Introduction
The kidneys have a crucial role in water and electrolyte homeostasis, acid–base balance, excretion of nitrogenous waste products and select endocrine functions. In terms of drug disposition, renal excretion of parent compounds or active metabolites represents a key route of drug elimination for many agents. This process involves a combination of glomerular filtration, renal tubular secretion and renal tubular reabsorption. Net renal drug elimination represents the sum of these processes.
Recent research has highlighted that renal antibacterial drug elimination (particularly renally cleared β-lactams) can be significantly elevated in the critically ill [
1], such that drug exposure is often suboptimal [
1]-[
3], and may lead to adverse clinical outcomes [
4]. This phenomenon has been termed augmented renal clearance (ARC) [
1], and is seen in association with an elevated urinary creatinine clearance (CL
CR) [
5]. Such measures have therefore been advocated as a useful tool to identify ARC in the critically ill [
6],[
7], the assumption being that elevated glomerular filtration is a key mechanism in this process.
However, measured CL
CR suffers from the limitations inherent to using any endogenous filtration marker, such that variations in diet, muscle mass and physical activity may also influence plasma CR concentrations. In addition, CL
CR provides limited mechanistic information on changes in renal tubular function. As such, renal drug handling in the critically ill requires additional research, ideally using exogenous indicators of renal function. One method involves the administration of multiple renal markers [
8], such that the elimination of each can be used to quantify changes in glomerular filtration, and tubular drug handling, simultaneously. Using such a ‘cocktail’, Gross and colleagues applied this approach in 12 healthy male subjects [
9], establishing useful baseline values for comparison.
The aims of this prospective observational study were therefore (1) to assess the feasibility and safety of administering multiple markers of renal function simultaneously in the intensive care unit (ICU), (2) to quantify glomerular filtration and renal tubular function in a cohort of critically ill patients at risk of ARC and (3) to compare these data with previously reported values in healthy volunteers.
Material and methods
Setting
This single-centre observational study was undertaken in a tertiary-level, university-affiliated ICU. Ethical approval was obtained from our institution’s Human Research Ethics Committee (HREC/09/QRBW/15), with written informed consent obtained from either the patient or the nominated substitute decision-maker.
Study population
Study participants had to be aged ≤60 years, have an expected ICU length of stay >24 hours, evidence of a systemic inflammatory response syndrome [
10] in the 24 hours prior to marker administration, a plasma CR concentration <120 μmol/L and no history of chronic kidney disease (CKD) or renal replacement therapy. Patients were excluded from receiving any study markers if (1) either invasive haemodynamic monitoring or an indwelling urinary catheter (IDC) was not employed as part of standard management; (2) they were <18 years of age; (3) they were pregnant; (4) rhabdomyolysis was clinically suspected or the plasma creatinine kinase level was >5,000 IU/L; (5) they were in the ‘risk’ category or greater for acute kidney injury, as defined by the risk, injury, failure, loss, and end-stage (RIFLE) kidney disease criteria [
11]; (6) there was a documented allergy and/or contraindication to one or more of the renal markers; (7) one or more markers were being employed clinically; or (8) the treating clinician considered the patient unsuitable for enrolment. Recruitment was carried out by convenience sampling.
Data collection, dosing administration and sampling protocol
Demographic data, including age, sex, height, weight, Acute Physiology and Chronic Health Evaluation (APACHE) II score and admission diagnosis were recorded prospectively. After confirming eligibility and obtaining consent, the following study markers were administered; sinistrin (Inutest; Laevosan, Linz, Austria) 2,500 mg in 10 ml intravenously (IV) over 30 seconds, 4% p-aminohippuric acid sodium salt (PAH; CFM Oskar Tropitzsch, Marktredwitz, Germany) 440 mg IV over 1 minute, rac-pindolol (Barbloc; Alphapharm, Millers Point, NSW, Australia) 5 or 15 mg orally or via an enteral feeding tube and fluconazole (Diflucan; Pfizer Australia Pty Ltd, West Ryde, NSW, Australia) 100 mg orally or via an enteral feeding tube.
