Relationship between nephrotoxicity and area under the concentration–time curve of vancomycin in critically ill patients: a multicenter retrospective study

ABSTRACT We aimed to assess the frequency of acute kidney injury (AKI) in different areas under the concentration–time curve (AUC) values of vancomycin (VAN) using a two-point blood collection method, allowing for accurate AUC assessment in critically ill patients. This multicenter retrospective observational study was conducted in eight hospitals. We retrospectively analyzed the data of patients who had received VAN in an intensive care unit (ICU) between January 2020 and December 2022. The primary outcome was the incidence of AKI. Patients were classified into three groups according to the AUC24–48h at the initial therapeutic drug monitoring (TDM) as follows: <500, 500–600, and ≥600 µg·h/mL. The AUC24–48h values were calculated using the Bayesian estimation software Practical AUC-guided TDM. Among 146 patients [median age (interquartile range), 67 (56–78) years; 39% women], the AUC24–48h <500 µg·h/mL had an AKI rate of 6.5% (7/107), the AUC24–48h 500–600 µg·h/mL had an AKI rate of 28.0% (7/25), and the AUC24–48h ≥600 µg·h/mL had an AKI rate of 42.9% (6/14). In multivariate Cox proportional hazard analysis, the AUC24–48h 500–600 µg·h/mL [hazard ratio 5.4, 95% confidence interval (CI) 1.64–17.63] and the AUC24–48h ≥600 μg·h/mL (hazard ratio 7.0, 95% CI 2.31–21.18) significantly correlated with a higher incidence of AKI compared with the AUC24-48h <500 μg·h/mL. In conclusion, we identified an association between AUC on day 2 and the risk of AKI in ICU patients, suggesting that not only AUCs above 600 µg·h/mL but also those between 500 and 600 µg·h/mL pose a risk for AKI. IMPORTANCE Vancomycin (VAN) is a glycopeptide antibiotic and one of the most commonly used antibiotics for severe infections caused by methicillin-resistant Staphylococcus aureus. However, higher VAN concentrations have been associated with an increased risk of acute kidney injury (AKI). Herein, we aimed to assess the frequency of AKI in different areas under the concentration–time curve (AUC) values of VAN using a two-point blood collection method, allowing for accurate AUC assessment in critically ill patients. We identified an association between AUC on day 2 and the risk of AKI in intensive care unit patients, suggesting that not only AUCs above 600 µg·h/mL but also those between 500 and 600 µg·h/mL pose a risk for AKI. Therefore, individualized dosing is feasible, with pharmacists being able to optimize VAN doses to attain appropriate targets.

V ancomycin (VAN) is a glycopeptide antibiotic and one of the most commonly used antibiotics for severe infections caused by methicillin-resistant Staphylococcus aureus (MRSA) (1,2).Based on pharmacodynamic considerations, the area under the serum concentration-time curve (AUC) divided by the minimum inhibitory concentra tion (MIC) is the preferred parameter for therapeutic monitoring (3,4).Regarding efficacy, maintaining the AUC/MIC value above 400 µg•h/mL is recommended (4)(5)(6).However, reports indicated a correlation between higher VAN concentrations and an increased risk of acute kidney injury (AKI) (7)(8)(9).The risk of AKI increases when VAN trough values exceed 15 or 20 µg/mL (10,11), which conforms with the findings from meta-analyses (5,12).AUC-guided monitoring significantly reduces the incidence of AKI compared to that associated with trough-guided monitoring (5,9,13,14).Therefore, recent guidelines recommend dosing designs indexed by the AUC instead of trough values (15,16).AUC values higher than 1,300 (8), 600 (11,17,18), 563 (19,20), and 550 µg•h/mL (21) have been associated with an increased risk of AKI.Zasowski et al. (22) reported that AUC values of 677 µg•h/mL or higher on day 1 and 683 µg•h/mL or higher on day 2 are associated with an increased risk of AKI.Another report indicated that an AUC of 515 µg•h/mL or higher on day 2 is associated with an increased risk of AKI (23).Meta-analysis identified values higher than 600 (5) or 650 µg•h/mL (24) as indicators for AKI risk.Although high AUC values are demonstrably associated with AKI, reports on the association between early intermediate AUC values and AKI are limited.Consequently, it remains unclear whether an early AUC ranging from 500 to 600 µg•h/mL can be considered as an area associated with AKI.Therefore, in clinical practice, the VAN dosage must be carefully adjusted, even within the reference range, considering both its effectiveness and safety.Particularly, in cases requiring intensive care, there is often a need to administer an adequate amount of VAN for treatment, despite the high risk of AKI.Therefore, determining the frequency of AKI based on AUC values would provide valuable information for the use of VAN in critically ill patients.This study aimed to assess the frequency of AKI for different AUC values using a two-point blood collection method, allowing accurate AUC assessment.

