Therapeutic drug monitoring of amikacin in septic patients

Aminoglycosides are an important therapeutic option for the treatment of life-threatening infections in critically ill patients [1]. These drugs are usually used in combination with β- lactams in infections cause by Gram-negative bacteria to extend the spectrum of antimicrobial activity [1, 2], and remain one of the few available therapies against multidrug-resistant pathogens, such as those producing extended-spectrum β-lactamases [3]. Recent guidelines on the treatment of sepsis have recommended using combination therapy, especially in patients with septic shock or if Pseudomonas aeruginosa is suspected [4, 5]. In a meta-analysis, Kumar et al. also suggested that combination therapy including aminoglycosides increased survival rates among patients suffering from septic shock of various etiologies and with an expected mortality exceeding 25% [6].

The antibacterial activity of aminoglycosides is related to the peak concentration (C_peak) obtained after drug injection [7]. In particular, a ratio between C_peak and the minimal inhibitory concentration (MIC) of the pathogen of greater than eight has been associated with increased clinical cure in uncomplicated infections [8]. Nevertheless, sepsis may significantly alter the pharmacokinetic (PK) properties of these drugs, with increased distribution volume (Vd) and altered clearance, resulting in insufficient drug concentrations for the empiric treatment of P. aeruginosa [9, 10]. In this respect, a higher than recommended loading dose (LD) of amikacin (AMK) was necessary to rapidly achieve therapeutic drug concentrations in patients with severe sepsis and septic shock [11]. However, the clinical impact of this strategy has not yet been studied among critically ill patients. Moreover, as these patients often suffer from acute kidney injury (AKI), which contributes to drug accumulation, the use of higher than recommended dose regimens may result in elevated trough concentrations (C_min), which are associated with an increased risk of toxicity [12].

Galvez et al. showed that a daily dose of at least 30 mg/kg of AMK was necessary to achieve optimal drug concentrations [13]; no increase in renal toxicity was reported compared to lower dose regimens, including 15 and 25 mg/kg. However, overall mortality in this study [13] was only 9% and one may argue that these patients did not really represent a critically ill population with different organ dysfunctions and poor outcomes. Moreover, in such patients, the use of large amounts of fluid, of vasopressors and/or of renal replacement therapy may significantly impact on PK drug changes over time, which would be unpredictable if therapeutic drug monitoring (TDM) is not routinely performed [14].

The aim of this study was, therefore, to evaluate the impact of an AMK regimen based on dose adjustment using daily TDM in an ICU population with severe sepsis and septic shock.

Sixty-three patients were included in this study. The patients' characteristics are shown in Table 1. Amikacin was combined with carbapenems (n = 21), cephalosporins (n = 14), piperacillin/tazobactam (n = 11), colistin (n = 8), trimethoprim/sulfamethoxazole (n = 3), ampicillin/sulbactam (n = 2), or other drugs (n = 4). The most common site of infection was the lung (n = 32) and 23 patients (44%) had positive blood cultures. The isolated Gram-negative bacteria were: P. aeruginosa (n = 19), Escherichia coli (n = 9), Acinetobacter baumanii (n = 7), Klebsiella pneumoniae (n = 6), Enterobacter cloacae (n = 4), Proteus mirabilis (n = 3) and others (n = 13). In 41 patients, AMK was initiated together with the first-line therapy; in the 22 other patients, AMK was started on the second day of therapy. The median MIC for amikacin was 4 (1 to 16) mg/L. Septic shock occurred in 20 patients (32%), with an overall ICU mortality of 25%.

Median PK parameters for the LD were: volume of distribution 0.42 L/kg, total clearance 3.3 mL/min and half-life 5.4 hrs (Table 2). Median duration of therapy was 5 [3‐6] days. The average AMK LD was 1500 (750 to 2400) mg; this resulted in a median C_peak of 42.6 (20.3 to 142.0) mg/L and a C_peak/MIC ≥8 in 40 (63%) patients. Patients with initial optimal C_peak/MIC had similar characteristics than others, with the exception of lower C-reactive protein and procalcitonin on the first day of therapy (Table 1). A similar proportion of patients receiving D1 had an optimal C_peak/MIC (26/39 (67%)) compared to those receiving D2 (14/24 (58%), P = 0.59). Among these 40 patients, 24 had C_peak/MIC >12, requiring daily dose reduction (median 1200 (600 to 1800) mg). Only 13 patients (21%) had a C_peak value greater than 64 mg/L, which corresponds to the target concentration to treat susceptible strains of P. aeruginosa. Notably, 5/39 (13%) patients receiving D1 and 8/24 (33%) receiving D2 achieved this target (P = 0.06). The PTA of optimal C_peak/MIC was ≥90% for MIC ≤3 mg/L, regardless of the dose initially given (Figure 1). The daily dose was increased in the 23 patients (37%) with C_peak/MIC <8 (median dose 2000 (1000 to 2400) mg), resulting in optimal C_peak/MIC in 15 (79%) (Table 3).