Enteral markers were administered only if the patient was documented to be tolerant of enteral nutrition. rac-Pindolol was withheld if there was (1) a history of severe reactive airway disease (asthma and/or chronic obstructive airway disease) or symptomatic bradyarrhythmia or (2) moderate vasopressor support (>10 μg/min infusion of noradrenaline or adrenaline) was being provided. Administration of all or a combination of markers was possible, depending on the patient profile.
Blood samples were then taken via the intra-arterial cannula at the following time points after marker administration: 5, 10, 15, 30, 60 and 120 minutes and 4, 6, 12 and 24 hours. Following collection, all blood samples were immediately placed on ice and centrifuged within 60 minutes at 3,000 rpm for 10 minutes. Plasma was then aliquoted off and stored for analysis at −80°C. All urine was collected via the IDC over the same 24-hour period, with a 10-ml sample stored at −80°C for later assays. The remaining urine was forwarded to the local hospital laboratory for biochemical analysis.
All marker administration was performed under medical supervision, with the patients receiving continuous cardiovascular and respiratory monitoring. All physiological parameters, fluid balance and any therapeutic interventions performed were recorded concurrently.
Bioanalysis
(
S)-pindolol, (
R)-pindolol, PAH and fluconazole in plasma and urine were measured by in-house liquid chromatography-tandem mass spectrometry methods validated in accordance with US Food and Drug Administration guidelines for bioanalysis. Separations were tailored for each analyte: a reverse-phase column was used for fluconazole; a hydrophilic interaction liquid chromatography column was used for PAH; and a chiral column was used for pindolol. Sinistrin was measured by using a commercially available enzyme-linked immunosorbent assay kit (FIT-GFR; BioPAL, Worcester, MA, USA). CR was measured in plasma and urine by using an isotope dilution mass spectrometry traceable assay through the institutional pathology laboratory. For comparison, estimated GFR (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula [
12].
Data treatment and statistical analysis
CL
CR was determined from the urinary CR concentration, 24-hour urine volume and plasma CR concentrations obtained from routine clinical testing. If multiple plasma values were available over the 24-hour study period, the mean was used in subsequent analysis. Plasma concentrations of all renal markers were plotted on a concentration/time graph, with the slope of the terminal portion calculated by linear regression. Suitability for use in further analysis was assessed by visual inspection. Backward extrapolation to time zero was performed for those markers administered IV (sinistrin and PAH). The area under the plasma concentration/time curve (AUC
0-∞) was determined using the linear trapezoidal rule and extrapolated to infinity by adding the product of the last measured plasma concentration multiplied by the terminal slope. As per previous recommendations, plasma concentrations to at least 120 minutes were required for accurate calculation of sinistrin AUC
0-∞ [
13]. Plasma CL for each marker was then calculated as dose/AUC
0-∞.
The total amount recovered in urine (Ae) of PAH, fluconazole and (
S)- and (
R)-pindolol was calculated by multiplying the urinary concentration by the 24-hour urine volume. Renal clearance (CL
R) was then calculated as Ae/AUC
0-∞, and non-renal clearance (CL
NR) as the difference of CL − CL
R. Sinistrin CL was taken as representing the GFR, and PAH CL
R was taken as effective renal plasma flow (ERPF). The filtration fraction was calculated as GFR/ERPF. Glomerular filtration of the unbound marker was calculated as
f u × GFR, where
f u is the unbound fraction in plasma. Tubular anion secretion was determined from the net tubular secretion of PAH as CL
R − (
f u × GFR). Tubular cation secretion was determined from the net tubular secretion of (
S)- and (
R)-pindolol as CL
R − (
f u × GFR). Tubular reabsorption was estimated from the net reabsorption of fluconazole as (
f u × GFR) − CL
R [
8]. The
f u of PAH was taken as 0.83 [
9]; for (
S)- and (
R)-pindolol, it was 0.45 [
14]; and for fluconazole, it was 0.83 [
9].
