Rifampin Pharmacokinetics and Safety in Preterm and Term Infants

Rifampin is active against methicillin-resistant staphylococcal species and tuberculosis (TB). We performed a multicenter, prospective pharmacokinetic (PK) and safety study of intravenous rifampin in infants of <121 days postnatal age (PNA).

ABSTRACT

Rifampin is active against methicillin-resistant staphylococcal species and tuberculosis (TB). We performed a multicenter, prospective pharmacokinetic (PK) and safety study of intravenous rifampin in infants of <121 days postnatal age (PNA). We enrolled 27 infants; the median (range) gestational age was 26 weeks (23 to 41 weeks), and the median PNA was 10 days (0 to 84 days). We collected 102 plasma PK samples from 22 of the infants and analyzed safety data from all 27 infants. We analyzed the data using a population PK approach. Rifampin PK was best characterized by a one-compartment model; drug clearance increased with increasing size (body weight) and maturation (PNA). There were no adverse events related to rifampin. Simulated weight and PNA-based intravenous dosing regimens administered once daily (<14 days PNA, 8 mg/kg; ≥14 days PNA, 15 mg/kg) in infants resulted in comparable exposures to adults receiving therapeutic doses of rifampin against staphylococcal infections and TB. (This study has been registered at ClinicalTrials.gov under identifier NCT01728363.)

TEXT

The majority of late-onset sepsis in the neonatal intensive care unit is due to Gram-positive organisms, including coagulase-negative Staphylococcus spp. (1, 2) and Staphylococcus aureus (2–8). S. aureus is associated with overwhelming septic shock, severe necrotizing pneumonia (2, 9–12), and a high risk of mortality (up to 40%) (13). The majority (95%) of coagulase-negative Staphylococcus isolates and 40% of S. aureus isolates are methicillin resistant (MRSA) (10, 11). Infants with these infections have prolonged hospitalizations and an increased risk of neurodevelopmental impairment (2, 14–16).

Rifampin is a semisynthetic derivative of rifamycin SV with a wide spectrum of antibacterial activity that includes methicillin-resistant staphylococcal species. Rifampin inhibits bacterial RNA polymerase but does not inhibit the mammalian enzyme (15). Rifampin is not an option for monotherapy given the high likelihood for the development of resistance, but it is often added to facilitate bacterial eradication among infants with persistent staphylococcal bacteremia (17).

Rifampin is approved by the U.S. Food and Drug Administration (FDA) for the treatment of tuberculosis (TB) and for the treatment of asymptomatic carriers of Neisseria meningitidis to eliminate meningococci from the nasopharynx (15). Rifampin exhibits in vitro activity against S. aureus (including MRSA) and Staphylococcus epidermidis (coagulase-negative Staphylococcus); however, the safety and effectiveness of rifampin in treating clinical infections due to these microorganisms have not been established in adequate and well-controlled trials.

Limited data suggest that there are important differences in drug disposition in the pediatric population, particularly infants, relative to adults (17–22). This is likely due to the ontogeny of CYP3A4, which appears in the first week of life and reaches adult levels by early childhood (1 to 10 years) (22). The primary metabolic pathway of rifampin is deacetylation in the liver, mainly via CYP3A4; however, CYP2C isoenzymes (CYP2C9, CYP2C8, and CYP2C18/19) also play a role (23–25). Rifampin undergoes enterohepatic circulation and is excreted in the bile and urine (23). Rifampin induces its own metabolism—the turnover of the inducible process has been estimated to have a half-life of approximately 6 to 8 days in adults, with full induction evident after 30 to 40 days (26).

The median (range) of area under the concentration curve from 0 to ∞ (AUC_∞) for adults treated for TB with rifampin doses within the label-recommended range (8.6 to 13 mg/kg, maximum 600 mg) was 72.6 μg ⋅ h/ml (range, 7.7 to 218.5) (23, 27, 28). The median (range) for C_max was 9.9 μg/ml (range, 0.7 to 19.0) (23, 27, 28). There are no established surrogate therapeutic targets for rifampin against Staphylococcus species. Data from phase I trials in infants suggest that peak rifampin concentrations of 2 to 5 μg/ml seem to be effective in the treatment of persistent staphylococcal bacteremia (29).

