ORIGINAL ARTICLE
Comparison of Equations With Estimate Renal Function to Predict Serum Vancomycin Concentration in Patients With Spinal Cord Injury—Does the Use of Cystatin C Improve Accuracy? Douglas D. DeCarolis, PharmD,* Joey G. Thorson, PharmD,* Rebecca A. Marraffa, PharmD,* Megan A. Clairmont, PharmD,* and Michael A. Kuskowski, PhD†‡
Background: Vancomycin dose selection is challenging in the spinal cord injury (SCI) population because of the difficulty in accurately estimating the renal function. Creatinine-based equations have been shown to be unreliable in this patient population. Adjusted equations designed for patients with SCI have not been well studied. Cystatin C is an alternative marker of renal function that is less affected by muscle mass and may offer improvement in estimating renal function leading to improved initial dose selection.
Objective: To compare the accuracy of serum creatinine- and serum cystatin C-based equations used in a pharmacokinetic (PK) model to predict steady-state serum vancomycin concentration in an SCI population. The rationale for this study is the need for an improved predictive model to guide initial vancomycin dose design before the availability of a measured steady-state serum concentration.
Methods: Patients with SCI receiving vancomycin with measured serum creatinine, cystatin C, and steady-state serum vancomycin concentration were identified. Serum creatinine- and cystatin C-based equations to estimate renal function were substituted into a population-based PK model to predict steady state-serum vancomycin concentration. Predictions using each equation in the model were compared with the measured steady-state serum vancomycin concentration. Predictive performances using each equation in the PK model were compared. Results: The final study population included 37 patients with SCI. The Chronic Kidney Disease Epidemiology Collaboration cystatin C equation provided significantly less bias, greater precision, and superior accuracy when used in the PK model.
Conclusions: In the SCI population, the use of Chronic Kidney Disease Epidemiology Collaboration cystatin C equation may Received for publication December 21, 2013; accepted February 21, 2014. From the *Minneapolis Veterans Affairs Health Care System, Minneapolis, Minnesota; †Geriatric Research Education and Clinical Center, Minneapolis Veterans Affairs Medical Center, Minneapolis, Minnesota; and ‡Department of Psychiatry, University of Minnesota Medical School, Twin Cities, Minnesota. The authors declare no conflict of interest. Correspondence: Douglas D. DeCarolis, PharmD, Minneapolis VA Health Care System, Pharmacy Service (119), One Veterans Drive Minneapolis, MN 55417 (e-mail: [email protected]). Copyright © 2014 by Lippincott Williams & Wilkins
632
improve initial vancomycin dosing. Further study into this potential is encouraged. Key Words: creatinine, cystatin C, vancomycin, spinal cord injury, pharmacokinetics (Ther Drug Monit 2014;36:632–639)
INTRODUCTION An accurate estimation of creatinine clearance (CrCl) or glomerular filtration rate (GFR) is vital to guide appropriate dose selection of pharmacological agents that rely on renal function for excretion. The Cockgroft–Gault (CG) equation, which uses the serum creatinine concentration to estimate CrCl (eCrCl), has been the standard of practice for drug dosing purposes for many years.1 More recently, other creatininebased equations such as the 4-variable Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration-creatinine (CKD-EPI creatinine) have been shown to be more accurate in estimating GFR (eGFR) and also recommended to guide dosing.2,3 Despite broad acceptance of these equations for use in the general population, their reliance on serum creatinine concentration limits accuracy in patients with significantly altered muscle mass,4 such as those with spinal cord injury (SCI). The National Kidney Disease Education Foundation states “neither eGFR or eCrCl will be accurate in individuals with extremes of body size or muscle mass” and recommends alternative methods to guide dose modification in renal impairment.5 Two alternative methods designed for dose modification in the SCI population have been published providing a creatinine “correction” for substitution into CG (hereafter referred to as CG-SCI creatinine-correction), however, neither has been externally validated.6,7 Cystatin C concentration provides another marker of renal function with properties advantageous to serum creatinine for dosing guidance in the SCI population. Cystatin C is among a class of cysteine protease inhibitors involved in intracellular protein catabolism. A low molecular weight and lack of protein binding allow free filtration by the glomeruli without tubular secretion.8 Reabsorption occurs at the proximal tubule where it is catabolized, thus negating reappearance in serum or clearance into urine.9 In contrast to creatinine, which relies on lean body mass for endogenous production, cystatin C is generated Ther Drug Monit Volume 36, Number 5, October 2014
Ther Drug Monit Volume 36, Number 5, October 2014
from all nucleated cells in the body. Thus, serum cystatin C concentration has been shown to be less affected by changes in muscle mass compared with creatinine.10,11 A meta-analysis of studies comparing the two concluded serum cystatin C concentration as a superior marker of GFR.12 Numerous cystatin C-based eGFR equations have since been developed. Despite positive study results, cystatin C-based eGFR has not shown consistent superiority over creatinine-based equations. The most rigorously tested equations were recently evaluated by the CKD-EPI group who concluded that cystatin C-based eGFR should not replace creatinine-based eGFR in general practice.13 However, it should be noted that studies evaluating cystatin C eGFR have not been performed in populations who may benefit most from its purported advantages, namely those with marked alterations of muscle mass as in SCI. Vancomycin is a glycopeptide antibiotic commonly used to treat serious gram-positive infection including those seen in the SCI population. It is excreted predominately through the renal route with clearance correlated with that of CrCl.14,15 Because of variable kidney function in treated patients, dose modification is often necessary to provide therapeutic antimicrobial concentrations while avoiding excessive exposure that may lead to potential adverse effects. The difficulty in estimating renal function in patients with SCI creates a challenge to the selection of an appropriate initial vancomycin dose design. General recommendations call for measurement of a steady-state trough serum vancomycin concentration to ensure that the dose provides the recommended target concentrations.16 However, such an assessment of proper dose is often delayed for several days because of the time required to achieve vancomycin steady-state conditions, particularly when renal impairment is present. This scenario emphasizes the importance of selecting an initial dose that is more likely to achieve therapeutic and safe target serum concentrations before the availability of a steady-state serum concentration. A more accurate estimation of renal function may provide this improved initial dose design. The objective of this study was to compare the accuracy of serum creatinine- and serum cystatin C-based equations used in a pharmacokinetic (PK) model to predict steady-state serum vancomycin concentration in an SCI population. The rationale for study is the need for an improved predictive model in this population to guide initial vancomycin dose design before the availability of a measured steady-state serum concentration. The serum creatinine- and serum cystatin C-based equations compared in the model include (formulae provided in Table 1): 1. Creatinine-based: CG using actual body weight, CG using ideal body weight (CG-IBW), MDRD, and CKD-EPI creatinine1–3; 2. CG-SCI creatinine correction–based: Mirahmadi et al6 and Lee–Wang7; 3. Cystatin C-based: CKD-EPI cystatin C and CKD-EPI creatinine-cystatin C.13
METHODS The medical center of the Minneapolis Veterans Affairs Health Care System (MVAHCS) consists of 199 acute care 2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
beds, 80 long-term care beds, an 18-bed polytrauma/rehabilitation unit, and a 30-bed SCI unit. A PK dosing service provides initial vancomycin dosing, serum concentration monitoring, and further dose adjustments if necessary for all patients. The initial dose is based on the clinical judgment of the PK dosing service with guidance from a 1-compartment population-based PK model (described below and Appendix 1). Doses are selected to provide a targeted trough vancomycin serum concentration of 10–20 mcg/mL or 15–20 mcg/mL for more serious infection.16 Trough vancomycin serum concentration is measured at steady state to confirm attainment of these target concentrations or as a basis for individualized dose adjustment. Patients receiving vancomycin between January 1, 2009 and December 31, 2012 were identified from a PK service database that tracks all institutional vancomycin use. The criteria for study inclusion consisted of a SCI history resulting in paraplegia or quadriplegia, receipt of vancomycin therapy, and the availability of measured serum creatinine, cystatin C, and trough vancomycin concentrations. Patients were excluded if renal function reflected end-stage renal disease or was considered unstable, defined as a change of serum creatinine greater than 0.3 mg/dL anytime in the 14 days before and after serum vancomycin measurement (total of 28 days). Patients were also excluded if the vancomycin serum concentration did not reflect steady-state conditions, defined as measurement before 5 estimated half-lives of a stable dose. Patients who met the criteria for multiple vancomycin courses were included only for the first course such that each patient is represented only once.
