Pharmacokinetics of Isoniazid in Low-Birth-Weight and Premature Infants

Abstract

Isoniazid (INH) is recommended for use as posttuberculosis exposure preventive therapy in children. However, no pharmacokinetic data are available for INH treatment in low-birth-weight (LBW) infants, who undergo substantial developmental and physiological changes. Our objectives in this study were to determine the pharmacokinetic parameters of INH at a dose of 10 mg/kg of body weight/day and to define its pharmacokinetics relative to the arylamine N-acetyltransferase-2 (NAT2) genotype. An intensive prospective pharmacokinetic sampling study was conducted at Tygerberg Children's Hospital, South Africa, in which we measured INH blood plasma concentrations at 2, 3, 4 and 5 h postdose. Twenty LBW infants (14 male, 16 exposed to HIV) were studied. The median birth weight was 1,575 g (interquartile range, 1,190 to 2,035 g) and the median gestational age was 35 weeks (interquartile range, 34 to 38 weeks). The NAT2 acetylation statuses of the infants were homozygous slow (SS) (5 infants), heterozygous intermediate (FS) (11 infants), and homozygous fast (FF) (4 infants). Using a noncompartmental analysis approach, the median maximum drug concentration in blood serum (C_max) was 5.63 μg/ml, the time after drug administration to reach C_maxT_max) was 2 h, the area under the concentration-time curve from 2 to 5 h (AUC_2–5) was 13.56 μg · h/ml, the half-life (t_1/2) was 4.69 h, and the elimination constant rate (k_el) was 0.15 h⁻¹. The alanine aminotransferase levels were normal, apart from 2 isolated values at two and three times above the normal levels. Only the three-times-elevated value was repeated at 6 months and normalized. All LBW infants achieved target INH blood plasma concentrations comparable to the adult values. Reduced elimination was observed in smaller and younger infants and in slow acetylators, cautioning against higher doses. The safety data, although limited, were reassuring. More data, however, are required for newborn infants.

INTRODUCTION

The HIV pandemic has been associated with a dramatic increase in tuberculosis (TB) rates among pregnant women (1). Maternal TB, regardless of HIV status, is associated with a high incidence of low birth weight (LBW) (defined as <2,500 g) and prematurity (a gestational age of <38 weeks) in infants (2–4). In South Africa, with its high burden of both TB and HIV, recent advances in neonatal care have improved infant survival, resulting in a considerable group of small and premature TB-exposed newborns requiring preventive therapy against TB. Isoniazid (INH) is the most widely recommended agent for preventing TB disease in children (5). Following Mycobacterium tuberculosis infection, up to 50% of infants (<12 months of age) will progress to have TB disease in the absence of INH preventive therapy (IPT) (6), while postexposure IPT reduces the risk of TB disease by 60 to 65% (7). There are limited pharmacokinetic data on the use of INH in young children and none in newborns or LBW infants requiring IPT due to maternal TB. INH is primarily metabolized through acetylation in the liver and intestines, and the acetylation rate is genetically determined (8). Developmental and physiological changes in the volume of distribution, maturity of liver enzymes, and the role of acetylation capacity may influence INH dosage requirements in this vulnerable population (9, 10). The aim of this study was to determine the pharmacokinetics of routinely administered INH at 10 mg/kg of body weight/day in TB-exposed LBW infants. The World Health Organization (WHO) recommends a dosage range of 10 to 15 mg/kg daily. We also describe the influence of the N-acetyltransferase 2 (NAT2) genotype on INH pharmacokinetics.

MATERIALS AND METHODS

Design and setting.

An intensive sampling pharmacokinetic study was conducted at Tygerberg Children's Hospital (TCH), Cape Town, South Africa, between May 2011 and May 2012. TCH is a tertiary referral hospital for 50,000 deliveries per annum. Here, the peak TB notification rate among young adults was >1,400 per 100,000 population in 2009 (11), and the provincial maternal HIV prevalence was 16.9% among public antenatal clinic attendees in 2009 (12). TCH manages an average of 6,000 high-risk complicated deliveries per year, 24% of which are LBW infants. The neonatal service has 136 neonatal beds and includes intensive- and high-care facilities.

Study procedures and drug administration.

