1 Introduction
Gentamicin is an aminoglycoside antibiotic that is frequently used in severe life-threatening infections. Aminoglycosides are widely used antibiotics, predominantly used empirically to expand Gram-negative coverage, although emerging aminoglycoside resistance is a widely recognized threat [
1]. Clearly, a favorable outcome can only be achieved with gentamicin if adequate exposure is ensured. For aminoglycosides, a distinct relation between aminoglycoside blood concentrations and both efficacy and toxicity has been reported [
2]. Many studies, mostly in vitro and animal in vivo studies, have shown that both the gentamicin maximum (peak) concentrations (
Cmax) relative to the minimal inhibitory concentration (MIC) (
Cmax/MIC) and the 24-h free drug area under the concentration–time curve (
fAUC
24)/MIC is predictive for effectiveness [
3‐
5]. While these pharmacodynamic indices are to some extent correlated, the general consensus now is that
fAUC
24/MIC is the primary pharmacodynamic index for aminoglycosides driving efficacy [
2,
6,
7]. Aminoglycoside (nephro- and oto-) toxicity correlates with minimum (trough) concentrations (
Cmin) > 1 mg/L [
8].
Obesity and morbid obesity, commonly defined as a body mass index (BMI) of > 40 kg/m
2, is known to influence different pharmacokinetic parameters such as clearance and volume of distribution, even though exact quantification is still warranted for many drugs [
9,
10]. This is especially true for gentamicin, which in normal weight patients is typically dosed on a mg/kg basis [
11]. For obese individuals, several dosing strategies have been proposed, mostly based on alternative body size descriptors such as adjusted body weight (ABW). ABW uses a scaling factor for correcting for limited drug diffusion in adipose body tissue [
12]. Several studies found that with increasing body weight, ABW was predictive for changes in aminoglycoside volume of distribution [
12‐
16] and therefore for
Cmax. More recently, lean body weight (LBW; represents fat-free mass consisting of bone tissue, muscles, organs, and blood volume calculated according to the Janmahasatian formula) was suggested for use in dosing gentamicin, also because of its correlation with volume of distribution [
17,
18]. However, as gentamicin
exposure drives efficacy, changes in gentamicin clearance are to be taken into account when optimizing drug dosing in the obese. Previous studies report an increase in total body clearance with increasing body weight [
12‐
14,
16], with two studies suggesting that ABW might be a predictive covariate for gentamicin clearance [
13,
14]. However, compared to current practice, the degree of obesity in these studies was limited, with average body weights that do not exceed 100 kg in most studies. Moreover, many studies rely on sparse sampling from therapeutic drug monitoring, in an era where aminoglycosides were typically dosed three times daily, and, as such, many studies obtained only a limited number of samples up to 8 h post infusion. As a consequence, the exact influence of obesity on the pharmacokinetics of gentamicin, especially clearance, remains yet to be quantified across the current body weights that we are facing in the clinic.
In this prospective clinical study, we study the pharmacokinetics of gentamicin in obese and morbidly obese individuals versus non-obese individuals in order to develop a dosing algorithm that can be used across the whole clinical population, and that will lead to similar exposure (area under the concentration–time curve from time zero to 24 h [AUC24]) and optimal Cmin values (< 1 mg/L) in obese individuals compared with their non-obese counterparts.
4 Discussion
In this study, we have successfully developed a population pharmacokinetic model for gentamicin based on full pharmacokinetic curves obtained in individuals with body weights ranging from 53 to 221 kg. Our study shows that in obese individuals, both gentamicin clearance and Vc are significantly influenced by body weight. These findings can be used as guide for dosing in the ever-increasing group of obese and morbidly obese patients.
Our study shows that gentamicin clearance increases with TBW. From the studies investigating the pharmacokinetics of aminoglycosides in obesity [
12,
14‐
17,
30,
31], four papers reported an increase in clearance in obese patients [
12‐
14,
16] and two studies found ABW as a predictive covariate [
13,
14]. In these studies participants were only moderately obese (average body weights around 80–100 kg with SDs around 15–20 kg). Moreover, at the time these studies were conducted, aminoglycosides were typically dosed in regimens of up to three times daily, and as such many studies obtained samples up to 8 h post infusion only, thereby limiting the estimation of gentamicin clearance and the prediction of 24-h exposure and
Cmin values. In this respect, we believe that our study is an important addition to the existing literature, since we were able to sample up to 24 h post infusion (instead of 8 h) in a wide range of body weights (53–221 kg) and, combined with using state-of-the-art modelling techniques, we could for the first time accurately assess gentamicin clearance and its covariates in the obese population.
