Association Between Vancomycin Area Under the Curve and Nephrotoxicity: a single center, retrospective cohort study in a veteran population

Abstract

Background

It is unclear which vancomycin area under the curve (AUC) values are most associated with risk for acute kidney injury (AKI).

Methods

This retrospective cohort study was undertaken to determine if vancomycin AUC >550 is associated with a higher rate of AKI than an AUC <550. Patients treated with vancomycin for at least 4 days at the VA St. Louis Health Care System from 1/1/2016 to 9/31/2018 were included. The primary outcome was AKI (defined as an increase in serum creatinine by 0.3 mg/dL or 50% from baseline). Secondary outcomes included length of stay, readmission in 30 days, and mortality in 30 days. A bivariate analysis was used to determine other potential factors affecting AKI rate, with significant variables (P < .2) to be included in the multivariate logistic regression analysis to determine independent risk for AKI.

Results

Two hundred patients were included in the analysis; 100 patients with an AUC ≥550 and 100 with an AUC <550. Only mean vancomycin dose (1722.50 mg vs 2361.25 mg; P < .05), mean AUC (465.88 vs 696.45; P < .05), and peak SCr (1.22 mg/dL vs 1.48 mg/dL; P = .015) were significantly different between groups (AUC <550 vs AUC ≥550, respectively). AKI occurred in 42% (42/100) of patients with AUC ≥550 compared with 2% (2/100) of patients with AUC <550 (P < .05). Secondary outcomes were not different between the groups. In the bivariate analysis, age ≥70, CrCl <50 mL/min, and AUC ≥550 (odds ratio, 49.5; 95% CI, 10.1–242.3; P < .05) were found to be independently associated with risk for developing AKI.

Conclusions

Patients with a vancomycin AUC ≥550 were found to have a significantly higher rate of AKI compared with those with an AUC <550.

In March 2020, the American Society of Health-System Pharmacists (ASHP), Infectious Diseases Society of America (IDSA), Pediatric Infectious Diseases Society (PIDS), and Society of Infectious Diseases Pharmacists (SIDP) published their updated consensus guidelines for the therapeutic monitoring of vancomycin for serious methicillin-resistant Staphylococcus aureus (MRSA) infections. These consensus guidelines no longer endorse trough monitoring as a surrogate for optimal dosing and safety, but state that vancomycin dosing for most infections should be targeted to an area under the curve (AUC)/minimum inhibitory concentration (MIC) ratio of 400 to 600 mg × h/L. The guidelines also endorse, based on national vancomycin susceptibility surveillance, that for most cases the MIC for vancomycin be assumed to be 1 mg/L [1]. These new consensus recommendations, in most instances, allow clinicians to focus on the AUC, which has been shown to be the pharmacodynamic parameter most associated with organism clearance for vancomycin [2–6].

Part of the rationale for shifting from trough-based dosing to AUC-based dosing has been the knowledge that vancomycin troughs of 15–20 mg/dL, the goal range outlined for most infections in the previous iteration of the guidelines, would frequently yield AUCs >600 mg × h/L and have been associated with elevated rates of acute kidney injury (AKI) [2, 7, 8]. The extent of vancomycin-induced AKI varies widely in the literature, ranging from 5% to 43% [9]. While the recommended AUC goal of 400–600 mg × h/L has been correlated with organism eradication in many studies, only a few small studies have investigated the rate of AKI in relation to the AUC. In addition, these studies have yielded conflicting results; older evaluations have found the breakpoint for AKI to be between 700 and 1300 mg × h/L, while some studies have suggested a threshold of 550–600 mg × h/L for AKI [1, 7–12].

We developed this retrospective cohort study to determine if there was a relationship between patients whose steady-state vancomycin dose yielded an AUC ≥550 mg × h/L, an AUC within the recommended goal range of 400–600 mg × h/L, and AKI.