Pharmacokinetic Model The model uses population-based averages to estimate the elimination rate constant and volume of distribution for each patient (Appendix 1). For purposes of this study, each of the creatinine- and cystatin-based equation results was substituted into the Rowland–Tozer equation to estimate the vancomycin half-life and elimination rate.17 Equations derived to estimated GFR adjusted for body surface area were unadjusted to more accurately reflect drug clearance.5,18 The institution-specific population average for vancomycin volume of distribution is 0.84 L/kg based on an analysis of previous multiple serum sample PK studies (unpublished observation). The estimated PK parameters are incorporated into the 1-compartment, individualized PK equations as described by Sawchuk–Zaske to predict steady-state serum vancomycin concentrations.19 Predictions using each creatinine- and cystatin C-based equation in the PK model were compared with the measured vancomycin concentration. Initial analysis identified 1 creatinine-based, 1 CG-SCI creatinine correction–based, and 1 cystatin C-based equation that provided the greatest accuracy when used in the PK model to predict the measured vancomycin concentration. Predictions using the 3 resulting equations were further compared against each other for bias, precision, and accuracy. All vancomycin doses used in the model were those being administered at the time of serum concentration measurement and not based on any study intervention.
633
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
TABLE 1. Equations for eCrCl and eGFR Name of the Equation Creatinine-based equations CG-ABW (actual body weight)† CG-IBW (ideal body weight)†‡ MDRD§
CKD-EPI creatinine§
Sex Male Female Male Female Male Female Male Male Female Female
CG-SCI creatinine correction–based equations Mirahmadi et al†
CKD-EPI creatinine-cystatin C§
Scr Scr Scr Scr
# . # .
0.9 0.9 0.7 0.7
Paraplegia Quadriplegia Scr , 1.0 Scr $ 1.0
Lee–Wang‡¶ Cystatin C-based equations CKD-EPI cystatin C§
Serum Creatinine* (Scr) (mg/dL); Serum Cystatin C (Scys) (mg/L)
Male Male Female Female Male
Scys # 0.8 Scys . 0.8 Scys # 0.8 Scys . 0.8 Scr # 0.9 and Scys # 0.8
Male
Scr # 0.9 and Scys . 0.8
Male
Scr . 0.9 and Scys # 0.8
Male
Scr . 0.9 and Scys . 0.8
Female
Scr # 0.7 and Scys # 0.8
Female
Scr # 0.7 and Scys . 0.8
Female
Scr . 0.7 and Scys # 0.8
Female
Scr . 0.7 and Scys . 0.8
Equation [(140 2 age) · ABW]/(72 · Scr) [(140 2 age) · ABW]/(72 · Scr) · 0.85 [(140 2 age) · IBW/(72 · Scr) [(140 2 age) · IBW/(72 · Scr) · 0.85 175 · Scr21.154 · age20.203 · 1.212 (if black) 175 · Scr21.154 · age20.203 · 0.742 · 1.212 (if black) 141 · (Scr/0.9)20.411 · 0.993Age [·1.159 if black] 141 · (Scr/0.9)21.209 · 0.993Age [·1.159 if black] 144 · (Scr/0.7)20.329 · 0.993Age [·1.159 if black] 144 · (Scr/0.7)21.209 · 0.993Age [·1.159 if black]
Use Use Use Use
“corrected” Scr in CG “corrected” Scr in CG “corrected” SCr in CG actual Scr in CG
133 · (Scys/0.8)20.499 · 0.996Age 133 · (Scys/0.8)21.328 · 0.996Age 133 · (Scys/0.8)20.499 · 0.996Age · 0.932 133 · (Scys/0.8)21.328 · 0.996Age · 0.932 135 · (Scr/0.9)20.207 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.207 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.601 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.601 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.248 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.248 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.601 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.601 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black]
CG-ABW, CG using actual body weight. *To convert values for serum creatinine to micromoles per liter, multiply by 88.4. †eCrCl in mL/min. ‡Ideal body weight (IBW): for males = 50 + 2.3 (height in inches 2 60); for females = 45 + 2.3 (height in inches 2 60). If the actual body weight is less than IBW, the actual body weight is used. §Estimates GFR in mL/min per 1.73 m2. To convert to units of mL/min: multiply result by BSA, then divide by 1.73. ¶Estimates a vancomycin clearance considered equivalent to CrCl in mL/min.
Laboratory Analysis Before March, 2012, the serum creatinine level was measured using Abbott Architect c8000 Clinical Chemistry analyzer (Abbott Diagnostics, Abbott Park, IL). After March 1, 2012, the samples were tested using the Siemens Dimension Vista analyzer (Siemens Healthcare Diagnostics Inc, Newark, DE). Both instruments use the enzymatic creatinine assay, traceable to the National Institute of Standards and Technology creatinine standard reference material 967.
634
Cystatin C was measured by particle-enhanced immunonephelometry using Siemens BNII analyzer. In the United States, a reagent calibrated against the international reference material was not available at the time of study. Cystatin C values represent those measured using the Siemens BNII analyzer in the United States. To reflect the International Federation of Clinical Chemistry, standardization requires conversion factor of 1.174.20 This factor is linear over the entire measurement range on Siemens BN 2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
systems (Siemens Healthcare Diagnostics Inc., personal communication, January 28, 2014). Serum concentrations of vancomycin (before March, 2012) were measured using the Abbott Architect c8000 Clinical Chemistry analyzer using a homogenous microparticle-enhanced turbidimetric immunoassay. After March 1, 2012, testing was performed on the Siemens ADVIA Centaur XP analyzer, which uses a competitive immunoassay using chemiluminescent technology.
Statistical Analysis Accuracy was defined as the ability of the PK model to predict within 30% of the measured serum vancomycin concentration (P30). This measure of accuracy is endorsed for investigations that compare estimates of GFR versus measured GFR.21,22 The authors adopted this as a clinically relevant measure for predicting eventual steady-state serum vancomycin concentration. Bias, also referred to as prediction error, is defined by the mean, across subjects, of the difference between the measured serum vancomycin concentrations from the predicted concentration. Bias measures the tendency and degree of over- or under-prediction with negative values reflecting underestimation. Precision is defined as the mean squared error (MSE) and conveys the size of spread or variation of predictions from the measured value. The MSE was also converted to the root MSE to reflect variation of predicted vancomycin serum concentration in the original units of mcg/mL for improved clinical interpretation. McNemar test was used to statistically compare accuracy as the success versus no success of each equation used in the PK model to achieve P30 accuracy. The Sheiner–Beal method was used to determine statistical differences for bias and precision.23 Computations were performed using the statistical software R (version 3.0.2, R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2013). The study was reviewed and approved by the Minneapolis VA Health Care System Institutional Review Board.