In 23 patients (37%), C_min was >5 mg/L after the LD, and the drug interval was increased (Table 3). Patients with CrCl ≥50 ml/min less frequently needed interval adjustment (8/43 (19%)) than those with CrCl <50 ml/min (15/20 (75%), P <0.001). There were no differences in the proportion of patients with C_min >5mg/L between those receiving D1 and D2 (14/39 (36%) vs. 9/24 (38%), P = 0.93). On day 2 of therapy, 10 patients (15%) still had a C_min >5mg/L, which returned below this threshold on day 3. Overall, only 11 patients (17%) required no dose or interval adjustment during AMK therapy.

Clinical cure was obtained in 51/60 (85%) patients and microbiological eradication in 43/58 (72%) patients; 3 patients could not be evaluated because they died before the end of therapy and 2 patients were weaned from mechanical ventilation and had no further microbiological samples taken to assess microbiological response. Specifically, clinical cure and microbiological eradication were achieved in 32/37 (86%) and 29/35 (83%) patients, respectively, with initial optimal C_peak/MIC, compared to 16/23 (70%) and 14/23 (61%), respectively, patients with insufficient initial C_peak/MIC (P = 0.18 and P = 0.07). The proportion of patients with clinical cure improved as the initial C_peak/MIC increased (Figure 2). Moreover, patients with initial optimal C_peak had significantly higher microbiological eradication and clinical cure rates than those achieving C_peak/MIC ≥8 after more than 3 days of therapy (Figure 3).

There was no particular difference among patients receiving AMK as the first-line therapy compared to others for microbiological eradication (31/40 vs. 12/18, P = 0.51), clinical cure (35/41 vs. 15/19, P = 0.71) and mortality (10/41 vs. 6/22, P = 1.00). Similarly, patients with infections due to P. aeruginosa or A. baumanii (n = 26), had similar eradication (15/23 vs. 28/35, P = 0.23) and clinical cure rates (22/26 vs. 29/34, P =1.00) than others, but lower C_peak/MIC (7.5 (5 to 11.5) vs. 13 (8 to 23.5), P = 0.005) and a trend toward fewer optimal C_peak/MIC (13/26 vs. 27/37, P = 0.07). Nevertheless, this did not affect ICU mortality (5/26 vs. 11/37, P = 0.39, respectively). No particular differences were found when patients with respiratory infections (n = 41) were compared to others or when patients concomitantly treated with carbapenems (n = 21) were compared to those receiving other additional drugs (data not shown).

AKI developed in 15 patients (24%) during AMK therapy; among these, 12 patients had CrCl <50 ml/min at the onset of therapy and 6 of them eventually underwent RRT. The median time to develop AKI after the administration of the LD was 3 [2‐6] days. Patients with AKI had similar C_peak values to patients without AKI (46.8 (39.1 to 56.2) vs. 40.7 (33.8 to 53.9) mg/L, P = 0.44), but the same proportion of optimal initial C_peak/MIC ratios (12/15 (80%) vs. 28/46 (61%), P = 0.27). C_min values were higher (10.8 (5.1 to 15.5) vs. 1.8 (0.6 to 4.7) mg/L, P <0.001), initial CrCl was lower (44 (25 to 51) vs. 92 (65 to 132) mL/min, P <0.001) while total AMK dose was similar (3.9 (1.5 to 7.5) vs. 5.0 (2.5 to 15) g, P = 0.25) in patients with AKI than in those without. Survivors (n = 47) had similar initial C_peak/MIC ratios (8 (6 to 18) vs. 13 (6 to 23), P = 0.26) than nonsurvivors, but lower C_min values (1.8 (0.6 to 5.3) vs. 7.6 (3.6 to 13.5) mg/L, P = 0.002).