There are a number of safety concerns previously reported with use of rifampin, such as bone marrow suppression, including thrombocytopenia (30). The rifampin FDA label also recommends that clinicians monitor liver function testing due to associations of rifampin use with elevated transaminases and bilirubin (31). Rifampin induces CYP-450 activity and accelerates the metabolism of several drugs frequently used in infants, including phenytoin, zidovudine, fluconazole, and methadone (31). The purpose of this study was to characterize the pharmacokinetics (PK) and safety of rifampin administered to preterm and term infants with suspected systemic infection.

RESULTS

Infant characteristics.

We enrolled 27 infants, and all received at least one dose of rifampin. The median (range) gestational age (GA) and postnatal age at enrollment (PNA) were 26 weeks (23 to 41 weeks) and 10 days (0 to 84 days), respectively (Table 1). We collected 102 PK samples, 39 of which were scavenged samples. Eighty-six samples (53 fresh, 33 scavenged) from 22 infants were used for the PK analysis; the reasons for exclusion were that (i) 8 samples had insufficient volume and (ii) 8 (<10%) were below the quantitation limit. In order to ensure that the scavenged samples did not significantly affect the PK results, the final model was also evaluated without the scavenged samples, and the PK parameter estimates did not significantly change. The weight-adjusted median (range) dose was 10.3 mg/kg (4.5 to 20.8 mg/kg), median (range) number of samples per subject was 4 (1 to 7), and the average (range) rifampin concentration was 6.59 μg/ml (0.05 to 38.51 μg/ml).

TABLE 1.

Characteristics of infants with plasma PK samples^a

Parameter	GA < 32 weeks		GA ≥ 32 weeks		Total days (n = 22)
	PNA < 14 days (n = 11)	PNA > 14 days (n = 6)	PNA < 14 days (n = 4)	PNA > 14 days (n = 1)
GA (wk)	27 (24–30)	25.5 (23–28)	38.5 (37–41)	37 (37–37)	27.5 (23–41)
PNA (days)	2 (0–12)	33.5 (21–84)	5.5 (2–12)	55	7.5 (0–56)
Postmenstrual age (wk)	27.6 (24.3–31.3)	29.9 (27.1–33.3)	39.93 (37.3–41.3)	44.9	29.9 (24.3–44.9)
WT (kg)	1.1 (0.65–1.7)	1.105 (0.81–1.68)	3.585 (3.39–3.7)	6	1.18 (0.65–6.0)
Serum creatinine (mg/dl)	0.87 (0.59–1.7)	0.51 (0.3–0.8)	0.55 (0.38–0.9)	0.3	0.7 (0.30–1.7)
AST (U/liter)	24 (11–43)	45 (22–323)	NS	NS	32 (11–323)
ALT (U/liter)	5.5 (4–7)	10.5 (8–134)	NS	NS	7.5 (4–134)
Total bilirubin (mg/dl)	4.9 (0.4–6.6)	5.55 (0.4–24.9)	12.2 (9.5–14.9)	NS	5.1 (0.4–24.9)
Albumin (g/dl)	2.35 (1.9–3)	2.35 (2–2.8)	2.6	3.7	2.45 (1.9–3.7)
Male	7 (63.6)	3 (50)	4 (100)	1 (100)	15 (68.2)
Race, white	6 (54.6)	3 (50)	2 (50)	1 (100)	12 (54.6)
Ethnicity, non-Hispanic	11 (100)	4 (66.7)	4 (100)	1 (100)	20 (90.9)

^aValues are medians (ranges) for continuous variables and no. (%) for categorical variables calculated based on values at the time of the first recorded dose. Laboratory values are baseline values. For categorical variables, the percentages were calculated as a function of the number of subjects in each category. GA, gestational age; PNA, postnatal age; WT, total body weight; AST, aspartate aminotransferase; ALT, alanine aminotransferase; NS, no subjects contributed data.

Population PK model development and evaluation.