RESULTS Fifty-six patients with SCI were identified as receiving vancomycin with measurements of serum creatinine, cystatin C and vancomycin concentration. Thirteen were excluded because of unstable renal function and 6 because of a serum vancomycin concentration reflecting non–steady-state conditions. The final population consisted of 37 patients with baseline characteristics provided in Table 2. The serum creatinine level ranged from 0.2 to 1.1 mg/dL (MVAHCS reference range, 0.7–1.2 mg/dL) with 36 of 37 patients having a value less than 1.0 mg/dL. The serum cystatin C level ranged more broadly from 0.5 to 2.7 mg/L (MVAHCS reference range, 0.5–1.0 mg/L). The calculated eCrCl and eGFR from each of the studied equations with corresponding estimated vancomycin half-lives are shown in Table 3. Estimates of renal function varied considerably ranging from an average estimated eCrCl of 48.1 mL/min using the Lee–Wang CG-SCI creatinine correction to an eGFR of 208.2 mL/min using MDRD. The median daily vancomycin dose at the time of serum concentration measurement was 2500 mg with a range 2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
TABLE 2. Patient Characteristics Characteristic
n = 37
SCI outcome Paraplegia, n (%) Quadriplegia, n (%) Years since injury, mean (6SD), yrs Sex—Male, n (%) Race, n (%) White Black Native American Asian Age, mean (6SD), yrs Height mean (6SD), inches Actual body weight, mean (6SD), kg Body mass index, mean (6SD)* Body surface area, mean (6SD), m2 Serum creatinine, mean (6SD), mg/dL† Serum cystatin C, mean (6SD), mg/L
15 22 14.4 37
(40.5) (59.5) (14.4) (100)
33 2 1 1 62.0 71.2 85.7 26.2 2.05 0.6
(89.2) (5.9) (2.7) (2.7) (69.0) (62.6) (622.3) (66.8) (60.25) (60.2)
1.1 (60.5)
*Body mass index is equal to weight (kg) divided by height squared (m). †To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
from 750 mg to 4500 mg (7.5–51.9 mg/kg). Dosing frequencies were divided as every 8 hours (n = 13), every 12 hours (n = 20), and every 24 hours (n = 4). The mean measured vancomycin concentration was 18.5 mcg/mL (SD 6 16.8– 20.1), obtained at an average of 17 minutes before the next dose, reflecting trough serum concentrations. The mean predicted vancomycin concentrations and respective measures of P30 accuracy using each equation in the PK model are shown in Table 4. Of the creatinine-based equations, CG-IBW and CKD-EPI creatinine provided the same P30 accuracy, both improved compared with that of
TABLE 3. eCrCl, eGFR, and Corresponding Estimated Vancomycin Half-life eCrCl or eGFR, Mean Estimated Vancomycin (6SD), mL/min Half-life, Mean, hrs Creatinine-based equations CG-ABW CG-IBW MDRD CKD-EPI creatinine CG-SCI creatinine correction–based equations Mirahmadi et al Lee–Wang Cystatin C-based equations CKD-EPI cystatin C CKD-EPI creatininecystatin C
187.0 156.4 208.2 135.4
(113.2) (85.8) (131.2) (32.5)
5.5 6.2 5.0 6.3
123.3 (66) 48.1 (5.4)
7.7 15.0
93.0 (33.9) 112.2 (31.4)
9.6 7.6
635
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
CG using actual body weight and MDRD. The CG-IBW was selected for further comparison because of its more frequent use in clinical practice. Of the 2 CG-SCI creatinine correction equations used in the PK model, Mirahmadi et al provided greater P30 accuracy than that reported by Lee–Wang and selected for further comparative analysis. The CKD-EPI cystatin C equation was chosen from the cystatin C-based equations because its accuracy in the model surpassed that of the CKD-EPI creatinine-cystatin C equation despite the latter equation using 2 markers of renal function. Predictions of the PK model using equations of CGIBW, Mirahmadi et al, and CKD-EPI cystatin C were further compared against the measured vancomycin serum concentrations for bias and precision (Table 5). The use of CKD-EPI cystatin C eGFR was found to be least biased and most precise. When predictive abilities were compared directly against each other (Table 6), the CKD-EPI cystatin C eGFR in the PK model was statistically superior compared with both CG-IBW and Mirahmadi et al. The comparative P30 accuracy was superior with the use of CKD-EPI cystatin C equation in the PK model compared with CG-IBW and Mirahmadi et al (Table 7). The use of CG-SCI creatinine correction of Mirahmadi et al in the model provided statistically improved predictive accuracy compared with CG-IBW, however, the P30 accuracy remained less than 50%. All results of bias, precision, and accuracy remained consistent regardless of years since SCI, paraplegia versus quadriplegia, age, baseline serum creatinine, and body weight.
TABLE 4. Measured and Predicted Serum Vancomycin Concentration With the Use of eCrCl or eGFR in PK Model Vancomycin Serum Accuracy (P30)% of Concentration, Mean, Predictions Within the (mcg/mL) (95%CI) 30% Measured Measured vancomycin concentration Predictions using creatinine-based equations in PK model CG-ABW CG-IBW MDRD CKD-EPI creatinine Predictions using CGSCI creatinine correction equations in PK model Mirahmadi et al Lee–Wang Predictions using cystatin C-based equations in PK model CKD-EPI cystatin C CKD-EPI creatininecystatin C
18.5 (16.8–20.1)
7.0 8.6 6.0 8.9
(7.0–8.4) (7.0–10.2) (4.7–7.3) (7.6–10.2)
10.8 21.6 8.1 21.6
11.9 (9.7–14.2) 29.3 (25.5–33.1)
40.5 16.2
14.7 (12.8–16.7) 11.2 (9.9–12.7)
67.6 35.1
CG-ABW, CG using actual body weight.
636
—
TABLE 5. Predictive Bias and Precision With the Use of eCrCl or eGFR in PK Model Compared to Measured Serum Vancomycin Concentration
Predictions using CG-IBW in PK model Predictions using Mirahmadi et al CG-SCI creatinine correction in PK model Predictions using CKD-EPI cystatin C in PK model
Bias* (95%CI), Mean, mcg/mL
Precision (RMSE),† Mean, mcg/mL
29.9 (211.9 to 27.9)
11.5 (9.6 to 13.2)
26.5 (29.0 to 24.1)
9.7 (7.7 to 11.3)
23.8 (25.2 to 2.4)
5.6 (4.5 to 6.5)
*Bias, which is defined as prediction error, is the mean, across subjects, of difference between the measured serum vancomycin concentrations from the predicted concentration. Lower values reflect less bias. †Precision, as measured by root mean squared error (RMSE), conveys the size of spread or variation of predictions from the measured value. Lower values reflect improved precision.
DISCUSSION Vancomycin dose selection in the SCI population can be problematic because of the unique challenges of accurately assessing renal function. In the absence of specific guidance for this population, general recommendations call for the monitoring of a serum trough concentration at steady-state.16 Because it may take several days to reach steady-state, assessment of the initial dose is delayed. This raises the importance of providing an initial dose most likely to deliver prompt and effective antimicrobial concentrations, particularly in an acutely ill patient. A more accurate initial dose will also minimize potentially excessive concentrations that may lead to adverse effect. Results of this study show, in an SCI population, a PK model using the CKD-EPI-cystatin C eGFR provided superior predictions of measured steady-state serum vancomycin concentrations than other equations. These results reflect a biologically plausible mechanism of a decreased influence of muscle mass on serum cystatin C concentration compared with serum creatinine. If correct, the use of a cystatin-C-based eGFR would improve estimation of renal function in SCI and lead to more accurate predictions of vancomycin clearance and improved dosing. An improved predictive ability would provide needed guidance for designing initial vancomycin dosing regimens for patients with SCI before the availability of measured serum concentrations. This concept is supported by other investigations of cystatin C in patients with SCI or reduced muscle mass. Two studies in SCI populations found a superior correlation of serum cystatin C than that of serum creatinine to measured CrCl or GFR.24,25 Comparable results in patients without SCI with reduced muscle mass have also been published.26,27 Other investigations have also found improved predictive ability of steady-state serum vancomycin concentration using cystatin C-based eGFR versus a creatinine-based equations in 2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
TABLE 6. Direct Comparative Predictive Ability Using eCrCl and eGFR Equations in PK Model: Bias and Precision Comparison
Difference in Bias, mcg/m
Difference in Precision (MSE),* mcg/mL
CKD-EPI cystatin C 6.1 (3.8–8.4)†‡ 2102.1 (289.3 to 2114.9)†§ versus CG-IBW CKD-EPI cystatin C 2.7 (0.36–5.1)†‡ 263.4 (232.6 to 294.2)†§ versus Mirahmadi et al *Per Sheiner–Beal statistical comparison, the MSE is used for statistical comparison of precision.23 †P , 0.05. ‡The greater the value, the greater the difference in prediction bias favoring the first comparator (CKD-EPI cystatin C use provided less bias than that of CG-IBW and Mirahmadi et al). §The more negative values reflect a greater difference in precision favoring the first comparator (CKD-EPI cystatin C use provided greater precision than that of CG-IBW and Mirahmadi et al).
