Anesth Analg 2006;102:1062-1069
© 2006 International Anesthesia Research Society
doi: 10.1213/01.ane.0000198626.67391.34
PEDIATRIC ANESTHESIA
Population Pharmacokinetics of Milrinone in Neonates with Hypoplastic Left Heart Syndrome Undergoing Stage I Reconstruction
Athena F. Zuppa, MD MSCE,
Susan C. Nicolson, MD,
Peter C. Adamson, MD,
Gil Wernovsky, MD,
John T. Mondick,
Nancy Burnham, RN, MSN,
Timothy M. Hoffman, MD,
J. William Gaynor, MD,
Lauren A. Davis, BS,
William J. Greeley, MD, MBA,
Thomas L. Spray, MD, and
Jeffrey S. Barrett, PhD
Division of Clinical Pharmacology and Therapeutics, Department of Pediatrics, Division of Critical Care, Department of Anesthesiology and Critical Care Medicine, Division of Cardiothoracic Anesthesia, Department of Anesthesiology and Critical Care Medicine, Division of Cardiology, Department of Pediatrics, Division of Cardiothoracic Surgery, Department of Surgery, Abramson Research Center, Philadelphia, Pennsylvania
Address correspondence and reprint requests to Athena F. Zuppa, MD, Abramson Research Center, Suite 916, 3615 Civic Center Blvd., Philadelphia, PA 19104-4318. Address e-mail to zuppa{at}email.chop.edu.
 |
Abstract
|
|---|
We performed a blinded, randomized pharmacokinetic study of milrinone in 16 neonates with hypoplastic left heart undergoing stage I reconstruction to determine the impact of cardiopulmonary bypass and modified ultrafiltration on drug disposition and to define the drug exposure during a continuous IV infusion of drug postoperatively. Neonates received an initial dose of either a 100 or 250 µg/kg of milrinone into the cardiopulmonary bypass circuit at the start of rewarming. Postoperatively, milrinone was infused to clinical needs. A mixed-effect modeling approach was used to characterize milrinone pharmacokinetics during cardiopulmonary bypass, modified ultrafiltration, and postoperatively using the NONMEM algorithm. All patients in this study demonstrated a modified ultrafiltration concentrating effect that occurred despite a modified ultrafiltration drug clearance of 3.3 mL · kg–1 · min–1. The infants in this study demonstrated an impaired renal clearance during the immediate postoperative period. A constant infusion of 0.5 µg · kg–1 · min–1 resulted in drug accumulation during the initial 12 h of drug administration. Postoperatively, milrinone clearance was significantly impaired (0.4 mL · kg–1 · min–1), improved by the 12th postoperative hour, and approached steady-state clearance (2.6 mL · kg–1 · min–1) by postoperative day 4. In the postoperative setting of markedly impaired renal function, an infusion rate of 0.2 µg · kg–1 · min–1 should be considered.
|
Introduction
|
|---|
The surgical management of hypoplastic left heart syndrome (HLHS), a congenital cardiac malformation that occurs in more than 2000 neonates in North America each year, consists of either staged reconstruction (1) or transplantation (2,3). Given the lack of organ availability and the improved outcomes with reconstructive surgery, patients with HLHS are often initially managed by the reconstructive sequence. Stage I reconstruction consists of an atrial septectomy, arch augmentation, and placement of a restrictive systemic or right ventricle to pulmonary artery shunt. Support techniques integral to this procedure include cardiopulmonary bypass (CPB) with or without deep hypothermic circulatory arrest (DHCA). Advances in both surgical technique and perioperative management have improved outcomes (4,5). Two such advances include the intraoperative and postoperative administration of milrinone and the use of modified ultrafiltration (MUF). MUF is a technique using ultrafiltration of the patient’s intravascular volume and hemofiltration of the bypass circuit after separation from CPB to reverse hemodilution (6). The hemoconcentrating effect of MUF may increase the plasma concentrations of drugs administered before its use.
Milrinone, an inodilating and lusitropic drug (7), is a bipyridine derivative of amrinone that is primarily used for the treatment of congestive heart failure. The drug is primarily (85%) cleared through renal secretion, with 15% undergoing glucuronidation, and is 70% protein bound (8). Milrinone is a drug commonly used to support cardiac output after congenital heart surgery in neonates, infants, and children. Limited pediatric data on milrinone pharmacokinetics are available, and no information is available for its perioperative use in neonates with HLHS. Pharmacokinetic studies suggest that milrinone clearance (CL) is greater and its volume of distribution (Vd) larger in children than in adults (9,10), but infants appear to have lower milrinone CL than children (9).