LBW infants born to HIV-infected and HIV-uninfected women were consecutively recruited if routinely receiving daily INH either for 6 months of IPT (10 mg/kg/day) or as in TB treatment, as per the local guidelines (13). The treatment for TB disease consisted of a 2-month intensive phase comprising daily INH, rifampin (RMP), and pyrazinamide (PZA), with or without ethionamide (ETH), followed by a 4-month continuation phase of INH and RMP alone. Multidrug-resistant TB (MDR-TB) preventive therapy (consisting of daily INH, ETH, and ofloxacin) was administered in consultation with a pediatric infectious diseases specialist. The eligibility criteria included weight of ≥1.2 kg on the day of pharmacokinetic sampling, being clinically stable in room air and tolerating oral preparations, and written informed consent from the mother/legal guardian. The maternal HIV infection status was routinely determined by an enzyme-linked immunosorbent assay, preceded by informed consent and appropriate pre- and posttest counseling. All HIV-exposed newborns (those born to HIV-infected mothers) received nevirapine (NVP) for 6 weeks postpartum or until cessation of breastfeeding, according to national prevention of mother-to-child transmission (PMTCT) guidelines (14). Exclusive breastfeeding was encouraged.

Data on the following characteristics were collected for the infants: gender, ethnicity, birth weight, gestational age, weight and age at pharmacokinetic sampling, feeding type (breastfeeding versus formula), HIV exposure, and concomitant medications. The gestational age was determined by the date of the mother's last menstrual period, early ultrasound findings, and/or a Ballard score performed by a single experienced neonatologist. Data on alanine aminotransferase (ALT), a proxy for drug-related hepatotoxicity, were collected at baseline (at pharmacokinetic sampling) and at 3- and 6-month follow-up visits. The Division of Microbiology and Infectious diseases (DMID) tables were used to grade potential hepatotoxicity (15). The infants required regular feeding (every 2 to 3 h), as per routine care. INH in powder form, obtained from Fluka Chemie AG (Buchs, Switzerland), was used on the pharmacokinetic sampling days. The INH powder was accurately weighed to administer a dose of 10 mg/kg according to the weight of a newborn when naked (weighed by the study nurse on the day prior to study drug administration). The INH powder was dissolved in 1 to 2 ml of sterile water, administered through a nasogastric tube, and flushed with 1 ml of water.

Pharmacokinetic sampling and laboratory analysis.

On the day of pharmacokinetic assessment, the INH dose, time of administration, and data obtained by phlebotomy were documented precisely. Four arterial blood specimens of 0.5 ml each were taken at 2, 3, 4, and 5 h postdose and collected in EDTA-coated tubes. The specimens were kept on ice and delivered to the laboratory within 30 min of collection. After centrifugation, the separation plasma fragment was assayed using the high-performance liquid chromatographic (HPLC) method described previously (16). The remaining blood cells were used to extract DNA for NAT2 genotyping. Genomic DNA (gDNA) was prepared with a simple salting-out procedure for extracting DNA from human nucleated cells (17). The gDNA was analyzed for the NAT2*5, NAT2*6, NAT2*7, NAT2*12, NAT2*13, and NAT2*14 alleles via a PCR-based strategy (18). Separate PCR aliquots were restricted with the MspI, FokI, KpnI, TaqI, DdeI, and BamHI restriction enzymes to delineate the polymorphisms at nucleotide positions 191, 282, 481, 590, 803, and 857, respectively. According to Vatsis nomenclature, the wild-type fast allele (F) was assigned as NAT2*4, NAT2*12, or NAT2*13 (19). These alleles confer normal enzyme activity on the NAT2 protein, while the mutant slow alleles (S), classified as NAT2*5, NAT2*6, NAT2*7, and NAT2*14 in humans, confer decreased enzyme activity on the NAT2 protein. Accordingly, the study participants were classified as homozygous fast (FF), heterozygous intermediate (FS), or homozygous slow (SS) acetylators, depending on the allele combination observed.

Pharmacokinetic parameters and statistical analysis.

