Vancomycin is a broad-spectrum antibiotic widely used to treat gram-positive bacterial infections. Because of interindividual variability in pharmacokinetics in children, ongoing therapy typically necessitates therapeutic drug monitoring in order to optimize the expected efficacy of vancomycin while minimizing toxicity. In children older than 1 month, the typical recommended initial dose is 40 to 60 mg/kg/day in 3 or 4 divided doses.1 Subsequent dosing adjustments are typically based on trough vancomycin serum or plasma concentrations at steady state.
In published studies, the pharmacodynamic surrogate that best predicts the clinical efficacy of vancomycin is the ratio between the 24-hour area under the serum or plasma concentration–time curve of vancomycin and the minimum inhibitory concentration of the targeted pathogen (AUC24/MIC). AUC24/MIC ratios from ≥200 to ≥400 have been associated with improved clinical and bacteriological outcomes in adults.2,3 In a 2009 multisociety consensus statement, Rybak et al4 reported that trough serum concentrations of 15–20 mg/L will correspond to an AUC24/MIC ≥400 for most adult patients when the pathogen MIC is <1 mg/L. Based on these data and because obtaining a single trough concentration is simpler than measuring and calculating the AUC24/MIC, the Infectious Diseases Society of America recommends to achieve steady-state trough serum concentrations of 15 to 20 mg/L for adults with serious methicillin-resistant Staphylococcus aureus infections if the MIC of the organism is ≤1 mg/L.5 Recommendations for pediatric patients mirror those of adults, suggesting trough vancomycin serum concentrations >10 mg/L when treating less serious infections, and 15–20 mg/L for infections such as endocarditis, osteomyelitis, meningitis and hospital-acquired pneumonia caused by S. aureus.4
Despite having clearly defined serum concentration goals for treatment with vancomycin, establishing a dosage schedule to achieve the desired therapeutic concentration remains challenging because of the large interindividual variability commonly observed in vancomycin pharmacokinetics in children. Given the potential benefit of achieving higher serum concentrations earlier in the course of treatment, the use of a single initial loading dose of vancomycin as previously reported in adult patients4,6 was deemed as an attractive option to facilitate a rapid attainment of target trough vancomycin concentrations in pediatric patients.
In the present study, we hypothesized that a loading dose of vancomycin would produce a significantly higher proportion of patients achieving target serum drug concentrations at 8 hours after the first dose.
PATIENTS AND METHODS
Setting and Patients
We conducted a double-blind randomized controlled trial of children aged 2 to 18 years hospitalized at Boston Children’s Hospital between February 1, 2011, and January 15, 2012, who required antimicrobial therapy with vancomycin (Hospira, Inc., Lake Forest, IL, lot #896188EO-4) for a suspected or documented infection. We excluded patients with a body weight above 67 kg (to limit the maximum loading dose to 2 g), preexisting severe renal dysfunction, defined as creatinine clearance <50 mL/min/1.73m2 using the original Schwartz equation,7 known hearing impairment, intravenous vancomycin treatment in the prior 7 days or undergoing a procedure with anticipated moderate to severe blood loss (eg, cardiac surgery or extensive orthopedic procedure).
For all participants enrolled in the study, relevant baseline demographic, medical history and safety data were recorded. Medical history data included primary and secondary diagnoses; other comorbidities such as obesity or cystic fibrosis; and presence of systemic inflammatory response syndrome, defined as 2 or more of the following: temperature >38.5°C or <36°C; mean heart rate >2 standard deviations above normal for age; mean respiratory rate >2 standard deviations above normal for age; or high or low white blood cell count for age.8,9
Randomization and Concealment
Participants were randomized in blocks of 2 and 4 to receive either a loading dose of 30 mg/kg of vancomycin as a single intravenous infusion over 2 hours (intervention group) or an initial vancomycin dose of 20 mg/kg intravenously over 2 hours (comparison group). The initial dose was administered over 2 hours in both groups to preserve allocation concealment. All patients subsequently received a 20 mg/kg dose every 8 hours as was the standard of care in our hospital for treatment of severe infections at the time of the study. Subsequent doses were administered over 1 hour, unless the patient developed red man syndrome (as identified by the clinical team), in which case the infusion time was increased to 2 hours. The investigators, family and primary care teams were blinded to group assignment, and the first dose of vancomycin for all participants was prepared so that the solution volumes were identical. The computer-generated randomization was concealed in a locked binder until the intervention was assigned.