METHODS

A retrospective chart review was conducted in patients admitted to the Veterans Affairs (VA) St. Louis Health Care System, a tertiary teaching hospital with 218 beds and a spinal cord injury unit, between January 1, 2016, and September 31, 2018, to determine if patients treated with vancomycin with a steady-state AUC calculated to be ≥550 had a higher incidence of AKI compared with vancomycin-treated patients with an AUC <550. Patients ≥18 to <90 years of age were included in the study if they received at least 4 days of vancomycin while inpatient, regardless of infectious source. Patients were excluded if they did not have a recorded weight, did not have a serum creatinine (SCr) collected before or during vancomycin treatment, received dialysis for chronic renal impairment, or received a single dose of any of the following agents 72 hours before vancomycin administration and up to 72 hours after the final vancomycin dose: amphotericin B, polymyxin, chemotherapy, vasopressors, or sulfamethoxazole-trimethoprim. This study was approved by the Institutional Review Board at the VA St. Louis Health Care System.

The primary outcome was the occurrence of AKI (up to 72 hours after the final dose of vancomycin) in patients with a steady-state vancomycin AUC ≥550 mg × h/L compared with <550 mg × h/L. AKI was defined as an SCr increase >0.3 mg/L or 50% from baseline in 2 or more consecutive measurements. AUC was calculated using a slightly modified version of the Moise-Broder equation [13]: vancomycin dose (mg/d) divided by vancomycin clearance (L/hr), where vancomycin clearance was calculated by multiplying 0.695 by the creatinine clearance (CrCl, mL/min) divided by weight (kg), and adding 0.05 to the sum [13]. All values to calculate vancomycin clearance and AUC were obtained after the drug had been given for 5 calculated half-lives (steady-state). Secondary outcomes included length of stay, readmission within 30 days of discharge, and mortality in 30 days.

Descriptive statistics were used to compare baseline characteristics of the cohort groups; the chi-square and Fisher exact tests were used to compare categorical variables, and the independent t test and Wilcoxon rank-sum test were used for continuous variables, as appropriate. Significance was determined using a 2-sided alpha of .05. A sample size of 100 patients in each group was needed to reach a power of 80%, based on an estimated 30% AKI rate in the AUC ≥550 group and a 10% rate in the AUC <550 group.

AUC ≥550, concomitant nephrotoxic diseases states (eg, diabetes, chronic kidney disease stage III or higher, or active malignancy), baseline CrCl <50 mL/min, age >70 years, vancomycin therapy ≥7 days, and co-administration of piperacillin-tazobactam, cefepime, or ceftriaxone at any point during therapy were factors input into a bivariate analysis to determine other factors affecting AKI rates. Factors with a P value <.2 were included in a multivariate regression analysis to determine if they were independent risk factors for the development of AKI. All statistical analyses were completed using IBM SPSS Statistics for Windows, version 27.0 (IBM Corp., Armonk, NY, USA).

Conflict of Interest Statement

Neither of the authors has conflicts to report regarding the contents of this manuscript.

Patient Consent Statement

This study was reviewed and approved by the Institutional Review Board at the VA St. Louis Health Care System. Because this was a retrospective study, patient consent was not required.

RESULTS

Two hundred patients were included in the analysis: 100 patients with an AUC ≥550 and 100 with an AUC <550. Demographic and clinical characteristics are shown in Table 1. Of these, only mean vancomycin dose (1722.5 mg vs 2361.25 mg; P < .05), mean AUC (465.88 vs 696.45; P < .05), and peak SCr (1.22 mg/dL vs 1.48 mg/dL; P = .015) were significantly different between groups (AUC <550 vs ≥550, respectively). The most common steady-state goal trough range was 15 to 20 mcg/mL in each group (40% vs 48% for AUC <550 vs ≥550, respectively). The most commonly treated infection was osteomyelitis, which was the indication for 48% of patients in each arm.

Table 1.