patients without SCI with reduced muscle mass.28–31 To the authors’ knowledge, this is the first report of improved predictive ability of serum vancomycin concentration using a cystatin C-based eGFR in an SCI population. Results also corroborate previous evidence of significant deficiencies of creatinine-based equations in the SCI population. SCI results in muscle denervation causing paralysis, immobility, and diffuse muscle atrophy. Because muscle mass is the major source for endogenous creatinine, studies have demonstrated significantly decreased creatinine production in patients with SCI compared with controls.6,32,33 These studies consistently cite lower serum creatinine concentrations in SCI compared with control patients with similar degrees of renal function. When used in creatinine-based equations, these lower serum creatinine concentrations lead to significant overestimation of renal function in the SCI population.34–36 This is consistent with results of this study in that serum creatinine concentrations were less than 1.0 mg/dL in 36 of the 37 subjects with 3 patients having concentrations as low as 0.2 mg/dL. All creatinine-based equations used in the PK model provided considerable under-prediction of the measured steady-state serum vancomycin concentrations reflecting high estimates of drug clearance. Unless clinical judgment supersedes, the use of creatinine-based equations may conceal clinically relevant reductions in renal function in SCI and lead to appreciably higher doses and potential adverse effect.37 Although these limitations of creatinine-based equations are well known in SCI, the authors thought it prudent to include them because of their common reference in drug dosing recommendations, automatic appearance in laboratory result
TABLE 7. Accuracy of the PK Model Using eCrCl and eGFR Methods to Predict Serum Vancomycin Concentration Within the 30% Measured (P30) P30
P value
67.6% versus 21.6% 67.6% versus 40.5%
,0.001 0.013
40.5% versus 21.6%
0.039
Equation CKD-EPI cystatin C versus CG-IBW CKD-EPI cystatin C versus Mirahmadi et al Mirahmadi et al versus CG-IBW
2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
reporting, and an absence of well-known alternatives. The more recent MDRD and CKD-EPI creatinine were also included because of an absence of data examining their use in the SCI population. This report provides new data regarding 2 alternative methods to assess renal function for drug dosing purposes in the SCI population. Both account for the SCI-related decreased creatinine production by providing a creatinine correction factor for substitution into CG. Mirahmadi et al use a correction based on their measurements of decreased endogenous creatinine production in paraplegia (20%) and quadriplegia (40%) compared with control subjects. The method of Lee–Wang relies on adjusting all serum creatinine concentrations less than 1.0 mg/dL upward to that value. This corrected creatinine is then substituted into a modified CG-IBW equation derived using linear regression from their SCI population to estimate vancomycin clearance and dose. The predictive ability of the PK model using the Mirahmadi et al CG-SCI creatinine correction was found more accurate. The authors presume the comparatively decreased accuracy of the Lee–Wang method to be due to the practice of adjusting varied serum creatinine concentrations to an identical value. For example, patients with a measured serum creatinine from 0.2 mg/dL to 1.0 mg/dL will use an identical corrected creatinine value of 1.0 mg/dL to estimate vancomycin clearance. In contrast, no correction is provided for a serum creatinine of 1.0 mg/dL or larger implying no change in the creatinine production despite the well-known reduction in patients with SCI. The authors note the Lee–Wang method seemed to provide better results with baseline serum creatinine values closer to 1.0 mg/dL but lost predictive abilities when lower values necessitated more substantial adjustment. Although the PK model predictions using Mirahmadi et al were superior to all other creatinine-based equations, its P30 accuracy was still less than 50%. Limitations of this study include its observational and retrospective nature. The sample size of 37 is small and consisted of only male patients. A definition of unstable renal function specifically for the SCI population is not available, and the one used could be questioned. In the absence of literature guidance, the authors adopted a strict definition of unstable renal function as any change of serum creatinine of 0.3 mg/dL over a 28-day period compared with the definition of the Kidney Disease-Improving Global Outcomes group of the same change over 48 hours in the general population.38 It must be emphasized that these results are not applicable when renal function is unstable. It is possible that a systematic bias of the population-based estimates used in the PK model may affect results; however, this would be expected to affect all equations to a similar degree. The MDRD equation is known to be less accurate when GFR is greater than 60 mL/min per 1.73 m2. The authors included MDRD because of its widespread availability in routine laboratory reports39 and the lack of data in SCI. Including a measured CrCl through a timed 24-hours urine collection would have been desirable but was not available for this analysis. However, measured CrCl has been shown to be unreliable in the SCI population.21,40,41 It also requires
637
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
24 hours for collection and additional time for processing further limiting usefulness for initial dose design. A full PK analysis with multiple vancomycin samplings would provide more robust data, however, our methods reflect current clinical practice and the latest guidelines for monitoring vancomycin trough concentration at steady-state. Of importance, the use of the population-based PK model is not intended to replace the individualized method of assessing serum vancomycin concentration at steady state, but rather as an adjunct to provide predictions to assist with dosing before its availability. There are limitations to the use of serum cystatin C measurement. There are non-GFR determinants of cystatin C such as inflammation, diabetes, and fat mass of unknown significance, which we did not control for.42 Despite its increasing use, its availability may be limited and costs may be higher compared with creatinine. At our institution, serum cystatin C measurement is available daily with a short turnaround time at an acquisition cost less than $10.00. Medicare reimbursement costs in 2013 are $18.69 for serum cystatin C and $7.04 for serum creatinine.43 It should be noted that a cystatin C reagent calibrated against the international certified reference material was not available at the time of the study. Actual results reflect cystatin C as measured by particle-enhanced immunonephelometry on the Siemens BNII analyzer in the United States. However, applying the conversion factor provided by Siemens Healthcare Diagnostics to reflect International Federation of Clinical Chemistry standardization did not significantly alter the results.
CONCLUSIONS In patients with SCI, the use of the CKD-EPI cystatin C equation in a population-based PK model provided significantly greater accuracy with less bias and greater precision for predicting steady-state serum vancomycin concentration compared with other equations that estimate renal function. This superior predictive ability may provide necessary guidance for improved initial dose selection in the SCI population before, or in the absence of a measured serum vancomycin concentration. If a serum cystatin C measurement is unavailable, the CG-SCI creatinine correction equation of Mirahmadi et al improved predictive ability compared with other creatinine-based equations. Both methods would require close monitoring of serum vancomycin concentration to assure optimal dosing. We encourage further study to validate this finding in additional SCI populations. REFERENCES 1. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41. 2. Levey AS, Coresh J, Greene T, et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006;145:247–254. 3. Levey AS, Stevens LA, Schmid CH, et al; for the Chronic Kidney Disease Epidemiology Collaboration. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. 4. Perrone RD, Madias NE, Levey AS. S.Cr. as an index of renal function: new insights into old concepts. Clin Chem. 1992;38:1933–1953.
