The independent association of vancomycin with Acute Kidney Injury (AKI) in the critically ill patient with sepsis or septic Shock is controversial.
The aim of this study was to evaluate the incidence of AKI in a cohort of patients with sepsis or septic Shock with an adequate and strict monitoring of vancomycin, guided by the area under the concentration-time curve in relation to the minimum inhibitory concentration (AUC/MIC ratio).
Material and methodsRetrospective cohort study on 106 patients admitted to the ICU with a diagnosis of sepsis or septic shock with vancomycin treatment, consecutively from January 2017 to December 2019.
AKI was defined according to Kidney Disease Improving Global Outcomes criteria. Risk factors associated with AKI were determined by multivariable logistic regression analysis.
ResultsIn our cohort, 28 patients out of 106 (26%) developed AKI. ICU and 30-day mortality were 18% and 22%, respectively. After multivariable logistic regression adjusted analysis, chronic liver disease was associated with AKI.
ConclusionIn our retrospective cohort study on critical patients with sepsis and septic shock, treated with vancomycin adjusting the dose guided by a pharmacokinetic/pharmacodynamic monitoring to achieve the target AUC0-24/CMI ratio, the incidence of AKI was 26%.
La asociación independiente de la vancomicina con la Lesión Renal Aguda (LRA) en el paciente crítico con sepsis o shock séptico es controvertida.
El objetivo de este estudio es evaluar la incidencia de LRA en una cohorte de pacientes con sepsis o shock séptico que recibieron vancomicina con ajuste de dosis de acuerdo a criterios farmacocinéticos/farmacodinámicos, como el área bajo la curva de concentración-tiempo en relación a la concentración mínima inhibitoria en las primeras 24&#¿;h (AUC0-24/CMI ratio).
Material y métodosCohorte retrospectiva de 106 pacientes ingresados en la UCI con diagnóstico de sepsis o shock séptico con tratamiento de vancomicina, de forma consecutiva desde enero de 2017 a diciembre de 2019.
La LRA se definió según criterios de Kidney Disease Improving Global Outcomes. Se determinaron los factores de riesgo asociados a LRA mediante análisis de regresión logística multivariable.
El ajuste de la dosis de vancomicina se hizo guiados mediante la monitorización farmacocinética para conseguir el objetivo de AUC0-24/CMI ≥ 400&#¿;mg&#¿;h /L.
ResultadosEn nuestra cohorte, 28 pacientes de los 106 (26%) desarrollaron LRA. La mortalidad en UCI y a los 30 días fue de 18% y 22% respectivamente. Tras el análisis multivariable ajustado mediante regresión logística, la enfermedad hepática crónica se asoció a LRA.
ConclusiónEn nuestra cohorte retrospectiva en pacientes críticos con sepsis o shock séptico, tratados mediante vancomicina con ajuste de la dosis mediante la monitorización farmacocinética/farmacodinámica para conseguir la AUC0-24/CMI objetivo, la incidencia de LRA ha sido del 26%.
Vancomycin is often used to treat serious infections in patients admitted to intensive care units (ICU) due to its cost-effectiveness and ease of administration. However, although this antibiotic has proved effective as a first line therapy for sepsis due to methicillin-resistant staphylococcus aureus (MRSA), controversies persist regarding its association with acute kidney injury (AKI).1,2 Nephrotoxicity has always been considered the most feared complication of vancomycin therapy, particularly when administered chronically and incorrectly, and above all in high-risk subpopulations such as critically ill patients in whom there are other AKI risk factors such as sepsis.3
Numerous pathophysiological mechanisms have been implicated in sepsis-induced AKI: immunomodulation, kidney and systemic inflammation, complement activation, renin angiotensin aldosterone system (RAAS) deregulation, mitochondrial dysfunction, metabolic reprogramming, macrocirculatory and microcirculatory changes, nephrotoxic drugs, hyperchloremia, and intra-abdominal hypertension.4
The latest guidelines for vancomycin use were published in 2020 by the Infectious Diseases Society of America (IDSA).4 Although it is an antibiotic with proven efficacy that is used as first-line therapy against MRSA-induced sepsis,5 there is as yet no consensus on the best dosing and monitoring strategy due to the potential risk of nephrotoxicity.