We have shown that higher than standard drug doses are initially necessary to obtain adequate C_peak/MIC in critically ill septic patients. Moreover, TDM appears mandatory during aminoglycoside therapy, because only 17% of patients required no drug dose adjustment. Clinical cure and microbiological eradication occurred more frequently, although not statistically different, in patients with an optimal initial C_peak/MIC than in other patients; however, there were no differences in ICU outcome between groups.

Because of widespread bacterial resistance to several groups of antimicrobials, AMK remains one of the more frequently prescribed aminoglycosides in critically ill patients, despite a very narrow therapeutic index and potential adverse events. Administration of a once daily dose of aminoglycosides in the treatment of severe infection has been widely accepted, because it achieves a higher C_peak than multiple daily doses and is associated with a lower risk of AKI [7, 12, 21]. However, PK changes occurring during sepsis require the use of a higher AMK dose (25 to 30 mg/kg) to rapidly achieve optimal C_peak/MIC, especially when treating less susceptible strains [11, 13]. In burn patients, even higher dose regimens (38 mg/kg) have been proposed to optimize drug therapy [22]. In our study, 63% of patients achieved this PD target after the LD. These data are similar to those reported in recent studies on large ICU cohorts [11, 13]. Nevertheless, important differences need to be underlined. First, we adapted the drug LD to the MIC of the pathogen, obtained before the onset of therapy. This approach was chosen to avoid the unnecessary risk of high AMK doses as we used this drug only for targeted therapy. When used as empirical therapy, drug regimens should be adapted rapidly to treat less susceptible strains, such as P. aeruginosa or A. baumanii, requiring a daily dose of at least 25 mg/kg and a C_peak target above 64 mg/L. Second, even patients with an MIC of 8 to 16 mg/L received AMK, although according to EUCAST recommendations [23], these pathogens are resistant to the drug and optimal C_peak can hardly be obtained in these situations. Although some authors have reported combining a very high AMK regimen (50 mg/kg) with high-flow CRRT to obtain adequate C_peak/MIC and rapidly reduce drug concentrations below the threshold of potential toxicity [24], this approach needs to be further validated and the choice of another antimicrobial should be considered for strains with MICs 8 to 16 mg/L. Finally, we calculated drug regimens according to IBW/ABW and adjusted dose for underweight (BMI <18) or obese (BMI >28) patients [18, 25]. Taccone et al. used a 25 mg/kg LD of AMK calculated using TBW; but the authors noted that there would have been a significant decrease in the proportion of patients with optimal C_peak if this regimen had been used based on IBW [11]. Similarly, Galvez et al. reported that an AMK LD of 30 mg/kg resulted in a larger number of patients with optimal C_peak than an LD of 25 mg/kg, if IBW was used to calculate drug dose regimens [13]. In our cohort, only 33% of patients receiving at least 25 mg/kg achieved optimal C_peak values to treat less susceptible strains. These data suggest a relative underestimation of AMK doses in ICU patients when IBW is applied and the need for even higher doses (30 to 35 mg/kg) to improve drug efficacy. Importantly, the risk of excessive drug dosing in obese patients and the exact role of IBW- or ABW-guided dose prescription needs to be further studied.

Aminoglycosides have been used for decades. However, monotherapy is effective only in urinary tract infections, and meta-analyses have failed to show the superiority of aminoglycoside-β-lactam combination therapy compared to β-lactams alone [1]. However, aminoglycosides were usually given as multiple daily injections and no peak monitoring was performed to optimize the drug regimen. Currently, β-lactams remain the first-line therapy in severe sepsis and septic shock [4]; furthermore, combination with an aminoglycoside may improve patient outcomes, especially in the most severe cases [6]. To the best of our knowledge, this study is the first to report on the clinical and microbiological effects of using high doses of aminoglycosides to reach therapeutic C_peak in critically ill septic patients. Recently, Delannoy et al. found that short-term combination therapy with an aminoglycoside for ICU-acquired bacteremias increased survival and reduced the duration of mechanical ventilation [26]. Our data suggest that early optimizing aminoglycoside peaks according to the MICs of the isolated pathogens may result in further increased therapeutic efficacy and in the clinical cure of severe infections.

