Pursuing the Real Vancomycin Clearance during Continuous Renal Replacement Therapy in Intensive Care Unit Patients: Is There Adequate Target Attainment?

Introduction: Vancomycin is used in intensive care unit (ICU) patients for the treatment of infections caused by gram-positive bacteria. The vancomycin pharmacokinetic/pharmacodynamic index is a ratio of the area under the concentration to the minimum inhibitory concentration ≥400–600 h*mg/L. This target can generally be achieved by a plasma concentration of 20–25 mg/L. Together with the pathophysiological alterations and pharmacokinetic variability associated with critical illness, the use of continuous renal replacement therapy (CRRT) may complicate the attainment of adequate vancomycin concentrations. The primary objective was the prevalence of attainment of vancomycin concentrations 20–25 mg/L after 24 h in adult ICU patients receiving CRRT. Secondary outcomes were to evaluate target attainment at days 2 and 3 and to calculate vancomycin clearance (CL) by CRRT and residual diuresis. Methods: We performed a prospective observational study in adult ICU patients on CRRT, which received at least 24 h continuous infusion of vancomycin. Between May 2020 and February 2021, daily vancomycin residual blood gas and dialysate samples were collected from 20 patients, every 6 h and if possible vancomycin urine samples. Vancomycin was analysed with an immunoassay method. The CL by CRRT was calculated by a different approach correcting for the downtime and providing insight into the degree of filter patency. Results: The proportion of patients with vancomycin concentrations <20 mg/L was 50% 24 h after starting vancomycin (n = 10). No differences were observed in patient characteristics. The target vancomycin concentration 20–25 mg/L was only achieved in 30% of the patients. On days 2 and 3, despite the use of TDM and albeit in lower percentages, sub- and supratherapeutic levels were still observed. Taking downtime and filter patency into account resulted in lower vancomycin CL. Conclusions: 50% of the studied ICU patients on CRRT showed subtherapeutic vancomycin concentrations 24 h after starting therapy. The results reveal that optimization of vancomycin dosage during CRRT therapy is needed.


Abstract
Introduction: Vancomycin is used in intensive care unit (ICU) patients for the treatment of infections caused by gram-positive bacteria. The vancomycin pharmacokinetic/ pharmacodynamic index is a ratio of the area under the concentration to the minimum inhibitory concentration ≥400-600 h*mg/L. This target can generally be achieved by a plasma concentration of 20-25 mg/L. Together with the pathophysiological alterations and pharmacokinetic variability associated with critical illness, the use of continuous renal replacement therapy (CRRT) may complicate the attainment of adequate vancomycin concentrations. The primary objective was the prevalence of attainment of vancomycin concentrations 20-25 mg/L after 24 h in adult ICU patients receiving CRRT. Secondary outcomes were to evaluate target attainment at days 2 and 3 and to calculate vancomycin clearance (CL) by CRRT and residual diuresis. Methods: We performed a prospective observational study in adult ICU patients on CRRT, which received at least 24 h continuous infusion of vancomycin. Between May 2020 and February 2021, daily vancomycin residual blood gas and dialysate samples were collected from 20 patients, every 6 h and if possible vancomycin urine samples. Vancomycin was analysed with an immunoassay method. The CL by CRRT was calculated by a different approach correcting for the downtime and providing insight into the degree of filter patency. Results: The proportion of patients with vancomycin concentrations <20 mg/L was 50% 24 h after starting vancomycin (n = 10). No differences were observed in patient characteristics. The target vancomycin concentration 20-25 mg/L was only achieved in 30% of the patients. On days 2 and 3, despite the use of TDM and albeit in lower percentages, sub-and supratherapeutic levels were still observed. Taking downtime and filter patency into account resulted in lower vancomycin CL. Conclusions: 50% of the studied ICU patients on CRRT showed subtherapeutic vancomycin concentrations 24 h after starting therapy. The results reveal that optimization of vancomycin dosage during CRRT therapy is needed.