After accounting for body size using weight (WT), inclusion of PNA, serum creatinine (SCR), and administration of a drug that is either an inhibitor, inducer or substrate of any CYP-P450 enzyme (CLINTE) reduced the objective function value (OFV) by 28.1, 13.5, and 8.0 points, respectively; however, during multivariable analysis, only PNA was retained in the final model. Inclusion of SCR to PNA reduced the OFV by 1.7, and inclusion of CLINTE to PNA increased the OFV by 6.6 points. No other covariates reached statistical significance. For the final model, the typical values for clearance (CL) and volume of distribution (V) can be expressed according to the following equations: CL (liters/h) = 3.6 ⋅ (WT/70)^0.75 ⋅ (PNA/7)^0.343 and V (liters) = 77.6 ⋅ (WT/70) (see Table S1 in the supplemental material).

Eta shrinkage estimates for CL and V were 13 and 29%, respectively, while epsilon shrinkage was 13% for the final model. Goodness-of-fit plots and a visual predictive check for all data included in the analysis are shown in Fig. S1 to S3 in the supplemental material. The model was evaluated using a 1,000-set bootstrap analysis: 98.9% of bootstrap data sets converged to three or more significant digits. The median of bootstrap fixed effects parameter estimates was within 1.0% of population estimates from the original data set for all parameters. The standardized visual predictive check revealed a reasonable fit between the observed and predicted rifampin concentrations. A uniform distribution of calculated observation percentiles was observed at earlier time points where the data points were relatively abundant (see Fig. S3 in the supplemental material). Overall, 5.8% (5/86) of observed concentrations were outside the 90% prediction interval. Empirical Bayesian estimates stratified by age are shown in Table 2. There was a relationship between PNA (days) and weight-normalized CL (CL values increased with PNA, Fig. 1). The parameter estimates between base and final models were very similar, but there was a significant change in interindividual variability (IIV) on CL between the base and final models. PNA explained an additional 50% of the IIV in CL (IIV drop from 54 to 26%) and the relationship observed in the Eta for CL versus PNA scatter plot disappeared (Fig. 2). To make sure the scavenged samples did not significantly affect the PK results, the final model was also evaluated without the scavenged samples, and the PK parameter estimates did not significantly change. Individual infant post hoc estimates normalized by WT showed that subjects with a PNA of ≥14 days had a median CL greater than subjects with subjects with a PNA of <14 days (Table 2).

TABLE 2.

Individual empirical Bayesian post hoc parameter estimates stratified by age^a

Parameter	Median (range)
	PNA < 14 days (n = 15)	PNA ≥ 14 days (n = 7)
CL (liters/h)	0.111 (0.061–0.549)	0.311 (0.12–0.773)
CL (liters/h/kg)	0.098 (0.055–0.223)	0.244 (0.129–0.328)
CL (liters/h/70 kg)	2.6 (1.4–5.4)	6.7 (3.4–8.2)
V (liters)	1.5 (0.712–4.3)	1.2 (1–4.1)
V (liters/kg)	1.1 (0.839–1.9)	1.1 (0.676–1.4)
V (liters/kg/70 kg)	77.8 (58.7–131.7)	80 (47.3–97.6)
Half-life (h)	7.1 (3–23.9)	3.5 (1.9–6.5)

^aPNA, postnatal age; CL, clearance; V, volume of distribution.

FIG 1.

Postnatal age versus clearance (CL). The black line and shaded area denote the Loess curve and an associated 95% confidence region, respectively.

FIG 2.

Comparing random effects on clearance (ETA_CL) versus postnatal age of the base (A) and final models (B). The black line and shaded area denote the Loess curve and an associated 95% confidence region, respectively.

Dosing simulations.

The 1,000 virtual patients generated from PK simulations matched the infant demographic data from this study (see Table S2 in the supplemental material). Doses of 8 mg/kg once daily resulted in an AUC_∞ at a steady state of ≥55.2 mg ⋅ h/liter in 92.3% of simulated infants <14 days PNA with a mean (standard deviation) C_max-ss of 8.3 ± 1.8 μg/ml. Doses of 15 mg/kg once daily resulted in an AUC_∞ at steady state of ≥55.2 mg ⋅ h/liter in 91.5% of infants ≥14 days PNA with a mean C_max-ss of 13.2 ± 3.3 μg/ml (Table 3). Overall, ≥90% of virtual infants with the proposed dosing achieved rifampin exposures comparable to adults treated for staphylococcal infections.