Baseline Characteristics

	AUC ≥ 550 (n = 100)	AUC < 550 (n = 100)	P Value
Age, mean ± SD, y	66.3 (10.2)	67.19 (9.4)	.516
Male, No. (%)	98 (98)	100 (100)
Weight, mean ± SD, kg	98.8 (22.6)	87.9 (27.5)	.099
Concomitant nephrotoxic agentsa	87	81	.247
WBC
Initial ± SD	11.79 (6.16)	12.15 (6.28)	.676
Peak ± SD	13.51 (6.17)	13.42 (7.18)	.989
Nadir ± SD	7.51 (3.43)	7.22 (2.59)	.499
SCr
Initial ± SD	1.18 (0.7)	1.11 (0.61)	.407
Peak ± SD	1.49 (0.9 0.99	1.22 (0.63)	.015
Nadir ± SD	(0.62)	0.88 (0.38)	.096
Initial vancomycin information, steady state
Dose ± SD, mg	2361.2 (838.0)	1722.5 (631.2)	<.005
Clearance ± SD, L/h	3.5 (1.1)	3.8 (1.3)	.09
AUC ± SD, mg*h/L	696.4 (166.1)	465.9 (76.8)	<.005
Steady state trough ranges
<10 mcg/mL	0	3
10–15 mcg/mL	16	35
15–20 mcg/mL	48	40
>20 mcg/mL	36	22
Average duration of therapy, d	23.46	24.46
Infectious source, No.
Osteomyelitis	48	48
Skin/soft tissue	22	20
Pneumonia	15	20
Bacteremia	7	8
Urinary	2	2
Intra-abdominal	4	1
Endocarditis	1	0
Other	1	1

Abbreviations: AUC, area under the curve; SCr, serum creatinine; WBC, white blood count.

^aConcomitant nephrotoxic agents included piperacillin-tazobactam, cefepime, aminoglycosides, angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, or a loop diuretic.

The primary outcome of acute kidney injury occurred in 22% (44/200) of all patients. Patients with an AUC ≥550 had an AKI rate of 42% (42/100) compared with those with an AUC <550 having an AKI rate of 2% (2/100; P < .05) (Figure 1). Length of stay was 26.41 compared with 26.32 days for AUC ≥550 vs <550 (Table 2). In patients with a vancomycin AUC ≥550, 30 patients were readmitted, and 12 patients died in 30 days; for those with an AUC <550, 22 patients were readmitted, and 6 patients died. No secondary outcome was statistically significant between the 2 groups.

Figure 1.

Primary outcome, acute kindey injury rate.

Table 2.

Secondary Outcomes

	AUC ≥550 (n = 100)	AUC <550 (n = 100)	P Value
Length of stay, mean ± SD, d	26.41 (35.10)	26.32 (37.21)	.673
Mortality in 30 d, No. (%)	12 (12)	6 (6)	.138
Readmission in 30 d, No. (%)	30 (30)	22 (22)	.099

Abbreviation: AUC, area under the curve.

In the bivariate analysis, the variables that met criteria for inclusion into the multivariate regression model were age >70 years, vancomycin AUC ≥550, CrCl <50 mL/min, concomitant piperacillin-tazobactam administration, and the presence of nephrotoxic comorbidities (Table 3). Of these, age >70 years, CrCl <50 mL/min, and AUC ≥ 550 (odds ratio [OR], 49.51; 95% CI, 10.117–242.334) were independently associated with risk for developing AKI (Table 4).

Table 3.

Bivariate Analysis to Identify Risk Factors Associated With AKI

	No AKI, No. (%)	AKI, No. (%)	P Value
AUC >550	58 (58)	42 (42)	<.005
Age >70 y	52 (67.5)	25 (32.5)	.005
Duration >7 d	130 (78.3)	36 (21.7)	.813
CrCl <50 mL/min	17 (63)	10 (37)	.043
Piperacillin-tazobactam	56 (71.4)	16 (28.6)	.162
Cefepime	62 (79.5)	16 (20.5)	.685
Ceftriaxone	33 (78.6)	9 (21.5)	.920
Nephrotoxic comorbidities	91(71.7)	36 (28.3)	.004

Abbreviations: AKI, acute kidney injury; AUC, area under the curve; CrCl, creatinine clearance.

Table 4.