Introduction
Rapid identification and optimal treatment of bacterial infections are essential in intensive care unit (ICU) patients. Vancomycin is used in critically ill patients for the treatment of severe infections, including methicillin resistant Staphylococcus aureus and enterococcus [1]. Like each antimicrobial drug, vancomycin has a pharmacokinetic/pharmacodynamic (PK/PD) index that is associated with effectiveness and toxicity. Vancomycin exhibits time-dependent bacterial killing and the current considered optimal vancomycin PK/PD index is a ratio of the area under the concentration (AUC) to the minimum inhibitory concentration (MIC) of 400-600 mg*h/L. This efficacy and toxicity target can be generally achieved by continuous infusion and keeping the vancomycin serum concentration between 20 and 25 mg/L [1]. Because of pathophysiological alterations and PK variability in ICU patients, target attainment of vancomycin concentrations is difficult to achieve [2]. As a result, vancomycin levels can be subtherapeutic in critically ill patients [3,4].
Besides infections, acute kidney injury (AKI) is a common complication of critical illness. Subsequently, of these patients about 25% need renal replacement therapy (RRT). The majority of these RRT procedures are continuous RRT (CRRT) methods [5]. Modern techniques have ensured that CRRT efficiently removes not only metabolic waste products but also hydrophilic antibiotics such as vancomycin. Several determinants during CRRT can influence vancomycin concentrations, like the delivered renal dose, type of CRRT technique, presence of residual diuresis and type and patency of the used filter [6][7][8]. Together with the pathophysiological alterations and PK variability associated with critical illness, the use of CRRT further complicates target attainment of vancomycin therapy. Data in literature to accurately calculate vancomycin clearance (CL) during CRRT are limited. Important details such as delivered renal doses based on weight, filter clotting, and external ICU procedures that reduce the total CRRT time, and the presence of residual diuresis are often poorly documented. Furthermore, due to filter clotting the filter patency and overall solute CL can be compromised.
Therefore, the primary objective was to determine the prevalence of target attainment vancomycin concentration 20-25 mg/L after 24 h in adult ICU patients with CRRT. Secondary outcomes were to calculate vancomycin CL by CRRT and residual diuresis and to evaluate target attainment at days 2 and 3.

Material and Methods
We performed a prospective observational study between May 2020 and February 2021 at the tertiary ICU of Erasmus MC University Medical Center (Rotterdam, The Netherlands). The medical research Ethics Committee approved the study and all study participants provided written informed consent (MEC 2020-80). Adult patients who received CRRT and who were treated with continuous infusion of vancomycin for at least 24 h were eligible for inclusion. Some patients were already on CRRT or receiving vancomycin, but they were not enrolled until both therapies had been concomitantly applied for at least 24 h.
Initiation of vancomycin was at the discretion of the attending physician. When vancomycin was started at the ICU, patients received a loading dose of 20 mg/kg followed by continuous infusion of 1,000 mg/day. After 24 h, therapeutic drug monitoring (TDM) was performed to adjust the regimen. Initiation and adjustment of CRRT were at the discretion of the attending physician according to local clinical practice. At start of the CRRT, the prescribed renal dose was 25-30 mL/kg/h. The prescribed renal dose was determined using the set total effluent flow of the dialysis machine (sum dialysate, substitute, and ultrafiltrate). Then, the prescribed renal dose was corrected for the downtime of the CRRT circuit (delivered renal dose). The most common cause of downtime is clotting of the extracorporeal circuit or transport of the patient [9]. An additional correction was performed in order to assess the effect of filter patency on the delivered renal dose expressed with urea ratio (corrected delivered renal dose) [9,10]. Urea diffuses completely across the membrane due to its low molecular weight in case of a patent filter. Below urea, ratio 0.80 has such an effect on CRRT-related CL that change of the membrane is required.
Data were extracted from the electronic medical records, including patient characteristics, Acute Physiology and Chronic Health Evaluation (APACHE) IV score Sequential Organ Failure Assessment (SOFA) score and CRRT modalities. A sustainable sampling method was carried out using arterial blood gas (ABG) tests. The residual blood from ABG tests has proved to be suitable for vancomycin analysis [11]. Additionally, at the same time dialysate samples were obtained. Lastly, when possible, urine samples were obtained from a pooled 24 h urine collection once daily at the same time the ICU nurse recorded the 24 h of urine output. Vancomycin concentrations were analysed by means of a validated particle-enhanced turbidimetric inhibition immunoassay auto analyser (Abbott Architect C4000, Chicago, IL, USA). The limit of quantification of the method was 1.1-100 mg/L.