Known risk factors for sepsis-induced AKI include septic shock, use of vasopressors and mechanical ventilation, gram-negative bacteraemia, use of renin-angiotensin-aldosterone system inhibitors, presence of chronic liver and kidney disease, pre-existing high blood pressure and diabetes, and smoking.3
Controversies persist with regard to the possible association between vancomycin and AKI in patients with adequately monitored serum levels 6–8.
The recommendation to titrate vancomycin for an AUC0-24/MIC target of ≥400&#¿;mg&#¿;h/L is based on the fact that although its efficacy is thought to be primarily time-dependent, an AUC0-24/MIC ratio of ≥400&#¿;mg&#¿;h/L is currently considered the optimal pharmacokinetic/pharmacokinetic (PK/PD) “efficacy” target for invasive MRSA infection.6
PK/PD-targeted vancomycin dosing based on AUC0-24/MIC has been associated with a lower incidence of AKI compared to trough level-based dosing strategies.9
However, a meta-analysis of 8 observational cohort studies with 2491 patients showed that the use of higher AUC0-24/MIC ratios is associated with a greater incidence of AKI compared to lower ratios; specifically, an AUC0-24/MIC ratio greater than 650&#¿;mg&#¿;h/L was associated with a higher incidence of AKI.10
In light of these findings, we performed this retrospective observational study to determine the incidence of AKI in a cohort of patients with sepsis and septic shock admitted to our ICU in whom vancomycin dosing was titrated according to PK/PD targets.
Material and methodA retrospective observational study was performed in consecutive patients admitted to the ICU of the Clínica Universidad de Navarra from January 2017 to December 2019 with sepsis or septic shock in whom vancomycin therapy was started.
The study was approved by the Research Ethics Committee of the Clínica Universidad de Navarra. The need for informed consent was waived due to the retrospective design of the study. Sepsis and septic shock were defined according to the Sepsis-3 consensus document.11 The SOFA (Sequential Organ Failure Assessment) score used to diagnose sepsis did not include creatinine-based kidney function, since patients with elevated creatinine prior to the start of treatment with vancomycin were excluded (n&#¿;=&#¿;25).
Vancomycin PK/PD variables obtained during therapeutic drug monitoring were used. Inclusion criteria were: patients >18 years of age with sepsis or septic shock starting vancomycin treatment in the ICU. Patients with chronic kidney injury on haemodialysis, AKI before starting vancomycin, patients who died in the first 48&#¿;h, those with incomplete data, and those that received vancomycin for ≤2 days were excluded (Fig. 1).
The primary objective was to determine the incidence and severity (according to KDIGO [Kidney Disease: Improving Global Outcomes] staging) of AKI in patients treated with vancomycin in whom dose titration was guided by PK/PD criteria to achieve an AUC0-24/MIC target of ≥400&#¿;mg&#¿;h/L without exceeding 600&#¿;mg&#¿;h/L.
The secondary objectives were to determine potential risk factors associated with AKI, including trough levels or AUC0-24/MIC ratio of vancomycin. In-ICU and 30-day mortality were also analysed.
Once we had determined the incidence of AKI, its associated with the main pharmacokinetic variables recorded was analysed, such as the minimum inhibitory concentration (MIC) or trough of vancomycin, the area under the concentration-time curve (AUC), and the target AUC/MIC ratio. AKI was defined and staged according to KDIGO guidelines.12
The sample size was determined by the number of consecutive patients included from January 2017 to December 2019, and was similar to the sample size used in a similar study published previously.1 A total of 78 AKI events gave an event:independent variable ratio of <10 in the multivariate logistic regression model adjusted for 6 variables.
The Kolmogorov-Smirnov test was used to confirm normal distribution of the data. A univariate analysis was performed using the Chi square test and Fischer's exact test for parametric and non-parametric categorical variables, respectively, and the Wilcoxon rank sum test and Student's t-test were used to analyse nonparametric and parametric continuous variables, respectively.
Unadjusted logistic regression and multivariate-adjusted logistic regression models were used to evaluate the association between trough levels and AUC0-24/MIC ratio of vancomycin and the incidence of AKI.
The following variables were added to the adjusted logistic regression model: age, sex, baseline serum vancomycin trough concentration, AUC0-24/MIC ratio of vancomycin, liver disease, and use of vasopressors. Significance was set at p&#¿;<&#¿;0.05. All statistical analyses were performed using STATA 16.0.
