Prediction Accuracy of Area under the Concentration–Time Curve of Vancomycin by Bayesian Approach Using Creatinine-Based Equations of Estimated Kidney Function in Bedridden Elderly Japanese Patients

Abstract

An administration plan for vancomycin (VCM) in bedridden elderly patients has not been established. This retrospective study aimed to evaluate the prediction accuracy of the area under the concentration–time curve (AUC) of VCM by the Bayesian approach using creatinine-based equations of estimated kidney function in such patients. Kidney function was estimated using the Japanese equation of estimated glomerular filtration rate (eGFR) and the Cockcroft–Gault equation of estimated creatinine clearance (eCCr). eCCr (serum creatinine (SCr) + 0.2) was calculated by substituting the SCr level +0.2 mg/dL into the Cockcroft–Gault equation. For eGFR/0.789, eGFR, eCCr, and eCCr (SCr + 0.2), the AUC values were calculated by the Bayesian approach using the therapeutic drug monitoring (TDM) software, BMs-Pod (ver 8.06) and denoted as AUC_eGFR/0.789, AUC_eGFR, AUC_eCCr, and AUC_{eCCr (SCr + 0.2)} respectively. The reference AUC (AUC_REF) was calculated by applying VCM’s peak and trough steady-state concentrations to first-order pharmacokinetic equations. The medians (range) of AUC_eGFR/0.789/AUC_REF, AUC_eGFR/AUC_REF, AUC_eCCr/AUC_REF, and AUC_{eCCr (SCr + 0.2)}/AUC_REF were 0.88 (0.74–0.93), 0.90 (0.79–1.04), 0.92 (0.81–1.07), and 1.00 (0.88–1.11), respectively. Moreover, the percentage of patients within 10% of the AUC_REF, defined as |Bayesian-estimated AUC − AUC_REF| < AUC_REF × 0.1, was the highest (86%) in AUC_{eCCr (SCr + 0.2)}. These results suggest that the Bayesian approach using eCCr (SCr + 0.2) has the highest prediction accuracy for the AUC_REF in bedridden elderly patients. Although further studies are required with more accurate determination methods of the CCr and AUC, our findings highlight the potential of eCCr (SCr + 0.2) for estimating VCM’s AUC by the Bayesian approach in such patients.

INTRODUCTION

Vancomycin (VCM) has antibacterial activity against Gram-positive bacteria and is an important drug used for treating infectious diseases, including methicillin-resistant Staphylococcus aureus (MRSA).¹⁾ However, therapeutic drug monitoring (TDM) is essential to ensure the therapeutic efficacy of the drug and avoid adverse drug reactions such as nephrotoxicity.²⁾ VCM is a renally excreted drug with a high urinary excretion rate of approximately 90% of the unchanged drug, and an increase in its trough concentration and/or area under the concentration–time curve (AUC) causes nephrotoxicity.²⁾ Several societies, including the Infectious Diseases Society of America, reported that an AUC/minimum inhibitory concentration (MIC) by a broth microdilution (BMD) ratio of 400 to 600 (assuming MICBMD of 1 mg/L) should be advocated as the target to achieve clinical efficacy while improving patient safety for patients with serious MRSA infections.²⁾ AUC-guided dosing is also recommended in Japan to ensure the efficacy and safety of VCM.^3–5) AUC estimation methods include the linear trapezoidal rule using multiple measured concentrations, 2-level estimation approach, and the Bayesian approach. Currently, the Bayesian approach is becoming mainstream.^6,7) An accurate assessment of kidney function is necessary for determining the pharmacokinetic parameters of patients based on the Bayesian approach, which provides us with the exact AUC.

A previous meta-analysis, which included evaluations from a total of 26 studies and 1234 patients, compared the creatinine-based equations of estimated kidney function and the cystatin C-based estimated glomerular filtration rate (eGFR) in predicting VCM clearance.⁸⁾ Okamoto et al.⁹⁾ reported that the serum concentration of cystatin C is a more credible indicator for predicting VCM clearance compared to serum creatinine (SCr) in elderly patients. However, bedridden elderly patients have never been the focus of any of these reports. Furthermore, assessing kidney function using cystatin C is effective in patients with conditions in which muscle mass is reduced, such as sarcopenia.¹⁰⁾ However, cystatin C is affected by thyroid function and some drugs, including steroids and cyclosporine.^11,12) It is also expensive to measure and is only permitted once every three months in the Japanese health insurance system. On the other hand, SCr is frequently used as an indicator of kidney function in clinical practice,¹³⁾ and is inexpensive to determine. However, it is difficult to accurately evaluate kidney function in patients with decreased muscle mass, such as bedridden elderly patients,¹⁰⁾ because their SCr concentrations remain the same; thus, resolving this issue would be very clinically beneficial. Otani et al.¹⁴⁾ reported that consistency between estimated creatinine clearance (eCCr) and measured CCr was highest when the SCr level was adjusted by adding 0.2 mg/dL to the enzymatically measured SCr value in bedridden elderly patients. Therefore, the Bayesian approach using the estimation of kidney function of this method may improve the prediction accuracy of VCM’s AUC in such population. However, there are few studies on the prediction accuracy of the AUC of VCM by the Bayesian approach using creatinine-based equations of estimated kidney function in such patients. The estimated kidney function using SCr tends to be overestimated in bedridden elderly patients,¹⁰⁾ leading to an overdose of VCM. Thus, there is an urgent need to establish an administration plan for VCM in such patients.