An important question is how the finding that clearance changes with body weight in obese individuals can be explained. The exponent of 0.73 (95% CI 0.57–0.90) we identified for the change with weight is comparable to the value of 0.75 which has been reported as a value that describes the influence of size on clearance in allometry theory [
32]. However, it is debatable whether an increase in weight resulting from obesity can be compared with an increase in weight because of an increase in size [
32]. For other drugs that were studied in obese patients, many show unchanged clearance with increasing weight, even when morbidly obese patients were included [
33‐
35]. The increase in gentamicin clearance with body weight we identify in this study could potentially be explained by a larger GFR in obese individuals and/or an increase in organic cation transporter 2 (OCT2) activity as gentamicin was reported to be a substrate for OCT2 [
36]. With respect to GFR, it is emphasized that in our study only individuals with a GFR > 60 mL/min were included. In our study, weight was the most important covariate, and after implementation of weight, no additional influence of GFR could be identified, even though the GFR range in our population was large (110–230 mL/min). While this does not preclude GFR being the explanation for the observed increase in gentamicin clearance in the obese, and also for other renally excreted drugs such as cefazoline, no increase in clearance with increasing weight was found when studied in morbidly obese and non-obese individuals [
35,
37]. As such, perhaps the increased activity in OCT2 that was reported in overfed rats and that led to increased gentamicin uptake in renal tubular cells [
36] may be considered as an explanation for the findings of our study. In line with this hypothesis, for metformin, which is known to be secreted by OCT2 in the tubulus, a larger clearance was found in obese adolescents (1.17 L/min) than in non-obese children (0.55 L/min), which was also explained by a higher OCT2-mediated tubular secretion of metformin in obese individuals [
38]. From these results it seems that more basic research is needed to identify the exact cause of our findings.
Furthermore, our study demonstrates that
Vc best correlates with body weight. Earlier studies with aminoglycosides in obese patients found ABW or LBW to correlate with volume of distribution [
12‐
14,
17,
30]. In our study we obtained a large number of samples over a 24-h window, including samples that were taken shortly after infusion (i.e., 5, 30, 60, and 90 min after infusion). This study design allows us to fully describe the pharmacokinetics of gentamicin in detail. Most of the previously published studies were performed with sparse (therapeutic drug monitoring) data with only a few samples taken shortly after infusion and consequently analyzed by non-compartmental analysis, thereby complicating exact estimation of the volume of distribution. While the detailed information resulting from our sampling scheme and advanced modelling strategy justifies the conclusions regarding changes of volume of distribution with weight, the results challenge the common assumption that only limited changes in volume of distribution are to be expected for hydrophilic drugs such as gentamicin. It therefore seems that lipophilicity alone is a poor predictor of how volume of distribution changes with increasing body weight, as was demonstrated in several recent reviews [
9,
39].
Based on the results of our study, we propose administration of gentamicin using a practical dose nomogram (Table
3) that is based on a body weight-derived allometric ‘dose weight’ [i.e., 70 × (TBW/70)
0.73] and is derived from the allometric relationship between clearance (driving AUC) and TBW (Table
2, Eq.
5). Considering
fAUC
24/MIC as the primary pharmacodynamic index for aminoglycoside treatment, our dosing nomogram yields a similar gentamicin exposure (AUC
24) across all weights with all
Cmin values < 1 mg/L (Fig.
4). In clinical practice, the nomogram can easily be implemented to select the initial gentamicin dosage, after which dose individualization may be employed by estimating the individual’s gentamicin clearance. This is typically done using therapeutic drug monitoring (where one or two samples are taken during the β-elimination phase, for instance between 2 and 8 h post infusion) in combination with Bayesian software employed with a suitable population pharmacokinetic model. The population pharmacokinetic model presented in the current paper could be used for this purpose. Alternatively, for example when such software is unavailable, other approaches have been suggested to individualize gentamicin drug treatment [
7].
Figure
4 also illustrates that ABW- and LBW-based dose regimens show trends towards a lower exposure with increasing body weight. Despite these trends across weight, it seems that 8 mg/kg LBW and 5–6 mg/kg ABW could be considered as alternatives for our nomogram because using these doses in the median range of the morbidly obese population leads to rather similar AUC
24 values. Implementation of LBW, and to a lesser extend ABW, has, however, been hampered by the complexity of the calculations, which is why we came up with our nomogram, as depicted in Table
3.
Some limitations may apply to our results. First, individuals in our study were, besides (some) being overweight, otherwise healthy, relatively young, and had no renal impairment. As a consequence, renal dysfunction in the obese could not be studied, while in non-obese patients gentamicin clearance has been reported to be dependent on renal function [
40]. Also, drug pharmacokinetics have been shown to be influenced by critical illness [
41]. Therefore, further refinement of our model is warranted for use in obese patients with renal impairment, critical illness, and/or older age. Still, we believe that the dose recommendations from the current study can be a valuable starting point for dosing of obese patients with renal impairment or critical illness. Second, in the current study we did not study the pharmacokinetics of gentamicin after significant reduction in body weight following bariatric surgery. It has been shown for the benzodiazepine midazolam that the pharmacokinetics in these individuals are different to those in individuals with the same body weight without a history of obesity [
42]. Third, we did not include individuals with a BMI 25–35 kg/m
2. However, based on the relationship between TBW and clearance and
Vc, as depicted in Fig.
2, we think it is justified to conclude that the pharmacokinetics will not be any different in these individuals. Last, the obese individuals in our study underwent bariatric surgery during the study procedures, which in theory might influence pharmacokinetics. In our hospital, bariatric surgery is performed laparoscopically, with a short procedure (usually 30–45 min) involving minimal blood loss (usually < 50 mL). Also, hemodynamics were tightly monitored and regulated during surgery. No major hemodynamic instability was recorded for any of the included individuals in our study. For this reason, we expect that the influence of surgery on the pharmacokinetics is negligible.