TABLE 3.

Simulated exposures by PNA group using different dosing regimens

PNA and parametera	Dose, duration
	8 mg/kg, 24 h	8 mg/kg, 12 h	10 mg/kg, 24 h	12 mg/kg, 24 h	15 mg/kg, 24 h	20 mg/kg, 24 h
PNA < 14 days
Cmax-fd (μg/ml)	7.0 ± 1.9	11.5 ± 2.5	8.8 ± 2.4	10.6 ± 2.8	13.4 ± 3.7	17.6 ± 4.7
Cmax-ss (μg/ml)	8.3 ± 1.8	16.2 ± 4.4	10.4 ± 2.3	12.5 ± 2.7	15.6 ± 3.4	20.7 ± 4.5
AUC∞ (μg ⋅ h/ml)	84.8 (30.2–193.0)	172.7 (70.0–405.0)	106.0 (37.8–241.3)	127.1 (45.3–289.5)	155.9 (49.5–361.4)	212.0 (75.5–482.5)
TA (%)	92.3	100	97.8	99.1	99.7	100
PNA ≥ 14 days
Cmax-fd (μg/ml)	6.7 ± 1.7	7.4 ± 1.5	8.4 ± 2.1	10.1 ± 2.5	12.9 ± 3.4	16.8 ± 4.1
Cmax-ss (μg/ml)	6.9 ± 1.6	8.2 ± 1.7	8.6 ± 2.0	10.3 ± 2.4	13.2 ± 3.3	17.1 ± 3.9
AUC∞ (μg ⋅ h/ml)	39.2 (15.1–96.7)	68.1 (30.0–141.8)	49.0 (18.9–120.8)	58.8 (22.7–145.0)	72.6 (28.1–165.8)	97.9 (37.8–241.6)
TA (%)	8.9	76.2	30.7	56.7	91.5	97.8

^aC_max-fd and C_max-ss, mean (standard deviation); AUC_∞, median (minimum to maximum); TA, % target attainment rate = (no. of subjects with AUC_∞ > 55.2 μg ⋅ h/ml/total number of subjects in the group) × 100.

Safety.

Ten (37%) infants experienced a total of 26 adverse events, none of which were attributed to the study drug. The 26 events included (n [%]): hyperbilirubinemia (3 [11%]), renal failure (2 [7%]), anemia (1 [3%]), tachycardia (1 [3%]), ventricular extrasytole (1 [3%]), patent ductus arteriosus (1 [3%]), constipation (1 [3%]), reflux (1 [3%]), pneumonia (1 [3%]), gastric residual (1 [3%]), decreased hematocrit (1 [3%]), hyperglycemia (1 [3%]), hypoglycemia (1 [3%]), hyperlipidemia (1 [3%]), hyponatremia (1 [3%]), metabolic acidosis (1 [3%]), agitation (1 [3%]), hypoxia (1 [3%]), respiratory distress syndrome (1 [3%]), pulmonary interstitial emphysema (1 [3%]), pulmonary edema (1 [3%]), tachypnea (1 [3%]), and skin lesion (1 [3%]). No adverse events were determined to be serious. There were no deaths during the study period. Median (range) values for baseline serum aspartate aminotransferase (AST), alanine aminotransferase (ALT), and total bilirubin were 30 U/liter (11 to 323 U/liter), 8 U/liter (4 to 134 U/liter), and 3.6 mg/dl (0.4 to 10.1 mg/dl), respectively. Values for serum AST, ALT, and total bilirubin at the end of dosing were 17 U/liter (12 to 39 U/liter), 12 U/liter (4 to 53 U/liter), and 4.8 mg/dl (0.6 to 9.5 mg/dl), respectively.