In neonates undergoing staged reconstruction of HLHS, CPB, DHCA, and MUF can impact drug disposition (11). Furthermore, because of the renal and hepatic immaturity of the neonate and the low cardiac output state that is common after a stage I reconstruction, it is anticipated that the pharmacokinetic profile of drugs would differ significantly from that observed in older infants. Therefore, novel and specific studies are required to understand drug disposition in this specialized population. In addition, nonlinear mixed-effects modeling techniques allow for the determination of pharmacokinetic parameters under fewer assumptions than classical pharmacokinetic approaches and from both small and sparsely sampled patient populations.
The objectives of this study were therefore to describe the plasma concentration-time profile of milrinone in infants with HLHS undergoing stage I reconstruction, to develop a population model to describe the impact of CPB and MUF on milrinone pharmacokinetics, and to describe milrinone CL in the postoperative setting. As there are no clinical pharmacological data available to guide dosing of milrinone in neonates undergoing stage I reconstruction, empirical loading doses on CPB of 100 µg/kg and 250 µg/kg that have been historically used at our institution were studied.
|
Methods
|
|---|
After IRB approval and parental permission, 16 neonates with a diagnosis of HLHS or its variants who underwent stage I reconstruction were enrolled in a blinded, randomized pharmacokinetic study of 2 doses of milrinone: 100 µg/kg or 250 µg/kg bolus, followed later by a standard dose infusion of 0.5 µg · kg–1 · min–1. Anesthetic technique consisted of fentanyl (50 µg/kg) and pancuronium. A volatile anesthetic was used as needed. The CPB circuit was primed with an average of 345 ± 26 mL of a combination of plasmalyte, 5% albumin, and whole blood to achieve a hematocrit of 30%. Hypothermia to 18°C–20°C and circulatory arrest were used in all patients. Study drug was administered into the venous reservoir of the CPB circuit 2 min into rewarming, after recommencing CPB on the completion of DHCA. After weaning from CPB with 3 µg · kg–1 · min–1 of dopamine infusing through a catheter in the common atrium, all patients underwent MUF. On transfer to the Cardiac Intensive Care Unit (CICU), a continuous IV infusion of milrinone at a standard dose rate (0.5 µg · kg–1 · min–1) was started 90 min after the intraoperative bolus. Postoperatively, each patient’s infusion was adjusted at the discretion of the medical team.
Three mL samples of blood were drawn from the CPB circuit or an arterial catheter immediately before and 5, 10, 15, and 20 min after the bolus dose. Additional samples were obtained 5 min after the initiation of MUF and at the end of MUF (Fig. 1, A). The milrinone concentration in the ultrafiltrate was measured. In the CICU, blood samples were obtained from an arterial catheter immediately before the start of the milrinone continuous infusion, at 3, 6, 9, and 24 h after the initiation of the infusion, and immediately before the discontinuation of the infusion (Fig. 1, B).
View larger version (19K):
[in this window]
[in a new window]
|
Figure 1. A, Schematic of body temperature related to cardiopulmonary bypass (CPB), deep hypothermic cardiac arrest (DHCA), and modified ultrafiltration (MUF) that is performed as standard of care. Arrows delineate time of pharmacokinetic sampling. B, Postoperative course and average duration of the milrinone infusion (EOI = end of infusion). Arrows delineate time of pharmacokinetic sampling.
|
|
Plasma was separated by centrifugation and stored at –70°C. Milrinone concentrations were determined using a previously described high performance liquid chromatography methodology with UV detection and a lower limit of detection of 5 ng/mL (12). The intraday coefficient of variance was 2.2%–8.3% for milrinone concentrations in the range of 5 to 500 ng/mL. The interday coefficient of variance was 3.9%–7.5% for the same concentration range.
Thirteen pharmacokinetic samples were to be collected for each patient. The population pharmacokinetic analysis was conducted using data from all 16 patients. NONMEM (NONMEM Project Group, University of California at San Francisco) version 5 level 1.1 (including PREDPP (ADVAN 5, TRANS 1)) was used for this analysis. Models were run with both the first-order estimation and the first-order conditional estimation with interaction method (depending on convergence outcome). Final models used the first-order conditional estimation method. S-Plus Version 6.2 (Mathsoft, Inc., Data Analysis Products Division, Seattle, WA) was used for goodness-of-fit diagnostics and graphical displays.