Multivariate Analysis to Define Independent Risk Factors for AKI

	OR (95% CI)	P Value
AUC >550	49.514 (10.117–242.334)	<.005
Age >70 y	2.427 (1.015–5.799)	.046
CrCl <50 mL/min	4.493 (1.061–19.031)	.041
Piperacillin-tazobactam	1.577 (0.646–3.851)	.318
Nephrotoxicity-related comorbidities	1.938 (0.73–5.143)	.184

Abbreviations: AKI, acute kidney injury; AUC, area under the curve; CrCl, creatinine clearance; OR, odds ratio.

DISCUSSION

This study found that patients with a vancomycin steady-state AUC ≥550 have an increased rate of nephrotoxicity and that age >70 years, initial CrCl <50 mL/min, and a vancomycin AUC ≥550 were independently associated with a risk of developing AKI. In this study, there was no difference in length of stay, 30-day readmission, or 30-day mortality between groups.

Our findings are somewhat at odds with earlier studies looking at similar outcomes but do tend to agree with more recent evaluations. A retrospective, single-center study conducted by Lodise et al. [14] in 2009 looked at patients treated with vancomycin for >48 hours. One hundred sixty-six patients were included, and 21 (12.7%) developed nephrotoxicity. Patients with an AUC of ≥1300 mg × h/L had a higher incidence of nephrotoxicity compared with patients with an AUC <1300 mg × h/L (25.9% vs 10.1%; P = .05) [14].

A more recent study with similar findings to the present analysis was a retrospective, single-center observational study at Liverpool Hospital in Sydney, Australia, published in 2017. This study was conducted from 2006 to 2012 and assessed AKI rate in 127 patients receiving vancomycin monotherapy to treat MRSA bacteremia for a minimum of 14 days. A steady-state vancomycin AUC threshold for nephrotoxicity of 563 mg × h/L was detected by classification and regression tree (CART) analysis, with nephrotoxicity occurring in 40% of patients above this threshold compared with 11.2% below (P = .002) [10].

A multicenter, retrospective evaluation of vancomycin exposure–toxicity thresholds was published by Zasowski et al. in 2017. Patients in this analysis received vancomycin from 2014 to 2015, and the definition of nephrotoxicity was the same as in the Australian study. Three hundred twenty-three patients were included, with 20 developing nephrotoxicity (6.2%). A CART analysis conducted showed that nephrotoxicity was significantly higher among patients with an AUC_0-24h ≥677 mg × h/L. Positive predictive values for all ranges tested were low, but the negative predictive value of the AUC_0-24h ≥677 mg × h/L was 96.7% (adjusted OR, 3.73; 95% CI, 1.646–8.470). These authors suggested an AUC of 700 mg × h/L as a daily limit [11].

A more recent prospective study published in 2018 had similar findings. This trial assessed 252 patients hospitalized at the University of Southern California Medical Center who received vancomycin from December 2012 to January 2016. The first year of the trial required dosing vancomycin based on goal trough calculations, whereas vancomycin was dosed based on AUC >400 mg × h/L in the 2 subsequent years. Over the course of the trial, the median initial vancomycin AUC was 625 mg × h/L in those who developed nephrotoxicity and 423 mg × h/L in those who did not. This trial also found a longer length of stay in those who developed nephrotoxicity compared with those who did not, with a median stay of 20 days compared with 6 days [7].

Finally, a meta-analysis, which included all the above studies, was published in 2019. An AKI definition of ≥0.5 mg/dL or a 50% increase from baseline was used in all included trials. Eight studies met inclusion–exclusion criteria, and 2491 patients were included. An AUC_0-24h <650 mg × h/L was associated with lower rates of AKI (OR, 0.36; 95% CI, 0.23–0.56). The authors additionally determined that the AKI was lower among patients undergoing AUC monitoring compared with trough-guided monitoring (OR, 0.68; 95% CI, 0.46–0.99) [12]. These trials, in combination with the findings of our present study, indicate that the vancomycin AUC threshold for nephrotoxicity could be lower than previously thought.