The vancomycin CL via CVVHD (CL CVVHD ) and CVVHDF (CL CVVHD ) were estimated from the saturation coefficient (Sd). An addition of both the sieving coefficient and the Sd in CVVHDF leads to an overestimation of the total CL [12,13]. Therefore, during CVVHDF is the Sd, the diafiltration equivalent of the sieving coefficient. The Sd is calculated as follows [8]: Next, the following equations were used to calculate the extracorporeal CL: The renal doses were expressed as mL/kg/h and CL in mL/min. When no saturation coefficient of vancomycin could be calculated, the urea ratio was used to estimate the saturation coefficient of vancomycin. When the urea ratio was >0.8, the CL of vancomycin was calculated using the median vancomycin saturation coefficient. When a patient had a residual urine output, the fraction CL of the total CL (CL residual renal function + CL by CRRT) was calculated.
Statistical analyses were performed using IBM SPSS for Windows version 28 (IBM Corp., Armonk, NY, USA). Continuous variables will be presented as median with an interquartile range and categorical variables will be given as counts with a percentage. Moreover, Fisher's exact t tests for categorical and Mann-Whitney tests for continuous variables were performed for different variables. The level of significance was set at p < 0.05.

Results
During the study period, 22 [11][12][13][14][15][16]). The median loading dose was 20 mg/kg and daily dose was 1,000 mg/day. Not every patient received a loading dose, as these patients were already treated with vancomycin prior to ICU admission. From the 20 patients, 18 were treated with CVVHD and 2 patients were treated with CVVHDF. Patients' characteristics are summarized in Table 1.
Of the patients, 50% had subtherapeutic vancomycin concentration of <20 mg/L 24 h after initiation. The target vancomycin concentration 20-25 mg/L was only achieved in 30% of the patients (online Suppl. Subsequently, this residual diuresis provided additional CLs during CRRT with CVVHD or CVVHDF. Table 2 presents the renal doses and calculated CLs by CRRT, the urea ratio, the Sd, the residual renal function, and the percentages of CL by residual diuresis contributing to the total CL.

Discussion
The aim of our study was to determine in ICU patients with continuous vancomycin and concomitant CRRT, the prevalence of target attainment vancomycin concentration 20-25 mg/L after 24 h. We found that only 30% of the patients achieved target attainment. Furthermore, 50% and 20% of the patients showed <20 mg/L and >25 mg/L vancomycin concentrations, respectively. This is in line with the study from Roberts et al. [14] who found that 72% of the patients failed to meet an optimal higher limit trough concentration, but in contrast with the studies from Covajes et al. [15] and Quinn et al. [16], who found a target attainment of 51% and 56%, respectively. However, these studies are not completely comparable given the use of both continuous and intermittent administration with other target concentrations of vancomycin or different types of CRRT.
To our knowledge, this is the first report that describes the renal doses based on body weight and sequential vancomycin CL by CRRT, which also corrects for the downtime of the CRRT and provides insight into the degree of filter patency during CRRT. We found for both CVVHD and CVVHDF a significant decrease in the prescribed renal dose versus the corrected delivered renal dose. The decrease is in line with a study from Claure-Del Granado et al. [10], who found that the prescribed renal CL overestimated the actual delivered CL of urea by 23.8% and that this was related to the decrease in filter patency expressed with urea ratio. The decrease in the delivered renal dose due to filter patency was less significant in the subsequent calculated CL of vancomycin. However, the Sd corrects already for the reduction in actual delivered CL. In addition, the CL calculated by using the prescribed renal dose is actually a fictitious CL and not an exact reflection of the actual CL during CRRT. Furthermore, theoretically it can be expected that there is a higher delivered renal dose in patients who did not achieve target attainment. A higher prescribed renal dose is associated with a lower vancomycin concentration. Covajes and colleagues showed that after similar loading doses, higher daily doses were needed in patients with prescribed renal doses of >40 mL/kg/h than in lower renal doses [15]. A possible explanation for not finding this relationship in our study might be related to the fixed prescribed renal dose of about 30 mL/kg/h and that our variability in CL was in the correction for downtime of the CRRT and the degree of the filter patency.