A two-compartment disposition model was deemed optimal to define the milrinone plasma and MUF concentrations based on results from the model building process, with the MUF process treated as a third compartment. Models were parameterized by weight-normalized CL (mL · kg–1 · min–1), volume of central compartment (V1, mL/kg), volume of peripheral compartment (V2, mL/kg) and intercompartmental clearance (Q, mL · kg–1 · min–1). V2 was assumed to be constant across all four intervals. Model building commenced with exploring changes in Vd and CL across intervals. Subsequent model building included the division of the study into four distinct phases: Interval 1 = CPB, Interval 2 = MUF, Interval 3 = postoperative infusion within 12 h postoperatively (INF0–12h) and Interval 4 = postoperative infusion after 12 h (INF12+h). The cut-off of 12 h was chosen based on the observance that urine output and drug CL increased for all patients by this time. CL and central Vd were further divided based on the four intervals. Intervals could be turned on and off by the assignment of indicator variables (0 if the interval was not occurring, and 1 if the interval was occurring). Parameter estimates were brought to the power of the dummy variable. Therefore, in the event that an interval was turned off, the parameter estimate would equal 1. Values of CL and Vd were determined by the following equations:
|
|
The parameter CLINF12+h was defined as time dependent and allowed to change dynamically as the interval progressed. The following equation was used to estimate CLINF12+h:
where CLss represents the parameter estimate for the steady-state milrinone CL that would occur once renal function returned to normal and Kss represents the rate constant at which CL is changing. It was assumed that volume did not change during the postoperative period and that VINF0–12h = VINF2.
First order elimination from the central compartment was characterized as K13 = CLMUF/V1 during interval 2 and as K10 = (CLINF0–12h *CLINF12+h)/V1 during intervals 3 and 4. As milrinone is primarily secreted unchanged in the urine, it was assumed that CL during Interval 1 (CPB) was negligible.
An exponential error model was used to describe the interpatient variability as (e.g., for CL): CLj = CL0jexp(
jCL) where exp(jCL) denotes the proportional difference between the true individual parameter (CLj) and the typical value (CL0j) predicted for an individual with covariates equal to those of patient j. Numerous error models were considered in the determination of the inter-individual variance-covariance matrix 
, including diagonal and full matrices. The dimensionality of the matrix was reduced in several steps to reflect the covariance structure of individual parameters and to eliminate estimation problems caused by over-parameterization.
Additive, proportional, and combined (additive and proportional) residual error models were considered during the model building process. In the end, random residual variability was modeled using an additive error model according to the following equation:
where CLij and CL0ij are the ith measure and model predicted milrinone plasma concentrations for the jth patient, respectively. Distinct parameters for 
were estimated for each interval. The analysis considered gestational age and weight as continuous variables and gender and race as categorical covariates.
Model building was performed in several steps. Initially, a compartmental model and the covariance matrix structure for inter-individual random effects were identified. Later models included log-transformed data, with back transformation of the final estimates. The likelihood ratio test at the significance level 
= 0.05 was used to discriminate between alternative hierarchical models. The Akaike Information Criteria was used to distinguish between non-hierarchical models. The level of 0.05 corresponds to a reduction (
) of 3.84 (
2, P < 0.05; 1 degree of freedom) in the minimum objective function when 1 parameter added to the model was used to examine significance. When more than 1 parameter was added, the critical reduction in the objective function corresponding to an = 0.05 was used (5.99, 7.81, 9.49 for df = 2, 3, and 4, respectively). In addition the minimum objective function, diagnostic goodness-of-fit plots were used for model building and selection.
|
Results
|
|---|
The two dose groups were not different with respect to gestational age, weight, age at surgery, baseline creatinine, total support time, bypass time, and circulatory arrest time (Table 1). Adverse events possibly or probably attributable to study drug were ventricular tachycardia (n = 1, 250 µg/kg bolus group) and hypotension (n = 1, 100 µg/kg bolus). Both events occurred in the CICU during continuous infusion. The end of infusion sample was not collected for patient 1.
All patients were successfully separated from CPB and begun on MUF. MUF was temporarily terminated in 2 patients from the 100 µg/kg bolus group and one patient from the 250 µg/kg bolus group secondary to the reinstitution of CPB for shunt revisions. No additional milrinone was administered during the second period of CPB. MUF was again performed on these patients after weaning from the second bypass period. Plasma concentrations were determined from the second MUF run for these 3 patients. The total volume of ultrafiltrate was collected to account for all milrinone elimination during MUF. The postoperative infusion was started at 125 ± 51 min after the bolus for these 3 patients compared with 93 ± 13 min for the remaining 13 patients.