Vancomycin Concentration Sampling and Analysis
Trough serum vancomycin concentrations were obtained within 60 minutes before the second (8-hour) and third (16-hour) vancomycin doses. In order to increase the likelihood of having a cloud of sparse data for population pharmacokinetic analysis, 1 or 2 additional serum vancomycin samples were obtained from each participant within the first 32 hours of therapy at a time coinciding with blood collection for clinical care. These samples were obtained only from participants with an indwelling catheter whose family provided written consent for additional sampling.
Vancomycin concentrations were measured using a fluorescence polarization immunoassay (Roche Diagnostics, Indianapolis, IN) on the Roche Integra 800 instrument. The assay had a limit of quantitation of 0.74 mg/L and an interassay coefficient of variability of <3%.
Our primary outcome was the proportion of patients whose serum vancomycin concentrations were within the target range (defined as 15–20 mg/L) 8 hours after the first dose (ie, first trough concentration). Our secondary study outcomes were: 1) a description of the vancomycin pharmacokinetic profile in a pediatric cohort employing population pharmacokinetic modeling; 2) evaluation of tolerability of the 30 mg/kg loading dose, focusing on the risk of nephrotoxicity defined as doubling in serum creatinine concentration compared with baseline4 within 7 days of administration of the loading dose of vancomycin, development of red man syndrome during the infusion or escalation of care defined as new use of mechanical ventilation or transfer to the intensive care unit within 7 days of the first dose of vancomycin. A participant was assumed to have received concomitant nephrotoxins if any of the following intravenous medications was administered within 24 hours of the first vancomycin dose: an aminoglycoside (eg, gentamicin, tobramycin, amikacin), a radiographic contrast agent, intravenous acyclovir, pentamidine, amphotericin B, cidofovir, furosemide, cyclosporine, foscarnet or ketorolac.
For study purposes, all subjects who had at least 1 documented vancomycin plasma concentration were included in the pharmacokinetic analysis.
Vancomycin pharmacokinetics have been successfully described before by 1-compartment models.10 Therefore, vancomycin plasma concentrations during the infusion (CpInf) were modeled or predicted if no blood samples were obtained, by employing the following 1-compartment pharmacokinetic equation:
where k0 is the infusion rate of vancomycin obtained by dividing the total infused dose of vancomycin by the duration of the intravenous infusion (tInf), kel is the elimination rate constant (h-1) and Vd is the apparent volume of distribution (mL/kg).
The concentrations of vancomycin during the postinfusion phase were modeled or predicted, using the following 1-compartment pharmacokinetic equation:
where CpEInf is the plasma concentration of vancomycin at the end of infusion, and t is the time at which postinfusion plasma concentrations were obtained or modeled.
Time-plasma concentration data were modeled in 3 steps. First, in order to overcome the limited and sparse vancomycin concentrations, the dataset was enriched by solving the 2 pharmacokinetic equations described above with Microsoft Office Excel 2007 (Microsoft Co., Redmond, WA), using the observed concentrations as the true data. Thereafter, enriched time-plasma concentrations from each patient were subjected to an individual pharmacokinetic analysis using SAAM II v. 1.2.1. (SAAM Institute, University of Washington, Seattle, WA). This analysis generated the individual Vd, assuming that the amount of drug present in the body at any given time in a 1-compartment model is described by the equation A(t) = Cp(t) Vd and kel; both of them were estimated by the software while fitting the data to the pharmacokinetic model described above.
Because of the limitations inherently associated with individual pharmacokinetic modeling, which in turn was based on enriched data generated by pharmacokinetic analysis of the original dataset with limited and sparse data, we proceeded with an iterated 2-stage population pharmacokinetic analysis including subjects from the 2 dosage regimens together using SAAM II. This final step generated the Bayesian Vd and kel estimates together with their corresponding 95% confidence intervals (CIs). The individual pharmacokinetic estimates were used as the initial parameters.
