Clearance of Amlodipine, Fentanyl, Fluconazole, Methylprednisolone, and Midazolam by Continuous Renal Replacement Circuits

Abstract

Critically ill pediatric patients on continuous renal replacement therapy (CRRT) have high mortality rates ranging from 30% to 70%, due in part to altered drug exposure from drug-CRRT circuit interactions. Drug loss within CRRT circuits can occur through both clearance by the hemofilter and adsorption to circuit components. Although these interactions are known to exist, their impact on the pharmacokinetics of most drugs is unknown, resulting in limited drug dosing guidance and increased risk for suboptimal drug exposure. In this study, we administered amlodipine, fentanyl, fluconazole, methylprednisolone, and midazolam individually and in combination with ex vivo, closed-loop, blood-primed CRRT circuits to quantify drug-circuit interactions. Circuits were dosed to drug-specific therapeutic concentrations, and drug concentrations in both plasma and effluent were measured over time. For all drugs administered individually, variable extraction by the CRRT circuit was observed (mean plasma recovery 0.4–49%). For drugs coadministered into a circuit, significant decreases in extraction and increases in drug recovery (2.5–109%) were found, suggesting dosing adjustments may be needed. This study highlights the need for additional studies of drug coadministration within CRRT circuits to describe complex drug-circuit and drug-drug interactions to provide dosing guidance in pediatric CRRT patients.

Continuous renal replacement therapy (CRRT) is a type of dialysis used in critically ill patients to replace kidney function during periods of acute renal failure. Although CRRT can be lifesaving, mortality rates within this population of patients remain high, averaging between 30% and 70%.¹ This high mortality may be due in part to altered drug exposure from drug interactions with the CRRT circuit. There are two main drug-circuit interactions that can alter the pharmacokinetics (PK) of drugs and lead to drug loss: 1) adsorption of drug to circuit components; and 2) clearance of drug through the hemofilter membrane via convective transport or diffusion.^2,3 The extent of this drug loss depends on the circuit component materials (e.g., tubing, filter type, and manufacturer), the drug’s physicochemical properties, and CRRT prescription.^2–8 Lipophilic and protein-bound drugs are typically adsorbed by circuit components and minimally cleared; in contrast, hydrophilic drugs with low protein bounding are minimally adsorbed and readily cleared.^2,3

Although it is known that drug PK is altered in patients supported with CRRT, the drug-circuit interactions of most drugs have not been quantified, resulting in limited CRRT-specific dosing guidance and increased risk for suboptimal drug exposure.⁴ Further complexities in drug-circuit interactions can arise from polypharmacy (i.e., drug-drug-circuit interactions). It is unclear if drugs can interact within the circuit through competition for albumin binding sites, saturation of adsorption, or clearance processes. Ex vivo studies have been used to identify the role of circuit components in drug loss and predict drug dosing changes for CRRT.^5–8 As the ex vivo circuit constitutes a closed loop lacking metabolic or elimination processes, drug loss results from circuit clearance, adsorption, or drug degradation.

To quantify drug extraction, we administered five drugs that are commonly used in pediatric CRRT patients (both individually and in combination) to ex vivo, closed loop, blood primed CRRT circuits.⁹ Study drugs were chosen based on their frequency of use, lack of dosing guidance, and potential for drug-drug interactions.⁹ We hypothesized that drugs with low protein binding and lipophilicity would be readily cleared by the CRRT circuits. The results of this study will improve the understanding of drug-circuit interactions and aid in optimizing drug dosing for these commonly used medications with a goal of identifying correlations between drug properties and filter clearance.

Materials and Methods

Continuous Renal Replacement Therapy Circuit Set Up

All ex vivo experiments were run on a PrisMax System (Baxter Healthcare, Deerfield, IL) using a TherMax heater (Baxter Healthcare) set to 37°C and an HF1000 filter set (Baxter Healthcare). The HF1000 filter sets were sterilized using ethylene oxide and comprised of a polyarylethersulfone hollow fiber dialysis membrane, and polyvinyl chloride (PVC) tubing for a total of 1.1 m² surface area. Circuits were connected in a closed loop to a 500 ml EXACTAMIX EVA bag (Baxter Healthcare) reservoir representing the patient (Figure 1). The reservoir was placed on an orbital shaker and mixed at 80 rpm throughout the duration of the ex vivo experiments. All drug circuits were run in triplicate using the continuous venovenous hemodiafiltration (CVVHDF) modality with PrismaSATE 4/2.5 Dialysis Solution (Baxter Healthcare). The CRRT prescription reflecting a 10 kg child was: blood flow rate 80 ml/min, pre blood pump replacement fluid rate 300 ml/hour, patient fluid removal 0 ml/hour, dialysate flow rate 400 ml/hour, and post-replacement fluid flow rate 100 ml/hour.

