B. G. Charles

A procedure for generating bootstrap samples for the validation of nonlinear mixed-effects population models

A method is presented for automated preparation of bootstrap data samples and their presentation to NONMEM for use in the validation of population pharmacokinetic or pharmacodynamic models, which have been developed with relatively small numbers of subjects. The bootstrap sampling procedure involves the use of an MS-DOS batch file and a script file for use with the public-domain text processing program, AWK, which is a single EXE file (48k in size for the version used in this report). A UNIX version of AWK is also available and UNIX users can adapt the process to their needs. Its use obviates the need for expensive, high-end statistical packages and associated script files written using in-built programming languages. The method can easily be adapted to bootstrap sampling requirements in other biomedical modelling applications.

Population pharmacokinetics of metformin in late pregnancy

The pharmacokinetic disposition of metformin in late pregnancy was studied together with the level of fetal exposure at birth. Blood samples were obtained in the third trimester of pregnancy from women with gestational diabetes or type 2 diabetes; 5 had a previous diagnosis of polycystic ovary syndrome. A cord blood sample also was obtained at the delivery of some of these women, and also at delivery of others who had been taking metformin during pregnancy but from whom no blood had been taken. Plasma metformin concentrations were assayed by a new, validated, reverse-phase HPLC method. A 2-compartment, extravascular maternal model with transplacental partitioning of drug to a fetal compartment was fitted to the data. Nonlinear mixed-effects modeling was performed in NONMEM using FOCE with INTERACTION. Variability was estimated using logarithmic interindividual and additive residual variance models; the covariance between clearance and volume was modeled simultaneously. Mean (range) metformin concentrations in cord plasma and in maternal plasma were 0.81 (range, 0.1-2.6) mg/L and 1.2 (range, 0.1-2.9) mg/L, respectively. Typical population values (interindividual variability, CV%) for allometrically scaled maternal clearance and volume of distribution were 28 L/h/70 kg (17.1%) and 190 L/70 kg (46.3%), giving a derived population-wide half-life of 5.1 hours. The placental partition coefficient for metformin was 1.07 (36.3%). Neither maternal age nor weight significantly influenced the pharmacokinetics. The variability (SD) of observed concentrations about model-predicted concentrations was 0.32 mg/L. The pharmacokinetics were similar to those in nonpregnant patients and, therefore, no dosage adjustment is warranted. Metformin readily crosses the placenta, exposing the fetus to concentrations approaching those in the maternal circulation. The sequelae to such exposure, eg, effects on neonatal obesity and insulin resistance, remain unknown.

High dose caffeine citrate for extubation of preterm infants: a randomised controlled trial

Objective: To compare two dosing regimens for caffeine citrate in the periextubation period for neonates born at less than 30 weeks gestation in terms of successful extubation and adverse effects. Design: A multicentre, randomised, double blind, clinical trial. Setting: Four tertiary neonatal units within Australia. Patients: Infants born less than 30 weeks gestation ventilated for more than 48 hours. Interventions: Two dosing regimens of caffeine citrate (20 v 5 mg/kg/day) for periextubation management. Treatment started 24 hours before a planned extubation or within six hours of an unplanned extubation. Main outcome measure: Failure to extubate within 48 hours of caffeine loading or reintubation and ventilation or doxapram within seven days of caffeine loading. Results: A total of 234 neonates were enrolled. A significant reduction in failure to extubate was shown for the 20 mg/kg/day dosing group (15.0% v 29.8%; relative risk 0.51; 95% confidence interval (CI) 0.31 to 0.85; number needed to treat 7 (95% CI 4 to 24)). A significant difference in duration of mechanical ventilation was shown for infants of less than 28 weeks gestation receiving the high dose of caffeine (mean (SD) days 14.4 (11.1) v 22.1 (17.1); p  =  0.01). No difference in adverse effects was detected in terms of mortality, major neonatal morbidity, death, or severe disability or general quotient at 12 months. Conclusions: This trial shows short term benefits for a 20 mg/kg/day dosing regimen of caffeine citrate for neonates born at less than 30 weeks gestation in the periextubation period, without evidence of harm in the first year of life.

Reactions of cisplatin hydrolytes with methionine, cysteine, and plasma ultrafiltrate studied by a combination of HPLC and NMR techniques.

