Comparison of Amikacin Pharmacokinetics in Neonates With and Without Congenital Heart Disease

Introduction

Aminoglycosides are commonly used in combination with ampicillin or vancomycin, as empiric antibiotic therapy for early or late-onset neonatal sepsis, respectively.¹ When comparing neonates to children and adults, they have a higher volume of distribution (Vd) and delayed renal elimination of aminoglycosides. In addition, these pharmacokinetic properties can also vary greatly when comparing preterm, late preterm, and term neonates.^2,3 Patients with congenital heart disease (CHD) may have further alteration of Vd secondary to development of edema and delayed renal elimination secondary to impaired cardiac output and decreased renal perfusion.⁴

Although most NICUs are using gentamicin or tobramycin as their aminoglycoside of choice for treatment of early or late-onset sepsis, there has been an change in resistance patterns of Escherichia coli to gentamicin and tobramycin.^5–7 Due to the increased resistance rates for this organism, some institutions have opted to change their first-line aminoglycoside to amikacin for empiric treatment of early and late-onset sepsis.⁸ There are a limited number of studies that have evaluated dosing and pharmacokinetics of amikacin in the neonatal population and some of these were published between the 1980s and 1990s.^8–18 However, the NICU patient demographics have changed dramatically over the last 30 to 40 years with a lower viable gestational age, improved respiratory support strategies, and development of medical and surgical treatment of neonates with CHD.^19,20 A more recent study compared published dosing recommendations to the institution's modified dosing protocol and noted greater attainment of goal peak concentrations with the modified dosing.⁸ However, previous studies have not specifically evaluated the effect of CHD on amikacin pharmacokinetics and dosing requirements. The purpose of this study was to compare amikacin pharmacokinetics in neonates with and without CHD, with the goal of optimizing initial dose selection of amikacin in neonates with CHD.

Methods

Study Design. This was an institutional review board-approved, retrospective review conducted at an academic medical center with a level IV, 96-bed NICU. Patients receiving amikacin in the NICU from January 1, 2013 through August 31, 2016 were screened using the institution's pharmacy database, Meditech (Medical Information Technology, Inc, Westwood, MA). They were included in the CHD group if they had cyanotic or acyanotic CHD; if surgery was required, patients were only included if their amikacin course occurred in the presurgical period. For each patient with CHD, a control patient was matched 1:1 according to postmenstrual age (PMA) during the same year of admission. Patients were included if they had an amikacin peak and trough concentration drawn appropriately, defined as a peak concentration obtained 30 minutes after completion of the infusion and amikacin trough concentration 30 to 60 minutes before the next scheduled dose. For concentrations not meeting the above criteria, peak and trough concentrations were extrapolated by the appropriate pharmacokinetic equation and were excluded only if the calculated Vd was ± 30% of the expected population Vd estimates 0.4 to 0.6 L/kg.⁸ Patients were excluded if they 1) had amikacin serum concentrations obtained prior to the third dose; 2) had acute kidney injury (AKI), defined as a reduction in urine output to < 0.5 mL/kg/hr for more than 8 hours, an absolute increase in serum creatinine (SCr) by 0.3 mg/dL, or an increase in SCr greater than 50% from baseline prior to receiving amikacin therapy;²¹ 3) received extracorporeal membrane oxygenation; 4) had congenital renal anomalies (e.g., hydronephrosis, multicystic kidney disease, renal agenesis); or 5) had undetectable serum trough concentrations <0.8 mg/L.