DISCUSSION

We conducted a dedicated PK and safety study of rifampin in infants under the FDA regulatory guidance (Investigational New Drug 130531). We included preterm and term infants at various PNAs. We used a variety of innovative pediatric study design strategies to optimize the study, including population PK analyses to analyze sparse samples, scavenged serum samples from the clinical lab, and “add-on” drug therapy. Add-on, or empirical, drug therapy allowed sites and clinicians to use rifampin when they suspected a Gram-positive infection. Using these strategies, we successfully recruited 27 infants over 22 months. Using simulations, we found that a rifampin dose of 8 or 15 mg/kg once daily for infants with a PNA of <14 days or 14 to <61 days, respectively, achieved exposures shown to be therapeutic in adults with S. aureus infections in ≥90% of infants. These results are particularly important in the era of increasing resistance to S. aureus, because the addition of appropriately dosed rifampin to staphylococcal bacteremia results in rapid sterilization of the bloodstream (21).

Our results are generally consistent with the three previous studies of rifampin that included at least some infants, with some important differences. In 9 pediatric patients aged 1 day to 18 years (mean age, 5.6 years) who received a single dose of intravenous (i.v.) rifampin (20 mg/kg), the V, CL, and half-life of rifampin were 1.1 liters/kg, 0.29 liters/kg/h, and 2.8 h, respectively (21). In 12 pediatric patients aged 3 months to 12 years (mean age, 4.6 years) dosed (280 ± 78 mg/m²), the V, CL, and half-life of rifampin were 0.63 liters/kg, 0.16 liters/kg/h, and 1.94 h, respectively (21). A PK study of rifampin (5 to 10 mg/kg/day i.v.) in 21 infants (median GA, 29 weeks [range, 26 to 41 weeks]; median PNA, 18 days [range, 11 to 55 days]) found that the mean rifampin half-life was 4.9 ± 1.7 h, and drug CL increased with increasing PNA (19). However, there was wide interpatient variability in drug concentrations, even though bacteria was cleared in all infants. Approximately 80% of rifampin is transported in blood bound to plasma proteins, mainly albumin; there is no evidence of variable binding, but this could contribute to large interindividual variability of the PK data (32). This study was limited by lack of inclusion of older infants who may require higher rifampin doses. Our PK parameters were most similar to this study (19), and in addition to PNA, we found that WT explained the majority of the IIV in CL. This is expected because body size and maturation (i.e., PNA) are the primary physiologic processes affecting CL in infants. Figure 1 shows the overall postnatal age effect for maturation of rifampin plasma clearance, especially within the first 20 days. More observations may be needed to show whether this trend continues between 20 and 60 days (the data are sparse, with high variability).

For many drugs, there are important PK differences between children and adults and between infants and children. Rifampin is likely one of those drugs (33, 34). Compared to older children or adults, infants have a larger V, likely because of the large amount of extracellular water compared to older children and adults (20). In healthy adults and following i.v. administration, rifampin distributes extensively into tissues and has a volume of distribution (V) at steady state of 1.6 liters/kg (18). Our parameter estimates for CL (3.6 liters/h, 70 kg) were much lower than values reported previously in adults (10 to 19.3 liters/h) (6). For rifampin, the differences are likely due to the ontogeny of metabolizing enzymes such as CYP3A4, which is expressed in the first week of life and reaches adult levels by early childhood (1 to 10 years) (34). Differences in extracellular fluid volume, plasma protein levels, protein affinity to antimicrobials, rate, and mechanism of elimination also may account for the PK discrepancies observed between infants and older populations (20). Of note, the presence of an interfering drug (CLINTE) was associated with lower objection function. While the difference was not statistically significant, the finding suggests that interfering drugs influence clearance.

There are no specific PK/pharmacodynamic targets against TB established for rifampin. There is also no safety margin associated with rifampin exposure, and therefore the simulated AUCs in this study were kept within the range of AUCs observed in adults without adverse events. According to the FDA-approved label, a single 600­mg oral dose of rifampin in healthy adults results in a mean peak serum level of 7 μg/ml (range, 4 to 32 μg/ml) (35). In children 0.25 to 12.8 years of age (n = 12), the mean peak serum concentration of rifampin at the end of a 30­min infusion of approximately 300 mg/m² was 25.9 ± 1.3 μg/ml (35). The C_max-fd and C_max-ss values predicted with the proposed dosing in infants are within the range of these concentrations.