The following pharmacokinetic parameters were generated using fixed sampling times for each patient through noncompartmental analysis (NCA): C_max (the maximum drug concentration observed), T_max (the time after drug administration to reach C_max), the AUC_2–5 (area under the time-concentration curve from 2 to 5 h), t_1/2 (the plasma half-life), and the k_el (elimination constant). The AUC was calculated according to the linear trapezoidal rule. The pharmacokinetic parameters were summarized using medians and interquartile ranges (IQR), except for T_max, which was summarized using means and standard deviations (SD). The pharmacokinetic parameters for the infants were compared by NAT2 genotype, birth weight, weight at the time of pharmacokinetic sampling, gestational age, age at sampling, gender, HIV exposure status, and feeding type. All covariates were analyzed dichotomously, except for NAT2, which was analyzed categorically into three levels: FF (fast), FS (intermediate), and SS (slow) acetylator types. Since the sample sizes were small, the dichotomous covariates were analyzed using the Wilcoxon rank-sum test. NAT2 was analyzed using the Kruskal-Wallis test, and if statistically significant, the trend test was used to assess for trends across ordered NAT2 groups. Dichotomous confounders, weight at pharmacokinetic sampling, and feeding type were assessed using Fisher's exact test. Any P value of <0.05 was considered statistically significant. All data were analyzed using Stata 12.1 special edition software (StataCorp, College Station, TX, USA).

Study approval was obtained from the Health Research Ethics Committee at Stellenbosch University (approval no. N10/07/232).

RESULTS

Twenty infants, 16 (80%) of whom were HIV exposed, were enrolled; 17 received IPT and 3 received treatment for TB. Ten infants (50%) had a birth weight of <1,500 g, and 17 (85%) were premature (Table 1). On the day of pharmacokinetic sampling, the median weight was 1,874 g (interquartile range, 1,361 to 2,120 g). Eight (40%) infants received theophylline for apnea of prematurity, and 15 (75%) of the 16 HIV-exposed infants received NVP for PMTCT (one HIV-infected infant received abacavir, lopinavir/ritonavir, and lamivudine).

TABLE 1.

Clinical characteristics of low-birth-weight infants at sampling (n = 20)^a

Clinical characteristics	Values
Sex (male) (no. [%])	14 (70)
Ethnicity (black) (no. [%])	12 (60)b
HIV exposed (no. [%])	16 (80)
Exclusive breastfeeding (no. [%])	9 (45)
Formula feeding (no. [%])	11 (55)
Birth wt (median [IQR]) (g)	1,575 (1,190–2,035)
>1,500 and <2,500 (no. [%])	10 (50)
<1,500 (no. [%])	10 (50)
Gestational age (median [IQR]) (wk)	35 (34–38)
Term (≥38 wk) (no. [%])	3 (15)
Premature (<38 wk) (no. [%])	17 (85)
Wt on day of PK sampling (median [IQR]) (g)	1,874 (1,361–2,120)
Age at time of PK sampling (median [IQR]) (days)	14 (9–31)
Concomitant medications (no. [%])c
Nevirapined	15 (75)
Theophylline	8 (40)
Combination of anti-TB drugs (RMP, PZA, ETH, ofloxacin)	4 (20)e
cART (abacavir, lopinavir/ritonavir, lamivudine) and cotrimoxazole	1 (5)f
Hydrochlorothiazide and spironolactone	1 (5)g

^aPK, pharmacokinetic; RMP, rifampin; PZA, pyrazinamide; ETH, ethionamide; cART, combination antiretroviral therapy.

^bSeven infants of mixed race, one white infant.

^cSome infants were on a different combination of the listed concomitant medications.

^dTwo infants on nevirapine (NVP) and isoniazid (INH) had a raised alanine aminotransferase (ALT) level at 3 months.

^eTwo infants on RMP and PZA; 1 infant on RMP, PZA, and ETH; 1 infant on ofloxacin and ETH.

^fOne infant with HIV infection was treated with cART and cotrimoxazole on day 59 of life.

^gUsed to control mild cardiac failure in infant with a ventricular septal defect.

The pharmacokinetic measures are shown in Table 2. The C_max, k_el, and consequently the t_1/2 could not be accurately determined for one infant in whom INH plasma concentrations remained high (2-h value, 4.6 μg/ml; 5-h value, 4.9 μg/ml). The t_1/2 ranged from 1.45 to 14.25 h. Figure 1 illustrates the INH plasma concentrations at 2, 3, 4, and 5 h postdose (n = 20). Nineteen infants achieved plasma concentrations above the reference of 3 μg/ml at 2 h (20), with a range of 2.9 to 10.7 μg/ml, while all 20 infants achieved target drug plasma concentrations of >1.5 μg/ml at 3 h postdose (21).