Study Objectives and Data Collection. Demographic data including postnatal age (PNA), PMA, and weight at time of amikacin administration were recorded. For those with CHD, the presence of a cyanotic or acyanotic defect was noted based on International Classification of Diseases (ICD)-9²² or ICD-10²³ codes and/or documentation in the medical record. Concomitant nephrotoxic agents (i.e., acyclovir, amphotericin, contrast, furosemide, ibuprofen, indomethacin, and vancomycin) and inotrope or vasopressor therapy received during amikacin therapy were noted. For those with positive cultures, microbiology results were recorded. Renal function markers, including SCr and urine output, were collected during amikacin therapy. The amikacin regimen, including dose and times of administration were collected. The Supplemental Table includes the dosing protocol used at the study institution.⁸ For each patient, pharmacokinetic data were calculated from amikacin concentrations using published pharmacokinetic equations, including elimination rate constant (K_e), half-life (t½), Vd, and clearance (CL).²⁴

The primary objective of this study was to compare the Vd, CL, K_e, and t½ of amikacin in neonates with cyanotic defects, acyanotic defects, and controls, adjusted for gestational and PNA. The secondary objective was to compare the incidence of AKI during amikacin treatment up to 24 hours following discontinuation between controls and the CHD group. An additional analysis was to identify potential risk factors associated with the development of AKI.

Statistical Methods. Descriptive statistics were computed for all demographic and clinical variables. Subjects with and without CHD were categorized and comparisons between these groups were of primary interest. Categorical variables, such as AKI and nephrotoxic agents, were compared between groups using asymptotic or exact Pearson χ² or Fisher exact tests. Interval and continuous variables such as Vd and CL were assessed for normality using the Shapiro-Wilk test (not reported) and compared between groups using Wilcoxon-Mann-Whitney tests. Logistic regression was used to determine the odds (95% CI) of AKI occurrence, controlling for PNA, use of inotropes/vasopressors, presence and number of nephrotoxins, and CHD status. Analysis of covariance (ANCOVA) was used to determine difference in CHD status least square means on pharmacokinetic variables after adjusting for gestational age and PNA. Subjects with CHD were further categorized as cyanotic or acyanotic, and similar descriptions and comparisons of categorical and continuous variables were made on this smaller subset of study participants. Tests of significance are 2-tailed and a 0.05 significance level was used. SAS STAT software for Windows version 9.4 (SAS Institute Cary, NC) was used for all analyses.

Results

Patient Demographics. Sixty patients with CHD met inclusion and exclusion criteria. The Vd for all of these patients were within ± 30% of the expected population Vd estimates; however, six with acyanotic CHD were excluded because a control patient could not be matched based on PMA during the admission year. Therefore, 54 patients with CHD who were awaiting surgical correction were matched with 54 controls according to PMA. Table 1 contains a comparison of demographics between patients with and without CHD. No significant differences were noted in gestational age and PMA; however, patients in the control group were significantly younger based on median PNA, 1.0 (IQR, 1.0–3.0) versus 3.0 (1.0–16.0) days, p = 0.016. In addition, the primary reason for NICU admission differed between groups, with cardiac (38.9%), prematurity (25.9%), and respiratory (24.1%) reasons being predominant in the CHD group and respiratory (31.5%), prematurity (27.8%), and surgical (22.2%) in the non-CHD group, p < 0.001. A significantly greater number of patients with CHD had positive cultures compared with the non-CHD group, 20.4% versus 1.8% (p = 0.002), respectively.

Table 1. CHD Versus Non-CHD Demographics

View Fullscreen

Of those with CHD, 33 (61.1%) had acyanotic CHD and 21 (38.9%) had cyanotic CHD. Table 2 provides an overview of the type of defects, and Table 3 compares demographics between groups. The cyanotic patients with CHD appeared to be older in both median gestational age and PMA compared with acyanotic patients, 38.9 (IQR, 36.1–39.3) weeks versus 34.9 (IQR, 27.0–37.4) weeks, p < 0.05 and 39.0 (IQR, 37.1–39.3) weeks versus 35.0 (IQR, 30.0–38.0) weeks, p < 0.05, respectively.

Table 2. Types of Acyanotic and Cyanotic Defects

View Fullscreen

Table 3. Comparison of Acyanotic Versus Cyanotic Congenital Heart Disease Demographics

View Fullscreen

Pharmacokinetic Analysis. Table 4 contains a comparison of pharmacokinetic parameters between patients with and without CHD, as well as comparisons between patients with acyanotic, cyanotic, and no CHD. Because there was a significant difference in PNA between groups, an ANCOVA was performed to control for this variable. Using the ANCOVA analyses, there was no difference in pharmacokinetic variables between groups after adjusting for gestational age and PNA.