Vancomycin monitoring techniquePatients received an initial dose of 25&#¿;mg/kg vancomycin over 2&#¿;h or 20&#¿;mg/kg in obese (BMI&#¿;>&#¿;30) patients, depending on total body weight. The maintenance dose was initially estimated on the basis of the glomerular filtration rate calculated using the CKD-EPI equation,13 adjusting for plasma albumin. All patients were monitored between the third and fourth maintenance dose by determining the MIC or trough concentration of vancomycin, and individual pharmacokinetic parameters were estimated using Bayesian methodology. These data were used to accurately titrate the maintenance dose.
Serum vancomycin concentration was determined using the particle-enhanced turbidimetric inhibition immunoassay (PETINIA) included in module C of the Abbott Scientific Alinity autoanalyzer.
PKS v1.10 software (Abbottbase Pharmacokinetics System) from Abbott Laboratories was used for Bayesian estimation of the pharmacokinetic parameters of the antibiotic. Rodvold’s linear two-compartment model included in this software was used as the population pharmacokinetic model.
In patients with unstable vancomycin clearance, further monitoring was scheduled after 24 or 48&#¿;h. The patient's clinical evolution determined the subsequent frequency of pharmacokinetic monitoring.
Vancomycin was titrated to achieve an AUC0-24/MIC target of ≥ 400&#¿;mg&#¿;h/L and to avoid exceeding 600&#¿;mg&#¿;h/L.
ResultsWe collected data from 162 patients admitted to the ICU with sepsis or septic shock who received intravenous vancomycin and whose serum levels were monitored during the study period. Fifty-six patients were excluded from the study for the following reasons: increase in creatinine within 24&#¿;h of starting treatment (n&#¿;=&#¿;25), KDIGO III AKI within 24&#¿;h of treatment (n&#¿;=&#¿;10), previous haemodialysis (n&#¿;=&#¿;8), creatinine not measured during the study period (n&#¿;=&#¿;6), died within 48&#¿;h of treatment (n&#¿;=&#¿;4), and missing data (n&#¿;=&#¿;3).
One hundred and six (106) patients met all the inclusion criteria (sepsis 61% and septic shock 39%) and were included for study and analysis (Fig. 1).
Of the 106 patients, 28 developed AKI (26.4%) and 3 required continuous renal replacement therapy (CRRT) (2.8%). In most cases, AKI was KDIGO I (16/28, 67%), with 25% and 7.1% of patients developing KDIGO II and III, respectively.
In the AKI group, most patients were women, the indication for ICU admission was usually not surgery-related, the length of ICU stay was longer, and the patients were more likely to present high blood pressure, chronic liver disease, heart failure, or chronic kidney disease, septic shock, and greater use of NSAIDs and contrast (Table 1).
Demographic, anthropometric, clinical, and analytical characteristics.
| AKI (n&#¿;=&#¿;28) | No AKI (n&#¿;=&#¿;78) | P value | |
|---|---|---|---|
| Men (n, %) | 19 (67.8%) | 52 (66%) | 0.83 |
| Age (median [IQR]), years | 60 [53−69] | 61 [50−72] | 0.83 |
| Height (cm, mean&#¿;±&#¿;SD) | 165&#¿;±&#¿;10 | 168&#¿;±&#¿;8 | 0.06 |
| Weight (median [IQR]), kg | 70 [63−82] | 68 [58−80] | 0.55 |
| BMI (median [IQR]), kg/m2 | 25 [23–28] | 24 [21–27] | 0.23 |