The calculated CL CVVHD 24.8 mL/min (18.7-30.2) from the delivered renal dose is in line with the reported CL CVVHD 22.1 mL/min (12.8-31.4) from the study by Joy and colleagues. Moreover, Joy et al. [6] found also a correlation between vancomycin CL and urea CL. Alternatively, van de Vijsel et al. [17] showed even a higher CL CVVHD of 40 mL/min (32.8-48.7). The reported CL is probably higher due to an estimate of total body CL, an unknown (higher) prescribed renal dose, and no corrections for downtime of the therapy and the extent to which the filter loses its patency during therapy. With respect to our calculated CL CVVHDF 36.4 mL/min (33.1-38.4) by the delivered renal dose, this is higher compared to the reported Continuous variables are presented as median with an interquartile range (IQR) and categorical variables are given as counts with a percentage. APACHE IV score, Acute Physiology and Chronic Health Evaluation IV score; CRRT, continuous renal replacement therapy; CVVHD, continuous venovenous hemodialysis; CVVHDF, continuous venovenous haemodiafiltration; ICU, intensive care unit; SOFA score, sequential organ failure assessment score; UF, ultrafiltration. a The level of significance was set at p < 0.05. b Other anticoagulation than citrate, e.g., heparin, epoprostenol, or defibrotide.  [18]. These researchers used a similar method by including a 4 h interval urea ratio into their study but found no significant differences between the filter patency over this interval. However, a possible explanation for the increased CL CVVHDF might be related to the differences in urea ratio and therefore a better filter patency (0.90 [0.79-0.95] vs. 0.80 [0.76-0.86]) for vancomycin CL.

Vancomycin Clearance and Target
Previous research concluded that the vancomycin CL is comparable in CVVHD and CVVHDF [17]. We did not observe this in the 2 patients who received CVVHDF. We noted a significantly higher CL CVVHDF compared to CL CVVHD (Table 2). A possible explanation for the increased CL in CVHHDF might be related to both the removal principles convection and diffusion and that with increasing molecular weight (vancomycin 1,449 Da) convection has a greater contribution to CL than diffusion [13]. This increased CL CVVHDF can cause subtherapeutic vancomycin concentrations when administering the same dose of vancomycin for both modalities. Moreover, our finding that residual diuresis has an increasing contribution to total CL of vancomycin depending on the amount of residual renal function, is in line with the previous study by Joy et al. [6].
Our results have implications for more personalized dosing of vancomycin in ICU patients with CRRT. Firstly, in order to attain a better target attainment  during CRRT our results advocate for a higher loading dose and initial maintenance dose. In literature, the use of higher loading doses (22-35 mg/kg) and higher daily doses (up to 2,000 mg/day) are advised [15,17,[19][20][21]. Chen and colleagues performed several Monte Carlo simulations to estimate the probability of target attainment (PTA) for different vancomycin loading and maintenance doses and prescribed renal doses [21]. Compared to our loading dose 20 mg/kg, a loading dose of at least 22 mg/kg provided a PTA of 90% for CVVHD and post dilution CVVHDF. In addition, assuming a MIC of 1 mg/L and to achieve a PTA of 90%, a higher maintenance daily dose of at least 1,500 mg and 2,000 mg compared to the studied 1,000 mg/day, is required during renal doses of 25 mL/kg/h and 30 mL/kg/h, respectively. On the other hand, Quinn et al. showed in their retrospective cohort of 160 patients with CVVHD and median body weight 88.6 kg that 53% of the patients had supratherapeutic trough concentrations with daily dosing of >10 mg/kg [16]. Again, in this study used intermittent vancomycin administration with a broad target attainment range of 10-25 mg/L. Therefore, our findings confirm the assumptions of previous studies that during CVVHD and CVVHDF a higher dose of 1,000 mg/ day and/or a higher loading dose of 20 mg/kg are required to achieve target attainment. Secondly, in clinical practice both recovery of diuresis and an adjustment or correction of the prescribed renal dose are not always taken into account. In order to avoid subtherapeutic vancomycin exposure, increasing the delivered renal dose and the presence of residual