Table 4. CHD Versus Non-CHD Comparison of Pharmacokinetics

View Fullscreen

AKI Occurrences. A total of 10 (18.5%) patients with CHD and five (9.3%) control patients developed AKI; this was not statistically different between groups. Table 5 contains a comparison of AKI occurrences and nephrotoxins between those with and without CHD. There were more patients in the CHD group that received a nephrotoxic medication compared with the non-CHD group, 34 (63.0%) versus 18 (33.3%), respectively, p = 0.0021. In addition, statistical differences were noted in the number of nephrotoxins that were concomitantly administered during use of amikacin (Table 5). A greater number of patients in the CHD group versus non-CHD group received vancomycin (42.6% versus 14.8%, p = 0.001) and furosemide (44.4% versus 9.5%, p < 0.001) concomitantly with amikacin. In addition, more patients in the CHD group received inotropes and/or vasopressors for cardiovascular support, 35.2% versus 14.8%, p = 0.015. In the regression analysis, use of inotropes and/or vasopressors was associated with a 4.5-times greater risk for development of AKI in CHD and non-CHD patients, OR 4.5 (95% CI, 1.43–13.8). Other factors such as PNA, acyanotic CHD, cyanotic CHD, and concomitant nephrotoxins were not associated with development of AKI.

Table 5. CHD Versus Non-CHD Comparison of Concomitant Nephrotoxins and AKI Occurrences

View Fullscreen

Discussion

This is the first study to evaluate pharmacokinetic parameters and incidence of AKI with amikacin in neonates with and without CHD. Although amikacin is not commonly used first-line in many NICUs, use may increase in the future as a result of changing resistance patterns for gentamicin and tobramycin. Therefore, it is important to evaluate amikacin pharmacokinetic differences in infants with CHD because concerns for altered pharmacokinetic parameters for gentamicin have been noted in this population.^25,26 Overall, we found no statistical difference in amikacin pharmaco-kinetic parameters between neonates with and without CHD. However, we did find neonates with CHD had a higher rate of AKI than those without CHD, though not statistically significant.

In the 2 previous studies that have evaluated gentamicin dosing in neonates and children with CHD, no comparator group was included.^25,26 Rather, they assessed potential differences in gentamicin pharmacokinetics in 42 patients with CHD, including 14 infants and 4 neonates.²⁵ They compared their values to previous population estimates and found similar values to patients without CHD. However, due to the small number of infants and neonates, the authors stated that definitive conclusions could not be made, and they were unable to control for other confounding variables. In our study, we attempted to control for confounding variables by including a control group of neonates without CHD that were matched to the patients with CHD by PMA, because this is one of the primary determinants for selection of dose and dosing interval. However, despite attempts to match patients, there were differences noted in PNA. Those in the CHD group were older, and this has implications on other variables that were assessed. For example, the reason that amikacin was initiated, and the antibiotics selected for empiric treatment would be different between groups. Because the median (IQR) age for the non-CHD group was 1 day (1–3), a majority of these were likely initiated on ampicillin and amikacin for rule out early-onset sepsis. Whereas the median (IQR) age for the CHD group was 3 days (1–16), and many of these were likely initiated on vancomycin and amikacin for late-onset sepsis. This is supported by the greater percentage of patients with CHD with culture positive sepsis and a greater number receiving vancomycin versus the non-CHD group. It is also not surprising due to pathophysiology that a greater percentage of the CHD group required use of furosemide and inotropes/vasopressors.

