvedilol concentration in adults. 18 It is indicated that a faster and altered metabolism of carvedilol is present in children, because significantly shorter half-lives of carvedilol have been reported. 19 Owing to the limited knowledge on the matter, pharmacokinetic investigation of enalapril and carvedilol-along with their metabolites-is necessary and essential for safe pharmacotherapy in children.
However, ethical constraints in studies on children regarding blood sampling (such as fear, discomfort and restricted blood volumes) impede meaningful paediatric clinical trials. To address these concerns, the use of (capillary) whole blood is a reasonable choice for paediatric pharmacokinetic investigations, because it reduces the amount of required blood compared with plasma or serum (no losses due to centrifugation) and is easier to sample than venous blood (finger prick). Difficulties arising from the use of whole blood as the matrix of choice are-amongst others-the effects of haemolysis, the presence of distinct haematocrits and a more complex matrix for sample preparation. Further, differences in drug concentration between plasma and whole blood are to be expected owing to the impact of the haematocrit and the blood-to-plasma ratio of the drugs and metabolites. 20,21 Because paediatric haematocrit reference values vary widely, between 29% and 50% in infants to adolescents, with even higher values in neonates (44%-70%), 22 these determinants have to be carefully evaluated during method development. Further, the investigation of red blood cell (RBC) partitioning is necessary for adequate pharmacokinetic modelling.
The use of liquid chromatography coupled to mass spectrometry (LC-MS/MS) offers a tool to quantify sensitively and targeted drugs and metabolites. Even if LC-MS/MS methods suitable for the detection of enalapril, enalaprilat and carvedilol have been developed for paediatric demands, 23,24 to our knowledge, no bioanalytical method is available to simultaneously assess enalapril, carvedilol and their metabolites. However, this is paramount to ensure safe and evidencebased pharmacotherapy in children suffering from heart failure. Therefore, the aim of this study was to develop an LC-MS/MS method using low volumes of whole blood or plasma facilitating the pharmacokinetic investigation of the drugs carvedilol and enalapril, along with their metabolites (enalaprilat, O-desmethyl carvedilol, 4-and 5-hydroxyphenyl carvedilol and 3-and 8-hydroxy carvedilol) within the framework of a national Dutch paediatric clinical study "CARS II" (CArvedilol Registry Study). 25

| Preparation of working solutions
Each analyte was dissolved in methanol to obtain stock solutions of approximately 100 μg/ml (depending on the exact weight of drug substance). Using these, one working solution containing exactly 10 μg/ml of all analytes and another one containing 1 μg/ml of the internal standards were prepared by dilution with methanol. The internal standard solution was further diluted with methanol on the day of analysis to obtain a working solution of 20 ng/ml.

| Preparation of QCs and calibration curve samples
The quality control (QC) and calibration curve samples were both prepared by spiking whole blood or plasma with the analyte working solution to obtain a concentration of 50.000 ng/ml (upper limit of quantification [ULOQ]).
Calibration curve samples were subsequently serially diluted in a 1:2 ratio to a final concentration of 0.024 ng/ml.
The QCs were prepared independently by serially diluting spiked whole blood or plasma. Five QC concentration levels were investigated during accuracy and precision runs. QC high (37.500 ng/ml) was obtained by mixing equal parts of the 50.000 and 25.000 ng/ml concentration levels. Further QC mid (1.563 ng/ml), two QC low (0.098 and 0.049 ng/ml) and the lower limit of quantification (LLOQ, 0.024 ng/ml) were assessed.

| Sample preparation
The combination of protein precipitation prior to mixed-mode strong cation exchange solid-phase extraction (SPE) was used for the sample preparation. A total of 100 μl of whole blood or plasma sample was mixed with 5 μl of the internal standard solution (20.000 ng/ml); subsequently, 5 μl of 2M aqueous trisodium citrate solution was added.
A Waters Quattro Premier XE (Milford, MA, USA) was used for the mass spectrometric detection. A desolvation gas flow of 900 L/h and a cone gas flow of 50 L/h were applied. Argon was utilised as the collision gas and its flow was set to 0.15 ml/min.
The source temperature was maintained at 135 C and the desolvation temperature at 500 C. The capillary voltage was 3.5 kV.
Substance-specific parameters are shown in Table 1, and the product ion spectra of all analytes and internal standards are displayed in Figure 1.

| Validation
Validation was carried out based on current bioanalytical method vali- (≤ 20% at the LLOQ). The signal-to-noise ratio was aimed to exceed 5 at the LLOQ.

| Matrix effect
One source was used for investigating the absolute matrix effect.
Thereby, a low (0.098 ng/ml in whole blood and 0.049 ng/ml in plasma), a mid (1.563 ng/ml) and a high (37.500 ng/ml) QC were applied.
Further, seven different human sources were used to assess the internal standard normalised matrix effect. Therefore, the extracted matrix was spiked with a low (0.049 ng/ml in plasma/0.098 ng/ml in whole blood) and a high concentration (37.500 ng/ml) in triplicates and compared with the spiked neat solution. The CV of the internal standard normalised matrix effect across all seven sources was restricted to ≤15%, according to the regulatory guidelines of the EMA. 26 T A B L E 1 Substance specific mass spectrometric parameters

| Recovery
To determine the recovery of the extraction process, low (0.049 ng/ml in plasma and 0.098 ng/ml in whole blood), mid (1.563 ng/ml) and high QCs (37.500 ng/ml) were analysed in triplicates using one source (n = 3). Because the guidelines do not specify target values for the recovery, it was aimed to be constant and maximised during method development.