This study evaluated the prediction accuracy of the Bayesian-derived AUC of VCM using creatinine-based equations for estimating kidney function in bedridden elderly patients.

MATERIALS AND METHODS

Study Design

The level of independence in daily living in older adults with disabilities and the severity were graded as follows: Independent, J1, J2, A1, A2, B1, B2, C1, and C2.¹⁵⁾ Level B (B1, B2) is defined as “patients requiring some care for living at home and who mainly remain on the bed during daytime.” Level C (C1, C2) is defined as “patients who are bedridden and require care for excretion, feeding, and dressing.”¹⁵⁾ In a previous study that evaluated the accuracy of predicting kidney function in bedridden patients, bedriddeness was graded as B1, B2, C1, and C2.¹⁴⁾ Therefore, in the present study, bedriddeness was graded as B1, B2, C1, and C2. We retrospectively analyzed the medical records of VCM-administered bedridden patients aged 65 and above. Data were collected at the Izumi Regional Medical Center between October 2009 and November 2018 and included the following parameters: sex, age, height, body weight, SCr, blood urea nitrogen, body mass index, infusion volume (mL/kg/d), combined administration of tazobactam/piperacillin, steady-state VCM serum concentration before infusion (VCM trough), steady-state VCM serum concentration 1 h after infusion (VCM peak). The following patients with factors that affect VCM clearance or kidney function were excluded from this study: cancer patients,^16,17) heart failure patients with ejection fraction less than 40%,¹⁸⁾ patients taking drugs such as trimethoprim-sulfamethoxazole combination,¹⁹⁾ cimetidine,²⁰⁾ and cobicistat,²¹⁾ patients with eGFR <30 mL/min/1.73 m², patients undergoing dialysis, patients with VCM dosing intervals >24 h, and patients lacking data. The SCr level measured on the same day as the VCM was used. If the SCr level was not measured on the same day as the VCM, the SCr level closest to the date of measurement of VCM concentration was used. However, patients whose SCr levels changed by more than 30% before and after the VCM measurement date were excluded from this study. SCr levels were determined using an enzymatic method with an autoanalyzer (Hitachi 7180; Hitachi, Tokyo, Japan). VCM concentrations were measured using a SIEMENS Dimension RXL Max System (SIEMENS Co., Ltd., Germany). This study was approved by the Ethics Committee of Izumi Regional Medical Center (No. 20211119-1).

Creatinine-Based Equations of Estimated Kidney Function

eGFR was obtained using equation (1).²²⁾

  

(1)

eGFR was presented in mL/min/1.73 m², and the body surface area (BSA) of each patient was calculated using the Du Bois equation (2)²³⁾ to remove the BSA correction (3):

  

(2)

  

(3)

Henceforth, mL/min, representing individual kidney function, was used as the unit for eGFR. The eGFR was converted to creatinine clearance by dividing by 0.789 and expressed as eGFR/0.789.²⁴⁾

Estimated creatinine clearance (eCCr) was obtained using equation (4).²⁵⁾

  

(4)

eCCr (SCr + 0.2) was calculated by substituting the SCr level +0.2 mg/dL into the Cockcroft–Gault equation (5).²⁶⁾

  

(5)

Reference AUC

In order to calculate patient-specific 24-h AUC value (AUC₂₄), serum concentrations of VCM’s peak and trough at steady-state were measured. In this study, the VCM’s peak serum concentration was obtained 1 h after the end of the infusion. The elimination rate constant (K_e) was calculated by Ln (VCM peak concentrations/VCM trough concentrations)/the time interval between these two concentrations.⁶⁾ The maximum VCM concentration was calculated by VCM peak concentrations × e^K_e×1.⁶⁾ The minimum VCM concentration was approximated as VCM trough concentrations. The AUC for one dosage interval was calculated by determining two values for the AUC. The first was the AUC during the infusion determined by the trapezoidal rule (AUC_inf = [maximum VCM concentration + minimum VCM concentration]/2× time of infusion). The second was the AUC during the elimination phase (AUC_elim) determined by the logarithmic trapezoidal rule (AUC_elim = [maximum VCM concentration - minimum VCM concentration]/K_e).^6,27,28) The AUC for one dosage interval was then calculated by adding AUC_inf and AUC_elim. Finally, the AUC₂₄ was calculated by multiplying this number by the number of daily doses.^6,27,28) The AUC₂₄ determined by this method was defined as the reference AUC (AUC_REF).