A difference between prior studies and our evaluation is the definition of AKI; most studies have defined AKI as an SCr increase of 0.5 or 50% from baseline, whereas our study defined AKI as an SCr increase of 0.3 or 50% from baseline. Despite this definition change, the AKI rate of 22% in our study is similar to that of previous studies, which have found the incidence of vancomycin-induced AKI to range from 5% to 40%. Another difference in our study is the calculation of AUC using total daily vancomycin dose, in mg/d, divided by the calculated vancomycin clearance (in L/h), whereas the above trials used Bayesian models or 2 vancomycin levels to calculate AUC. Bayesian models are stated in the current guidelines as the preferred method of AUC determination [1]. Within our population, the modified Moise-Broder equation [13] to calculate vancomycin AUC has corresponded closely with Bayesian models and 2 vancomycin levels used to calculate an AUC; we believe this calculation accurately represents vancomycin AUC in our patient population.

Some strengths of our study include the exclusion of patients with certain nephrotoxic factors at baseline and that some factors were additionally assessed (eg, diabetes, CKD stage III or higher, active malignancy, or concomitant administration of certain antimicrobials) to determine alternative potential factors affecting AKI rate. Additionally, the results presented here correspond with the updated vancomycin monitoring guidelines recommending a vancomycin AUC minimum of 400 mg × h/L and maximum of 600 mg × h/L to optimize efficacy and safety. There were multiple patients in our study with an AUC of 550 to 600 mg × h/L who experienced an AKI, but patients with a vancomycin AUC between 400 to 550 mg × h/L rarely experienced this adverse outcome. Furthermore, the lack of difference in secondary outcomes, such as readmission rate, mortality rate, and length of stay, could have similar efficacy for the patients treated with vancomycin at a lower AUC, although our study was not powered to detect a difference for these secondary outcomes. The relatively long length of stay of around 26 days in our study is likely due to the high rate of osteomyelitis treatment in our population and the presence of a spinal cord unit within our hospital.

Another strength for this study was that all infectious indications were included. Most patients were treated for osteomyelitis, which is indicative of the most frequent usage of vancomycin within our health care system. Nonetheless, there remained a variety of indications treated with vancomycin in this time frame, notably skin and soft tissue infections, pneumonia, and bacteremia.

The present evaluation is not without limitations. First, only the initial steady-state vancomycin dose and AUC were assessed—nephrotoxicity may have occurred while a patient was on a higher or lower vancomycin dose if the dose was adjusted during therapy. This method was chosen as many patients did not have a large change in dose without a corresponding change in renal function. It must also be mentioned that clinical pharmacy specialists at the VA St. Louis Health Care System monitor all patients receiving vancomycin and were incorporating vancomycin AUC in dosing decisions throughout the duration of this study.

Second, this was a retrospective, single-center chart review, so we only had access to data while patients were admitted to the VA St. Louis Health Care System. However, many of our patients completed treatment while remaining inpatient at our acute care hospital or while at our affiliated nursing home/rehabilitation division; we thus were able to continue monitoring many of the patients for an increase in SCr during the entirety of vancomycin therapy.

Finally, patients were determined to have received concomitant nephrotoxic agents if they received just 1 dose of the agent; patients included in the bivariate analysis may only have received 1 dose of piperacillin-tazobactam. As background for our population, the combination of piperacillin-tazobactam plus vancomycin is typically changed to less nephrotoxic, equivalent therapy. Patients will often receive a single dose or short course of piperacillin-tazobactam before being transitioned to metronidazole plus cefepime or ceftriaxone, based on risk factors; the trend toward significance in nephrotoxicity with piperacillin-tazobactam plus vancomycin may have been more significant if patients had been continued on this combination of antibiotics for longer durations.

In conclusion, we found that there was a significantly higher rate of nephrotoxicity in veteran patients receiving vancomycin therapy when their steady-state AUC was found to be ≥550 mg × h/L. Patients with an AUC above this threshold were 49.51 times more likely to develop AKI. Additionally, patients who were over the age of 70 years or had a baseline CrCl <50 mL/min were also found to have an elevated risk of developing AKI. Vancomycin doses yielding an AUC <550 mg × h/L were not associated with a shorter length of stay, lower 30-day mortality rate, or lower 30-day readmission rate. These results suggest that there may be a benefit in targeting an AUC ≤550 mg × h/L for certain patients when dosing vancomycin.