| SAPS-3 (mean&#¿;±&#¿;SD) | 50&#¿;±&#¿;15 | 53&#¿;±&#¿;15 | 0.38 |
| Vancomycin loading dose (median [IQR]), mg/kg | 27 [16−31] | 26 [20−33] | 0.36 |
| PCT ng/mL (mean&#¿;±&#¿;SD) | 11&#¿;±&#¿;33 | 12&#¿;±&#¿;28 | 0.32 |
| Baseline eGRF (median [IRQ]), ml/min | 66 [36−94] | 60 [40−82] | 0.26 |
| Nadir Hb (mean&#¿;±&#¿;SD) | 8.61&#¿;±&#¿;1.4 | 8.64&#¿;±&#¿;1.3 | 0.93 |
| Indication for admission (n, %) | 0.23 | ||
| Surgical | 3 (11%) | 29 (37%) | |
| Medical | 9 (32%) | 11 (14%) | |
| Respiratory | 4 (14%) | 14 (18%) | |
| Cardiology | 3 (11%) | 13 (17%) | |
| Other | 9 (32%) | 11 (14%) | |
| Length of stay (median [IQR]), days | 8 [5–11] | 8 [5–13] | 0.71 |
| Duration of vancomycin treatment (median [IQR]), days | 6 [5–10] | 7 [5–10] | 0.48 |
| Creatinine prior to vancomycin (median [IQR]), mg/dL | 0.8 [0.65−1.05] | 0.9 [0.7−1.1] | 0.39 |
| Baseline creatinine (median [IQR]), mg/dL | 0.9 [0.7−1.3] | 0.9 [0.7−1.1] | 0.39 |
| Creatinine 48&#¿;h after vancomycin (median [IQR]), mg/dL | 1.2 [0.9−1.6] | 0.75 [0.6−1.0] | 0.0001 |
| Peak creatinine (median [IQR]), mg/dL | 1.65 [1.1−2.1] | 0.9 [0.7−1.3] | 0.0001 |
| High blood pressure (n, %) | 15 (53%) | 35 (44%) | 0.42 |
| Diabetes mellitus (n, %) | 5 (17%) | 16 (20%) | 1 |
| CHF (n, %) | 2 (7.1%) | 4 (5.1%) | 0.65 |
| CKF (n, %) | 4 (14%) | 8 (10%) | 0.72 |
| Liver disease (n, %) | 9 (32%) | 9 (11%) | 0.01 |
| Vancomycin AUC0-24/MIC ratio (median [IQR]) | 503 [428−544] | 503 [428−544] | 0.49 |
| Vancomycin trough levels (median [IQR]) | 10.7 [3.3−16.2] | 9.6 [5.2−12.7] | 0.26 |
| Hypotension with vancomycin (n, %) | 14 (50%) | 29 (37%) | 0.23 |
| Septic shock (n, %) | 12 (42%) | 29 (37%) | 0.59 |
| SOFA (median [IRQ]) | 3 [2–5] | 3 [1–6] | 0.72 |
| Vasopressors (n, %) | 12 (42%) | 29 (37%) | 0.59 |
| Nephrotoxic drugs (n, %) | 26 (92.8%) | 65 (83%) | 0.04 |
| ICU mortality | 10 (36%) | 9 (11.5%) | 0,008 |
| 30-day mortality | 10 (36%) | 14 (18%) | 0.04 |
Abbreviations: BMI&#¿;=&#¿;body mass index; CHF&#¿;=&#¿;congestive heart failure; CKF&#¿;=&#¿;chronic kidney failure; eGFR&#¿;=&#¿;estimated glomerular filtration rate; Hb&#¿;=&#¿;haemoglobin; Hb nadir&#¿;=&#¿;minimum haemoglobin during vancomycin treatment; IQR&#¿;=&#¿;interquartile range; PCT&#¿;=&#¿;procalcitonin; SD&#¿;=&#¿;standard deviation.
Regarding the pharmacokinetic data, in the vast majority of patients trough levels ranged from 11 to 18&#¿;mg/L (58/106, 54.7%) or were below 10&#¿;mg/L (41/106, 38.7%), and AUC0-24/MIC ranged from 401−600&#¿;mg&#¿;h/L (78%) (Table 2).
Pharmacological characteristics.
| AUC0-24/MIC | Total (n&#¿;=&#¿;106) | AKI (n&#¿;=&#¿;28) | No AKI (n&#¿;=&#¿;78) |
|---|---|---|---|
| AUC0-24/MIC | |||
| <401 | 20 (19%) | 5 (18%) | 15 (19%) |
| 401–500 | 56 (53%) | 16 (57%) | 40 (51%) |
| 501–600 | 26 (25%) | 7 (25%) | 19 (24%) |
| >600 | 4 (4%) | – | 4 (5%) |
| Vancomycin (MIC or trough concentration) | |||
| 0–10 | 41 (39%) | 10 (36%) | 31 (40%) |
| 11–14 | 34 (32%) | 9 (32%) | 25 (32%) |
| 15–18 | 24 (22%) | 6 (21%) | 18 (24%) |
| 19–22 | 4 (4%) | 2 (7%) | 2 (2%) |
| >22 | 3 (3%) | 1 (4%) | 2 (2%) |
Abbreviations: AKI: acute kidney injury; AUC/MIC: Area under the curve/minimum inhibitory concentration.