AUC Based on the Bayesian Approach

In this study, population pharmacokinetic parameters reported by Yasuhara et al. were used.²⁹⁾ For the four estimated kidney functions, eGFR/0.789, eGFR, eCCr, and eCCr (SCr + 0.2), AUC₂₄ values were calculated by the Bayesian approach using the TDM software, BMs-Pod (ver 8.06)³⁰⁾ and denoted as AUC_eGFR/0.789, AUC_eGFR, AUC_eCCr, and AUC_{eCCr (SCr + 0.2)}, respectively. VCM’s steady-state peak and trough concentrations were used for the Bayesian approach.

Endpoint

Linear regression analysis was carried out on AUC_REF and AUC_eGFR/0.789, AUC_REF and AUC_eGFR, AUC_REF and AUC_eCCr, and AUC_REF and AUC_{eCCr (SCr + 0.2)}. The four groups of AUC_eGFR/0.789/AUC_REF, AUC_eGFR/AUC_REF, AUC_eCCr/AUC_REF, and AUC_{eCCr (SCr + 0.2)}/AUC_REF were compared. The percentage of patients within 20% of the AUC_REF was calculated as |Bayesian-estimated AUC − AUC_REF| < AUC_REF × 0.2. The percentage of patients within 10% of the AUC_REF was calculated as |Bayesian-estimated AUC − AUC_REF|< AUC_REF × 0.1.

Statistical Analysis

Nonparametric multiple comparisons were performed using the Wilcoxon signed-rank test with Bonferroni correction. Significance was set at p < 0.05 for all analyses. Statistical analyses were performed using Excel 2010 (Microsoft Corp., Redmond, WA, U.S.A.) with the add-in software Ekuseru-Toukei 2012 (Social Survey Research Information Co., Ltd., Tokyo, Japan). Statistical power analyses were conducted using the G*Power software program (version 3.1.9.7). The statistical power of the linear regression analyses between AUC_REF and the Bayesian-derived AUC was calculated at a significance (alpha) level of 0.05 (two-tailed) and an effect size (correlation coefficient). The statistical power of the Wilcoxon signed-rank test with Bonferroni correction was calculated at a significance (alpha) level of 0.0083 (two-tailed) and an effect size f². The effect size f² was calculated from the means, standard deviations, and correlation coefficient in both groups.

RESULTS

Patient Characteristics

A total of 131 elderly patients were treated with VCM during the study period. Of these 131 patients, 39 had data on serum concentrations of VCM before and 1 h after administration of steady-state VCM infusion. Among the 39 patients, 2 with cancer, 1 with eGFR <30 mL/min/1.73 m², 1 with >30% change in SCr level before and after the VCM measurement date, 3 with ejection fraction <40%, and 2 with VCM dosing intervals >24 h were excluded. Of the remaining 30 patients, 21 bedridden patients were included in the analysis, excluding nine who were not bedridden. Patient characteristics are presented in Table 1. Their median (range) age was 81 (72–92) years. The median (range) SCr level was 0.54 (0.35–1.08) mg/dL.

Table 1. Characteristics of the Patients

Factor	N = 21
Sex (male/female)	11/10
Age (year)	81 (72–92)
Height (cm)	155 (140–176)
Body weight (kg)	47.8 (34.9–60.0)
Body mass index (kg/m2)	18.77 (14.74–29.34)
Body surface area (m2)	1.42 (1.17–1.69)
Blood urea nitrogen (mg/dL)	13.8 (4.4–38.4)
Serum creatinine (mg/dL)	0.54 (0.35–1.08)
Serum creatinine (mg/dL) <0.6 (Yes/No)	12/9
eCCr (mL/min)	68.4 (24.9–106.9)
eCCr (SCr + 0.2) (mL/min)	49.3 (20.5–74.0)
eGFR (mL/min/1.73 m2)	92.8 (36.8–135.4)
eGFR (mL/min)	78.4 (29.5–123.4)
eGFR/0.789 (mL/min)	99.4 (37.3–156.4)
Steady-state VCM trough concentrations (µg/mL)	14.6 (8.7–23.8)
Steady-state VCM peak concentrations (µg/mL)	28.9 (18.7–41.8)
Dosage of VCM per day (mg)	1200 (650–2000)
Dosing intervals of VCM (12h/24h)	13/8
AUCREF (µg·h/mL)	524.2 (329.1–778.8)
Infusion volume (mL/kg/d)	26.4 (1.7–62.8)

AUC, area under the concentration–time curve; eCCr, estimated creatinine clearance; eGFR, estimated glomerular filtration rate; VCM, vancomycin.