Baseline clinical characteristics
This study included 146 patients after applying the exclusion criteria (Fig. 1).The patients had a median age of 67 years [interquartile range (IQR): 56-78 years], and 39% of the patients were female (Table 1).The median body weight and body mass index (BMI) were 60 kg (IQR: 50-69 kg) and 22.3 kg/m 2 (IQR: 19.3-25.5 kg/m 2 ), respectively.The median sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation II (APACHE II) scores were 6 (IQR: 2-9) and 18 (IQR: 15-24), respec tively.Fifty-five (38%) patients had sepsis, 34 (23%) had septic shock, and 64 (44%) had bacteremia (Table 2).Of the 146 patients, 107 had an AUC 24-48h at initial TDM of less than 500 µg•h/mL, 25 had an AUC 24-48h of between 500 and 600 µg•h/mL, and 14 had an AUC 24-48h of greater than 600 µg•h/mL.There were no significant differences in age, sex, SOFA score, APACHE II score, or renal function among the three groups (Table 1).In terms of VAN doses, the loading dose was not significantly different among the three groups; however, the maintenance dose was significantly different, with a trend toward higher doses in the low-, intermediate-, and high-AUC groups (Table 1).There was a significant difference in the AUC 24-48h at the dosing design among the three groups, with the intermediate-and high-AUC groups having higher values than those of the low-AUC group (Fig. S1; Table S1).

Relationships of the VAN AUC 24-48h level with the frequency of AKI
The primary outcome of AKI was observed in 20 (13.7%) patients.The high-AUC group had an AKI rate of 42.9% (6/14), the intermediate-AUC group had an AKI rate of 28.0% (7/25), and the low-AUC group had an AKI rate of 6.5% (7/107).Patients with an AUC 24- Month XXXX Volume 0 Issue 0 10.1128/spectrum.03739-23 2 48h of ≥500 µg•h/mL had a significantly higher incidence of AKI than that of patients with an AUC 24-48h of <500 µg•h/mL (P < 0.001).In the low-AUC group, the dose of VAN was increased after the initial TDM in 44 (41%) patients, decreased in 9 (8%), kept at the same dose in 38 (36%), and discontinued in 16 (15%).In the intermediate-AUC group, the dose was reduced after the initial TDM in 15 (60%) patients, kept at the same dose in 4 (16%), and discontinued in 6 (24%).In the high-AUC group, the dose was reduced after the initial TDM in 7 (50%) patients, kept at the same dose in 4 (29%), and discontinued in 3 (21%).

Validation of factors associated with AKI
The predictive performances of the CART-derived and other toxicity threshold candi dates are listed in Table 3. Nephrotoxicity was significantly higher among patients with an AUC 24-48h of ≥462 µg•h/mL and AUC ss of ≥651 µg•h/mL, while no threshold was discovered for blood urea nitrogen (BUN)/serum creatinine (Scr) (Table 3; Fig. S2).In the analyses in which all patients were included, an AUC 24-48h of 462 µg•h/mL had an accuracy of 72%, a sensitivity of 80%, and a specificity of 71% for the prediction of AKI; that of 500 µg•h/mL had an accuracy of 77%, a sensitivity of 65%, and a specificity of 79% for the prediction of AKI; whereas that of 600 µg•h/mL had an accuracy of 85%, a sensitivity of 30%, and a specificity of 94% for the prediction (Table 3).

Relationship between AUC 24-48h level and time to AKI onset
Kaplan-Meier curves revealed that the cumulative rate of AKI was the highest in the high-AUC group (Fig. 2).Compared to those observed in the low-AUC group, the medium-and high-AUC groups had significantly higher AKI rates (log-rank P = 0.010 and P < 0.001, respectively; Fig. 2).