Figure 1.

CRRT circuit configuration. Drug was dosed at location A and samples were collected at B (plasma) and C (effluent). CRRT, continuous renal replacement therapy.

Circuits were primed with ~150 ml Plasma-Lyte A, one unit of human red blood cells (adenine saline added leukocytes reduced [~300 ml]), 125 ml thawed human plasma frozen within 24 hours after phlebotomy, 350 units heparin sulfate, 7 mEq sodium bicarbonate, 180 mg calcium gluconate, 6.25 g albumin 25%, and 1.6 g Tris(hydroxymethyl)aminomethane (THAM; Fisher Scientific, Pittsburgh, PA). Before dosing and at every hourly sample collection, pH was measured using an i-STAT 1 Analyzer (300-G; Flextronics Manufacturing, Singapore) and EG6+ cartridge (03P77–25; Abbott Point of Care, Abbott Park, IL) to target physiological pH (7.2–7.4) throughout the duration of the experiment.

Controls

A control sample was paired with each circuit replicate to quantify drug degradation. For individually dosed circuits, approximately 50 ml of circulating prime was removed after physiological pH had been verified, but before drug dosing. The prime fluid was then transferred to a 50 ml PVC plastic tube (229426; CELLTREAT, Pepperell, MA). For coadministered controls, the total circuit prime volume was doubled to permit five samples per time point, gently mixed, and divided into two 500 ml EXACTAMIX EVA bags. One bag was connected to the circuit as above and the other used for the control. The coadministered circuit control was placed in a 37°C water bath and physiological pH was verified before dosing. Controls were kept in a preheated 37°C water bath after dosing and for the remainder of the experiment.

Drug Dosing

Study drugs were provided by the University of Utah Hospital pharmacy. Each drug was dosed through an arterial port directly below the reservoir bag to achieve therapeutic concentration within the circuit as follows: amlodipine 10 ng/ml, fentanyl 2 ng/ml, fluconazole 20 mg/L, methylprednisolone 0.66 mg/L, and midazolam 0.15 mg/L (Figure 1).^10–22 Coadministered drugs were injected into the circuit in the following order: fluconazole, methylprednisolone, amlodipine, midazolam and fentanyl. Total time for administration of coadministered drugs was between 23 and 27 seconds. Dosing was completed at time = 0. Control samples were dosed directly to the 50 ml tube or EXACTAMIX EVA bag to achieve the same target concentration at time = −5 minutes and rotated for 5 minutes before being returned to the 37°C water bath. Before sample collection, the control was mixed by gently inverting the container five times.

Blood samples of 1.5 ml for individual or 5 ml for coadministered drug circuits were collected from both the circuit and a paired control at the following times: 1, 5, 15, 30 minutes, and 1, 2, 3, 4, 5, 6 hours. Sample collection volume consisted of 3% for individual and 10% for coadministered circuits over the entirety of the ex vivo experiments. Coadministered samples were then split equally into five microcentrifuge tubes (one tube per drug). Before sample collection, approximately 1 ml of blood was drawn as waste before being returned to the circuit following sample collection. Blood samples were immediately centrifuged at 3,000 g for 10 minutes at 4°C. Plasma was then aliquoted and placed in cryovials (Fisher Scientific). A paired effluent sample was collected from directly above the effluent bag at all time points, and immediately transferred to cryovial(s) (Fisher Scientific). All samples were stored at −20°C for no longer than 8 hours before being transferred to −80°C for long-term storage.