TABLE 2.

Summary of pharmacokinetic parameters in low-birth-weight infants (n = 20)

Pharmacokinetic parametersa	Median (IQR)b
Cmax (μg/ml)	5.63 (4.86–7.53)
Tmax (h)	2 (2–2)
AUC2–5 (μg · h/ml)	13.56 (11.75–19.10)
kel (h−1)c	0.15 (0.10–0.23)
t1/2 (h)c	4.69 (3.08–7.60)

^aC_max, maximum drug concentration; T_max, time to C_max; AUC_2–5, area under the concentration-time curve; k_el, first-order elimination rate constant; t_1/2, half-life.

^bIQR, interquartile range.

^ck_el and t_1/2 not calculated for 1 patient due to continuous high plasma drug concentrations.

FIG 1.

Individual isoniazid drug concentrations in low-birth-weight infants (n = 20). Symbols show the progressions for individual infant subjects.

The pharmacokinetic parameters were compared by demographic and clinical covariates (Table 3). The distribution of acetylation status was as follows: 5 were slow, 11 were intermediate, and 4 were fast acetylators. Statistically significant differences in the C_max, AUC_2–5, t_1/2, and k_el were shown (P = 0.024, 0.020, 0.013, and 0.013, respectively) between slow, intermediate, and fast acetylators. Slow acetylators had a higher median C_max (6.5 μg/ml), larger AUC_2–5 (16.93 μg · h/ml), and longer t_1/2 (6.56 h) than those of intermediate and fast acetylators, in decreasing order. Figure 2 illustrates INH concentrations in relation to the NAT2 genotype.

TABLE 3.

Effect of clinical and other characteristics on isoniazid pharmacokinetic parameters in low-birth-weight infants (n = 20)

Subject characteristics	Cmax parametera			AUC2–5 parameterb			t1/2 parameterc
	No. of infants	Median (IQR) value (μg/ml)	P	No. of infants	Median (IQR) value (μg · h/ml)	P	No. of infants	Median (IQR) value (h)	P
NAT2 genotype
SS (slow)	5	6.54 (5.60–8.05)		5	16.93 (14.59–21.87)		5	6.56 (4.86–9.57)
FS (intermediate)	11	6.41 (5.03–7.66)		11	13.21 (12.03–21.26)		11	4.52 (3.48–6.27)
FF (fast)	4	3.90 (2.95–4.86)	0.0235	4	6.78 (4.96–11.31)	0.0199	3	1.78 (1.45–2.04)	0.0134
Birth wt
Very low (<1,500 g)	10	6.58 (5.60–8.05)		10	15.35 (13.21–21.62)		10	4.56 (3.74–6.27)
Low (≥1,500 to <2,500 g)	10	4.99 (4.11–6.41)	0.0284	10	12.61 (8.14–14.59)	0.0696	9	4.52 (1.95–6.56)	0.4142
Wt at PK time (g)
<1,750	9	7.66 (5.60–8.05)		9	21.26 (13.21–21.62)		9	5.93 (4.12–9.15)
≥1,750	11	5.03 (4.11–6.54)	0.0304	11	13.19 (8.14–14.59)	0.0441	10	3.60 (1.95–5.09)	0.0864
Gestational age
<38 wk	17	5.66 (5.03–7.66)		17	14.48 (13.19–21.26)		16	4.69 (3.61–6.42)
≥38 wk	3	4.11 (3.02–6.41)	0.1530	3	11.46 (5.41–12.03)	0.0502	3	2.04 (1.95–8.64)	0.4338
Corrected gestational age at PK time
<36 wk	11	5.66 (4.94–8.05)		11	14.48 (13.09–21.62)		10	4.39 (3.74–6.27)
≥36 wk	9	5.33 (4.11–6.54)	0.1837	9	13.19 (11.46–14.59)	0.1837	9	4.86 (2.04–6.56)	0.6242
Gender
Female	6	5.14 (3.02–8.05)		6	14.54 (5.41–21.62)		5	6.27 (2.04–6.56)
Male	14	6.04 (5.03–7.40)	0.4579	14	13.27 (12.03–16.93)	0.8690	14	4.39 (3.48–5.93)	0.7812
HIV status
Exposed	16	5.45 (4.45–6.97)		16	13.49 (10.19–16.91)		15	4.52 (2.04–6.27)
Negative	4	7.34 (6.14–9.39)	0.0472	4	17.60 (13.27–25.08)	0.1306	4	6.53 (3.08–11.91)	0.4237
Feeding type
Breastfed	9	7.66 (5.66–8.05)		9	21.26 (13.33–21.62)		9	5.93 (4.12–9.15)
Formula fed	11	4.94 (3.35–6.41)	0.0027	11	12.03 (8.14–14.59)	0.0135	10	4.13 (1.95–5.09)	0.1208