DISCUSSIONS
Our results suggest an association between the AUC on day two and the risk of AKI in intensive care unit (ICU) patients, indicating that not only AUCs above 600 µg•h/mL but also those between 500 and 600 µg•h/mL pose a risk for AKI.From the perspec tive focused on AKI prevention, these results provide novel insights into the previous recommendation of not exceeding a steady-state AUC of 600 µg•h/mL.For the dosing design, an AUC of less than 500 µg•h/mL on day 2 is recommended.
Our study had several strengths: (1) We enrolled patients from eight institutions in Japan, all of which studied ICU patients (2).We calculated the accurate AUC using a two-point blood collection method (3).We conducted additional analysis excluding TZP.
The first strength of this study was that we enrolled ICU patients from eight Japanese centers to determine the association between AUC and the risk of AKI.A systematic review reported a mean range of AKI occurrence of 4.3-17 days (12) after the initiation of VAN, with onset reported as early as 2-3 days after the initiation of therapy (26,27).The incidence of VAN-induced AKI varies from 5% to 43% (12) depending on various factors such as the definition of AKI, VAN exposure, concomitant use of TZP or diuretics, disease severity, and other potential risk factors.ICU patients (16%-22%) are particularly at a higher risk of AKI than non-ICU patients (3%-7%) (8,25,28).Our overall AKI incidence of 13.7% was similar to that of recent large retrospective clinical studies that reported incidences of 11.7% in critically ill patients (29) and 15.8% in critically ill Japanese patients (25).However, some reports have indicated higher AKI frequencies in ICU patients using trough-based dosing designs, such as 20% (8) or 24% (10).In a study of 1,882 patients using VAN, Hashimoto et al. ( 25) calculated the optimal cutoff values for groups with AKI risk factors (e.g., CKD, concomitant use of TZP, or diuretics).However, in the ICU cases, the optimal cutoff values could not be calculated because of the small number of cases (25).AUC-guided monitoring significantly reduces the incidence of AKI compared to that associated with trough-guided monitoring (5,9,13,14).Therefore, indexing ICU patients by AUC may be useful in reducing the risk of AKI, and the results of this study may provide additional information on AUC cutoff values and AKI frequency in ICU patients.For ICU patients at high risk of AKI, an AUC of 500-600 µg•h/mL on day 2 was found to be associated with an increased risk of AKI.Since the diagnoses of infection and isolated bacteria are important factors in the treatment of VAN, we incorporated factors, such as septic shock, sepsis, and MRSA bacteremia, in addition to VAN AUC and dosage, with similar results (Table 4 and S4).A recent study reported that patients with AUCs between 400 and 550 µg•h/mL rarely experienced AKI (21).Because an early AUC of 400 µg•h/mL or higher has been associated with early efficacy (30), a target of 400-500 µg•h/mL should be considered for efficacy and safety in the initial dosing design.In ICU patients, CART analysis suggested that an AUC 24-48h below 462 µg•h/mL may reduce the frequency of AKI (Fig. 3).Depending on the type of isolate (or if the bacteria are undetectable), VAN may be discontinued early, but this study revealed the importance of early AUC, and the initial dosing design of VAN may be important even for early discontinuation.
Second, we calculated the AUC using a two-point blood draw method.Conducted in ICU patients, this study suggests that evaluating the AUC using trough and peak values is significant, considering the effects of circulatory dynamics and volume of distribution (15,31,32).Previous AUC studies mostly used AUCs calculated from trough values alone (19,22,23).Although trough values can predict AUCs to some extent, the AUCs calculated using two-point blood sampling are clearly more accurate (33).Therefore, we are confident that the AUC values obtained in this study more accurately explain the relationship between AUC and AKI risk.A recent meta-analysis reported a reduced probability of AKI and a higher probability of achieving pharmacokinetic targets when VAN is administered as a continuous infusion compared to the probabilities associated with intermittent infusion without affecting total mortality, suggesting that continuous administration is also an important treatment option (34).However, Japanese guidelines do not recommend continuous VAN administration due to lack of evidence (15).Therefore, in this study, no cases were treated with continuous administration.Continuous administration is an important treatment option, and further accumulation of evidence in Japan is required.
Third, the additional analysis excluded TZP.Although literature regarding TZPinduced AKI is still incomplete, its combination with VAN carries a higher risk of AKI than those associated with other beta-lactams (such as carbapenems and cefepime) (35,36).Presumably, the concomitant use of TZP is an independent risk factor unrelated to the serum concentrations of VAN (35,36), and several studies have reported the additive nephrotoxic effects of acute interstitial nephritis and direct cell necrosis (37,38).This study also revealed that the combination of TZP and VAN was an independent risk factor for AKI regardless of the AUC value.Univariate logistic regression and CART analyses were performed to calculate the probability of AKI in the absence of concomitant TZP.The absence of the TZP combination resulted in a 2% reduction in the frequency of AKI according to the cutoff value indicated for the AUC (Fig. 3) and a wider safe range of AUCs corresponding to the respective AKI frequencies (Table S2).These results support the findings of Oda et al. (39) that the frequency of AKI decreases with interventions that prevent the use of VAN and TZP combination therapy.Further validation of the association between the combination of TZP-VAN and AKI is needed; both TZP and VAN can bind to renal transporters that mediate creatinine secretion (40,41).Furthermore, a recent study examining changes in creatinine in patients treated with VAN and TZP suggested that the increase in creatinine may be pseudo-nephrotoxicity (42).Animal models suggest that TZP may reduce the nephrotoxicity of VAN (43).Since this study also evaluated creatinine-based AKI, pseudo-nephrotoxicity should be considered.To clarify these problems, the renal injury with cystatin C needs to be evaluated in combination.
Nevertheless, our study had several limitations.First, as this was a retrospective observational study, some selection bias might have occurred in the study population.However, because patient data were extracted from eight different Japanese hospitals, the selection bias was likely smaller than that of a single-hospital study.Second, owing to the limited number of cases, we were unable to validate the probability of AKI using only patients with a combination of TZP and VAN.Third, interpretating these results is difficult because it is unclear whether an elevated AUC increases the risk of AKI or whether AKI increases the likelihood of an elevated AUC (36).However, our results suggest that the intermediate-and high-AUC groups were designed with higher doses than those associated with the low-AUC group from the initial dosing design stage (Table S1), which may explain why an elevated AUC increases the risk of AKI.Fourth, this study did not include cases in which VAN was administered continuously.Fifth, this study did not examine VAN preparation and conservation methods.Sixth, due to the limited number of AKI cases, we could not examine the association between AUC and severity of AKI and subsequent transition to maintenance dialysis.Finally, this study did not evaluate the efficacy of VAN.These aspects should be addressed in future studies.
In conclusion, we found an association between AUC on day 2 and the risk of AKI in ICU patients, suggesting that not only AUCs above 600 µg•h/mL but also those between 500 and 600 µg•h/mL pose a risk for AKI.Furthermore, an AUC of less than 462 µg•h/mL on day 2 may minimize the risk of AKI.We suggest that the initial dosing be designed so that the AUC on day 2 does not exceed 500 µg•h/mL.Our data indicate that individu alized dosing is feasible, with pharmacists being able to optimize VAN doses to attain appropriate PK/PD targets.An additional benefit of such an approach is the refinement of the guidelines for MRSA infections, with the respective AUC targets set according to the risk of AKI.