638
5. National Kidney Disease Education Program. Chronic kidney disease and drug dosing: information for providers (Revised January 2010). Available at: http://www.nkdep.nih.gov/resources/ckd-drug-dosing-508. pdf. Accessed August 30, 2013. 6. Mirahmadi MK, Byrne C, Barton C, et al. Prediction of creatinine clearance from serum creatinine in spinal cord injury patients. Paraplegia. 1983;21:23–29. 7. Lee JP, Wang YJ. Testing the predictive ability of the “spinal cord injury equation” in estimating vancomycin clearance. Am J Health Syst Pharm. 2013;70:669–674. 8. Simonsen O, Grubb A, Thysell H. The blood serum concentration of cystatin C (gamma-trace) as a measure of the glomerular filtration rate. Scand J Clin Lab Invest. 1985;45:97–101. 9. Jacobsson B, Lignelid H, Bergerheim US. Transthyretin and CysC are catabolized in proximal tubular epithelial cells and the proteins are not useful as markers for renal cell carcinomas. Histopathology. 1995;26: 559–564. 10. Vinge E, Lindergard B, Nilsson-Ehle P, et al. Relationships among serum cystatin C, serum creatinine, lean tissue mass and glomerular filtration rate in healthy adults. Scand J Clin Lab Invest. 1999;59: 587–592. 11. Baxmann AC, Ahmed MS, Marques NC, et al. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clin J Am Soc Nephrol. 2008;3:348–354. 12. Dharnidharka VR, Kwon C, Stevens G. Cystatin C is superior to S.Cr. as a marker of kidney function: a meta-analysis. Am J Kidney Dis. 2002;40: 221–226. 13. Inker LA, Schmid CH, Tighiouart H, et al. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med. 2012;367: 20–29. 14. Cheung RPF, DiPiro JT. Vancomycin: an update. Pharmacotherapy. 1986;6:153–169. 15. Matzke GR, McGory RW, Halstenson CE, et al. Pharmacokinetics of vancomycin in patients with various degrees of renal function. Antimicrob Agents Chemother. 1984;25:433–437. 16. Rybak M, Lomaestro B, Rotschafer JC, et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am J Health Syst Pharm. 2009;66:82–98. 17. Rowland M, Tozer TN. Clinical Pharmacokinetics, Concepts and Applications. 2nd ed. Philadephia, PA: Lea & Febiger; 1989: 238–254. 18. Stevens LA, Nolin T, Levey AS. In reply to ‘estimated GFR for drug dosing: a bedside formula,’ ‘drug dose adjustments in patients with renal impairment,’ ‘use of the MDRD study equation for drug dosing,’ and ‘estimated GFR vs creatinine clearance for drug dosing’. Am J Kidney Dis. 2009;54:985–986. 19. Sawchuk RJ, Zaske DE. Pharmacokinetics of dosing regimens which utilize multiple intravenous infusions: gentamicin in burn patients. J Pharmacokinet Biopharm. 1976;4:183–195. 20. Grubb A, Blirup-Jensen S, Lindstrom V, et al. First certified reference material for cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med. 2010;48:1619–1621. 21. National Kidney Foundation. KDOQI clinical practice guidelines for chronic kidney disease: evaluation, classification and stratification. part 5. Evaluation of laboratory measurements for clinical assessment of kidney disease. Am J Kidney Dis. 2002;39:S76–S110. 22. Earley A, Miskulin D, Lamb EJ, et al. Estimating equations for glomerular filtration rate in the era of creatinine standardization: a systematic review. Ann Intern Med. 2012;156:785–795. 23. Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm. 1981;9:503–512. 24. Thomassen SA, Johannesen IL, Erlandsen EJ, et al. Serum cystatin C as a marker of the renal function in patients with spinal cord injury. Spinal Cord. 2002;40:524–528. 25. Jenkins MA, Brown DJ, Ierino FL, et al. Cystatin C for estimation of glomerular filtration rate in patients with spinal cord injury. Ann Clin Biochem. 2003;40:364–368. 26. Beringer PM, Hidayat L, Heed A, et al. GFR estimates using cystatin C are superior to serum creatinine in cystic fibrosis. J Cyst Fibros. 2009;8: 19–25.
2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
27. Delanaye P, Cavalier E, Radermecker RP, et al. Cystatin C or creatinine for detection of stage 3 chronic kidney disease in anorexia nervosa. Nephron Clin Pract. 2008;110:c158–c163. 28. Tanaka A, Suemaru K, Otsuka T, et al. Estimation of the initial dose setting of vancomycin therapy with use of cystatin C as a new marker of renal function. Ther Drug Monit. 2007;29: 261–264. 29. Tanaka A, Aiba T, Otsuka T, et al. Population pharmacokinetic analysis of vancomycin using serum cystatin C as a marker of renal function. Antimicrob Agents Chemother. 2010;54: 778–782. 30. Suzukia A, Imanishi Y, Nakanob S, et al. Usefulness of serum cystatin C to determine the dose of vancomycin in critically ill patients. J Pharm Pharmacol. 2010;62:901–907. 31. Okamoto G, Sakamoto T, Kimura M, et al. Serum cystatin C as a better marker of vancomycin clearance than serum creatinine in elderly patients. Clin Biochem. 2007;40:485–490. 32. Mohler JL, Barton SD, Blouin RA, et al. The evaluation of creatinine clearance in spinal cord injury patients. J Urol. 1986; 136:366–369. 33. Sawyer WT, Hutchins K. Assessment and predictability of renal function in spinal cord injury patients. Urology. 1982;19:377–380. 34. Mohler JL, Ellison MF, Flanigan RC. Creatinine clearance prediction in spinal cord injury patients: comparison of 6 prediction equations. J Urol. 1988;139:706–709. 35. Kaji D, Strauss I, Kahn T. Serum creatinine in patients with spinal cord injury. Mt Sinai J Med. 1990;57:160–164. 36. MacDiarmid SA, McIntyre WJ, Anthony A, et al. Monitoring of renal function in patients with spinal cord injury. BJU Int. 2000;85: 1014–1018. 37. Vaidyanathan S, Watt JWH, Singh G, et al. Dosage of once-daily gentamicin in spinal cord injury patients. Spinal Cord. 2000;38: 197–198. 38. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Inter Suppl. 2012;2:1–138. 39. Accetta NA, Gladstone EH, DiSogra C, et al. Prevalence of estimated GFR reporting among US clinical laboratories. Am J Kidney Dis. 2008; 52:778–787. 40. Walser M. Assessing renal function from creatinine measurements in adults with chronic renal failure. Am J Kidney Dis. 1998;32: 23–31. 41. Sepahpanah F, Burns SP, McKnight B, et al. Role of creatinine clearance as a screening test in persons with spinal cord injury. Arch Phys Med Rehabil. 2006;87:524–528. 42. Stevens LA, Schmid CH, Greene T, et al. Factors other than glomerular filtration rate affect serum cystatin C levels. Kidney Int. 2009;75: 652–660. 43. Centers for Medicare and Medicaid Services. Clinical laboratory fee schedule. Available at: http://www.cms.gov/Medicare/MedicareFee-for-Service-Payment/ClinicalLabFeeSched/clinlab.html. Accessed August 30, 2013.
2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
APPENDIX 1. One-Compartment, PopulationBased PK Model for the Prediction of SteadyState Serum Vancomycin Concentration Step 1. Estimate half-life and convert to elimination rate constant (k) a. Calculate eCrCl or eGFR in mL/min and place the result in equation, Estimated t1=217 ¼
vancomcyin half -life in normal renal functionð7:9 hoursÞ15 ; 1 2 ffe · ½1 2 ðeGFR=100Þg
fe = fraction of drug excreted unchanged in urine (0.9). b. Determine k from the estimated half-life. k¼
0:693 : estimated half -life
Step 2. Estimate Volume of Distribution (Vd) MVAMC population average Vd = 0.84 L/kg.
Step 3. Prediction of steady-state vancomycin serum concentration19: a. Predict the peak concentration (Cpmax)(mcg/mL)
Cpmax ¼
0 dose=infusion timeðmg=hrÞ 1 2 e 2 kt · ðk · Vd Þ ð1 2 e 2 kT Þ
T = dose interval, t0 = infusion time. b. Predict the concentration at trough concentration (for study, this time was the same as the time of the actual measured vancomycin concentration): 0
Cpmin ¼ Cpmax · e 2 kðT 2 t Þ
639
Comparison of Equations With Estimate Renal Function to Predict Serum Vancomycin Concentration in Patients With Spinal Cord Injury—Does the Use of Cystatin C Improve Accuracy? Douglas D. DeCarolis, PharmD,* Joey G. Thorson, PharmD,* Rebecca A. Marraffa, PharmD,* Megan A. Clairmont, PharmD,* and Michael A. Kuskowski, PhD†‡
Background: Vancomycin dose selection is challenging in the spinal cord injury (SCI) population because of the difficulty in accurately estimating the renal function. Creatinine-based equations have been shown to be unreliable in this patient population. Adjusted equations designed for patients with SCI have not been well studied. Cystatin C is an alternative marker of renal function that is less affected by muscle mass and may offer improvement in estimating renal function leading to improved initial dose selection.
Objective: To compare the accuracy of serum creatinine- and serum cystatin C-based equations used in a pharmacokinetic (PK) model to predict steady-state serum vancomycin concentration in an SCI population. The rationale for this study is the need for an improved predictive model to guide initial vancomycin dose design before the availability of a measured steady-state serum concentration.