Linear Regression Analysis

The linear regression analysis is presented in Fig. 1. In the linear regression analysis of AUC_REF and AUC_eGFR/0.789, the R² value was 0.926 (p < 0.001); the value for AUC_REF and AUC_eGFR was 0.866 (p < 0.001); the value for AUC_REF and AUC_eCCr was 0.857 (p < 0.001); the value for AUC_REF and AUC_{eCCr (SCr + 0.2)} was 0.911 (p < 0.001).

Fig. 1. The Linear Regression Analysis between (A) AUC_REF and AUC_eGFR/0.789, (B) AUC_REF and AUC_eGFR, (C) AUC_REF and AUC_eCCr, and (D) AUC_REF and AUC_{eCCr (SCr + 0.2)}

y = x (dotted line) and regression line (solid line). AUC, area under the concentration–time curve; eCCr, estimated creatinine clearance; eGFR, estimated glomerular filtration rate; SCr, serum creatinine.

Bayesian-Estimated AUC to AUC_REF Ratio

A boxplot with an overlaid beeswarm plot of Bayesian-derived AUC to AUC_REF ratio is presented in Fig. 2. The medians (range) of AUC_eGFR/0.789/AUC_REF, AUC_eGFR/AUC_REF, AUC_eCCr/AUC_REF, and AUC_{eCCr (SCr + 0.2)}/AUC_REF were 0.88 (0.74–0.93), 0.90 (0.79–1.04), 0.92 (0.81–1.07), and 1.00 (0.88–1.11), respectively. Results of the Wilcoxon signed-rank test with Bonferroni correction revealed that AUC_eGFR/0.789/AUC_REF was significantly lower than AUC_eGFR/AUC_REF (p = 0.016), AUC_eCCr/AUC_REF (p < 0.001), and AUC_{eCCr (SCr + 0.2)}/AUC_REF (p < 0.001); AUC_eGFR/AUC_REF was significantly lower than both AUC_eCCr/AUC_REF (p = 0.01) and AUC_{eCCr (SCr + 0.2)}/AUC_REF (p < 0.001); AUC_eCCr/AUC_REF was significantly lower than AUC_{eCCr (SCr + 0.2)}/AUC_REF (p < 0.001).

Fig. 2. A Boxplot with an Overlaid Beeswarm Plot of the Bayesian-Estimated AUC to AUC_REF Ratio

The bottom and top of the box show the 25 and 75% rankings and, therefore, the interquartile range. The minimum and maximum rankings are denoted by the lower and upper whiskers. A range within 20% of the AUC_REF is shown in gray. The four groups were compared using the Wilcoxon signed-rank test with Bonferroni correction. * p < 0.05, ** p < 0.01, *** p < 0.001. AUC, area under the concentration–time curve; eCCr, estimated creatinine clearance; eGFR, estimated glomerular filtration rate; SCr, serum creatinine.

The Mean of AUC to AUC_REF Ratio and Accuracy of AUC Using the Bayesian Approach

The mean of AUC to AUC_REF ratio and accuracy of AUC using the Bayesian approach are presented in Table 2. The mean ± standard deviation of AUC_eGFR/0.789/AUC_REF, AUC_eGFR/AUC_REF, AUC_eCCr/AUC_REF, and AUC_{eCCr (SCr + 0.2)}/AUC_REF were 0.86 ± 0.05, 0.90 ± 0.08, 0.94 ± 0.08, and 1.00 ± 0.07, respectively. One hundred percent of the values of AUC_eCCr and AUC_{eCCr (SCr + 0.2)} were within 20% of AUC_REF while 90% of AUC_eGFR/0.789 and AUC_eGFR were within 20% of AUC_REF. Furthermore, 86% of AUC_{eCCr (SCr + 0.2)}, 71% of AUC_eCCr, 52% of AUC_eGFR, and 29% of AUC_eGFR/0.789 values were within 10% of the AUC_REF.

Table 2. The Mean of AUC to AUC_REF Ratio and Accuracy of AUC Using the Bayesian Approach

Equation	Mean ± S.D.	Within 20% of AUCREF (%)	Within 10% of AUCREF (%)
eGFR/0.789	0.86 ± 0.05	90	29
eGFR	0.90 ± 0.08	90	52
eCCr	0.94 ± 0.08	100	71
eCCr (SCr + 0.2)	1.00 ± 0.07	100	86

AUC, area under the concentration–time curve; eCCr, estimated creatinine clearance; eGFR, estimated glomerular filtration rate; SCr, serum creatinine; within 20% of AUC_REF (%), percentage of patients difference between Bayesian-estimated AUC and AUC_REF were within 20% of AUC_REF; within 10% of AUC_REF (%), percentage of patients difference between Bayesian-estimated AUC and AUC_REF were within 10% of AUC_REF; S.D., standard deviation.