Postoperatively, the clinical team caring for these patients increased the milrinone infusion in 2 patients from the 250 µg/kg bolus group and 4 patients in the 100 µg/kg bolus group to a maximal dose rate of 1 µg · kg–1 · min–1. One patient from each of the 100 and 250 µg/kg bolus groups had the dose decreased in the postoperative period.
Milrinone plasma concentrations at specific time points post-bolus are shown in Table 2. A bolus dose of 100 and 250 µg/kg resulted in peak plasma concentrations of 287 ± 34 ng/mL and 662 ± 129 ng/mL, respectively. Concentration-time profiles for the intraoperative data are represented in Figure 2A. All patients demonstrated an increase in plasma concentration during MUF. Concentration-time profiles for the intraoperative data are represented in Figure 2B. Patients included in this figure received a continuous infusion of 0.5 µg · kg–1 · min–1. This plot indicates that drug accumulation is observed over the first 12 h postoperatively. The urine output and serum creatinine for the first 5 postoperative days are shown in Figure 3.
View larger version (18K):
[in this window]
[in a new window]
|
Figure 2. A, Plasma concentrations during intraoperative period, including modified ultrafiltration (MUF). Squares represent the 100 µg/kg bolus group; triangles represent the 250 µg/kg bolus group. B, Plasma concentrations during postoperative period. Squares represent the 100 µg/kg bolus group; triangles represent the 250 µg/kg bolus group. Subjects included in this figure received a continuous infusion of 0.5 µg · kg–1 · min–1.
|
|
View larger version (16K):
[in this window]
[in a new window]
|
Figure 3. Overview of postoperative fluid balance and serum creatinine for the first 5 postoperative days. The dotted line represents urine output on each postoperative day. The solid line represents serum creatinine on each postoperative day.
|
|
Table 3 contains the chronology of the models explored during the population pharmacokinetic evaluation of milrinone. The final structural model from which the covariate evaluation was performed was a two-compartment model with MUF as a third compartment using natural log transformed data. This model allowed for CLINF12+h to dynamically change with time during this interval to eventually achieve the population-estimated steady-state CL. CLINF12+h was therefore derived by estimating CLss and Kss. Changes in CLINF12+h for each individual patient are represented in Figure 4. The final model did not fix parameter estimates for CLss or Kss. As the CL on CPB was assumed to be negligible, CLCBP was fixed at 0. The removal of the inter-individual variability estimate on CLMUF did not affect the objective function or the other parameter estimates. The removal of an inter-individual variability term on CLss increased the objective function by 10, but improved the standard error of the inter-individual variability estimate for Kss. Therefore, the base model did not include an inter-individual variability term on CLss or CLMUF (Table 3, model 90d). Individual and population estimates compared with measured plasma concentrations are represented in Figures 5 and 6, respectively. The structural model included an additive error model, with distinct parameters for error estimated for each interval. This model represented the model from which the covariates were added in a stepwise fashion. Final parameter estimates are represented in Table 4, with the respective standard errors of the individual parameter estimates. Final weight-normalized population estimates by interval are represented in Table 5.
View larger version (31K):
[in this window]
[in a new window]
|
Figure 4. Change in clearance in second postoperative phase by patient. Each panel represents an individual patient. The x-axis represents time in interval 4 (days); the y-axis represents the individual weight normalized clearance estimate.
|
|
Gestational age and weight were evaluated for correlation with VMUF, VINF0–12h, CLMUF, CLINF0–12h, and CLINF12+h. Positive correlations were found between weight and VINF0–12h and CLINF0–12h. Body weight was determined to be a significant covariate on the Vd during the continuous infusion and CL during the first 12 h of the continuous infusion. However, given the narrow weight range of the study subjects (2.4–3.96 kg), weight on VINF0–12h and CL NF0–12h was not added to the final model expression.
|
Discussion
|
|---|
After a loading dose of 100 µg/kg on CPB, plasma peak and trough milrinone concentrations are similar to those achieved with 50 µg/kg loading doses in other clinical settings (13). All patients in this study demonstrated a MUF concentrating effect that occurred despite a MUF drug CL of 3.3 mL · kg–1 · min–1. Because of impaired renal clearance during the immediate postoperative period, a standard infusion of 0.5 µg · kg–1 · min–1 resulted in drug accumulation during the initial 12 hours of drug administration. Immediately postoperatively, milrinone CL was significantly impaired (0.4 mL · kg–1 · min–1); it improved by the 12th postoperative hour and approached steady-state CL by postoperative day 4. The weight normalized population estimate for steady-state CL in this population was 2.6 mL · kg–1 · min–1, which is less than reported for older infants after cardiac surgery.
