Decreasing the time to achieve therapeutic vancomycin concentrations in critically ill patients: developing and testing of a dosing nomogram

The emergence of multidrug-resistant bacteria has been associated with the inappropriate use and the inadequate dosing of antibiotics [1]. The EPIC II study showed that 51% of the patients admitted to ICU had infections and that 71% of all patients were receiving antibiotics [2]. In addition, this study demonstrates that infection by methicillin-resistant Staphylococcus aureus (MRSA) is particularly problematic. Compared to methicillin-susceptible S. aureus (MSSA), MRSA is independently associated with an almost 50% higher likelihood of hospital death [3] and has been considered as a serious threat by the Centers for Disease Control and Prevention (CDC) [4]. Despite its widespread use, vancomycin remains the first-line agent in the treatment of patients with MRSA infection, including those in the critical-care setting. In Portugal, MRSA prevalence is one of the highest in Europe and the emergence of vancomycin-resistant enterococci and vancomycin-resistant S. aureus is an area of particular concern [5],[6].

Though there are limited data to support its routine use in patient care [7], the administration of vancomycin by continuous infusion (CI) has been used for the treatment of critically ill septic patients, because of its practical advantages: 1) rapid achievement of steady-state target concentrations; 2) lower variability in drug exposure; 3) simplicity of interpretation of therapeutic drug monitoring (TDM) and dose adjustment; 4) ease of administration; 5) lower rates of nephrotoxicity, 6) lower costs and 7) lower mortality [8]-[11]. However, achieving the desired serum concentration can still be difficult in this group of patients [11]-[14]. Amongst the several factors that contribute to the difficulties in establishing adequate dosing regimens [15], augmented renal clearance (ARC) is emerging as a new and crucial factor, since vancomycin is predominantly eliminated by the kidneys [16],[17]. ARC refers to an enhanced elimination of circulating solute including drugs and is defined as a creatinine clearance exceeding 130 ml/minute. ARC appears to be quite common in sub-populations of critically ill patients and can lead to very low concentrations of renally cleared drugs like vancomycin.

The aim of this study was to develop a dosing nomogram for the administration of vancomycin by CI for the first 24 h of therapy based on an 8-h measured urinary creatinine clearance (8 h CL_CR); and second, to evaluate its efficiency in a separate cohort of critically ill septic patients.

The main demographic characteristics of the patients belonging to groups 1 and 2 (79 and 25 patients, respectively) are shown in Table 1 and the dosing characteristics and observed pharmacokinetics of vancomycin treatment are described in Table 2. The predominant foci of the infection were lung (63.2%), skin and soft tissues (7.5%), bloodstream (6.3%) and abdominal (6.3%). Globally, the frequency of achievement of adequate vancomycin serum concentrations (20 to 30 mg/L) was 51% (n = 40/79) in group 1 versus 84% (n = 21/25) in group 2 (P <0.005). Of note, the population of group 1 showed a wide range of renal function, between 25 and 335 ml/minute/1.73 m². The incidence of ARC in group 1 was 36% (n = 29/79) and the CL_vanco was 6.8 and 4.2 L/h in patients with and without ARC, respectively. Within these 29 patients showing ARC, only 28% achieved adequate vancomycin serum concentrations (20 to 30 mg/L); 74% of the remaining 50 patients achieved therapeutic concentrations. The 8 h CL_CR and CL_vanco on day 1 in group 1 was significantly linearly correlated (r ² = 0.66; P <0.001) (Figure 1).

The equation from the linear regression was as follows:

Using equation 3, a new equation was developed for calculating a continuous infusion vancomycin dose per day, considering 25 mg/L as the preferred target:

We then used equation 4 to develop a dosing nomogram for vancomycin dosing in the first 24 h, after a loading dose (Figure 2). As a result of the application of this nomogram on group 2 (n = 25 patients), we observed that 21 (84%) achieved serum concentrations between 20 and 30 mg/L on day 1. Two patients (8%) exceeded these limits (34.3 and 33.7 mg/L) and another two patients did not meet the target interval (16.8 and 16.7 mg/L). Of note, all the patients with ARC belonging to group 2 were in the target concentration range (n = 10/10), meaning that all the under- and over-treated patients were non-ARC patients (4/15, 26.6%). The observed vancomycin serum concentrations within the 25 patients, and the respective visual interrelation with the target interval (20 to 30 mg/L) and preferred target concentration (25 mg/L) is showed in the Figure 3. With the exception of one patient who had an increase of over 0.3 mg/dL of serum creatinine concentration in two consecutive days without needing treatment interruption (1/25, 4.0%), no clinical or laboratory vancomycin-related side effects were noted during the period of treatment at the ICU within group 2. On the other hand, the incidence of nephrotoxicity during vancomycin treatment in group 1 was 6.3% (5/79) (Table 1).