There are several strengths and weaknesses of our study. In general, the FDA allows extrapolation of efficacy if the disease process is similar in infants and adults. However, as mentioned above, there are no specific PK/pharmacodynamic targets against S. aureus or safety thresholds established for rifampin in adults. Our simulated AUCs are within the range of AUCs observed in adults without adverse events. We also found a lack of clinically (and statistically) significant differences in weight-scaled estimates between premature infants (GA <32 weeks) and full-term infants (≥32 weeks). This may be due to the small number of full-term infants included in the study.

Conclusion.

Rifampin PK were well characterized by a one-compartment PK model with WT and PNA as significant covariates for CL. A rifampin dose of 8 or 15 mg/kg once daily for infants with PNA <14 days or 14 to <61 days, respectively, achieved therapeutic exposures against S. aureus infection in ≥90% of infants. This is also therapeutic for TB infections. These data will help optimize the dose of rifampin in vulnerable infants.

MATERIALS AND METHODS

Patient population.

PK and safety data were collected as part of the Pharmacokinetics of Anti-Staphylococcal Antibiotics in Infants clinical trial (Staph Trio; NICHD-2012-STA01, ClinicalTrials.gov NCT01728363; IND 115,396). This was a multicenter (n = 8 centers enrolling for rifampin), prospective, multiple-dose PK and safety study. Infants <120 days of age with a suspected systemic infection or receiving rifampin per local standard of care were eligible. Infants meeting any of the following criteria were excluded: history of allergic reactions to rifampin, urine output <0.5 ml/h/kg over the prior 24 h, SCR ≥1.7 mg/dl, and any condition that in the judgment of the investigator precluded participation because of safety concerns. The study was approved by each local institutional review board. Informed consent was obtained from the legal guardian for each infant. The first and last infants were enrolled on 29 June 2013 and 27 April 2015, respectively.

Drug dosing and sample collection.

Infants received i.v. rifampin based on their GA and PNA, unless they were prescribed rifampin per standard of care, in which case dosing was at the discretion of the caregiver. The protocol dosing was 10 mg/kg every 24 h for four doses if <32 weeks GA/<14 days PNA, 15 mg/kg every 24 h for 4 doses if <32 weeks GA/14 to 120 days PNA, 15 mg/kg every 24 h for 4 doses if ≥32 weeks GA/<14 days PNA, and 20 mg/kg every 24 h for 4 doses if ≥32 weeks GA/14 to 120 days PNA. Rifampin was administered over 30 min. Up to 8 timed plasma PK samples and up to 10 scavenged PK samples (leftover plasma from the clinical lab) per infant were planned for collection after single and multiple doses of rifampin. Predefined PK sample time collection windows were outlined in the protocol (see Table S3 in the supplemental material); however, samples collected outside these windows were included in the analysis.

Analytical methods.

Whole blood was collected (200 μl) in an EDTA-K2 microtainer and processed immediately prior to freezing at the study sites. Plasma samples were stored at or below −70°C within 8 h of collection. PK samples were sent to the Pediatric Trials Network central laboratory (OpAns, LLC, Durham, NC) for storage and analysis. Rifampin concentrations were quantified using a validated liquid chromatography-tandem spectrometry assay. The chromatography system and mass spectrometer used for sample analysis were the Agilent 1200 series high-performance liquid chromatography system and an Agilent 6400 series triple quadrupole system, respectively. The Pursuit XRS Ultra C₁₈ column (50 by 2 mm [i.d.], 2.8 μm; Agilent) and a gradient mobile phase (water containing 0.5% [vol/vol] formic acid; methanol containing 0.1% [vol/vol] formic acid) were used. The validation range for the assay was 0.05 to 50 μg/ml. Quality control samples included the following nominal concentrations: 0.05, 0.15, 4, and 40 μg/ml. The lower limit of quantification was 0.05 μg/ml. Accuracy and precision were assessed using five determinations at theoretical levels 0.15, 4, and 40 μg/ml, using FDA bioanalytical assay validation criteria (e.g., ±15%). All samples run for the study met acceptance criteria.