Statistical Powers

All the statistical powers of the linear regression analyses between AUC_REF and AUC_eGFR/0.789, AUC_REF and AUC_eGFR, AUC_REF and AUC_eCCr, and AUC_REF and AUC_{eCCr (SCr + 0.2)} amounted to 100%. The statistical powers of the Wilcoxon signed-rank test with Bonferroni correction amounted to 59.5–100.0%. The statistical powers of the Wilcoxon signed-rank test with Bonferroni correction were <80% for between AUC_eGFR/0.789/AUC_REF and AUC_eGFR/AUC_REF and between AUC_eGFR/AUC_REF and AUC_eCCr/AUC_REF with values of 70.4 and 59.5%, respectively.

DISCUSSION

This study demonstrated that the Bayesian approach using eCCr (SCr + 0.2) has the highest prediction accuracy for the AUC_REF in bedridden elderly patients.

The creatinine-based equation for the eGFR of Japanese patients was reported in 2008 and is currently used in many institutions.²²⁾ eGFR is a more accurate measurement of kidney function than eCCr because it considers the individual’s height. However, SCr level is strongly affected by muscle mass, leading to overestimation of kidney function in elderly sarcopenic patients.¹⁰⁾ In this study, there was a significant correlation between AUC_REF and AUC_eGFR/0.789 (p < 0.001) where AUC_eGFR/0.789 tended to be lower in relation to AUC_REF (Fig. 1). In addition, the median of AUC_eGFR/0.789/AUC_REF was 0.88 (Fig. 2). These results suggest that determining the AUC of VCM by the Bayesian approach using eGFR/0.789 may lead to an overdose. Hirata et al. reported²⁴⁾ that although eCCr is higher than eGFR in younger patients, it is lower in late elderly patients, and since kidney function is often underestimated with age, obtaining the GFR by multiplying eCCr by 0.789 should only be done in younger patients. In this study, the median of eGFR/0.789 was higher than the median of eCCr (Table 1). This suggests that using the eGFR/0.789 may lead to an overestimated CCr in bedridden elderly patients. Therefore, we recommend against using eGFR/0.789 equations while planning the administration of VCM in such patients. Nakatani et al.¹⁰⁾ reported that the prediction accuracy of kidney function could be improved by considering the muscle mass in the eGFR equation. Accordingly, further studies are needed to determine whether the Bayesian approach using the eGFR/0.789, which considers muscle mass, can improve the prediction accuracy for the AUC.

The Cockcroft–Gault equation of eCCr is frequently used in medical institutions.²⁵⁾ However, eCCr tends to underestimate kidney function with age and low body weight, so its prediction accuracy may be higher than that of eGFR in elderly frail patients.²⁴⁾ In this study, AUC_eCCr/AUC_REF was closer to 1 than AUC_eGFR/0.789/AUC_REF (Fig. 2). We consider that the eCCr would be a better fit than the eGFR equation for the elderly and underweight population in this study (Table 1). However, AUC_eCCr was also underestimated, suggesting that eCCr may be overestimated for kidney function in the population of this study (Fig. 2). Essentially, the Cockcroft–Gault equation needs to use the SCr level by the Jaffe method,²⁵⁾ such that if the CCr is calculated using the SCr level by the enzymatic method in the Cockcroft–Gault equation, it will be higher.²⁴⁾ Since the SCr level measured by the Jaffe method is about 0.2 higher than the SCr level measured by the enzyme method, the CCr value using the Scr level by the Jaffe method is similar to that of the eCCr (SCr + 0.2).²⁴⁾ In this study, the median difference between eCCr and eCCr (SCr + 0.2) was as large as 19.1 mL/min (Table 1). Therefore, adding 0.2 to the SCr level in the Cockcroft–Gault equation may solve the overestimation of CCr, which may lead to the improved predictability of CCr. Otani et al.¹⁴⁾ reported that eCCr (SCr + 0.2) was more accurate in predicting CCr than eCCr in bedridden elderly patients. Therefore, we investigated whether the Bayesian-estimated AUC using eCCr (SCr + 0.2) could have better predictability. In this study, the correlation with the AUC_REF was better for AUC_{eCCr (SCr + 0.2)} than AUC_eCCr (Fig. 1). Moreover, AUC_{eCCr (SCr + 0.2)}/AUC_REF was closer to 1 compared to AUC_eCCr/AUC_REF (p < 0.001) (Fig. 2). Hence, AUC_{eCCr (SCr + 0.2)} was higher than AUC_eCCr when both were within 10% of AUC_REF (Table 2). These results suggest that eCCr (SCr + 0.2) may be more suitable for determining the dosage of VCM than eCCr in bedridden elderly patients. Tsutsuura et al.⁴⁾ reported that when using an AUC/MIC cutoff of 400 (400  ±  15%, 392.7–451) as an indicator of effectiveness, high AUC/MIC ratios, as opposed to low ratios, had significantly lower treatment failure rates in a meta-analysis. In addition, when using an AUC cut-off of 600 (600  ±  15%, 550–683) as an indicator of safety, the VCM-induced acute kidney injury (AKI) incidence rates were significantly higher for high AUC values than for low AUC values.⁴⁾ We believe that an administration plan with a less predictive bias regarding AUC is essential in maintaining the efficacy of VCM and preventing VCM-induced AKI. In this study, the correlation between AUC_REF and AUC_{eCCr (SCr + 0.2)} was strong (Fig. 1), and the median (range) of AUC_{eCCr (SCr + 0.2)}/AUC_REF was 1.00 (0.88–1.11) (Fig. 2). Therefore, we believe that the Bayesian approach using eCCr (SCr + 0.2) is expected to improve the efficacy and safety of VCM in bedridden elderly patients. Further studies are needed to determine the effect of AUC-guided dosing using eCCr (SCr + 0.2) on the efficacy and safety of VCM in such population.