Methods: Patients with SCI receiving vancomycin with measured serum creatinine, cystatin C, and steady-state serum vancomycin concentration were identified. Serum creatinine- and cystatin C-based equations to estimate renal function were substituted into a population-based PK model to predict steady state-serum vancomycin concentration. Predictions using each equation in the model were compared with the measured steady-state serum vancomycin concentration. Predictive performances using each equation in the PK model were compared. Results: The final study population included 37 patients with SCI. The Chronic Kidney Disease Epidemiology Collaboration cystatin C equation provided significantly less bias, greater precision, and superior accuracy when used in the PK model.
Conclusions: In the SCI population, the use of Chronic Kidney Disease Epidemiology Collaboration cystatin C equation may Received for publication December 21, 2013; accepted February 21, 2014. From the *Minneapolis Veterans Affairs Health Care System, Minneapolis, Minnesota; †Geriatric Research Education and Clinical Center, Minneapolis Veterans Affairs Medical Center, Minneapolis, Minnesota; and ‡Department of Psychiatry, University of Minnesota Medical School, Twin Cities, Minnesota. The authors declare no conflict of interest. Correspondence: Douglas D. DeCarolis, PharmD, Minneapolis VA Health Care System, Pharmacy Service (119), One Veterans Drive Minneapolis, MN 55417 (e-mail: [email protected]). Copyright © 2014 by Lippincott Williams & Wilkins
632
improve initial vancomycin dosing. Further study into this potential is encouraged. Key Words: creatinine, cystatin C, vancomycin, spinal cord injury, pharmacokinetics (Ther Drug Monit 2014;36:632–639)
INTRODUCTION An accurate estimation of creatinine clearance (CrCl) or glomerular filtration rate (GFR) is vital to guide appropriate dose selection of pharmacological agents that rely on renal function for excretion. The Cockgroft–Gault (CG) equation, which uses the serum creatinine concentration to estimate CrCl (eCrCl), has been the standard of practice for drug dosing purposes for many years.1 More recently, other creatininebased equations such as the 4-variable Modification of Diet in Renal Disease (MDRD) and Chronic Kidney Disease Epidemiology Collaboration-creatinine (CKD-EPI creatinine) have been shown to be more accurate in estimating GFR (eGFR) and also recommended to guide dosing.2,3 Despite broad acceptance of these equations for use in the general population, their reliance on serum creatinine concentration limits accuracy in patients with significantly altered muscle mass,4 such as those with spinal cord injury (SCI). The National Kidney Disease Education Foundation states “neither eGFR or eCrCl will be accurate in individuals with extremes of body size or muscle mass” and recommends alternative methods to guide dose modification in renal impairment.5 Two alternative methods designed for dose modification in the SCI population have been published providing a creatinine “correction” for substitution into CG (hereafter referred to as CG-SCI creatinine-correction), however, neither has been externally validated.6,7 Cystatin C concentration provides another marker of renal function with properties advantageous to serum creatinine for dosing guidance in the SCI population. Cystatin C is among a class of cysteine protease inhibitors involved in intracellular protein catabolism. A low molecular weight and lack of protein binding allow free filtration by the glomeruli without tubular secretion.8 Reabsorption occurs at the proximal tubule where it is catabolized, thus negating reappearance in serum or clearance into urine.9 In contrast to creatinine, which relies on lean body mass for endogenous production, cystatin C is generated Ther Drug Monit Volume 36, Number 5, October 2014
Ther Drug Monit Volume 36, Number 5, October 2014
from all nucleated cells in the body. Thus, serum cystatin C concentration has been shown to be less affected by changes in muscle mass compared with creatinine.10,11 A meta-analysis of studies comparing the two concluded serum cystatin C concentration as a superior marker of GFR.12 Numerous cystatin C-based eGFR equations have since been developed. Despite positive study results, cystatin C-based eGFR has not shown consistent superiority over creatinine-based equations. The most rigorously tested equations were recently evaluated by the CKD-EPI group who concluded that cystatin C-based eGFR should not replace creatinine-based eGFR in general practice.13 However, it should be noted that studies evaluating cystatin C eGFR have not been performed in populations who may benefit most from its purported advantages, namely those with marked alterations of muscle mass as in SCI. Vancomycin is a glycopeptide antibiotic commonly used to treat serious gram-positive infection including those seen in the SCI population. It is excreted predominately through the renal route with clearance correlated with that of CrCl.14,15 Because of variable kidney function in treated patients, dose modification is often necessary to provide therapeutic antimicrobial concentrations while avoiding excessive exposure that may lead to potential adverse effects. The difficulty in estimating renal function in patients with SCI creates a challenge to the selection of an appropriate initial vancomycin dose design. General recommendations call for measurement of a steady-state trough serum vancomycin concentration to ensure that the dose provides the recommended target concentrations.16 However, such an assessment of proper dose is often delayed for several days because of the time required to achieve vancomycin steady-state conditions, particularly when renal impairment is present. This scenario emphasizes the importance of selecting an initial dose that is more likely to achieve therapeutic and safe target serum concentrations before the availability of a steady-state serum concentration. A more accurate estimation of renal function may provide this improved initial dose design. The objective of this study was to compare the accuracy of serum creatinine- and serum cystatin C-based equations used in a pharmacokinetic (PK) model to predict steady-state serum vancomycin concentration in an SCI population. The rationale for study is the need for an improved predictive model in this population to guide initial vancomycin dose design before the availability of a measured steady-state serum concentration. The serum creatinine- and serum cystatin C-based equations compared in the model include (formulae provided in Table 1): 1. Creatinine-based: CG using actual body weight, CG using ideal body weight (CG-IBW), MDRD, and CKD-EPI creatinine1–3; 2. CG-SCI creatinine correction–based: Mirahmadi et al6 and Lee–Wang7; 3. Cystatin C-based: CKD-EPI cystatin C and CKD-EPI creatinine-cystatin C.13
METHODS The medical center of the Minneapolis Veterans Affairs Health Care System (MVAHCS) consists of 199 acute care 2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
beds, 80 long-term care beds, an 18-bed polytrauma/rehabilitation unit, and a 30-bed SCI unit. A PK dosing service provides initial vancomycin dosing, serum concentration monitoring, and further dose adjustments if necessary for all patients. The initial dose is based on the clinical judgment of the PK dosing service with guidance from a 1-compartment population-based PK model (described below and Appendix 1). Doses are selected to provide a targeted trough vancomycin serum concentration of 10–20 mcg/mL or 15–20 mcg/mL for more serious infection.16 Trough vancomycin serum concentration is measured at steady state to confirm attainment of these target concentrations or as a basis for individualized dose adjustment. Patients receiving vancomycin between January 1, 2009 and December 31, 2012 were identified from a PK service database that tracks all institutional vancomycin use. The criteria for study inclusion consisted of a SCI history resulting in paraplegia or quadriplegia, receipt of vancomycin therapy, and the availability of measured serum creatinine, cystatin C, and trough vancomycin concentrations. Patients were excluded if renal function reflected end-stage renal disease or was considered unstable, defined as a change of serum creatinine greater than 0.3 mg/dL anytime in the 14 days before and after serum vancomycin measurement (total of 28 days). Patients were also excluded if the vancomycin serum concentration did not reflect steady-state conditions, defined as measurement before 5 estimated half-lives of a stable dose. Patients who met the criteria for multiple vancomycin courses were included only for the first course such that each patient is represented only once.
Pharmacokinetic Model The model uses population-based averages to estimate the elimination rate constant and volume of distribution for each patient (Appendix 1). For purposes of this study, each of the creatinine- and cystatin-based equation results was substituted into the Rowland–Tozer equation to estimate the vancomycin half-life and elimination rate.17 Equations derived to estimated GFR adjusted for body surface area were unadjusted to more accurately reflect drug clearance.5,18 The institution-specific population average for vancomycin volume of distribution is 0.84 L/kg based on an analysis of previous multiple serum sample PK studies (unpublished observation). The estimated PK parameters are incorporated into the 1-compartment, individualized PK equations as described by Sawchuk–Zaske to predict steady-state serum vancomycin concentrations.19 Predictions using each creatinine- and cystatin C-based equation in the PK model were compared with the measured vancomycin concentration. Initial analysis identified 1 creatinine-based, 1 CG-SCI creatinine correction–based, and 1 cystatin C-based equation that provided the greatest accuracy when used in the PK model to predict the measured vancomycin concentration. Predictions using the 3 resulting equations were further compared against each other for bias, precision, and accuracy. All vancomycin doses used in the model were those being administered at the time of serum concentration measurement and not based on any study intervention.