Population PK model development.

Rifampin plasma PK data were collected following administration and were analyzed with a nonlinear mixed effects modeling approach using NONMEM software (version 7.2; Icon Solutions, Ellicott City, MD). The first-order conditional estimation method with interaction was used for all model runs. Run management was performed using Pirana (version 2.8.1) (36). Visual predictive checks and bootstrap methods were performed with Perl-speaks-NONMEM (PsN; version 3.6.2) (37). Data manipulation and visualization was performed using software R (version 3.0.2; R Foundation for Statistical Computing, Vienna, Austria) and RStudio (version 0.97.551; RStudio, Boston, MA); with the packages lattice, Xpose, and ggplot2 used for the latter (38–40). STATA software (release 14; StataCorp LP, College Station, TX) was used for descriptive statistics of the patient demographic and lab data.

Based on visual inspection of the PK data and a review of the primary literature, one- and two-compartment PK models were evaluated using the ADVAN1 TRANS2 and ADVAN3 TRANS4 subroutines, respectively, in NONMEM. Between-subject variability was assessed for PK model parameters using an exponential relationship (equation 1):

$$P_{\mathit{ij}}={\theta }_{\mathit{Pop},j}\times \text{exp}({\eta }_{\mathit{ij}})$$ (1)

where P_ij denotes the estimate of parameter j in the ith individual, θ_Pop,j is the population value for parameter j, and η_ij denotes the deviation from the average population value for parameter j in the ith individual. The random variable η is assumed to be normally distributed with a mean zero and variance ω². Proportional, additive, and combined (proportional plus additive) residual error models were evaluated (equations 2 to 4):

$$C_{\mathit{obs},\mathit{ij}}=C_{\mathit{pred},\mathit{ij}}\times (1+{\epsilon }_{\mathit{prop},\mathit{ij}})$$ (2)

$$C_{\mathit{obs},\mathit{ij}}=C_{\mathit{pred},\mathit{ij}}+{\epsilon }_{\mathit{add},\mathit{ij}}$$ (3)

$$C_{\mathit{obs},\mathit{ij}}=C_{\mathit{pred},\mathit{ij}}\times (1+{\epsilon }_{\mathit{prop},\mathit{ij}})+{\epsilon }_{\mathit{add},\mathit{ij}}$$ (4)

where C_obs,ij is the jth-observed rifampin concentration in the ith individual, C_pred,ij is the jth-predicted concentration in the ith individual, and ε_prop,ij and ε_add,ij are random variables with mean zero and variance σ_prop,ij² and σ_add,ij², respectively.

WT was assumed to be a significant covariate for CL and V, and was included in the base model. The relationship between WT and PK parameters was characterized using a fixed exponent (0.75 and 1) allometric and linear relationship for CL and V parameters (scaled to a 70-kg standardized WT), respectively. The following covariates were then also explored: SCR, total bilirubin levels, PNA in days, GA in weeks, postmenstrual age (defined as the sum of the GA [weeks] plus PNA in weeks [days/7]), administration of a drug that is either an inhibitor, inducer or substrate of any CYP-P450 enzyme (CLINTE: cefepime, cefotaxime, clindamycin, clonidine, dexamethasone, fluconazole, fentanyl, gentamicin, hydrocortisone, midazolam, morphine, propranolol, ranitidine, vancomycin, and piperacillin), ethnicity, and sex. With the exception of WT, other continuous covariates were normalized to the population median value as described in equations 5 or 6 for all continuous variables, and only those that minimized successfully were reported, whereas for categorical covariates such as gender, a relationship as shown in equation 7 was used:

$$P_{\mathit{ij}}={\theta }_{\mathit{Pop},j}\times {\left(\frac{{\text{cov}}_i}{{\text{cov}}_m}\right)}^{{\theta }_{\text{cov}}}$$ (5)

$$P_{\mathit{ij}}={\theta }_{\mathit{Pop},j}\times e^{[{\theta }_{\text{cov}}\times \left(\frac{{\text{cov}}_i}{{\text{cov}}_m}\right)]}$$ (6)

$$P_{\mathit{ij}}={\theta }_{\mathit{Pop},j}\times \mathit{CATVAR}$$ (7)

where cov_i denotes the individual covariate value, cov_m is the population median covariate value, θ_cov is a parameter that represents the covariate effect, and CATVAR is a categorical variable.