Drug Assays

Assays to measure plasma and effluent drug concentrations were developed according to U.S. Food and Drug Administration guidance.²³ Plasma and effluent samples for midazolam, fentanyl, and fluconazole were analyzed at OpAns Laboratory (Durham, NC) using HPLC/MS-MS. The lower limit of quantification was 100 pg/ml, 0.02 ng/ml, and 10 ng/ml with calibration curve range of 0.1–200, 0.02–50, and 10–10,000 ng/ml for midazolam, fentanyl, and fluconazole, respectively. Intra- and inter-day precision (%CV) ranged from 2.26 to 8.48 and 3.47 to 10.5 for midazolam, 0.6 to 17.1 and 4.3 to 10.5 for fentanyl, and 0.7 to 8.5 and 2.9 to 6.8 for fluconazole. Because of the shape of the calibration curve, fentanyl samples were assessed using a modified regression model incorporating a power function. Plasma and effluent samples for amlodipine and methylprednisolone were analyzed at the Center for Human Toxicology at the University of Utah. The lower limit of quantification was 0.5 and 1 ng/ml with calibration curve ranges of 0.5–50 and 1–1,000 ng/ml for amlodipine and methylprednisolone, respectively. Intra- and inter-day precision (%CV) ranged from 1.6 to 2.2 and 1.6 to 2.2 for amlodipine, and 1.6–11.9 and 3.3–10.8 for methylprednisolone (Table 1, Supplemental Digital Content, http://links.lww.com/ASAIO/B463).

Drug Recovery and Elimination

Drug recovery for circuit and control plasma samples was calculated using the following equation:

$$\mathrm{Recovery}(\%)=\frac{C_t}{C_i}\times 100$$ (1)

where C_t is the concentration at time t and C_i is the initial concentration measured at time = 5 minutes.

Paired circuit plasma and effluent samples from time = 15 minutes to time = 6 hours were used to calculate saturation coefficients and transmembrane clearances for all drugs. The following equations were used:

$$SA=\frac{C_{eff}}{C_p}$$ (2)

$$CL=Q_{\text{eff}}\times SA$$ (3)

where SA is the saturation coefficient, C_eff is the effluent concentration, C_P is the plasma concentration, CL is the transmembrane clearance, Q_eff is the effluent flow rate, respectively. Because of a predominance of concentrations below the limit of quantification (BLQ) in amlodipine effluent samples, the lower limit of quantification (0.5 ng/ml) was used for SA calculations.

Statistical Analysis

Mean percent recovery from paired circuit and control samples were compared at the latest sampling time using a two-sample t-test. Linear mixed effects models were estimated for each drug to test for differences in transmembrane clearance over the experimental time period with circuit number used as a random effect to account for between circuit variability. Fixed effects included time, circuit type (single drug versus coadministered) and the interaction between circuit type and time. This model was used to test differences in transmembrane clearance between single drug and coadministered circuits from 15 minutes to 6 hours.

Results

Physiologic pH was maintained in all circuits except individual amlodipine circuit 1. In this circuit, pH increased above 7.4 at 2 hours and remained high reaching a peak of 7.67 at 6 hours. Fluconazole control samples were not collected at 6 hours; instead, samples collected at hour 8 were used to calculate percent recovery (n = 3). A system failure occurred 1 hour into coadministered circuit 2, therefore a fourth coadministered circuit was run to replace circuit 2. Plasma and effluent concentrations for all drugs can be found in Table 2, Supplemental Digital Content, http://links.lww.com/ASAIO/B464, and Table 3, Supplemental Digital Content, http://links.lww.com/ASAIO/B465.

Single Drug Circuits

Fluconazole was rapidly extracted by the CRRT circuit with mean percent recovery of 0.4% in the circuits at 6 hours and 102% in the controls (p < 0.001) (Figure 2, Figure 1, Supplemental Digital Content, http://links.lww.com/ASAIO/B461, Figure 2, Supplemental Digital Content, http://links.lww.com/ASAIO/B462). Fluconazole was freely cleared by the hemofilter with a mean saturation coefficient and transmembrane clearance of 0.979 and 750 ml/hour, respectively (Table 1). Midazolam mean percent recovery was 48% in the circuits and 90% in the controls at 6 hours (p = 0.006). The saturation coefficient (0.046) and transmembrane clearance (0.611 ml/minute) of midazolam were reflective of minimal hemofilter clearance. Methylprednisolone mean percent recovery was 11.4% in the circuits compared to 113% in controls at 6 hours (p = 0.009). The saturation coefficient and clearance of methylprednisolone were 0.35 and 278.76 ml/hour, respectively. The mean plasma recovery of fentanyl was 22% in the circuits and 96% in controls at 6 hours (p = 0.004). The fentanyl saturation coefficient was 0.33 with a clearance rate of 266.41 ml/hour. At 6 hours, amlodipine mean percent recovery was 49% in the circuits and 109% in the controls at 6 hours (p = 0.001). The saturation coefficient of amlodipine was 0.140, and transmembrane clearance was 112 ml/hour.