Our results show that in a broad population of adult ICU patients treated with continuously infused vancomycin, the use of a dosing nomogram significantly increased the achievement of therapeutic concentrations in the first 24 h of treatment, particularly in the patients exhibiting ARC. Application of this nomogram was found to be easy, user-friendly and effective. The nomogram requires the availability of an 8 h CL_CR and vancomycin serum concentration monitoring, which is not available in all ICUs, but these results demonstrate how beneficial such tests can be to ensure more accurate antibiotic dosing.

Vancomycin is a glycopeptide antibacterial agent that is predominantly excreted unchanged in urine by glomerular filtration and as such, kidney function is a determinant factor of vancomycin pharmacokinetics and dosing. The best measure of kidney function is the glomerular filtration rate (GFR). Urinary creatinine clearance seems to be the best clinical surrogate of renal function, taking into account its applicability at the bedside, its reliability and the negligible costs associated, provided that clinicians are aware of its limitations [20]-[22]. Estimates of GFR using equations based on serum creatinine concentrations are flawed in the critically ill patient [21],[23]. Taking all of this information together, the measurement of the renal clearance of creatinine in this setting could be used, from a clinical point of view, as the single most accessible parameter allowing appraisal of pharmacokinetic characteristics of the critically ill patient.

Clinicians are used to adjusting the dosing of antibiotics according to acute or chronic renal failure, however the adjustment to elevated function of the kidneys is considered to be quite rare, and there are no recommendations on this clinical issue. Moreover, the incidence of ARC (here defined as CL_CR >130 ml/minute/1.73 m²) is probably high and ubiquitous in every ICU with an underestimated incidence [24]. Indeed ARC is being increasingly described, with previously reported rates, varying widely between 18% to as high as 57% in critically ill patients without renal dysfunction [23],[24].

In the context of severe infection, a major consequence of ARC is the high renal clearance of hydrophilic antibiotics leading to a risk of inducing sub-therapeutic concentrations, therapeutic failure, emergence of multi-resistant bacterial strains, and even potentially increased mortality [25]. To date, different antibiotics have been studied in the presence of ARC and these reports have shown a strong association between ARC and sub-therapeutic concentrations [12],[26],[27]. Importantly for vancomycin, low serum concentrations are associated with decreases in susceptibility and in treatment failure of patients with MRSA infections [28]. A recent large-scale multicentre point-prevalence study revealed that a substantial proportion of critically ill patients treated with vancomycin did not achieve the target vancomycin concentration and showed high variability in pharmacokinetics parameters, supporting a re-evaluation of vancomycin dosing recommendations in this particular setting [29]. Furthermore, a recent consensus review recommended more aggressive vancomycin dosing to ensure achievement of the pharmacodynamic index associated with efficacy [16]. Though vancomycin is widely used in the ICU, there are relatively few studies focused on the early optimization of serum concentrations where CI is the prescribed mode of administration [18],[30]-[35]. Surprisingly, only three studies among these used measured renal clearance of creatinine [32]-[34], three used mathematical estimates of renal function [18],[31],[35] and one study did not provide this information [30]. Among those studies evaluating the serum vancomycin concentration on day 1 or day 2, the achievement of target concentrations of 20 to 30 mg/L ranged between 48 and 52% [30],[32].