A forward inclusion (P < 0.05 and change in objective function value of <3.8) and backward elimination (P < 0.001 and a change in objective function value of <10.8) approach was used to assess the statistical significance of relevant covariates. Missing covariate values were imputed using the median value for the sample.

Population PK model evaluation.

During the population PK model building process, successful minimization, diagnostic plots, the plausibility and precision of parameter estimates, and OFV and shrinkage values were used to assess model appropriateness. Parameter precision for the final population PK model was evaluated using nonparametric bootstrapping (1,000 replicates) to generate the 95% confidence intervals for parameter estimates. Visual predictive checks were performed whereby the base and final models were used to generate 1,000 Monte Carlo simulation replicates per time point of rifampin exposure, and simulated results were compared to those observed in the study. The dosing and covariate values used to generate the simulations in the visual predictive check were the same as those used in the study population.

Exposure-dose evaluation.

Because there are no established surrogate therapeutic targets for rifampin against Staphylococcus species, therapeutic target attainment rates in infants were obtained by matching rifampin exposures to therapeutic exposures against S. aureus observed in adults. PK-Sim (version 5.3.2; Bayer Technology Services GmbH, Leverkusen, Germany) software was used to generate a virtual population of 1,000 patients that matched the distribution of this population based on PNA, GA, and WT. Multiple dosing regimens were evaluated using Monte Carlo simulations based on PK parameters and associated variability generated from the final population PK model and compared to therapeutic exposures for MRSA in adults summarized from several publications (AUC_∞ = 55.2 μg ⋅ h/ml) (29, 32–34, 41–45). The AUC was calculated using equation 8:

$${\text{AUC}}_{{\infty }_i}=\frac{{\text{Dose}}_i}{{\text{CL}}_i}$$ (8)

Intermittent infusion equations used to calculate the maximum concentration of the end of infusion after the first dose C_max-fd (equation 9) and steady state C_max-ss (equation 10) were as follows:

$$C_{\text{max-fdi}}=\frac{{\text{Dose}}_i}{V_{i}k{e}_i{T}_i}\times (1-e^{-k{e}_i{T}_i})$$ (9)

$$C_{\text{max-ssi}}=\frac{{\text{Dose}}_i}{V_{i}k{e}_i{T}_i}\times \left(\frac{1-e^{-k{e}_i{T}_i}}{1-e^{-k{e}_i{\tau }_i}}\right)$$ (10)

where Dose is the dose of rifampin, ke is the elimination rate constant, T is infusion duration, V is volume of distribution, and τ is the dosing interval. The proportion of infants achieving these exposures was calculated. Multiple dosing regimens were simulated from the virtual population using the final population PK model described here and compared to therapeutic adult exposure (AUC_∞ at steady state = 55.2 μg ⋅ h/ml). The median (range) of the simulated AUC_∞ and the mean (standard deviation) C_max-fd and C_max-ss were stratified by PNA <14 days and ≥14 days. The following dosing regimens were evaluated: 8, 10, 12, 15, and 20 mg/kg once per day and 8 mg/kg twice per day infused over 60 min. Optimal rifampin dosing was selected when >90% of infants achieved therapeutic exposures.

Safety.

Adverse events and serious adverse events were assessed through 72 h and 7 days after the last dose of rifampin, respectively. The safety data were reviewed on a quarterly basis by the Best Pharmaceuticals for Children Act Data Monitoring Committee. An adverse event was defined as any untoward medical occurrence that took place during the study regardless of relatedness to study drug. A serious adverse event was defined as an adverse event that resulted in any of the following: (i) death, (ii) life-threatening adverse event, (iii) persistent or significant incapacity or disruption in the ability to perform normal life functions, (iv) prolonged hospitalization, or (v) important medical event.