Figure 2.

Mean plasma percent recovery for single and coadministered circuits amlodipine (A, B), fentanyl (C, D), fluconazole (E, F), methylprednisolone (G, H), and midazolam (I, J). Error bars represent standard deviation across replicates. Circuit recovery is denoted in blue (light blue single circuit recovery, dark blue coadministered circuit recovery) and control in red. *A 6 hour fluconazole control sample was not collected. The 8 hour control sample was used to calculate percent recovery with a mean plasma recovery of 102% (98–110%).

Table 1.

Saturation Coefficient (SA) and Transmembrane Clearance (CL) for All Drugs

	Single Circuits		Coadministered Circuits		ΔCL
Drug	SA	CL (ml/Hour)	SA	CL (ml/Hour)	ΔCL (ml/Hour)
Amlodipine	0.23 (0.050)	182 (40.3)	0.18 (0.080)	148 (68.7)	−14.99***
Fentanyl	0.33 (0.070)	266 (59.3)	0.26 (0.060)	210 (49.8)	−21.64**
Fluconazole	0.98 (0.010)	782 (5.82)	0.98 (0.010)	780 (9.51)	−16.62*
Methylprednisolone	0.35 (0.010)	278 (10.2)	0.34 (0.030)	273 (26.3)	+0.8619
Midazolam	0.049 (0.005)	39.0 (4.17)	0.034 (0.000)	27.3 (3.87)	+1.067

Δ CL is the change in clearance between single and coadministered circuits.

^*p < 0.05.

^**p < 0.01.

^***p < 0.001.

Coadministered Circuits

The clearance of all drugs was lower in coadministered circuits (Figure 2, Table 1, Figure 1, Supplemental Digital Content, http://links.lww.com/ASAIO/B461). The decrease in transmembrane clearance was statistically significant for amlodipine, fentanyl, and fluconazole over the experimental time period (Table 1). A corresponding increase in percent recovery was observed for all drugs except fentanyl (Table 2). Amlodipine and midazolam had the most substantial increase in recovery of 60.6% and 15.1% at 6 hours, respectively.

Table 2.

Percent Recovery at 6 Hours

Drug	Single Circuit Recovery (%)	Coadministered Circuit Recovery (%)	Δ Recovery (% Change)
Amlodipine	48.7 (15.5)	109 (23.0)	+60.6
Fentanyl	21.9 (1.17)	20.5 (6.64)	−1.38
Fluconazole	0.370 (0.340)	2.46 (3.88)	+2.09
Methylprednisolone	11.4 (1.00)	12.8 (1.28)	+1.40
Midazolam	48.4 (0.100)	63.5 (4.40)	+15.1

Δ Recovery is the change in recovery between single and coadministered circuits.

Discussion

Interactions between drugs and CRRT circuits can occur through adsorption of drug to circuit components or clearance via the hemofilter membrane. Physicochemical properties can be predictive of a drug’s interactive behavior within CRRT circuits, but are not conclusive.^2,3 Ex vivo experiments provide an approach to quantitively determine drug extraction through drug-circuit interactions. In this study, we examined individual and coadministered drug-circuit interactions using an ex vivo approach and quantified the extraction of five drugs commonly administered to pediatric CRRT patients (Table 3).

Table 3.

Physicochemical Properties of Drugs

Drug	Type	LogP	Plasma Protein Binding (%)	Molecular Weight (g/mol)
Amlodipine	Calcium channel blocker	3.0024	9325	40924
Fentanyl	Synthetic opioid	4.0524	80–8524	33624
Fluconazole	Antifungal	0.5026	1227	30624
Methylprednisolone sodium succinate	Corticosteroid	2.6024	7728	49624
Midazolam	Benzodiazepine	3.8929	9724	32524