The half-life (t½; in hours) of vancomycin was derived directly from kel, that is,
, for each participant in the case of the individual pharmacokinetic analysis, or from the population kel data in the case of the population pharmacokinetic analysis. As part of the individual pharmacokinetic analysis, AUC0–24h was computed automatically by SAAM II according to the following equation:
Study data were collected and managed using REDCap electronic data capture tools hosted at Boston Children’s Hospital.11 Statistical analyses were carried out using Stata Software release 12 (StataCorp, College Station, TX). Scatterplots were constructed using GraphPad Prism version 5.01 for Windows (GraphPad Software, San Diego, CA).
A Student t test was used to compare continuous variables between groups, and the χ2 and Fisher’s exact tests were used to compare categorical variables. The 2-sample Wilcoxon rank sum (Mann–Whitney) test was used to compare medians of the time to trough at 8 and 16 hours. Baseline and follow-up creatinine concentrations were measured for 40 patients at the discretion of the provider, and the remaining missing data for 19 patients were treated as missing and excluded from the analysis. An exploratory regression analysis was performed to investigate the relationship between study arm and vancomycin concentration before the second vancomycin dose while adjusting for potential confounders such as age, body mass index ≥85th percentile-for-age (“overweight”) and presence of systemic inflammatory response syndrome. A negative binomial regression model was chosen in view of the observed over-dispersion in the data set.
Using pharmacokinetic simulations and previously published pediatric pharmacokinetic parameters for vancomycin,12 we estimated that 36% of patients in the conventional dose group and 73% patients in the loading dose group would achieve a vancomycin serum concentration above 15 mg/L at 8 hours. At initiation of the trial, sample size calculations projected 84 enrolled participants to obtain 90% statistical power to detect this difference in proportion, setting type I error at 0.05. The trial was stopped for analysis before predetermined sample size was reached due to slower enrollment rate than expected.
The study was approved by the Committee on Clinical Investigation at Boston Children’s Hospital; written informed consent was obtained from all participants’ parents, and verbal assent was obtained for children capable of providing it. An Investigational New Drug exemption was obtained from the US Food and Drug Administration.
One hundred fifteen families were approached for enrollment (Fig. 1). Of 61 patients who consented to participate, 59 were enrolled and randomized (in the other 2, the clinical team decided to not administer vancomycin). All randomized participants received the initial vancomycin dose they were assigned to. Table 1 displays the demographic and clinical characteristics of randomized participants at the time vancomycin was prescribed (n = 59). There were no significant differences in baseline characteristics between patients in the 2 study arms.
A vancomycin trough concentration at 8 hours (the primary study outcome) was measured in 46 participants, whereas no sample was obtained in 11 participants in the loading dose group and for 2 participants in the conventional dose group (P = 0.01). Of them, 8 children in the loading dose group and 1 child in the conventional dose group who did not have the 8-hour trough level measured, developed red man syndrome or had vancomycin discontinued before receiving a second dose. The 13 patients without the 8-hour trough vancomycin concentration measurement were excluded from the primary outcome analysis (ie, comparison of the observed 8-hour through concentrations between the 2 study groups).
Vancomycin concentrations observed at 8-hour and 16-hour time points (trough values) are depicted in Figure 2 using box plot representations where the line within the box marks the median value. Two study participants (7%) of 30 in the loading dose group and no participant of 29 in the conventional group had an 8-hour trough concentration between 15 and 20 mg/L before the second dose of vancomycin (P = 0.17). In a negative binomial regression model, age, weight and the presence of systemic inflammatory response syndrome were not independent predictors of vancomycin trough concentrations at 8 hours (P = 0.72, P = 0.58 and P = 1.00, respectively). Two of 30 (7%) participants who received a loading dose and 2 of 29 (7%) participants who received a conventional dose had a vancomycin trough concentration higher than 20 mg/L before the second dose. Four of 30 (13%) participants who received a loading dose and 1 of 29 (4%) participants who received a conventional dose had a vancomycin trough concentration higher than 20 mg/L before the third dose.