The GFR by the enzymatic method and the CCr by the Jaffe method, i.e., CCr (SCr + 0.2), are considered to have similar values.²⁴⁾ However, in this study, the correlation analysis between eGFR and eCCr (SCr + 0.2) showed that eGFR was found to be higher than eCCr (SCr + 0.2) in all patients (Supplementary Fig. 1). This is because eCCr (SCr + 0.2) tends to be lower than eGFR in underweight and elderly patients.²⁴⁾ In this study, the correlation with the AUC_REF was better for AUC_{eCCr (SCr + 0.2)} compared to AUC_eGFR (Fig. 1). Moreover, AUC_{eCCr (SCr + 0.2)}/AUC_REF was closer to 1 compared to AUC_eGFR/AUC_REF (p < 0.001) (Fig. 2). Hence, AUC_{eCCr (SCr + 0.2)} was higher than AUC_eGFR when both were within 10% of AUC_REF (Table 2). These results suggest that eCCr (SCr + 0.2) is better than eGFR in assessing kidney function to determine the dosage of VCM in bedridden elderly patients.

In this study, the statistical powers of the Wilcoxon signed-rank test with Bonferroni correction between AUC_eGFR/0.789/AUC_REF and AUC_eGFR/AUC_REF and between AUC_eGFR/AUC_REF and AUC_eCCr/AUC_REF were <80%, respectively. However, the statistical powers of the Wilcoxon signed-rank test with Bonferroni correction were above 80% for all comparisons with AUC_{eCCr (SCr + 0.2)}/AUC_REF. Therefore, we believe that this study has enough statistical power to conclude that eCCr (SCr + 0.2) is the best in calculating AUC using Bayesian approach in bedridden elderly patients, compared to the other creatinine-based equations analyzed.

This study has some limitations. First, due to the study’s retrospective nature, data collection relied only on evaluating clinical progress notes, laboratory test results, and other documentation. Second, we could not calculate the AUC_REF by the relatively more accurate methods, such as the linear trapezoidal rule using multiple VCM concentrations.³¹⁾ Instead, we calculated it by applying steady-state peak and trough serum concentrations of VCM to first-order pharmacokinetic equations, as described in Materials and Methods. When comparing the AUC_REF calculated by this method and the Bayesian-estimated AUC using the full concentration–time profile, the median error was found to be ≤2% for our method, which is not clinically significant.⁶⁾ Therefore, we used the first-order pharmacokinetic equations to calculate the AUC_REF in this study. Third, since this study lacks data on measured CCr by urine collection, the predictive ability of the estimated kidney functions could not be evaluated. In addition, cystatin C, an indicator of kidney function in patients with reduced muscle mass, was not measured. Otani et al.¹⁴⁾ examined the prediction accuracy of creatinine-based equations for estimating kidney function using CCr measured by urine collection as an index in bedridden elderly patients. They reported that 75.6% of eCCr (SCr + 0.2) fell within the range of the measured CCr ±30%.¹⁴⁾ The creatinine-based equation for the eGFR of Japanese patients in 2008 was reported to be accurate enough that 75% of all cases fell within the range of measured GFR ±30%,²²⁾ which is comparable to the measured CCr ±30% of eCCr(SCr + 0.2) in bedridden elderly patients. Based on these findings, eCCr (SCr + 0.2) was used as an index of kidney function in this study, and the Bayesian-estimated AUC using this equation was found to be highly predictive in these patients. Although further studies are needed, our findings may assist in selecting kidney function estimation equations to determine the dosage of VCM in bedridden elderly patients.