Of the five study drugs, fluconazole was extracted to the greatest extent with essentially 0% recovery after 6 hours. Low protein binding (12%) and lipophilicity (logP 0.4) are predictive of minimal adsorption and high hemofilter clearance. The high saturation coefficient, transmembrane clearance, and lack of drug degradation within control samples support this conclusion. High clearance rates and similar saturation coefficients were also found in an adult ex vivo study.⁷ This study used a crystalloid mixture and therefore did not account for plasma protein binding or red blood cell interactions. The similarities in saturation coefficients may be caused by the low plasma protein binding of fluconazole. Patient studies have shown differences in clearance depending on flow rates and CRRT modality.^10–12 Regardless of flow rate or modality, fluconazole was readily cleared.^10–12 The decreased fluconazole concentrations resulting from high clearance during CRRT are likely sufficient to prevent patients from reaching therapeutic concentrations. Indeed, one study found that adult hemodialysis patients taking up to 800 mg of fluconazole daily were unable to reach appropriate trough concentrations.¹¹

Similar to fluconazole, methylprednisolone was extensively extracted by the circuit (11% recovery at 6 hours). Methylprednisolone extraction was caused by a combination of adsorption and hemofilter clearance. The saturation coefficient and transmembrane clearance suggest that methylprednisolone is not substantially cleared by the hemofilter. Based on methylprednisolone’s protein binding (77%), lipophilicity (logP 2.60), and lack of degradation in control samples, the remainder of drug loss can be attributed to adsorption. Studies investigating interactions between CRRT and methylprednisolone are scarce. One study from 1977 measured the clearance of methylprednisolone in adult intermittent hemodialysis patients.¹³ The high flow rates and antiquated dialysis equipment render this data inapplicable to current pediatric CRRT patients.

Similar to methylprednisolone, fentanyl extraction can be attributed to both adsorption and hemofilter clearance. The high lipophilicity (logP 4.05) and protein binding (77%) are predictive of adsorption and minimal hemofilter clearance. In adult patient studies, fentanyl was not found in the effluent for intermittent and continuous venovenous hemodialysis, and no change was detected in blood concentrations before, during, and after hemodialysis.^14–17 Conversely, a pediatric ex vivo study using extracorporeal membrane oxygenation circuits found high rates of adsorption.¹⁸ There is some evidence that certain filter types are more prone to absorption and may trap fentanyl, thus reducing blood concentrations.^15,16 Our ex vivo studies indicate that fentanyl is minimally cleared by the hemofilter; with loss in drug concentration most likely caused by adsorption. Fentanyl is usually administered as a continuous infusion with additional bolus dosing as needed. This provides the potential for saturation of circuit surfaces because of the high adsorption of fentanyl. It is not known if saturation is possible or how long this process would take to occur. Future ex vivo studies using continuous infusions of fentanyl are needed to determine if saturation occurs and the potential effects of circuit saturation on drug concentrations and dosing.

In contrast to fentanyl, the loss of midazolam within the circuit was predominantly caused by adsorption. Although the ex vivo experiments presented in this study found limited midazolam clearance via the hemofilter, there are conflicting data in patients.^19–22 Discrepancies may be due to flow rates. Studies conducted in patients with lower CRRT prescription flow rates were associated with no elimination. The disease state of the patient could also affect clearance of midazolam. In critically ill patients, low albumin levels can increase the ratio of unbound drug and subsequently, clearance of the drug through the hemofilter. In studies where some clearance was noted, calculated sieving coefficients and clearance rates were comparable to those found in this study.^20,21 Midazolam, like fentanyl, is often dosed as a continuous infusion with boluses as needed to achieve sedation. It is unclear if saturation of circuit surfaces is possible or could affect the steady-state concentration of midazolam. Ex vivo experiments are needed to determine the clearance and adsorption of midazolam during continuous infusion.

The mean plasma recovery of amlodipine was similar to midazolam, however, hemofilter clearance was higher. Drug loss in the CRRT circuit was likely caused by both adsorption and filter clearance. The protein binding and lipophilicity of amlodipine is lower than midazolam. Published studies on amlodipine PK in CRRT patients are lacking. A study in intermittent hemodialysis adult patients found low levels of amlodipine in effluent reaching 1.3% of patient serum concentrations.³⁰ Because of the high protein binding of amlodipine, it is unlikely this drug would be dialyzable. In contrast, our study found amlodipine effluent concentrations to be around 12% of plasma concentrations, suggesting this drug is minimally dialyzable when dosed individually. Most effluent samples were BLQ and more studies are needed to determine the hemofilter clearance of amlodipine.