Conventional pharmacokinetic approaches are difficult to apply to the dynamic changes that occur in the HLHS setting. Although milrinone pharmacokinetics generally can be described by linear pharmacokinetic processes, the application of CPB and MUF negate the application of less stringent methodologies and assumptions (i.e., noncompartmental analysis and superposition principle). Specifically, both Vd and CL are changing dynamically, owing to the time-dependent application of vascular circuitry, fluid removal and filtration, and renal impairment. Complicating these fixed, nonrandom effects are patient-specific random effects that dictate the individual differences in patient response to pharmacotherapy. These effects are evident in the time to regain equilibrium in the pharmacokinetic ‘handling‘ of the drug. Hence, a mixed-effect modeling approach is critical to correctly define the intervals during which drug disposition behaves similarly, as well as to appropriately partition variance about the parameters that define such intervals. Benefits of these techniques included the ability to include all patients in the final analysis regardless of frequent changes in milrinone dosing, estimation of Vd and CL across intervals despite limited sampling, with a time-sensitive expression of changes in CL over the second postoperative phase, and a final estimate of CL at steady-state. In addition, the intra-individual variance was estimated for each interval and thus more appropriately reflects the sensitivity of sampling and model specification by interval. As a result of the small sample size, model validation was not feasible.
A recent study of pediatric patients less than 6 years of age who received milrinone as a slow loading dose followed by a constant-rate infusion after cardiac surgery used population modeling to describe milrinone’s pharmacokinetic properties (14). The pharmacokinetics were best described by a weight-normalized one-compartment model. The Vd 482 mL/kg was independent of age, whereas CL was linearly related to age according to the expression, CL = 2.42 mL · kg–1 · min–1 [1 + 0.396*age]. A comparable two-compartment model had a central Vd of 66 mL/kg, peripheral Vd of 269 mL/kg, CL of 6.29 mL · kg–1 · min–1, and an inter-compartmental CL of 4.75 mL · kg–1 · min–1. The two-compartmental model for this study estimated a central Vd of 159 mL/kg and a peripheral Vd of 343 mL/kg in the postoperative setting. In pediatric patients who underwent biventricular cardiac surgery, the Vd and CL of milrinone was reported as 900 mL/kg and 3.8 mL · kg–1 · min–1 for infants and 700 mL/kg and 5.9 mL/kg/min for children, respectively (9). In this neonatal population, postoperative steady-state CL was estimated at 2.6 mL · kg–1 · min–1.
Our pilot study was designed to describe the drug disposition of milrinone in this patient population. Given the need to first characterize the pharmacokinetics of milrinone, the sample size and design of this trial were optimized to achieve pharmacokinetic end-points. The results of this study will be used to perform a larger study in which sampling, methodology, and sample size will be designed to achieve clinically relevant pharmacodynamic end-points. This study was therefore not powered to compare hemodynamic variables between dosing groups, and did not control for hemodynamic confounders such as vasoactive infusions, fluid balance, and sedation. In addition, we had hypothesized that the Vd while on CPB would be larger than that in the postoperative setting. However, our study estimated similar Vd for these two settings. This may be a result of potentially influential covariates that were not measured or included in the model (e.g., protein binding).
Current labeling recommends a milrinone loading dose of 50 µg/kg and a continuous infusion of 0.375–0.75 µg · kg–1 · min–1. These doses achieve milrinone concentrations of approximately 200 ng/mL, a concentration that is considered to be therapeutic in other patient settings (15). In our study, as renal functionimproved, milrinone CL improved, approaching previously reported CL determined in infants by postoperative day 4. Based on CL data, infusion rates of 0.2 µg · kg–1 · min–1 during the first day would be estimated to result in steady-state plasma concentrations of 200 ng/mL. However, the therapeutic steady-state concentrations for this patient population have not been defined. Therefore, infusions should be titrated to effect, with dose increases to be expected as renal function improves.