In this study, there were no pharmacokinetic differences noted between those with and without CHD for K_e, t½, Vd, or CL when adjusting for gestational age and PNA. Significant renal maturation occurs in the first week of life and can affect amikacin K_e, t½, and CL; therefore, it was important to control for PNA because there was a statistical difference noted between groups.²⁷ Also, no pharmacokinetic differences were noted when specifically looking at those patients with acyanotic or cyanotic CHD versus those without CHD. These pharmacokinetic differences were explored because previous studies have noted that children with more hemodynamically significant CHD have a higher Vd due to congestive heart failure and delayed renal CL.⁴ The K_e, t½, Vd, and CL noted in this study is similar to previously reported pharmacokinetic parameters for amikacin in neonates.^8,13

Nearly 20% of patients with CHD in this study developed AKI; however, this was not statistically different when compared with those without CHD. It was also noted that nearly two-thirds of patients with CHD also received another nephrotoxic agent during administration of amikacin, and half of these received more than 1 nephrotoxic agent concomitantly. Most commonly, these nephrotoxic agents were furosemide and vancomycin; this is not surprising given the frequent use of furosemide for diuresis in patients with CHD. Also, as stated previously, the empiric use of vancomycin with an aminoglycoside for late-onset sepsis is common practice. In their evaluation of gentamicin use in patients with CHD, Moffett et al²⁵ noted that 56% of their patients received at least 1 other nephrotoxic agent, with a majority of these receiving furosemide. However, this study did not include data regarding incidence of AKI in the patients with CHD receiving gentamicin, so it is difficult to compare our findings. A previous study assessing AKI in neonates receiving amikacin found an overall rate of 7.7%, which is similar to our control group. Previous studies have noted that children with CHD have higher rates of AKI versus children without CHD.²⁸

There were several limitations to our study. First, due to the retrospective design, we were unable to determine the causality of AKI. Second, in September 2016, there was a change from an off-site laboratory to an on-site facility and a different assay method was used for the analysis of the amikacin concentrations. With this change in the assay, the lower limit for detectability for the amikacin trough differed from the off-site to the on-site laboratory facilities, 0.8 to 2.5 mg/L, respectfully. With this higher threshold, many CHD and control patients had trough concentrations < 2.5 mg/L, and pharmacokinetic calculations were unable to be performed. Therefore, the decision was made to include only those with samples obtained prior to the laboratory change in September 2016. Although this resulted in a smaller sample size than expected, we believe that if we did not exclude those samples post-laboratory changes, the data would be skewed because those with faster CL would have undetectable trough concentrations < 2.5 mg/L. This is further supported by the fact that using the same dosing strategy as in the present study, Hughes et al⁸ found a mean amikacin trough of 2 mg/L. Third, an attempt was made to compare pharmacokinetic parameters between patients with acyanotic and cyanotic CHD because it has been noted that children with more hemodynamically significant CHD may have altered pharmacokinetic parameters that affect dosing.⁴ However, subdividing the CHD group resulted in even smaller samples sizes, 33 in the acyanotic and 21 in the cyanotic group. A post hoc power analysis was conducted and determined that we would have needed 330 and 1645 patients in the CHD and non-CHD groups to detect a difference between Vd and CL, respectively. If any differences exist between CHD and non-CHD patients, they may be clinically negligible. Last, we note that the CHD group had a higher number of positive cultures. Unfortunately, we are unable to explain these findings, but we feel that this does not impact our pharmacokinetic and AKI analysis.

At this time, we would recommend using the dosing strategy listed in the Supplemental Table for a majority of patients with CHD. Based on our previous study evaluating this dosing strategy, a mean peak and trough concentration of approximately 28 and 2 mg/L, respectively, would be achieved.⁸ The exceptions would be for neonates who have not established adequate urine output (i.e., <1 mL/kg/hr), those neonates undergoing repair with cardiopulmonary bypass, and also those neonates with significant hemodynamic instability requiring a significant number of inotropes or vasopressors. For all of these neonates, clinicians may need to choose an extended dosing interval listed in the Supplemental Table and monitor trough concentrations more vigilantly to adjust the dosing interval if needed and avoid development of AKI.

Conclusions

There was no statistically significant difference in pharmacokinetic parameters and AKI between CHD and non-CHD groups. Although the incidence of AKI was slightly greater in the CHD group, this difference may have been due to the greater use of nephrotoxic agents and vasopressors/inotropes in those with CHD. Clinicians may consider using standard dosing of amikacin for neonates with CHD, with careful monitoring for AKI in those receiving concomitant nephrotoxic agents.