Alterations in clearance and percent recovery can occur because of changes in plasma protein binding, electrostatic interactions, direct drug-drug interactions, and adsorption. This study provides evidence for differential drug-CRRT circuit interactions when drugs are coadministered compared to individually dosed circuits. Our ex vivo experiments identified changes in percent recovery and clearance between single and coadministered circuits. Most drugs within the coadministered circuit had decreased clearance to some degree which could be suggestive of reduced flow across the filter membrane. However, these decreases were not consistent or correlated with physicochemical properties such as protein binding and lipophilicity. Protein binding can influence clearance and adsorption of drugs. Both amlodipine and midazolam can initiate conformational changes in albumin that can displace other drugs, such as fentanyl.^25,31,32 Conformational changes can also alter binding site accessibility, drug binding affinity, and speed of complex formation. As we did not find increased drug clearance in coadministered circuits, alterations in protein binding are unlikely. Direct drug-drug interactions could also result in changes to protein binding and other drug-circuit interactions.

Changes in adsorption and the potential for saturation are the most likely explanation for differences observed between single and coadministered circuits. Amlodipine, midazolam, and fentanyl are highly lipophilic drugs with moderate to substantial protein binding, as demonstrated by extensive adsorption in single circuits. Fentanyl is the most lipophilic and is positively charged at physiologic pH. Midazolam is neutral and amlodipine is slightly positive.^33,34 Although the HF1000 filter membrane is neutral, the tubing and PVC components are polarized, resulting in partial negative charges.³⁵ These electrostatic interactions could account for fentanyl’s enhanced binding affinity. Increased percent recovery of midazolam and amlodipine could be caused by competition with fentanyl and midazolam on circuit component surfaces. Polyvinylpyrrolidone used during PAES membrane preparation to increase hydrophilicity can leach out over the duration of treatment.³⁶ Leaching leads to increased hydrophobicity within the membrane and could increase binding of highly lipophilic drugs such as fentanyl over time. The ability of drugs to saturate extracorporeal life support circuits, such as CRRT, and effect of changes in membrane surfaces over time is unclear. Ex vivo studies including multiple doses or continuous infusions would be informative. The changes observed in coadministered circuits may be products of potentially confounding and mixed effects.

There were several limitations to this study. Ex vivo experiments were conducted using a single circuit configuration (eg, one filter, modality, prescription). Modality, flow rate, and circuit component materials can significantly affect the rate of clearance.^4–6 Although, a recent ex vivo study found that the clearance rate of cefepime, a small molecule drug was not dependent on modality.⁸ All drugs in our study were small molecule, with varied physicochemical properties. Ex vivo experiments with varying circuit configurations and prescriptions will be needed to fully elucidate these interactions. Future studies should include supratherapeutic concentrations and pairwise drug dosing to identify the potential for saturation and any effect of drug-drug interactions on adsorption. Although albumin levels were representative of children supported with CRRT, we could not account for impaired albumin binding caused by accumulation of fatty acids and uremic toxins that occurs during kidney dysfunction.^37,38 Uremic toxins are cleared by CRRT, however fatty acids are not. Future studies should include varying levels of albumin to determine the effect on drug hemofilter clearance. Despite these limitations, this study provides quantified drug-circuit interactions for five drugs of varying physicochemical properties that can be used to aid dosing in pediatric CRRT patients.

Although informative, these ex vivo results cannot be directly translated into patient dosing. Physiologically based pharmacokinetic (PBPK) models can be used to translate these data into dosing recommendations. These models are mathematical tools that integrate drug-specific and population-specific information to predict the effect of different factors (eg, age and critical illness) on drug exposure.^39,40 A virtual compartment can be built to represent the CRRT circuit and parameterized using ex vivo data. This compartment can then be linked to a full PBPK model to predict drug concentrations in children supported with CRRT.

Conclusions

We examined the effect of CRRT on the clearance and adsorption of five commonly used drugs in pediatric patients supported with CRRT. All five drugs were cleared to some extent by the CRRT circuit when administered individually; however, significant decreases in clearance were found when drugs were coadministered into a single circuit. These findings highlight how both drug-circuit and drug-drug-circuit interactions can alter PK during CRRT.

Supplementary Material

Supplemental digital content is available for this article. Direct URL citations appear in the printed text, and links to the digital files are provided in the HTML and PDF versions of this article on the journal’s Web site (www.asaiojournal.com).