Overall mortality in the ICU and at 30 days was 18% and 22%, respectively, and was higher in the group with AKI than in the group without AKI (36% vs 12% and 36% vs 18%, respectively) (Table 3).
Descriptive analysis of AKI KDIGO stage.
| KDIGO 0 (n&#¿;=&#¿;78) | KDIGO I (n&#¿;=&#¿;19) | KDIGO II–III (n&#¿;=&#¿;9) | Total (n&#¿;=&#¿;106) | |
|---|---|---|---|---|
| Suspected AKI (n, %) | 52 (76%) | 9 (47%) | 3 (33.3%) | 64 (60%) |
| Diagnosed AKI (n, %) | 40 (51%) | 11 (58%) | 5 (55.5%) | 56 (53%) |
| Aminoglycosides (AMG) (n, %) | 26 (33%) | 3 (16%) | 3 (43%) | 32 (30%) |
| NSAIDs (n, %) | 32 (41%) | 11 (58%) | 3 (43%) | 46 (43%) |
| Contrast (n, %) | 11 (14%) | 4 (21%) | 1 (14%) | 16 (15%) |
| Diuretics (n, %) | 57 (73%) | 13 (68%) | 8 (88.8%) | 78 (74%) |
| Isolation (n, %) | 50 (64%) | 15 (79%) | 8 (88.8%) | 73 (69%) |
| Died in ICU (n, %) | 9 (12%) | 7 (37%) | 3 (33.3%) | 19 (18%) |
| Died at 30 days (n, %) | 14 (18%) | 7 (37%) | 3 (33.3%) | 24 (22%) |
| N. nephrotoxic drugs | ||||
| 0 (n, %) | 10 (12%) | – | – | 10 (9%) |
| 1 (n, %) | 49 (63%) | 17 (89%) | 7 (77.7%) | 73 (69%) |
| 2 (n, %) | 16 (21%) | 2 (11%) | 2 (22.2%) | 20 (19%) |
| 3 (n, %) | 2 (3%) | – | – | 2 (2%) |
| 4 (n, %) | 1 (1%) | – | – | 1 (1%) |
| Nephrotoxic drug | ||||
| Yes | 68 (88%) | 19 (100%) | 9 (100%) | 96 (91%) |
Abbreviations: NSAIDs: Nonsteroidal anti-inflammatory drugs.
Most patients received intravenous diuretics (78/106, 74%), with no differences observed between the AKI vs non-AKI group (75% vs 73%). The AKI group received more nephrotoxic drugs (100% vs 87%) (Table 3).
The unadjusted logistic regression analysis showed a correlation between liver disease and AKI (OR 3.6, 95% CI 1.2–10.4, p&#¿;=&#¿;0.01) (Table 4).
Unadjusted logistic regression analysis for AKI.
| Odds ratio | 95% CI | P value | |
|---|---|---|---|
| AKI | |||
| Age | 1.006 | 0.97–1.03 | 0.70 |
| Sex | 1.06 | 0.43–2.58 | 0.89 |
| Nephrotoxic drugs | 5.2 | 0.64–41.7 | 0.12 |
| CHF: | 1.45 | 0.40–5.28 | 0.56 |
| Vasopressors | 1.26 | 0.52–3.04 | 0.59 |
| Liver disease | 3.6 | 1.2–10.4 | 0.01 |
| SOFA | 0.95 | 0.82–1.1 | 0.59 |
| Vancomycin trough levels | 01:03 | 0.96–1.1 | 0.27 |
| Vancomycin AUC0-24 /MIC ratio | 1.001 | 0.97–1.1 | 0.52 |
The adjusted multivariate logistic regression model showed an association between chronic liver disease and AKI (OR 3.48, 95% CI 1.19–10.13, p&#¿;=&#¿;0.02). Patients with chronic liver disease had a 3.48-fold higher risk of developing AKI compared to patients without liver disease.
No association was observed between the use of vasopressors and AKI, and vancomycin trough levels and AUC0-24/MIC ratio were not associated with AKI (Table 5).
After calculating the variance inflation factor (VIF) and observing the correlation matrix of the variables, we found no multicollinearity among the independent variables of the logistic regression model. The VIF of the variables was between 1.05 and 1.5. The correlation coefficient of the variable correlation matrix was < 0.4 for all variables.