A recent study described a new regimen for CI of vancomycin during continuous renal replacement therapy, which allowed the achievement of target drug concentrations in 63% of patients at 24 h [36]. Our study, using a dosing regimen guided by a dosing nomogram that is based on the 8 h CL_CR and after an adequate loading dose, permitted us to reach the target drug concentration in most patients on day 1 (n = 21/25, 84%), providing optimal and early antibiotic exposure in the septic patient, with negligible secondary effects (only one patient with a minor increase in serum creatinine concentration, without evolution to renal failure or need for interruption of the treatment). Roberts et al. conducted a population pharmacokinetic analysis of vancomycin CI in a large cohort of critically ill patients, using a Monte Carlo dose simulation for different total body weight, for different creatinine clearances and for different weight-based dosing vancomycin CI regimens [35]. The authors found that higher-than-recommended loading and daily doses of vancomycin seem to be necessary to rapidly achieve therapeutic serum concentrations in these patients. In addition, they state that a patient with a CL_CR of 100 ml/minute/1.73 m² would require at least 35 mg/kg per day by CI to maintain target concentrations. Curiously, when we use the average weight of our 25 patients belonging to group 2 (75 kg), we found very similar results: 2.625 mg versus 2.600 mg (according to Table 3) in a period of 24 h, respectively. Both approaches seem to exhibit some complementarity: a retrospective development of a model and, on the other hand, a clinical prospective validation of a nomogram, respectively.

To the best of our knowledge, the study by Pea et al. is the only to provide and validate two user-friendly dosing nomograms for the treatment of critical ill patients with vancomycin by CI [18]. They described a significant correlation between CL_vanco and creatinine clearance (r ² = 0.56, P <0.001) and between the observed and the predicted serum drug concentration (r ² = 0.64, P <0.001), confirming the dependency of vancomycin elimination on the renal function. However, as acknowledged by the authors, the renal performance was evaluated by the Cockroft-Gault formula, which is a limitation of the study [21],[23]. Of interest and despite this, when we created a new dosing nomogram based on the formula described by these authors, but using the same ideal target that we used in our study (25 mg/L), we obtained similar results for the calculation of the daily vancomycin dosage by CI to achieve target drug concentration (Table 3), confirming in two independent studies the need for a higher dosage to achieve adequate concentrations in the first 24 h of treatment with vancomycin by CI. The similar methodology applied in both studies, similar populations, exclusion of patients with prolonged ICU admission, and choice of the Cockroft-Gault formula by Pea and coworkers (showing higher accuracy when compared with other estimates of renal creatinine clearance [23]) may be possible explanations for the similarities between the two nomograms. On the other hand, the lower incidence of ARC in both cohorts (approximately 15%) when compared to those in our study (36.7 and 40.0% in group 1 and 2, respectively) and the recent literature describing lower accuracy of Cockroft-Gault estimates in both extremes of the normal range of 8 h CL_CR [23],[37] may be possible explanations for the lower agreement between the two nomograms, particularly when considering low and very high values of 8 h CL_CR.

Altogether, our study shows that it is possible to increase the likelihood of target attainment in the first 24 h of treatment with vancomycin, with potential benefits including better outcome and reduction of the development of bacterial resistance. Our study strengths lie in the considerable number of patients included in the retrospective cohort, the significant correlation obtained between measured 8 h CL_CR and CL_vanco (r ² = 0.66, P <0.001), and the wide range of 8 h CL_CR exhibited by these patients (25 to 335 ml/minute/1.73 m²). Therefore, this study is based on a representative cohort of septic critically ill patients, making it applicable in various levels of renal function, including patients with ARC - a sub-group with particular risk of under-treatment with vancomycin [12].

However, some limitations should be acknowledged. First, this was a single-center study, and therefore an extrapolation of the findings to other settings must be done with caution. Second, the second cohort (group 2) was smaller, giving less certainty to our conclusions. Third, 8 h CL_CR cannot be accepted as a gold-standard method to assess kidney function: its determination requires creatinine concentrations to be at steady-state, which is a rarely reached condition in critically ill patients. Finally, pharmacokinetic studies require a constant physiological status, leading us again to the absence of physiological stability in the critical-care setting. In addition, it is possible that not all patients were at actual steady-state, given that 11% of patients in group 1 had an interval between commencement of vancomycin infusion and sampling for TDM less than 16 h.

A novel and easy-to-use vancomycin dosing nomogram for the first 24 h of treatment, based on the 8-h renal clearance of creatinine has been developed and prospectively shown to be effective in septic and critically ill patients at a teaching hospital.