The results of the pharmacokinetic analyses are shown in Tables 2 (individual pharmacokinetics) and 3 (population pharmacokinetics). The half-life of vancomycin, derived from the population kel values, was 3.6 hours (95% CI: 3.2–4.0 hours). The vancomycin plasma concentration–time curve was generated based on the means for concentration measurements obtained at different time points (Fig. 3). There was a tendency toward higher systemic exposure in the intervention group within the first dose interval (0–8 hours) of vancomycin administration; this difference was not statistically significant, with an overlap between the 95% CIs at all time points. Goodness-of-fit indicated that there was an adequate fit between the equations and stepwise process and the observed concentrations of vancomycin (data not shown).
Table 4 displays adverse outcomes. Data for baseline and follow-up creatinine measurements within 7 days of first dose were available for 40 participants. One patient with cystic fibrosis developed acute kidney injury 7 days after administration of a conventional dose while receiving concomitant nephrotoxic medications; renal function recovered when these medications and vancomycin were discontinued. The serum creatinine doubled within 7 days in 4 loading dose recipients; creatinine measurements were normalized by day 12 in all participants, except for 1 who received several nephrotoxic medications, in whom the serum creatinine normalized a month later.
Other adverse events during the first 3 doses included phlebitis (n = 1), pruritus (n = 1) and transient swelling above the eyebrows in a participant with mild to moderate red man syndrome. One patient with a history of red man syndrome developed anaphylaxis upon receiving the 30 mg/kg loading dose of vancomycin. No participant died within 7 days of initiation of vancomycin.
In this clinical trial, an intravenous loading dose of 30 mg/kg of vancomycin administered in children aged 2–18 years with suspected or confirmed infection did not result in a significant increase in the proportion of patients reaching target serum concentration at first trough (8 hours after first dose), compared with patients receiving standard therapy (20 mg/kg). Despite the fact that a loading dose of 30 mg/kg represented a 50% increase to the standard initial dose of vancomycin, our study findings contrast to studies of adult patients reporting that administration of a loading dose of 25–30 mg/kg in critically ill patients effectively achieves early target serum concentrations.4,6,13,14
Our study also suggests that conventional dosing guidelines may not be optimal for achieving currently recommended trough concentrations in children. Recently, leading professional societies, including Infectious Diseases Society of America, have recommended to target higher vancomycin trough concentrations in the treatment of serious infections, compared with previous guidelines,4 suggesting that an initial dose of 40–60 mg/kg/day of intravenous vancomycin in children may not be adequate. Frymoyer et al15 showed that a daily dose of 60 mg/kg, which is the highest initial dosage recommendation listed on the package insert for vancomycin, is inadequate to achieve trough concentrations of 15–20 mg/L or even 10–15 mg/L in children. In our study, the median trough concentrations before the second and third doses were 7.7 mg/L at 8 hours and 9 mg/L at 16 hours, suggesting that without dosage adjustment, recommended concentrations will not be achieved in a timely fashion in most pediatric patients. Thus, our results support the adjustment of vancomycin dosing based on early trough measurement (ie, before the second or third dose) in the pediatric clinical setting, where available, in contrast with adult literature that suggests waiting until the fourth or fifth dose before measuring trough concentration. In our study, vancomycin concentrations were measured in blood samples drawn at times that would be expected to be before achievement of its steady state, and at the time of the study, in our institution dose adjustment was typically done based on a steady-state concentration. Subsequent levels were obtained as part of routine clinical care, and changes to the dosing regimen were made at the primary provider’s discretion.
It is possible that despite administering a 30 mg/kg as an initial loading dose, our strategy was not effective because of the relatively slow infusion rate over 2 hours (ie, up to 25% of the dosing interval), which in the context of a relatively short half-life led to a slow rise in serum concentrations, resulting in no statistical difference in plasma drug concentrations between the 2 studied groups. In the context of high interpatient variability in pharmacokinetic parameters and a trend toward higher risk for development of red man syndrome in children receiving a loading dose, we cannot recommend a loading dose of vancomycin based on our results.