| Carry-over
Carry-over was evaluated by analysing the ULOQ of the calibration curve and a consecutive blank sample without matrix. This setting was measured five times and the mean response in the blank sample was determined. The carry-over should not exceed the signal of the LLOQ by more than 20% for the analytes and 5% for the internal standards. 26,27 2.6.6 | Dilution integrity Whether concentrations above the ULOQ can be adequately determined was evaluated by diluting a 120.000 ng/ml of whole blood or plasma sample with blank whole blood/plasma (in a 1:4 ratio).
This dilution approach was applied to fit the concentration in the calibration curve range. These samples were measured using a threefold approach. The accuracy was allowed to deviate by ±15% (RE) from the nominal concentration, and precision was limited to ≤15% (CV).
F I G U R E 1 Product ion spectra of all analytes and internal standards. m/z: mass-to-charge ratio 2.7 | Pharmacokinetic investigations and applicability

| Evaluation of the influence of the haematocrit
Because distinct haematocrits might influence the matrix effects or recoveries of analytes, haematocrits of 30%, 40%, 50% and 60% were investigated. These values were chosen, as they reflect the reported haematocrit range in children between 2 weeks and 18 years of age. 28 Therefore, blood was separated into plasma and the RBC fraction.
Subsequently, RBCs were lysed by vortexing and multiple freezethaw cycles. Plasma was then spiked with 30%, 40%, 50% or 60% RBCs. A concentration of 10.000 ng/ml was analysed in triplicates for each haematocrit using prespiked and postspiked samples, along with neat solutions.

| Determination of the blood-to-plasma ratio
The blood-to-plasma ratio (K B P ) was determined to investigate the drug and metabolite binding to erythrocytes. Therefore, freshly drawn whole blood was spiked with 10.000 ng/ml of the analytes, cautiously shaken to avoid lysis and left at room temperature. After 30 min, the spiked and unspiked whole blood was centrifuged for 10 min at room temperature and 2,000×g. Plasma and RBC fractions were separated for both samples. The obtained blank plasma and RBCs were also spiked with 10.000 ng/ml of the analyte mix (reference samples). All four obtained plasma and RBC samples were analysed in triplicates, and the blood-to-plasma ratios were calculated (Equation 1). A modified calculation of that used by Hinderling et al. 1997 andYu et al. 2005 was applied to correct for recovery or possible matrix effects in the distinct matrices. 29,30 K B P = area ratio RBC fraction area ratio reference RBC = area ratio plasma fraction area ratio reference plasma where K B P is the calculation of the blood to plasma ratio, RBC is the red blood cell, and H is the haematocrit.

| Continued method performance verification
The method was applied to the determination of carvedilol, enalapril and their metabolites in whole blood and plasma samples of the paediatric CARS II study, a nationwide prospective study in children with dilated cardiomyopathy in the Netherlands. More detailed information on this study is given in the Supporting Information. To continuously monitor the reproducibility of the assay during the analysis of study samples, ISR was applied based on previously established in-house bioanalytical QC systems. 31,32 Following regulatory bioanalytical guidelines, reanalysis of 10% of the first 1,000 study samples with concentrations close to the maximum concentration (c max ) and during the elimination phase is recommended. 26

Plasma
In plasma, a linear range from 0.024 to 50.000 ng/ml was found, which included-in addition to the assessed analytes in whole  26,27 The signal-to-noise ratio at the LLOQ ranged between 26 (5-hydroxyphenyl carvedilol) and 177 (enalapril).
The accuracy and precision results for the QC levels and the LLOQ complied with international bioanalytical guidelines of the EMA and FDA. 26,27 Detailed results for all analytes and both matrices are shown in Table 2. Representative chromatograms for the blank, LLOQ and QC high in whole blood and plasma are displayed in Figure 3.

Plasma
In plasma, mean recoveries were higher than those in whole blood, at

| Carry-over
Whole blood/plasma The carry-over was below 20% of the LLOQ for the following analytes   However, when spiking the internal standards to whole blood samples before precipitation, analyte and internal standard variations between haematocrits resembled each other. Increasing haematocrits from 30% to 60% led to a decrease in recovery of up to −37%, which was especially pronounced for carvedilol ( Figure 4a). However, the normalisation to the internal standard compensated for these losses substantially (Figure 4b). The absolute matrix effects were not influenced by haematocrit variations.
Haematocrit-independent determination of the analytes is important in face of interindividual haematocrit variations and a broader haematocrit range in children. 22 Further, it facilitates conducting the calibration curve in adult whole blood, which is indispensable considering the ethical restrictions regarding paediatric blood sampling. 33

| Blood-to-plasma ratio
The distribution of the drugs between the RBC and plasma fraction was evaluated by determining the blood-to-plasma ratio.  (Figure 5a). Of these, at least 87.5% met the EMA and FDA guideline requirements of ≤20% deviation from the initial determined value for all analytes (Figure 5b). 26

| CONCLUSION
An LC-MS/MS method for the quantification of enalapril, carvedilol and their metabolites tailored for paediatric demands was validated based on key features of bioanalytical guidelines of the FDA and EMA regarding linearity, accuracy, precision, sensitivity, dilution integrity, matrix effect and recovery in whole blood and plasma. The developed method was used to investigate the influence of different haematocrits as well as the drug and metabolite distribution between whole blood and plasma and proved to be suitable for use within these investigations. Next, the method was applied to the measurement of paediatric study samples, where guideline-conforming ISR results were obtained, which confirmed the reproducibility of the method beyond validation. Thus, the assay facilitates the reliable generation of data, which therefore supports the implication of evidencebased pharmacotherapy in children.

FUNDING INFORMATION
This work was in part supported by a combined grant of The Netherlands Heart Foundation (grant number 2013T087) and the Foundation Stichting Hartedroom, The Netherlands.

ACKNOWLEDGEMENT
Open access funding enabled and organized by Projekt DEAL.