In summary, for neonates undergoing stage I reconstruction of HLHS, an initial loading dose of 100 µg/kg on CPB resulted in plasma concentrations similar to those observed in other therapeutic settings. The net effect of MUF is to increase plasma milrinone concentrations by approximately 35%. Assuming that renal CL is minimal during this time, it is possible that MUF provides both hemoconcentration and a ‘second bolus effect‘ because blood that was returned to the patients from the venous reservoir contained milrinone. This is in contrast to adults on CPB who did not undergo MUF in whom drug concentrations decreased over time (13). Postoperative renal dysfunction, in some cases severe, is common but transient after stage I reconstruction of HLHS. In the postoperative setting of markedly impaired renal function, an infusion rate of 0.2 µg · kg–1 · min–1 should be considered, based on clinical needs. The population model reported here may result in a more rational pharmacologic dosing of milrinone in this specialized pediatric subpopulation and can provide the basis for a future study that explores the relationship between drug exposures and drug effect.
The authors would like to thank Robert J. Mullen PharmD, James Bailey MD, Marc Gastonguay PhD, Peter Moate and Raymond C. Boston PhD for their assistance with the conduct and analysis of this study.
|
Footnotes
|
|---|
Supported, in part, by an unrestricted educational grant from Sanofi Synthelabo and the Pediatric Pharmacology Research Unit grant number U01 HD37255-02.
Accepted for publication October 28, 2005.
|
References
|
|---|
- Norwood WI, Lang P, Hansen DD. Physiologic repair of aortic atresia-hypoplastic left heart syndrome. N Engl J Med 1983;308:23–6.[Web of Science][Medline]
- Ikle L, Hale K, Fashaw L, et al. Developmental outcome of patients with hypoplastic left heart syndrome treated with heart transplantation. J Pediatr 2003;142:20–5.[Web of Science][Medline]
- Chang RK, Chen AY, Klitzner TS. Clinical management of infants with hypoplastic left heart syndrome in the United States, 1988–1997. Pediatrics 2002;110:292–8.[Abstract/Free Full Text]
- Mahle WT, Wernovsky G. Long-term developmental outcome of children with complex congenital heart disease. Clin Perinatol 2001;28:235–47.[Web of Science][Medline]
- Tweddell JS, Hoffman GM, Mussatto KA, et al. Improved survival of patients undergoing palliation of hypoplastic left heart syndrome: lessons learned from 115 consecutive patients. Circulation 2002;106:I82–9.
- Gaynor JW, Kuypers M, van Rossem M, et al. Haemodynamic changes during modified ultrafiltration immediately following the first stage of the Norwood reconstruction. Cardiol Young 2005;15:4–7.[Medline]
- Shipley JB, Tolman D, Hastillo A, Hess ML. Milrinone: basic and clinical pharmacology and acute and chronic management. Am J Med Sci 1996;311:286–91.[Web of Science][Medline]
- Young RA, Ward A. Milrinone: a preliminary review of its pharmacological properties and therapeutic use. Drugs 1988;36:158–92.[Medline]
- Ramamoorthy C, Anderson GD, Williams GD, Lynn AM. Pharmacokinetics and side effects of milrinone in infants and children after open heart surgery. Anesth Analg 1998;86:283–9.[Abstract]
- Hoffman TM, Wernovsky G, Atz AM, et al. Efficacy and safety of milrinone in preventing low cardiac output syndrome in infants and children after corrective surgery for congenital heart disease. Circulation 2003;107:996–1002.[Abstract/Free Full Text]
- Boldt J, Brenner T, Lehmann A, et al. Is kidney function altered by the duration of cardiopulmonary bypass? Ann Thorac Surg 2003;75:906–12.[Abstract/Free Full Text]
- Edelson J, Koss RF, Baker JF, Park GB. High-performance liquid chromatographic analysis of milrinone in plasma and urine: intravenous pharmacokinetics in the dog. J Chromatogr 1983;276:456–62.[Medline]
- Bailey JM, Levy JH, Kikura M, et al. Pharmacokinetics of intravenous milrinone in patients undergoing cardiac surgery. Anesthesiology 1994;81:616–22.[Web of Science][Medline]
- Bailey JM, Hoffman TM, Wessel DL, et al. A population pharmacokinetic analysis of milrinone in pediatric patients after cardiac surgery. J Pharmacokinet Pharmacodyn 2004;31:43–59.[Web of Science][Medline]
- Bailey JM, Miller BE, Lu W, et al. The pharmacokinetics of milrinone in pediatric patients after cardiac surgery. Anesthesiology 1999;90:1012–8.[Web of Science][Medline]
Top
Abstract
Introduction