Our findings offer a new paradigm, suggesting that currently recommended target trough concentrations may not be optimal in children. An AUC24/MIC ratio >400 has been suggested to be the most accurate predictor of favorable clinical outcomes.2,14 In our pediatric population, the median estimated AUC24/MIC was above 400 in both study arms despite vancomycin trough levels were lower than 15 mg/L in almost all patients. Therefore, as suggested by Gordon et al16 in their retrospective study and Le et al17 in their pharmacokinetic analysis, our data raise the concern that the currently recommended trough vancomycin concentrations may not be relevant in pediatric patients. Further research, however, is required to determine the optimal dosing schedule and therapeutic vancomycin monitoring method to achieve the desired pharmacodynamics goals and clinical outcomes in this population.
Our study has several limitations. First, because our sample size was smaller than anticipated, our trial was underpowered to detect an increase of ≥40% in target serum concentrations of vancomycin between study arms. Second, the number of cases excluded from the primary outcome analysis was larger in the high-dose group (n = 11 of 30) than in the reference group (n = 2 of 29). This could introduce analysis bias, although its direction is unknown. Finally, the increase in red man syndrome in the high-dose group could indicate that longer infusion times are warranted to administer 30 mg/kg in children. Anecdotally, we noticed that most red man syndrome reactions occurred toward the end of the vancomycin infusion, at the time a normal saline flush was administered. Following this observation, the standard practice at our institution was changed to avoid this adverse event as a result of a flush at the end of infusion.
The compartmental pharmacokinetic analysis was performed with SAAM II. This is a user-friendly software for developing simple and complex pharmacokinetic models, and it generates robust parameters that are compatible with other more sophisticated pharmacokinetic software packages.18 The pharmacokinetic parameters were similar between the 2 study groups, with the exception of the mean Vd, which was slightly larger (by 11%) in the loading dose group. Vd represents a nonphysiologic space. It is a nonmeasurable, calculated value and is affected by the individual’s body composition, such as fat and water content. Therefore, a 11% difference in mean Vd values between the 2 groups is modest and acceptable, as the groups were not matched on the above parameters. Moreover, both the Vd values derived from the individual study groups and from the entire cohort (779.2 mL/kg; 95% CI: 739.0–813.2 mL/kg) sit well within the reported ranges of 262 to 1046 mL/kg.10 The population half-life of vancomycin of 3.6 hours in our study is within the range of 2–3 hours in children >4 years old and 5–8 hours in adults previously reported by others.19 In relation to clearance (CL), if we use the population t½ and Vd obtained in our study to extrapolate our results using the standard equation of t½ = (Ln 2 × Vd)/CL, the resulting value of 149.9 mL/h/kg is close to the CL of 20–112 mL/h/kg previously reported.10 Furthermore, the use of SAAM II allowed us to provide a brief stepwise description of the analysis of sparse vancomycin data performed in the present study, which can be easily replicated, as well as implement derived pharmacokinetic parameters obtained in the clinical setting.
The administration of a loading dose of 30 mg/kg of vancomycin did not result in earlier achievement of therapeutic trough concentrations in children enrolled in this study. Our data confirm that current dosing recommendations are inadequate to achieve published goal trough concentrations of vancomycin in the majority of children. However, despite low trough concentrations, the median AUC24/MIC was greater than 400, suggesting that 60 mg/kg/day dosing is likely to achieve appropriate therapeutic targets in a large proportion of patients. This finding questions the appropriateness of currently recommended target trough concentrations of vancomycin in children. Larger prospective studies are needed to determine whether a target vancomycin concentration lower than currently recommended would correspond to favorable clinical outcomes and lower risk of toxicity.