^aC_max, maximum drug concentration; IQR, interquartile range.

^bAUC_2–5, area under the concentration-time curve.

^ct_1/2, half-life. The t_1/2 not calculated for 1 patient due to continuous high drug plasma concentrations.

FIG 2.

Isoniazid concentrations in relation to NAT2 genotyping by acetylator type. Each line represents a different infant subject (n = 20).

Infants weighing <1,750 g on the day of pharmacokinetic sampling had a higher C_max and AUC_2–5 than those of heavier infants (7.66 μg/ml versus 5.03 μg/ml [P = 0.030] and 21.26 μg · h/ml versus 13.19 μg · h/ml [P = 0.044], respectively). The data in Fig. 3 suggest increased INH absorption and reduced clearance in smaller infants. Similar findings were observed in infants with younger gestational age; however, these differences were not statistically significant. Feeding also affected the C_max and AUC_2–5 values: exclusively breastfed infants had a higher C_max and AUC_2–5 than did formula-fed infants (7.66 μg/ml versus 4.94 μg/ml [P = 0.003] and 21.26 μg · h/ml versus 12.03 μg · h/ml [P = 0.014], respectively). Since weight at the time of pharmacokinetic sampling was associated with feeding status (P = 0.001), feeding and weight were likely confounded, with smaller babies more likely to receive breast milk. Gender did not influence the pharmacokinetic parameters.

FIG 3.

Isoniazid concentrations relative to current weight in low-birth-weight infants. Each line represents a different infant subject (n = 20).

ALT levels were determined in 19 (95%) infants at baseline, in 14 (70%) at 3 months, and in 11 (55%) at 6 months. Two 3-month values were abnormal; 1 was mildly elevated (<2.5 times elevated [DMID grade 1]) and one moderately elevated (<5 times elevated [DMID grade 2]). The mildly elevated value was not repeated, and the moderately elevated ALT value normalized at 6 months. Both raised ALT values occurred in HIV-exposed infants receiving IPT and NVP.

DISCUSSION

This is the first study describing the pharmacokinetics of INH and correlating with NAT2 genotypes in LBW and premature infants. The INH plasma concentrations in LBW infants, administered at a dose of 10 mg/kg, compared well to the published target adult values (20–22). An increased although variable half-life was observed in LBW infants, cautioning against higher-INH-dosing strategies. Markedly reduced clearance was present in smaller infants and in slow acetylators. The most important pathway for INH metabolism in humans is dependent on the trimodal NAT2 acetylation (18) already apparent in this young age group.

Optimal and safe INH dosing for TB prevention and treatment are especially relevant for LBW infants. We administered INH at the lower end of the WHO recommended dosage of 10 to 15 mg/kg (23). The desirable pharmacokinetic targets for children include either a 2-h blood plasma concentration of 3 to 5 μg/ml (20) or a 3-h value of >1.5 μg/ml (21), both correlating with good clinical response in adults. We found good absorption of INH in all infants, and the adult pharmacokinetic target values were achieved. All but one infant achieved a 2-h drug plasma concentration of ≥3 μg/ml; the 2-h value for that infant was 2.9 μg/ml. All infants achieved a 3-h value of ≥1.5 μg/ml. The effect of routine feeding every 2 to 3 h for LBW infants did not influence the C_max. However, food intake is known to decrease the bioavailability of INH (24), and we were unable to evaluate pharmacokinetics in the absence of feeding. Concomitant medicines frequently used in LBW infants included theophylline and NVP, neither of which should impact the pharmacokinetic parameters of INH (25, 26).