CONCLUSION

This study demonstrated that the Bayesian approach using eCCr (SCr + 0.2) has the highest prediction accuracy for the AUC_REF in bedridden elderly patients, compared to the other creatinine-based equations analyzed. Our findings highlight the potential of eCCr (SCr + 0.2) for estimating VCM’s AUC by the Bayesian approach in bedridden elderly patients. However, future studies are required with more accurate CCr determination methods such as the urine collection method as well as more accurate methods for AUC calculation such as the multiple-measurement trapezoidal method.

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

This article contains supplementary materials.

REFERENCES

• 1) Gardete S, Tomasz A. Mechanisms of vancomycin resistance in Staphylococcus aureus. J. Clin. Invest., 124, 2836–2840 (2014).
• 2) Rybak MJ, Le J, Lodise TP, Levine DP, Bradley JS, Liu C, Mueller BA, Pai MP, Wong-Beringer A, Rotschafer JC, Rodvold KA, Maples HD, Lomaestro BM. Executive summary: therapeutic monitoring of vancomycin for serious methicillin-resistant staphylococcus aureus infections: a revised consensus guideline and review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, the Pediatric Infectious Diseases Society, and the Society of Infectious Diseases Pharmacists. Pharmacotherapy, 40, 363–367 (2020).
• 3) Oda K, Jono H, Nosaka K, Saito H. Reduced nephrotoxicity with vancomycin therapeutic drug monitoring guided by area under the concentration–time curve against a trough 15–20 µg/mL concentration. Int. J. Antimicrob. Agents, 56, 106109 (2020).
• 4) Tsutsuura M, Moriyama H, Kojima N, Mizukami Y, Tashiro S, Osa S, Enoki Y, Taguchi K, Oda K, Fujii S, Takahashi Y, Hamada Y, Kimura T, Takesue Y, Matsumoto K. The monitoring of vancomycin: a systematic review and meta-analyses of area under the concentration–time curve-guided dosing and trough-guided dosing. BMC Infect. Dis., 21, 153 (2021).
• 5) Hashimoto N, Kimura T, Hamada Y, Niwa T, Hanai Y, Chuma M, Fujii S, Matsumoto K, Shigemi A, Kawamura H, Takahashi Y, Takesue Y. Candidates for area under the concentration–time curve (AUC)-guided dosing and risk reduction based on analyses of risk factors associated with nephrotoxicity in vancomycin-treated patients. J. Glob. Antimicrob. Resist., 27, 12–19 (2021).
• 6) Pai MP, Neely M, Rodvold KA, Lodise TP. Innovative approaches to optimizing the delivery of vancomycin in individual patients. Adv. Drug Deliv. Rev., 77, 50–57 (2014).
• 7) Aljefri DM, Avedissian SN, Rhodes NJ, Postelnick MJ, Nguyen K, Scheetz MH. Vancomycin area under the curve and acute kidney injury: a meta-analysis. Clin. Infect. Dis., 69, 1881–1887 (2019).
• 8) Mohd Tahir NA, Mohd Saffian S, Islahudin FH, Abdul Gafor AH, Makmor-Bakry M. A meta-analysis on the performance of cystatin C- versus creatinine-based eGFR equations in predicting vancomycin clearance. J. Korean Med. Sci., 35, e306 (2020).
• 9) Okamoto G, Sakamoto T, Kimura M, Ukishima Y, Sonoda A, Mori N, Kato Y, Maeda T, Kagawa Y. Serum cystatin C as a better marker of vancomycin clearance than serum creatinine in elderly patients. Clin. Biochem., 40, 485–490 (2007).
• 10) Nakatani S, Maeda K, Akagi J, Ichigi M, Murakami M, Harada Y, Utsumi S, Fukunaga M, Narita Y, Kondo Y, Ishitsuka Y, Irie T, Kadowaki D, Hirata S. Coefficient of determination between estimated and measured renal function in Japanese patients with sarcopenia may be improved by adjusting for muscle mass and sex: a prospective study. Biol. Pharm. Bull., 42, 1350–1357 (2019).
• 11) Cimerman N, Brguljan PM, Krasovec M, Suskovic S, Kos J. Serum cystatin C, a potent inhibitor of cysteine proteinases, is elevated in asthmatic patients. Clin. Chim. Acta, 300, 83–95 (2000).
• 12) Jayagopal V, Keevil BG, Atkin SL, Jennings PE, Kilpatrick ES. Paradoxical changes in cystatin C and serum creatinine in patients with hypo- and hyperthyroidism. Clin. Chem., 49, 680–681 (2003).
• 13) Gounden V, Bhatt H, Jialal I. Renal Function Tests. StatPearls. StatPearls Publishing, Treasure Island, FL (2021).