DiscussionIn our retrospective cohort, we observed an incidence of AKI of 26%, an ICU mortality rate of 18%, and a 30-day mortality rate of 22%. This is below the rate previously described in a population with sepsis or septic shock with a SAPS-3 of 52&#¿;±&#¿;12, which predicts an in-hospital mortality rate of 22%.14
Our incidences of AKI and mortality in the ICU are lower than expected, considering that a meta-analysis of 47 observational studies in patients with sepsis and septic shock3 published in 2020 found an AKI incidence of 42% among 22 observational studies in patients with sepsis (AKI, n&#¿;=&#¿;16,399/39067, 42%), an AKI incidence of 60.47% among 12 observational studies in patients with septic shock (AKI, n&#¿;=&#¿;12,678/20965, 60.47%), and an in-hospital mortality rate of 42% and 55% in patients with sepsis and septic shock, respectively. A plausible explanation for our low incidence of AKI is the optimal use of vancomycin, since 78% of the patients in our cohort met the recommended PK/PD criteria (AUC0-24/MIC 400−600&#¿;mg&#¿;h/L).
Recent studies have shown that the incidence of vancomycin-associated AKI can vary from 5% to 50%. The pathophysiology is believed to be multifactorial, and although the exact mechanism remains unknown,1,2 inflammation, oxidative stress, mitochondrial dysfunction, and cellular apoptosis are involved. The most common type of kidney injury is direct tubular toxicity, followed by interstitial nephritis.15,16 Other suggested mechanisms are lysosomal dysfunction due to accumulation of vancomycin in tubular cells, the formation of vancomycin-associated tubular casts, and the effect on renal blood flow.17
Proximal tubular secretion of creatinine mediated by organic anion transporters (OAT1 and OAT3) and p-gP (ABCB1) accounts for 40% of urinary creatinine, leading some authors to suggest that vancomycin could inhibit this tubular mechanism and cause a spurious increase in plasma creatinine or “pseudotoxicity”, because tubular filtration remains unaltered. The same would occur with the use of piperacillin-tazobactam (PT).18 In patients with sepsis or septic shock, vancomycin is frequently combined with broad-spectrum antibiotics such as carbapenems and, more frequently, PT.19 Numerous studies have warned that vancomycin&#¿;+&#¿;PT can increase the risk of AKI, but not vancomycin&#¿;+&#¿;carbapenem.20,21 Nevertheless, a recent study found that vancomycin&#¿;+&#¿;PT was not associated with kidney decline, worse prognosis, or higher mortality, suggesting that the increase in plasma or serum creatinine is not in itself an indication of toxicity.22 These authors recommend the use of more reliable biomarkers, such as cystatin or KIM-1 (Kidney Injury Molecule-1) to diagnose AKI.23
To achieve this goal, when MIC is ≥1.0&#¿;mg/L, dosage should be in the upper range. According to Rybak et al. in their 2009 consensus on therapeutic monitoring of vancomycin, high trough concentrations (15−20&#¿;mg/L) are needed to achieve target exposure.24 However, following warnings from numerous professionals about the increase in kidney decline in the US due to an increased use of high-dose vancomycin, they published a new consensus in 2020 in which they advised against routine monitoring of serum levels of vancomycin.6 This recommendation is consistent with our clinical experience, in which levels of 400 mcg x h/L are usually achieved with trough concentrations of between 11 and 12&#¿;mg/L.
A retrospective observational study on the use of continuous infusion of vancomycin to achieve high trough levels (20−30&#¿;mg/L) showed an AKI incidence of 29% and an association between AKI and peak serum vancomycin levels > 40&#¿;mg/L.1
In our cohort, we observed a difference between patients with trough or baseline levels >19&#¿;mg/L (AKI 7%) vs. ≤19&#¿;mg/L (AKI 4%) (Table 2), although this difference did not reach statistical significance due to the low number of patients with trough levels >19&#¿;mg/L (n&#¿;=&#¿;7, 6.6%), of which 3 (2.8%) had levels >22&#¿;mg/L. A more recent study in Japan in a larger cohort (n&#¿;=&#¿;346) found an AKI incidence of 11%; however, these patients were not severely ill, since only 14% were admitted to the ICU.25
Up to 60% of patients hospitalized for any cause receive antibiotics,19 and among these the incidence of drug-related AKI is a high as 60%, especially in the elderly. The characterises of antibiotic-related AKI vary from interstitial nephritis due to beta-lactams to acute tubular necrosis due to aminoglycosides.26
In all our patients, plasma creatinine was stable before starting antibiotic treatment, and all received vancomycin in combination with other antibiotics. The most widely used combination was vancomycin&#¿;+&#¿;meropenem (60%), followed by piperacillin, tazobactam (30%), aminoglycosides (5%), and others (5%).