The reactions of cis-[PtCl(NH3)2(H2O)]+ with L-methionine have been studied by 1D 195Pt and 15N NMR, and by 2D[1H, 15N] NMR. When the platinum complex is in excess, the initial product, cis-[PtCl(NH3)2(Hmet-S)]+ undergoes slow ring closure to [Pt(NH3)2(Hmet-N,S)]2+. Slow ammine loss then occurs to give the isomer of [PtCl(NH3)(Hmet-N,S)]+ with chloride trans to sulfur. When methionine is in excess, a reaction sequence is proposed in which trans-[PtCl(NH3)(Hmet-S)2]+ isomerises to the cis-isomer, with subsequent ring closure reactions leading to cis-[Pt(Hmet-N,S)2]2+. Near pH 7, methionine is unreactive toward cis-[PtCl(OH)(NH3)2]. By contrast, L-cysteine reacts readily with cis-[PtCl(OH)(NH3)2] at pH 7, but there were many reaction products, including bridged species. Cis-[PtCl(OH)(NH3)2] reacts with reduced thiols in ultrafiltered plasma but these are oxidized if the plasma is not fresh or appropriately stored. With very low concentrations of the platinum complexes (35.5 microM), HPLC experiments (UV detection at 305 nm) indicate that the thiolate (probably cysteine) reactions become simpler as bridging becomes less important.

Population pharmacokinetics of intravenous caffeine in neonates with apnea of prematurity

To study the population pharmacokinetics of caffeine after intravenous administration to premature infants with apnea.

Caffeine citrate treatment for extremely premature infants with apnea: population pharmacokinetics, absolute bioavailability, and implications for therapeutic drug monitoring.

The objective of this study was to develop a population model of the pharmacokinetics (PK) of caffeine after orogastric or intravenous administration to extremely premature neonates with apnea of prematurity who were to undergo extubation from ventilation. Infants of gestational age <30 weeks were randomly allocated to receive maintenance caffeine citrate dosing of either 5 or 20 mg/kg/d. Four blood samples were drawn at prerandomized times from each infant during caffeine treatment. Serum caffeine was assayed by enzyme-multiplied immunoassay technique. Concentration data (431 samples, median: 4 per subject) were obtained from 110 (52 male) infants of mean birth weight of 1009 g, current mean weight (WT) of 992 g, mean gestational age of 27.6 weeks, and mean postnatal age (PNA) of 12 days. Of 1022 doses given, 145 were orogastric, permitting estimation of absolute bioavailability. A 1-compartment model with first-order absorption was fitted to the data in NONMEM. Patient characteristics were screened (P < 0.01) in nested models for pharmacokinetic influence. Model stability was assessed by nonparametric bootstrapping. Clearance (CL) increased nonlinearly with increasing PNA, whereas volume of distribution (Vd) increased linearly with WT, according to the following allometric models: CL (L/h) = 0.167 (WT/70)0.75 (PNA/12)0.358; Vd (L) = 58.7 (WT/70)0.75. The mean elimination half-life was 101. Interindividual variability (IIV) of CL and Vd was 18.8 % and 22.3 %, respectively. Interoccasion variability (IOV) of CL and Vd was 35.1% and 11.1%, respectively. This study established that the elimination of caffeine was severely depressed in extremely premature infants but increased nonlinearly after birth up to age 6 weeks. Caffeine was completely absorbed, which has favorable implications for switching between intravenous and orogastric routes. The interoccasion variability about CL was twice the interindividual variability, which, among other factors, indicates that routine serum concentration monitoring of caffeine in these patients is not warranted.

Population Pharmacokinetic Modeling in Very Premature Infants Receiving Midazolam during Mechanical Ventilation: Midazolam Neonatal Pharmacokinetics

BACKGROUND Midazolam is used widely as a sedative to facilitate mechanical ventilation. This prospective study investigated the population pharmacokinetics of midazolam in very premature infants. METHODS Midazolam (100 microg/kg) was administered as a rapid intravenous bolus dose every 4-6 h to 60 very premature neonates with a mean (range) gestational age of 27 weeks (24-31 weeks), a birth weight of 965 g (523-1,470 g), and an age of 4.5 days (2-15 days). A median (range) of four (one to four) blood samples, 0.2 ml each, were drawn at random times after the first dose or during continuous treatment, and concentrations of midazolam in serum were assayed by high-performance liquid chromatography. A population analysis was conducted using a two-compartment pharmacokinetic model using the NONMEM program. RESULTS Average parameter values (interpatient percent coefficient of variation) for infants with birth weights 1,000 g or less were total systemic clearance (Cl(T)) = 0.783 ml/min (83%), intercompartmental clearance (Cl(Q)) = 6.53 ml/min (116%), volume of distribution of the central compartment (V1) = 473 ml (70%), and volume of distribution of the peripheral compartment (V2) = 513 ml (146%). For infants with birth weights more than 1,000 g they were as follows: Cl(T) = 1.24 ml/min (78%), Cl(Q) = 9.82 ml/min (98%), V1 = 823 ml (43%), and V2 = 1,040 ml (193%). The intrapatient variability (percent coefficient of variation) in the data was 4.5% at the mean concentration midazolam in serum of 121 ng/mL. CONCLUSIONS Serum concentration-time data were used in modeling the population pharmacokinetics of midazolam in very premature, ventilated neonates. Clearance of midazolam was markedly decreased compared with previous data from term infants and older patients. Infants weighing less than 1,000 g at birth had significantly lower clearance than those weighing more than 1,000 g.