diuresis should lead to incentives to perform TDM. However, even with TDM after 24 h it is possible you are still lagging behind target attainment. For instance, despite TDM with some dose adjustments after 24 h, we still observed subtherapeutic vancomycin concentrations at days 2 and 3. Thirdly, to avoid supratherapeutic exposure and risk of nephrotoxicity, frequent clotting of the filter and a prolonged downtime can also be factors to perform extra TDM. In our study, 4 patients had vancomycin concentrations >25 mg/L (25.6 mg/L, 25.8 mg/L, 26.3 mg/L and 31.0 mg/L). The patient with 31.0 mg/L had a 13-h downtime period in the first 24 h of CVVHD. Presumably, the other 3 high therapeutic concentrations were the result of a higher starting dose of 1,500 mg/day or 2,000 mg/ day. Although data are limited, vancomycin exposure is related to the probability of AKI [22,23]. Based on the current best available evidence, the upper limit vancomycin exposure is set at 600 h*mg/L [1].
However, from a critical care perspective the need for rapid adequate source control for sepsis trumps the concern for the risk of AKI in the first 24-48 h and therefore advocates for a higher initial dosing regimen.
Our results and the current literature emphasize the complexity of dosing in critically ill patients with vancomycin and CRRT. Model-informed precision dosing (MIPD) might be a valuable tool to improve and individualize the dosing of vancomycin. MIPD is a dosing tool in which a mathematical model in combination with individually measured patient and disease characteristics are used to calculate the optimal dose [24]. MIPD may help start with a "first dose right" regimen and earlier further adaption based on parameters like residual diuresis, filter patency, and delivered renal dose. Previous studies about population PK vancomycin models in CRRT patients showed that SOFA score, the CRRT effluent flow, and residual urine production were significant covariates to be predictive of vancomycin CL [19,25,26]. However, a possible issue is the generalizability of the existing CRRT models available in MIPD software. Since many of the published models have been developed within a specific patient population and type of CRRT, caution must be applied when extrapolating dosing recommendations.
Some limitations of this study should be noted. The data were limited with respect to the number of patients, some missing data and not every patient received the same initial loading dose or daily dose. However, 6 out of 9 patients without or an altered loading dose resulted still in target attainment due to probably steady-state concentrations prior to ICU admission. Moreover, the possible higher daily doses in these patients advocate an adjustment of the current dosing regimen. Lastly, the outcome measure of target attainment by vancomycin concentration rather than the pharmacodynamic parameter AUC/MIC may underor overestimate the daily exposure of vancomycin treatment. However, applying a simplified AUC calculation (vancomycin concentration x 24 h and assuming a MIC of 1 mg/L), 50% of the patients still failed to reach the target AUC 24h/MIC 400-600 mg*h/L. In addition, it is important to note that the vancomycin PK/PD target is derived from retrospective, observational studies of patients with methicillin resistant Staphylococcus aureus bloodstream infections and that in a recent study, a modest and inconsistent AUC/MIC performance was found regarding positive clinical outcome [27][28][29][30]. Therefore, further research is required to optimize the PK/PD target for vancomycin in critically ill patients with CRRT.

Conclusions
In conclusion, 50% of the studied ICU patients on CRRT showed subtherapeutic vancomycin concentrations 24 h after start of the infusion. Furthermore, the general approach of calculating the vancomycin CL by CRRT is usually done by only using the effluent flow or the prescribed renal dose. In this study, we presented an additional correction to the CL for the downtime and gave insight into the degree of filter patency during CRRT. Our findings indicate the complexity of dosing vancomycin during CVVHD or CVVHDF and show that optimization of vancomycin dosage is necessary during CRRT treatment. Further research is warranted to relate the different CL calculation approach to clinical outcome and target attainment.