Continuous data are presented as the mean (95% confidence interval (CI)). Categorical data are presented as counts (%). For bivariate correlation between continuous variables, we used a Pearson correlation coefficient (r). A paired Student’s t-test was used to compare intra-patient data. Bland-Altman analysis was employed to explore the agreement between sinistrin CL, CLCR and CKD-EPI eGFR values. A double-sided P-value <0.05 was considered as statistically significant, and SPSS version 22 software (IBM, Armonk, NY, USA) was used for all analyses.
Discussion
In comparison with previous data reported in healthy volunteers, by using identical exogenous markers of renal function, we have demonstrated elevated glomerular filtration and renal tubular anion secretion in a selected cohort of critically ill patients at risk of ARC. As these mechanisms are central to β-lactam renal elimination, the potential for inadequate drug exposure is considerable in this setting [
7], an assertion consistent with recent literature [
2]-[
4],[
15].
Overall, the study protocol was well tolerated by all participants, with no adverse events observed. Furthermore, suitable concentration/time profiles could be generated in most instances, limited primarily by our strict criteria for determining sinistrin AUC
0-∞ [
13]. Sinistrin CL in our study was higher (see Table
3) than previously reported in non–critically ill subjects [
9],[
16],[
17]. In keeping with this, CL
CR measures were also elevated, with good correlation (
r =0.7) between these values.
PAH CL was also higher in our cohort, although mechanisms other than renal excretion of unchanged drug accounted for a significant fraction. This process reflects acetylation of PAH [
18], which has previously been reported to account for 15% to 30% of total drug elimination [
18]-[
20]. Fluconazole CL in the critically ill has been reported to be between 0.88 to 2.17 L/hr [
21]. The findings of the present study are consistent with these previous reports, although fluconazole CL
NR was significantly elevated in our study. This metabolic pathway involves glucuronide conjugation in the liver, a process that appears to decline in parallel with deteriorating renal function [
22]. Of note, total CL and CL
R of both enantiomers of pindolol were lower than previously reported in healthy volunteers [
14].
The reduced CL
R of
rac-pindolol and the suggestion of impaired renal cation secretion were unexpected findings. A likely explanation may involve variation in urinary pH, as previously described by Ujhelyi
et al. [
23]. In their study, (
S)- and (
R)-pindolol CL
R was determined in eight healthy male subjects before and after urinary acidification with ammonium chloride. At a mean urinary pH of 5.0, a twofold increase in CL
R of both isomers was noted [
23] and thought to be principally related to an increase in organic transporter activity. In our study, the mean (95% CI) urinary pH was 6.16 (5.75 to 6.57), considerably more alkaline than might have been expected and which may account for the reduced CL
R observed.
Another potential mechanism impairing
rac-pindolol tubular secretion may involve inhibition of the organic cation transport system, as has been demonstrated with co-administration of cimetidine [
24]. Of note, this is thought primarily to involve competitive inhibition of the luminal efflux process rather than basolateral uptake [
25]. Although none of the study participants were concurrently receiving cimetidine, other medications have been potentially implicated [
26], resulting in varying degrees of inhibition. As such, whereas co-administration of the study markers themselves do not generate any interactions [
9],[
27],[
28], a potential additional drug–drug interaction in our study cohort cannot be excluded.
The observed changes in CL
NR of both PAH and fluconazole are notable findings. Acetylation of PAH occurs in the liver and kidneys, with the metabolite (
N-acetyl-PAH) then excreted in urine [
29]. In contrast to previous work [
18]-[
20], plasma PAH CL was significantly elevated in our study, with CL
NR accounting for, on average, 48% of total PAH elimination. Similarly, fluconazole CL
NR was also elevated, a finding disparate from prior research in the critically ill [
30]. This may be a consequence of greater tubular reabsorption of fluconazole, a process driven by increased glomerular filtration. Notwithstanding this, as non-renal elimination of PAH and fluconazole primarily involve conjugation, these observations remain consistent with increased solute delivery to other drug eliminating organs, such as the liver. Whether augmented hepatic clearance also occurs in this setting remains uncertain, although, given the potentially significant impact of such changes on PK parameters, this represents an essential area for future study.