Haemodynamic responses and population pharmacokinetics of midazolam following administration to ventilated, preterm neonates

Objective: To evaluate the effects of intravenous midazolam on haemodynamic variables and cerebral blood flow velocity (CBFV) and to determine the pharmacokinetics using a population approach in very low birthweight (VLBW) ventilated infants.

Pharmacokinetics and Therapeutic Drug Monitoring of Gentamicin in the Elderly

Gentamicin is frequently used in elderly patients as serious infection, particularly Gram-negative bacilli, remains one of the major health problems experienced by this age group. A number of physiological changes in drug disposition occur with ageing and potentially these can affect gentamicin pharmacokinetics. In particular, there is a measurable decline in renal function, especially after the aged of 65. Any differences in drug distribution with age are apparently not reflected in gentamicin disposition data, as patients of varying ages have similar volumes of distribution.Seriously ill patients with infections frequently require treatment with many different drugs. Of note, the combination of gentamicin and a β-lactam antibacterial can result in inactivation. However, there appears to be no published data describing detrimental or beneficial pharmacokinetic interactions between gentamicin and drugs used in the elderly. Nonetheless, gentamicin should be used cautiously in order to prevent potential exacerbation of its nephrotoxicity and/or ototoxicity. Such problems may occur as a result of coadministration with, for example, amphotericin, cisplatin, vancomycin, foscarnet, nonsteroidal anti-inflammatory drugs or furosemide (frusemide). The presence of concurrent disease in aged patients (e.g. malignancy, fluid balance disorders and sepsis) may cause problems. In sepsis, for example, the volume of distribution of gentamicin may be increased; however, other pharmacokinetic data are contradictory and inconclusive.Like other aminoglycosides, gentamicin has a narrow therapeutic index and therapeutic drug monitoring has proven to be beneficial, particularly in vulnerable populations such as the elderly. Moreover, there is substantial pharmacokinetic variability in these patients. Recent data support the use of extended interval or once daily doses of gentamicin. It has been suggested that because of a lack of studies for this regimen in the elderly, specific recommendations cannot yet be made. We would argue that some recommendations for its cautious adoption in aged patients could be justified. Suggested procedures for the once daily administration of gentamicin include the use of the ‘Hartford’ nomogram and the targeted area under the concentration-time curve.The susceptibility of the elderly to aminoglycoside-related nephrotoxicity (and probably ototoxicity) may arise from a decline in renal function and an impaired capacity for cellular repair and regeneration. However, of greater importance is the duration of aminoglycoside therapy and the concomitant use of other nephrotoxic drugs.Further confirmation of the utility and tolerability of the once daily regimen and other possible approaches to gentamicin therapy in the elderly are essential.

Population pharmacokinetics of hydroxychloroquine in patients with rheumatoid arthritis

Hydroxychloroquine (HCQ) is an antimalarial drug that is also used as a second-line treatment of rheumatoid arthritis (RA). Clinically, the use of HCQ is characterized by a long delay in the onset of action, and withdrawal of treatment is often a result of inefficacy rather than from toxicity. The slow onset of action can be attributed to the pharmacokinetics (PK) of HCQ, and wide interpatient variability is evident. Tentative relationships between concentration and effect have been made, but to date, no population PK model has been developed for HCQ. This study aimed to develop a population PK model including an estimation of the oral bioavailability of HCQ. In addition, the effects of the coadministration of methotrexate on the PK of HCQ were examined. Hydroxychloroquine blood concentration data were combined from previous pharmacokinetic studies in patients with rheumatoid arthritis. A total of 123 patients were studied, giving the data cohort from four previously published studies. Two groups of patients were included: 74 received hydroxychloroquine (HCQ) alone, and 49 received HCQ and methotrexate (MTX). All data analyses were carried out using the NONMEM program. A one-compartment PK model was supported, rather than a three-compartment model as previously published, probably because of the clustering of concentrations taken at the end of a dosing interval. The population estimate of bioavailability of 0.75 (0.07), n = 9, was consistent with literature values. The parameter values from the final model were:EQUATION Clearance was not affected by the presence of MTX, and, hence, steady-state drug concentrations and maintenance dosage requirements were similar. A population PK model was successfully developed for HCQ.