633
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
TABLE 1. Equations for eCrCl and eGFR Name of the Equation Creatinine-based equations CG-ABW (actual body weight)† CG-IBW (ideal body weight)†‡ MDRD§
CKD-EPI creatinine§
Sex Male Female Male Female Male Female Male Male Female Female
CG-SCI creatinine correction–based equations Mirahmadi et al†
CKD-EPI creatinine-cystatin C§
Scr Scr Scr Scr
# . # .
0.9 0.9 0.7 0.7
Paraplegia Quadriplegia Scr , 1.0 Scr $ 1.0
Lee–Wang‡¶ Cystatin C-based equations CKD-EPI cystatin C§
Serum Creatinine* (Scr) (mg/dL); Serum Cystatin C (Scys) (mg/L)
Male Male Female Female Male
Scys # 0.8 Scys . 0.8 Scys # 0.8 Scys . 0.8 Scr # 0.9 and Scys # 0.8
Male
Scr # 0.9 and Scys . 0.8
Male
Scr . 0.9 and Scys # 0.8
Male
Scr . 0.9 and Scys . 0.8
Female
Scr # 0.7 and Scys # 0.8
Female
Scr # 0.7 and Scys . 0.8
Female
Scr . 0.7 and Scys # 0.8
Female
Scr . 0.7 and Scys . 0.8
Equation [(140 2 age) · ABW]/(72 · Scr) [(140 2 age) · ABW]/(72 · Scr) · 0.85 [(140 2 age) · IBW/(72 · Scr) [(140 2 age) · IBW/(72 · Scr) · 0.85 175 · Scr21.154 · age20.203 · 1.212 (if black) 175 · Scr21.154 · age20.203 · 0.742 · 1.212 (if black) 141 · (Scr/0.9)20.411 · 0.993Age [·1.159 if black] 141 · (Scr/0.9)21.209 · 0.993Age [·1.159 if black] 144 · (Scr/0.7)20.329 · 0.993Age [·1.159 if black] 144 · (Scr/0.7)21.209 · 0.993Age [·1.159 if black]
Use Use Use Use
“corrected” Scr in CG “corrected” Scr in CG “corrected” SCr in CG actual Scr in CG
133 · (Scys/0.8)20.499 · 0.996Age 133 · (Scys/0.8)21.328 · 0.996Age 133 · (Scys/0.8)20.499 · 0.996Age · 0.932 133 · (Scys/0.8)21.328 · 0.996Age · 0.932 135 · (Scr/0.9)20.207 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.207 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.601 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 135 · (Scr/0.9)20.601 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.248 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.248 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.601 · (Scys/0.8)20.375 · 0.995Age [·1.08 if black] 130 · (Scr/0.7)20.601 · (Scys/0.8)20.711 · 0.995Age [·1.08 if black]
CG-ABW, CG using actual body weight. *To convert values for serum creatinine to micromoles per liter, multiply by 88.4. †eCrCl in mL/min. ‡Ideal body weight (IBW): for males = 50 + 2.3 (height in inches 2 60); for females = 45 + 2.3 (height in inches 2 60). If the actual body weight is less than IBW, the actual body weight is used. §Estimates GFR in mL/min per 1.73 m2. To convert to units of mL/min: multiply result by BSA, then divide by 1.73. ¶Estimates a vancomycin clearance considered equivalent to CrCl in mL/min.
Laboratory Analysis Before March, 2012, the serum creatinine level was measured using Abbott Architect c8000 Clinical Chemistry analyzer (Abbott Diagnostics, Abbott Park, IL). After March 1, 2012, the samples were tested using the Siemens Dimension Vista analyzer (Siemens Healthcare Diagnostics Inc, Newark, DE). Both instruments use the enzymatic creatinine assay, traceable to the National Institute of Standards and Technology creatinine standard reference material 967.
634
Cystatin C was measured by particle-enhanced immunonephelometry using Siemens BNII analyzer. In the United States, a reagent calibrated against the international reference material was not available at the time of study. Cystatin C values represent those measured using the Siemens BNII analyzer in the United States. To reflect the International Federation of Clinical Chemistry, standardization requires conversion factor of 1.174.20 This factor is linear over the entire measurement range on Siemens BN 2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
systems (Siemens Healthcare Diagnostics Inc., personal communication, January 28, 2014). Serum concentrations of vancomycin (before March, 2012) were measured using the Abbott Architect c8000 Clinical Chemistry analyzer using a homogenous microparticle-enhanced turbidimetric immunoassay. After March 1, 2012, testing was performed on the Siemens ADVIA Centaur XP analyzer, which uses a competitive immunoassay using chemiluminescent technology.
Statistical Analysis Accuracy was defined as the ability of the PK model to predict within 30% of the measured serum vancomycin concentration (P30). This measure of accuracy is endorsed for investigations that compare estimates of GFR versus measured GFR.21,22 The authors adopted this as a clinically relevant measure for predicting eventual steady-state serum vancomycin concentration. Bias, also referred to as prediction error, is defined by the mean, across subjects, of the difference between the measured serum vancomycin concentrations from the predicted concentration. Bias measures the tendency and degree of over- or under-prediction with negative values reflecting underestimation. Precision is defined as the mean squared error (MSE) and conveys the size of spread or variation of predictions from the measured value. The MSE was also converted to the root MSE to reflect variation of predicted vancomycin serum concentration in the original units of mcg/mL for improved clinical interpretation. McNemar test was used to statistically compare accuracy as the success versus no success of each equation used in the PK model to achieve P30 accuracy. The Sheiner–Beal method was used to determine statistical differences for bias and precision.23 Computations were performed using the statistical software R (version 3.0.2, R Core Team, R Foundation for Statistical Computing, Vienna, Austria, 2013). The study was reviewed and approved by the Minneapolis VA Health Care System Institutional Review Board.
RESULTS Fifty-six patients with SCI were identified as receiving vancomycin with measurements of serum creatinine, cystatin C and vancomycin concentration. Thirteen were excluded because of unstable renal function and 6 because of a serum vancomycin concentration reflecting non–steady-state conditions. The final population consisted of 37 patients with baseline characteristics provided in Table 2. The serum creatinine level ranged from 0.2 to 1.1 mg/dL (MVAHCS reference range, 0.7–1.2 mg/dL) with 36 of 37 patients having a value less than 1.0 mg/dL. The serum cystatin C level ranged more broadly from 0.5 to 2.7 mg/L (MVAHCS reference range, 0.5–1.0 mg/L). The calculated eCrCl and eGFR from each of the studied equations with corresponding estimated vancomycin half-lives are shown in Table 3. Estimates of renal function varied considerably ranging from an average estimated eCrCl of 48.1 mL/min using the Lee–Wang CG-SCI creatinine correction to an eGFR of 208.2 mL/min using MDRD. The median daily vancomycin dose at the time of serum concentration measurement was 2500 mg with a range 2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
TABLE 2. Patient Characteristics Characteristic
n = 37
SCI outcome Paraplegia, n (%) Quadriplegia, n (%) Years since injury, mean (6SD), yrs Sex—Male, n (%) Race, n (%) White Black Native American Asian Age, mean (6SD), yrs Height mean (6SD), inches Actual body weight, mean (6SD), kg Body mass index, mean (6SD)* Body surface area, mean (6SD), m2 Serum creatinine, mean (6SD), mg/dL† Serum cystatin C, mean (6SD), mg/L
15 22 14.4 37
(40.5) (59.5) (14.4) (100)
33 2 1 1 62.0 71.2 85.7 26.2 2.05 0.6
(89.2) (5.9) (2.7) (2.7) (69.0) (62.6) (622.3) (66.8) (60.25) (60.2)
1.1 (60.5)
*Body mass index is equal to weight (kg) divided by height squared (m). †To convert values for serum creatinine to micromoles per liter, multiply by 88.4.