The authors thank Michael Giarrusso, PharmD, Jessica Miller, PharmD, and Hayden Schwenk, MD, for their help with participant enrollment; the Pharmacy Department for help with study implementation and supply of vancomycin; the nursing and physician teams and the clinical laboratory at Boston Children’s Hospital; Mari Nakamura, MD, MPH, and Robert Wright, MD, MPH, for assistance with data safety and monitoring; the Clinical Research Program at Boston Children’s Hospital for help with statistical support; the Program for Patient Safety and Quality at Boston Children’s Hospital for partial funding of the study; and the Division of Infectious Diseases at Boston Children’s Hospital. Participation of Dr. AA Nava-Ocampo was supported by PharmaReasons, Toronto, Ontario, Canada.
1. Package. Insert. Vancomycin Package Insert. Revised September 2007 Lake Forest, IL Hospira, Inc.
2. Moise-Broder PA, Forrest A, Birmingham MC, et al. Pharmacodynamics of vancomycin and other antimicrobials in patients with Staphylococcus aureus lower respiratory tract infections. Clin Pharmacokinet. 2004;43:925–942
3. Brown J, Brown K, Forrest A. Vancomycin AUC24/MIC ratio in patients with complicated bacteremia and infective endocarditis due to methicillin-resistant Staphylococcus aureus and its association with attributable mortality during hospitalization. Antimicrob Agents Chemother. 2012;56:634–638
4. Rybak MLB, Rotschafer JC, Moellering R Jr, et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am J Health Syst Pharm. 2009;1:82–98
5. Liu C, Bayer A, Cosgrove SE, et al. Clinical practice guidelines by the infectious diseases society of america for the treatment of methicillin-resistant Staphylococcus aureus infections in adults and children. Clin Infect Dis. 2011;52:e18–55
6. Truong J, Levkovich BJ, Padiglione AA. Simple approach to improving vancomycin dosing in intensive care: a standardised loading dose results in earlier therapeutic levels. Intern Med J. 2012;42:23–29
7. Schwartz GJ, Gauthier B. A simple estimate of glomerular filtration rate in adolescent boys. J Pediatr. 1985;106:522–526
8. Septic Shock: Initial Evaluation and Management in Children [Internet]. 2012 Waltham, MA: UpToDate, Inc
9. Fleming S, Thompson M, Stevens R, et al. Normal ranges of heart rate and respiratory rate in children from birth to 18 years of age: a systematic review of observational studies. Lancet. 2011;377:1011–1018
10. Marsot A, Boulamery A, Bruguerolle B, et al. Vancomycin: a review of population pharmacokinetic analyses. Clin Pharmacokinet. 2012;51:1–13
11. Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381
12. Murphy JE Clinical Pharmacokinetics. 20084th ed Bethesda, MD American Society of Health-System Pharmacists
13. Wang JT, Fang CT, Chen YC, et al. Necessity of a loading dose when using vancomycin in critically ill patients. J Antimicrob Chemother. 2001;47:246
14. Mohammedi I, Descloux E, Argaud L, et al. Loading dose of vancomycin in critically ill patients: 15 mg/kg is a better choice than 500 mg. Int J Antimicrob Agents. 2006;27:259–262
15. Frymoyer A, Hersh AL, Benet LZ, et al. Current recommended dosing of vancomycin for children with invasive methicillin-resistant Staphylococcus aureus infections is inadequate. Pediatr Infect Dis J. 2009;28:398–402
16. Gordon CL, Thompson C, Carapetis JR, et al. Trough concentrations of vancomycin: adult therapeutic targets are not appropriate for children. Pediatr Infect Dis J. 2012;31:1269–1271
17. Le J, Bradley JS, Murray W, et al. Improved vancomycin dosing in children using area-under-the-curve exposure. Pediatr Infect Dis J. 2013;32:e155–e163
18. Finkelstein Y, Nava-Ocampo AA, Schechter T, et al. Discrepancies in pharmacokinetic analysis results obtained by using two standard population pharmacokinetics software programs. Fundam Clin Pharmacol. 2009;23:53–57
19. Broome L, So TY. An evaluation of initial vancomycin dosing in infants, children, and adolescents. Int J Pediatr. 2011;2011:470364
Keywords:© 2013 by Lippincott Williams & Wilkins, Inc.
vancomycin; pediatrics; pharmacokinetics; methicillin-resistant Staphylococcus aureus; loading dose