In our study, we observed an increased but variable half-life (1.45 to 14.25 h) using a dose of 10 mg/kg of INH, in line with an earlier study of two neonates in whom the half-lives were 7.8 and 19.8 h, respectively (27). This is not surprising, since most drugs in newborns have a prolonged elimination half-life in general. This finding, however, cautions against using a high dose of INH in small infants. Furthermore, a study describing the INH pharmacokinetics according to phenotype, performed in 34 children (two infants < 1 month of age), showed a definite decrease of half-life with increasing age, suggesting slower elimination of INH in young children (28). Although desirable INH pharmacokinetic targets are essential for efficacy in infants, caution should be applied when dosing at the higher range to prevent potential toxicity.

In this study, markedly reduced INH clearance was noted in the smaller and younger LBW infants. INH has a significant first-pass metabolism, with the rate of drug metabolism depending largely on the maturation of hepatic enzymes. Impaired elimination in the smaller and younger LBW infants is therefore probably due to immature hepatic enzymes. The development of drug-metabolizing enzymes varies widely between neonates and may be prolonged in premature infants (29). Pharmacokinetic studies show that the grade of maturation of enzymes is the most important factor in determining the rate of metabolism of a drug, with most liver enzymes maturing after the first year of life (30, 31). The effect of reduced INH elimination was even more pronounced in breastfeeding infants. However, this was confounded by the fact that smaller infants were more likely to receive breast milk. Antituberculosis drugs, including INH, are secreted in the breast milk of women on TB treatment, with levels ranging between 0.05 and 28% (32). The levels of antituberculosis drugs in breast milk are inadequate to prevent or treat infants but may increase the exposure to these medications. Therefore, Tran et al. recommended dosing at the lower end of the therapeutic range (i.e., 10 mg/kg/day of INH) in order to decrease the risk of potential toxicity (32).

Age and acetylator status influence INH pharmacokinetics in children (28, 33). INH is acetylated to acetylisoniazid by a hepatic and intestinal enzyme, N-acetyltransferase 2, which is coded for genetically. In our study, trimodal clearance of INH as a function of the NAT2 genotype was already apparent, even in this young age group. Our results showed that all compared parameters were significantly different for all three acetylator groups, with slow acetylators having decreased elimination compared to intermediate and fast acetylators. Recent data from an IPT trial conducted in children 3 to 24 months of age illustrated not only the difference for each genotype group but also immature NAT2 activity, particularly in fast acetylators, with the acquisition of activity to adult values occurring over the first 2 years of life (34). This is in keeping with the maturation of genetically determined NAT2 activity over time. The impact of enzyme maturity on the INH dosing for LBW infants requires further study.

The transient elevation of blood serum transaminases occurs commonly with INH, but clinically manifested hepatotoxicity is rare. No formal relationship has been demonstrated between INH plasma concentrations and hepatotoxicity. Previous observations indicate that hepatotoxicity may be dose related (35, 36). In our study, only two ALT results were slightly raised at month 3, with the thrice-elevated value returning to normal at month 6. The limited safety data collected were reassuring, and no jaundice was observed in any infant. A limitation of the study was the few pharmacokinetic sampling points, mainly because of the small blood volume available in LBW infants. An earlier time point may have assisted with determining the C_max more accurately, as previous pharmacokinetic studies in children indicate that the T_max occurs any time between 1 and 2 h (28, 37).

In conclusion, a sound understanding of the pharmacokinetic properties of currently used antituberculosis drugs is essential for optimal use in newborns and infants. LBW infants receiving 10 mg/kg of INH had desirable blood drug concentrations, which were comparable to the adult target values. However, a prolonged half-life and reduced elimination of INH were noted in smaller and younger infants, especially in those genetically determined to be slow acetylators. Therefore, we caution against exceeding a dosage of 10 mg/kg in this population. Although no serious adverse effects were observed, more data on safety are needed. Also, more research is needed on the appropriate dosing requirements for TB drugs in newborns and infants.