Study participants
This multicenter retrospective observational study was conducted at Sapporo Medi cal University Hospital, Kyorin University Hospital, Tokyo Women's Medical University Hospital, Showa University Fujigaoka Hospital, Yokohama General Hospital, Kitasato University Hospital, Showa University Northern Yokohama Hospital, and Tohoku Kosai Hospital.We retrospectively analyzed data from patients who had received at least 24 h of VAN in the ICU from January 2020 to December 2022.The enrollment protocol for this study is shown in Fig. 1.Patients younger than 18 years, those undergoing hemodialysis or continuous renal replacement therapy, and those with missing data for peak or trough VAN were excluded.Patients were classified into three groups according to the AUC 24- 48h at the initial TDM as follows: the low-AUC group (<500 µg•h/mL), intermediate-AUC group (500-600 µg•h/mL), and the high-AUC group (≥600 µg•h/mL).

Outcomes
The primary outcome was the incidence of AKI, which was defined according to the Kidney Disease Improving Global Outcomes criteria [i.e., an increase in Scr of ≥0.3 mg/dL (within 48 h), of ≥50% from the most recent pre-treatment data during therapy or continuation of a urine volume of 0.5 mL/kg/h for 6 h or longer] (44).We calculated the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy to evaluate the diagnostic abilities of AUC 24-48h , AUC ss , and BUN/Scr.The pharmacoki netic profiles of VAN were retrospectively analyzed from the trough and peak levels using the Bayesian estimation software Practical AUC-guided TDM (PAT) (ver.3.0) (33).