• 14) Otani T, Kase Y, Kunitomo K, Shimooka K, Kawazoe K, Sato Y, Yamauchi A. What is the correct adjustment protocol for serum creatinine value to reflect renal function in bedridden elderly patients? The Japanese Journal of Nephrology and Pharmacotherapy, 7, 3–12 (2018).
• 15) Maruta M, Tabira T, Makizako H, Sagari A, Miyata H, Yoshimitsu K, Han G, Yoshiura K, Kawagoe M. Impact of outpatient rehabilitation service in preventing the deterioration of the care-needs level among Japanese older adults availing long-term care insurance: a propensity score matched retrospective study. Int. J. Environ. Res. Public Health, 16, 1292 (2019).
• 16) Alqahtani S, Almatrafi A, Bin Aydan N, Alqahtani M, Alzamil F, Alsultan A, Asiri Y. Optimization of vancomycin dosing regimen in cancer patients using pharmacokinetic/pharmacodynamic modeling. Pharmacotherapy, 40, 1192–1200 (2020).
• 17) Izumisawa T, Wakui N, Kaneko T, Soma M, Imai M, Saito D, Hasegawa H, Horino T, Takahashi N. Increased vancomycin clearance in patients with solid malignancies. Biol. Pharm. Bull., 43, 1081–1087 (2020).
• 18) Shimamoto Y, Fukuda T, Tominari S, Fukumoto K, Ueno K, Dong M, Tanaka K, Shirasaka T, Komori K. Decreased vancomycin clearance in patients with congestive heart failure. Eur. J. Clin. Pharmacol., 69, 449–457 (2013).
• 19) Berglund F, Killander J, Pompeius R. Effect of trimethoprim-sulfamethoxazole on the renal excretion of creatinine in man. J. Urol., 114, 802–808 (1975).
• 20) Roubenoff R, Drew H, Moyer M, Petri M, Whiting-O’Keefe Q, Hellmann DB. Oral cimetidine improves the accuracy and precision of creatinine clearance in lupus nephritis. Ann. Intern. Med., 113, 501–506 (1990).
• 21) Lepist E-I, Zhang X, Hao J, Huang J, Kosaka A, Birkus G, Murray BP, Bannister R, Cihlar T, Huang Y, Ray AS. Contribution of the organic anion transporter OAT2 to the renal active tubular secretion of creatinine and mechanism for serum creatinine elevations caused by cobicistat. Kidney Int., 86, 350–357 (2014).
• 22) Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, Yamagata K, Tomino Y, Yokoyama H, Hishida A. Collaborators developing the Japanese equation for estimated GFR. Revised equations for estimated GFR from serum creatinine in Japan. Am. J. Kidney Dis., 53, 982–992 (2009).
• 23) Du Bois D, Du Bois EF. A formula to estimate the approximate surface area if height and weight be known. 1916. Nutrition, 5, 303–311, discussion, 312–313 (1989).
• 24) Hirata S, Shibata A, Miyamura S, Kadowaki D. Theory and practice of accurately assessing the renal function of individual patients. The Japanese Journal of Nephrology and Pharmacotherapy, 5, 3–18 (2016).
• 25) Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron, 16, 31–41 (1976).
• 26) Ando Y, Minami H, Saka H, Ando M, Sakai S, Shimokata K. Adjustment of creatinine clearance improves accuracy of Calvert’s formula for carboplatin dosing. Br. J. Cancer, 76, 1067–1071 (1997).
• 27) Finch NA, Zasowski EJ, Murray KP, Mynatt RP, Zhao JJ, Yost R, Pogue JM, Rybak MJ. A quasi-experiment to study the impact of vancomycin area under the concentration–time curve-guided dosing on vancomycin-associated nephrotoxicity. Antimicrob. Agents Chemother., 61, e01293-17 (2017).
• 28) Meng L, Wong T, Huang S, Mui E, Nguyen V, Espinosa G, Desai J, Holubar M, Deresinski S. Conversion from vancomycin trough concentration-guided dosing to area under the curve-guided dosing using two sample measurements in adults: implementation at an academic medical center. Pharmacotherapy, 39, 433–442 (2019).
• 29) Yasuhara M, Iga T, Zenda H, Okumura K, Oguma T, Yano Y, Hori R. Population pharmacokinetics of vancomycin in Japanese adult patients. Ther. Drug Monit., 20, 139–148 (1998).
• 30) Oda K. Development of software for antimicrobial PK/PD simulation incorporating montecarlo simulation based on Microsoft^® Office Excel. Iryou Yakugaku (Japanese Journal of Pharmaceutical Health Care and Sciences), 37, 335–344 (2011).
• 31) Curry SH, Whelpton R. Appendix: mathematical concepts and the trapezoidal method. Drug Disposition and Pharmacokinetics: from principles to applications. John Wiley & Sons, Ltd., pp. 345–350 (2010).