Elevated CL
CR has been reported in numerous subgroups, including traumatic brain injury [
31], polytrauma [
32],[
33], burns [
34], sepsis [
35], ventilator-associated pneumonia [
36], meningitis [
37] and major surgery [
38]. Lower illness severity scores [
39], younger age [
35], male sex [
4] and systemic inflammation [
40] are common features. Indeed, young male patients admitted post-trauma and without the requirement for vasopressor therapy formed the majority of our study cohort. Activation of renal reserve represents a possible explanation for these findings [
41], although data from our study reinforce that changes in GFR in this context are likely more complex. Specifically, for a given ERPF, the filtration fraction varied considerably, accounting for the increased sinistrin CL observed in some patients (Figure
3). These data indicate that changes in intra-glomerular filtration pressure (due to variable afferent and efferent arteriolar tone) are potentially a key mechanism warranting further in-depth study.
CKD-EPI eGFR values were significantly lower than CL
CR measures, reinforcing the limited utility of these estimates in such patients [
42]. CL
CR measures displayed less bias in comparison to sinistin CL, although precision was poor with either estimate (Figure
2). Nonetheless, low-normal plasma CR concentrations in young trauma patients without oliguria should alert the clinician to the possibility of ARC, following which a urinary CL
CR test could be used to reinforce this finding. Of note, the use of CL
CR as a surrogate outside a population at risk of ARC remains uncertain, particularly given the increasing inaccuracy of this measure with declining renal function [
43].
We wish to declare the following limitations. We chose to employ study markers previously validated in healthy volunteers, allowing for direct comparison with a non–critically ill cohort. Our use of enteral pindolol and fluconazole mirror the approach used by Gross
et al. [
9], although gastrointestinal absorption in the critically ill is often more variable. Importantly, both fluconazole [
44] and pindolol [
45] have excellent oral bioavailability, and none of the study participants receiving these markers were intolerant of enteral nutrition.
Our sampling interval was also limited to 24 hours, as opposed to longer periods that have been used to characterize fluconazole pharmacokinetics [
46]. However, fluconazole CL
R over this shorter period is comparable with that based on sampling for 120 hours [
9], allowing for a shorter study period. Finally, we chose to measure ERPF by using a PAH bolus IV approach rather than constant infusion. Although there has been some criticism of this technique [
47], peak concentrations were substantially lower than those known to saturate anion transport in the kidney [
19], such that our analyses are likely to provide a reliable measure of PAH CL
R.
Competing interests
JL has acted as a consultant to AstraZeneca and Janssen-Cilag; he has received honoraria from AstraZeneca, Janssen-Cilag and Wyeth Australia; and he was involved with the Bayer European Society of Intensive Care Medicine Advisory Board. JAR has previously consulted for Janssen-Cilag, AstraZeneca, Pfizer and Gilead; he has been involved in advisory boards for Janssen-Cilag and AstraZeneca; and he has received unrestricted grants from Janssen-Cilag, AstraZeneca and Novartis. AAU has received an honorarium and travel assistance from Pfizer. AstraZeneca and Edwards Lifesciences provide an annual unrestricted donation to the Burns, Trauma and Critical Care Research Centre at The University of Queensland. The remaining authors declare that they have no potential conflicts of interest in relation to this article.
Authors’ contributions
AAU and JL conceived of the study. AAU, PJ, JAR, MLS and TS were involved in protocol development, ethical approval and implementation. AAU, PJ, JS, MLS, TS and RD collected the data. SCW completed the laboratory analysis and quality assurance. AAU and JAR provided the statistical and pharmacokinetic analysis. AAU wrote the initial manuscript draft, with all of the remaining authors contributing to subsequent revisions. AAU takes responsibility for archiving the data and guarantees the integrity of the paper from inception to publication. All of the authors read and approved the final manuscript.