Data collection
Information regarding the patient's age, sex, body weight, height, BMI, Scr, creatinine clearance (CCr), eGFRcre, VAN dosage, VAN concentrations (trough and peak levels), AUC, infectious diagnosis, causative bacteria, SOFA and APACHE II scores, and concomitant medications was obtained from the patient's medical records.The absence of bacterial growth after 7 days of appropriate incubation was defined as "absence of blood cultures." Blood samples were collected immediately (within 30 min) before VAN administration to obtain trough concentrations.Samples for peak concentrations were collected 0.5-4 h after an intravenous VAN infusion.CCr was calculated using the Cockcroft-Gault formula (Eq. 1) (45).The eGFRcre was calculated using an equation provided by the Japanese Society of Nephrology (Eq.2) (46).(Eq.2) eGFRcre mL/min/1.73m 2 = 194 × Scr −1.094 × Age −0.287 × 0.739 if female

Statistical analyses
Data were presented as medians (IQR: 25th-75th percentile) and expressed as frequen cies and percentages.Welch's test was used to compare continuous variables between groups.Differences in categorical variables between the two groups were examined using the chi-squared test and Fisher's exact test.Patient backgrounds of the three groups divided by AUC 24-48h were compared using the Kruskal-Wallis test.A classification and regression tree (CART) analysis was applied to identify the threshold value for the AKI onset day and the AUC 24-48h of the VAN cutoff capable of predicting an increased risk of AKI.Receiver operating characteristic curves were used to calculate the area under the curve and cutoff values for predicting AKI.Kaplan-Meier curves and log-rank tests were used to assess the cumulative frequency of AKI.Univariate and multivariate Cox proportional hazard analyses were performed to identify potential factors that may have influenced the clinical outcomes.To account for the increased alpha error, the Bonferroni correction was applied to the comparison of the three groups, and P-values less than 0.0167 were considered significant.The software programs JMP Pro 17.0 (SAS Institute Inc., Cary, NC, USA) and EZR version 1.54 (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (The R Foundation for Statistical Computing, Vienna, Austria), were used for the statistical analyses in this study.

FIG 2
FIG 2Relationship between the AUC and the cumulative rate of AKI Log-rank: low-AUC vs. intermediate-AUC group, P = 0.010; low-AUC vs high-AUC group, P < 0.001; intermediate-AUC vs. high-AUC group, P = 0.284.To account for the increased alpha error, the Bonferroni correction was applied to the comparison of the three groups, and a P-value of less than 0.0167 was considered significant.Abbreviations: AKI, acute kidney injury; AUC, area under the concentration-time curve.

FIG 3
FIG 3 Vancomycin exposure-toxicity curve.The relationship between the AKI risk and the vancomycin AUC 24-48h is represented using a best-fit curve.The CART threshold is indicated by the vertical dashed lines.(A) All patients (n = 146) and (B) patients without concomitant TZP (n = 121).Abbreviations: AKI, acute kidney injury; AUC, area under the concentration-time curve; TDM, therapeutic drug monitoring; CART, classification and regression tree; TZP, tazobactam/piperacillin.

TABLE 2
Infectious diseases and causative bacteria a a The number of cases included overlaps.Data are presented as numbers (percentages).Statistical significance was set at P < 0.05.Abbreviations: CRBSI, catheter-related bloodstream infection; MRSA, methicillin-resistant Staphylococcus aureus; MR-CNS, methicillin-resistant coagulase-negative staphylococci.

TABLE 3
Diagnostic accuracy of each factor for AKI b a Optimal cut-off values.bPredictiveperformance of CART-derived and other candidate AUC toxicity thresholds.Abbreviations: AKI, acute kidney injury; AUC, area under the concentration-time curve; AUC 24-48h , AUC on day 2; AUC SS , AUC at steady state; BUN/Scr, ratio of blood urea nitrogen to serum creatinine; PPV, positive predictive value; NPV, negative predictive value; TP, true positive; TN, true negative; FP, false positive; FN, false negative; CART, classification and regression tree.Research Article Microbiology SpectrumMonth XXXX Volume 0 Issue 0 10.1128/spectrum.03739-235

TABLE 4
Cox proportional hazard analyses of factors associated with AKI a

Univariate model Multivariate model 1 Multivariate model 2 HR (95% CI) P value HR (95% CI) P value HR (95% CI) P value
a Statistical significance was set at P < 0.05.To account for the increased alpha error, the Bonferroni correction was applied to the comparison of the three groups, and P-values less than 0.0167 were considered significant.Abbreviations: AKI, acute kidney injury; HR, hazard ratio; CI, confidence interval; BMI, body mass index; SOFA, sequential organ failure assessment; APACHE II, acute physiology and chronic health evaluation II; VAN, vancomycin; AUC, area under the concentration-time curve; AUC 24- 48h , AUC on day 2; eGFRcre, creatinine-based estimated glomerular filtration rate; BUN/Scr, ratio of blood urea nitrogen to serum creatinine; TZP, tazobactam/piperacillin.Research Article Microbiology SpectrumMonth XXXX Volume 0 Issue 0 10.1128/spectrum.03739-236