If using higher than standard AMK doses is mandatory to optimize its antibacterial activity, the main concern is related to the potential development of renal damage. Aminoglycosides are eliminated by the kidneys and drug uptake by tubular cells is a saturable mechanism when drug concentrations exceed 15 μg/mL [27]. In our study, 24% of patients developed AKI, independent of the initial LD they received. Similarly, Gerlach et al. found AKI in 12% of critically ill surgical patients treated with AMK, especially if combined with other nephrotoxins [28]. In contrast, Galvez et al. found no increased risk of AKI between a drug regimen using 30 mg/kg compared to others using lower doses [13]. Clinical studies have suggested that AKI is more prevalent when there is pre-existing renal impairment or diabetes mellitus or in the case of prolonged therapy [27]. Our data also suggest that the risk of drug accumulation is dependent on baseline renal function, with higher C_min values in patients with CrCl <50 mL/min. Indeed, 12/15 patients who developed AKI during AMK therapy already had reduced CrCl at the onset of therapy, making it difficult to separate the toxic effects of the drug from the kidney damage induced by sepsis. In patients without critical illness, the use of high aminoglycoside dose regimens did not result in AKI, provided that C_min was monitored [29, 30]. Nevertheless, ICU patients have several risk factors for AKI, including hypovolemia, shock, use of diuretics or contrast medium and the presence of chronic kidney disease prior to ICU admission [31], and aminoglycosides could further aggravate the development of renal dysfunction in this population. Among 360 consecutive patients treated with aminoglycoside therapy in an ICU, AKI occurred in 58% and was independently associated with a lower glomerular filtration rate, hypotension, a higher prevalence of diabetes and a more frequent association with other nephrotoxic drugs [32]. Simulation of the clinical effects of amikacin given once daily also showed that, for an MIC of 16 μg/mL and a desired cure rate of at least 90%, the probability of renal failure was expected to be 100% [33]. Thus, an increase in sCr should be expected among ICU patients with pre-existing renal dysfunction, high disease severity, and infections with pathogens with high MICs when high-dose aminoglycosides adapted to the optimal C_peak/MIC are used.

Our study has some limitations. First, we did not evaluate the effectiveness of high vs. standard AMK doses. Although it is possible that this strategy would provide some benefits when compared to standard regimens, because of an increased postantibiotic effect and a potentially reduced risk of drug resistance development, we cannot draw any conclusions regarding the superiority of optimized peak concentrations in this setting. Second, we used the Cockroft-Gault formula to estimate renal function, although this approach has several limitations when applied to critically ill patients. Third, we could not demonstrate any survival benefits for those patients who rapidly achieved optimal C_peak. This may be because of the higher proportion of optimal initial C_peak/MIC ratios among patients who subsequently developed AKI, which may have blunted any potential beneficial effect of optimized therapy.

Moreover, even empirical aminoglycoside therapy with optimized C_peak has never been demonstrated to improve outcome among critically ill patients. Further studies are needed to address this point in large ICU cohorts. Fourth, in one-third of the patients, AMK was initiated only when MIC was obtained by Etest. Thus, this could be considered as being in contrast to common recommendations, which suggest early administration of appropriate antibiotic therapy to improve survival rates in septic patients [6]. Nevertheless, the benefits of this approach have not been clearly demonstrated for combination therapy and this strategy would have resulted in patients with pathogens resistant to the drug being exposed to potentially undesirable adverse events of the drug without likely benefit. In addition, for patients without positive microbiological samples, no microbiological assessment would have been possible. Fifth, we could not provide any data on the appropriateness of the first-line therapy (that is β-lactams) concentrations, which could have been an additional determinant in the evaluation of microbiological eradication and clinical cure. Sixth, the ranges of delivered dose of AMK were quite large in each group. Thus, it is possible that more precise regimens (that is 20 mg/kg vs. 30 mg/kg) would have resulted in different results. Also, the quite limited number of patients included over a 3-year period was related to the strict inclusion and exclusion criteria; this may have introduced a selection bias in this study. Finally, we did not specifically calculate a sample size to estimate the number of patients to be included in this study, as we could not identify any previous data describing the effects of initial optimal C_peak values on clinical and microbiological response in critically ill patients.

TDM resulted in adjustment of AMK therapy in most of our septic patients. Early achievement of an optimal C_peak/MIC ratio may have an impact on clinical and microbiological responses, but not on outcome. In patients with impaired renal function prior to treatment, AMK therapy may be associated with a further decline in renal function.

No funding sources.