According to the latest IDSA 2020 guidelines, vancomycin levels should be monitored in severely ill patients, including critically ill patients, in order to optimize efficacy and minimize toxicity.6 They also suggest achieving target trough levels of 15−20&#¿;mg/L in the treatment of gram-positive infection, and increasing this range in more serious infections. These recommendations could explain the higher incidence of vancomycin-related toxicity observed in dozens of recent studies and meta-analyses, which determine the use of this antibiotic.26,27
One of the challenges involved in vancomycin therapy is the need for a loading dose. Failure to administer such a dose delays achievement of the therapeutic target and has a negative effect on outcomes, particularly in critically ill patients with sepsis. Furthermore, administering a loading dose does not guarantee achievement of a high target concentration of over 15&#¿;mg/L. This was evidenced in a prospective study performed in two Norwegian hospitals, in which only 40% of patients achieved therapeutic levels (defined as MIC&#¿;>&#¿;15&#¿;mg/L) in the first 72&#¿;h despite administering loading and maintenance doses.7 We believe that the trough concentration ranges recommended to achieve effectiveness are unnecessarily high.
Although the efficacy of vancomycin is thought to be primarily time-dependent, an AUC0-24/MIC ratio of ≥400&#¿;mg/h/L is currently considered the optimal PK/PD “efficacy” target in invasive MRSA infection.6
However, there is no definitive evidence to support this hypothesis, except for the findings of the preliminary studies on pneumonia published by Schentag.28 Considering the pharmacodynamics of vancomycin, these authors believe AUC0-24/MIC to be the most suitable parameter to ensure efficacy and define therapeutic goals.7
The risk of AKI varies widely depending on drug concentration levels, drug behaviour, and the population studied.
In our cohort, vancomycin was associated with possible sepsis-induced AKI, making it difficult to clearly establish a causal relationship between vancomycin and AKI.
In our opinion, there is no correlation between the AUC/MIC ratio of vancomycin and AKI in our cohort, because 78% of patients met the target AUC/MIC ratio of 400–600&#¿;mg&#¿;h/L.
Our study has the following limitations: (a) it has a small sample size of 106 patients; (b) it is a retrospective, single-centre study without a control group; and finally, (c) some clinically important variables were not analysed, such as the use of mechanical ventilation. Despite our small sample, our total of 78 AKI events gave us an adequate event:independent variable ratio of <10 in the multivariate logistic regression model adjusted for 6 variables.
One of the advantages of our study is our use of Bayesian methodology to estimate individual pharmacokinetic parameters, and our strict monitoring not only of trough levels, but also of the AUC0-24/MIC ratio. Bayesian forecasting is currently considered the best method for estimating vancomycin dosage because it is less prone to prediction bias and total error than other methods given the limited number of samples per patient available in clinical practice. A recent study showed that Bayesian dosing estimates together with vancomycin monitoring in patients with acute kidney injury (sepsis only in 19%) produced a significantly lower rate of nephrotoxicity compared to controls (AKI according to RIFLE criteria 8% vs 21% p&#¿;=&#¿;0.03).8
Another strength of our study is the use of strict monitoring not only of trough levels, but also of the AUC0-24/MIC ratio. This complies with the latest IDSA guidelines that recommend using pharmacokinetic model analysis to monitor subpopulations considered at high risk of AKI, such as critically ill patients, elderly and paediatric patients, obese patients, and those with previous kidney impairment.6
In conclusion, in our retrospective cohort of critically ill patients with sepsis or septic shock in whom vancomycin dosage was adjusted using pharmacokinetic criteria to achieve the target AUC0-24/MIC, 26% developed AKI, a lower incidence than that previously described in the same population.
CRediT authorship contribution statementPM, AA, ADG: Conceptualisation, Methodology. MV: data analysis, writing: original draft preparation. PM, AA, MV: Supervision, revision.
FundingThis study has received no funding.