from 750 mg to 4500 mg (7.5–51.9 mg/kg). Dosing frequencies were divided as every 8 hours (n = 13), every 12 hours (n = 20), and every 24 hours (n = 4). The mean measured vancomycin concentration was 18.5 mcg/mL (SD 6 16.8– 20.1), obtained at an average of 17 minutes before the next dose, reflecting trough serum concentrations. The mean predicted vancomycin concentrations and respective measures of P30 accuracy using each equation in the PK model are shown in Table 4. Of the creatinine-based equations, CG-IBW and CKD-EPI creatinine provided the same P30 accuracy, both improved compared with that of
TABLE 3. eCrCl, eGFR, and Corresponding Estimated Vancomycin Half-life eCrCl or eGFR, Mean Estimated Vancomycin (6SD), mL/min Half-life, Mean, hrs Creatinine-based equations CG-ABW CG-IBW MDRD CKD-EPI creatinine CG-SCI creatinine correction–based equations Mirahmadi et al Lee–Wang Cystatin C-based equations CKD-EPI cystatin C CKD-EPI creatininecystatin C
187.0 156.4 208.2 135.4
(113.2) (85.8) (131.2) (32.5)
5.5 6.2 5.0 6.3
123.3 (66) 48.1 (5.4)
7.7 15.0
93.0 (33.9) 112.2 (31.4)
9.6 7.6
635
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
CG using actual body weight and MDRD. The CG-IBW was selected for further comparison because of its more frequent use in clinical practice. Of the 2 CG-SCI creatinine correction equations used in the PK model, Mirahmadi et al provided greater P30 accuracy than that reported by Lee–Wang and selected for further comparative analysis. The CKD-EPI cystatin C equation was chosen from the cystatin C-based equations because its accuracy in the model surpassed that of the CKD-EPI creatinine-cystatin C equation despite the latter equation using 2 markers of renal function. Predictions of the PK model using equations of CGIBW, Mirahmadi et al, and CKD-EPI cystatin C were further compared against the measured vancomycin serum concentrations for bias and precision (Table 5). The use of CKD-EPI cystatin C eGFR was found to be least biased and most precise. When predictive abilities were compared directly against each other (Table 6), the CKD-EPI cystatin C eGFR in the PK model was statistically superior compared with both CG-IBW and Mirahmadi et al. The comparative P30 accuracy was superior with the use of CKD-EPI cystatin C equation in the PK model compared with CG-IBW and Mirahmadi et al (Table 7). The use of CG-SCI creatinine correction of Mirahmadi et al in the model provided statistically improved predictive accuracy compared with CG-IBW, however, the P30 accuracy remained less than 50%. All results of bias, precision, and accuracy remained consistent regardless of years since SCI, paraplegia versus quadriplegia, age, baseline serum creatinine, and body weight.
TABLE 4. Measured and Predicted Serum Vancomycin Concentration With the Use of eCrCl or eGFR in PK Model Vancomycin Serum Accuracy (P30)% of Concentration, Mean, Predictions Within the (mcg/mL) (95%CI) 30% Measured Measured vancomycin concentration Predictions using creatinine-based equations in PK model CG-ABW CG-IBW MDRD CKD-EPI creatinine Predictions using CGSCI creatinine correction equations in PK model Mirahmadi et al Lee–Wang Predictions using cystatin C-based equations in PK model CKD-EPI cystatin C CKD-EPI creatininecystatin C
18.5 (16.8–20.1)
7.0 8.6 6.0 8.9
(7.0–8.4) (7.0–10.2) (4.7–7.3) (7.6–10.2)
10.8 21.6 8.1 21.6
11.9 (9.7–14.2) 29.3 (25.5–33.1)
40.5 16.2
14.7 (12.8–16.7) 11.2 (9.9–12.7)
67.6 35.1
CG-ABW, CG using actual body weight.
636
—
TABLE 5. Predictive Bias and Precision With the Use of eCrCl or eGFR in PK Model Compared to Measured Serum Vancomycin Concentration
Predictions using CG-IBW in PK model Predictions using Mirahmadi et al CG-SCI creatinine correction in PK model Predictions using CKD-EPI cystatin C in PK model
Bias* (95%CI), Mean, mcg/mL
Precision (RMSE),† Mean, mcg/mL
29.9 (211.9 to 27.9)
11.5 (9.6 to 13.2)
26.5 (29.0 to 24.1)
9.7 (7.7 to 11.3)
23.8 (25.2 to 2.4)
5.6 (4.5 to 6.5)
*Bias, which is defined as prediction error, is the mean, across subjects, of difference between the measured serum vancomycin concentrations from the predicted concentration. Lower values reflect less bias. †Precision, as measured by root mean squared error (RMSE), conveys the size of spread or variation of predictions from the measured value. Lower values reflect improved precision.
DISCUSSION Vancomycin dose selection in the SCI population can be problematic because of the unique challenges of accurately assessing renal function. In the absence of specific guidance for this population, general recommendations call for the monitoring of a serum trough concentration at steady-state.16 Because it may take several days to reach steady-state, assessment of the initial dose is delayed. This raises the importance of providing an initial dose most likely to deliver prompt and effective antimicrobial concentrations, particularly in an acutely ill patient. A more accurate initial dose will also minimize potentially excessive concentrations that may lead to adverse effect. Results of this study show, in an SCI population, a PK model using the CKD-EPI-cystatin C eGFR provided superior predictions of measured steady-state serum vancomycin concentrations than other equations. These results reflect a biologically plausible mechanism of a decreased influence of muscle mass on serum cystatin C concentration compared with serum creatinine. If correct, the use of a cystatin-C-based eGFR would improve estimation of renal function in SCI and lead to more accurate predictions of vancomycin clearance and improved dosing. An improved predictive ability would provide needed guidance for designing initial vancomycin dosing regimens for patients with SCI before the availability of measured serum concentrations. This concept is supported by other investigations of cystatin C in patients with SCI or reduced muscle mass. Two studies in SCI populations found a superior correlation of serum cystatin C than that of serum creatinine to measured CrCl or GFR.24,25 Comparable results in patients without SCI with reduced muscle mass have also been published.26,27 Other investigations have also found improved predictive ability of steady-state serum vancomycin concentration using cystatin C-based eGFR versus a creatinine-based equations in 2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
TABLE 6. Direct Comparative Predictive Ability Using eCrCl and eGFR Equations in PK Model: Bias and Precision Comparison
Difference in Bias, mcg/m
Difference in Precision (MSE),* mcg/mL
CKD-EPI cystatin C 6.1 (3.8–8.4)†‡ 2102.1 (289.3 to 2114.9)†§ versus CG-IBW CKD-EPI cystatin C 2.7 (0.36–5.1)†‡ 263.4 (232.6 to 294.2)†§ versus Mirahmadi et al *Per Sheiner–Beal statistical comparison, the MSE is used for statistical comparison of precision.23 †P , 0.05. ‡The greater the value, the greater the difference in prediction bias favoring the first comparator (CKD-EPI cystatin C use provided less bias than that of CG-IBW and Mirahmadi et al). §The more negative values reflect a greater difference in precision favoring the first comparator (CKD-EPI cystatin C use provided greater precision than that of CG-IBW and Mirahmadi et al).
patients without SCI with reduced muscle mass.28–31 To the authors’ knowledge, this is the first report of improved predictive ability of serum vancomycin concentration using a cystatin C-based eGFR in an SCI population. Results also corroborate previous evidence of significant deficiencies of creatinine-based equations in the SCI population. SCI results in muscle denervation causing paralysis, immobility, and diffuse muscle atrophy. Because muscle mass is the major source for endogenous creatinine, studies have demonstrated significantly decreased creatinine production in patients with SCI compared with controls.6,32,33 These studies consistently cite lower serum creatinine concentrations in SCI compared with control patients with similar degrees of renal function. When used in creatinine-based equations, these lower serum creatinine concentrations lead to significant overestimation of renal function in the SCI population.34–36 This is consistent with results of this study in that serum creatinine concentrations were less than 1.0 mg/dL in 36 of the 37 subjects with 3 patients having concentrations as low as 0.2 mg/dL. All creatinine-based equations used in the PK model provided considerable under-prediction of the measured steady-state serum vancomycin concentrations reflecting high estimates of drug clearance. Unless clinical judgment supersedes, the use of creatinine-based equations may conceal clinically relevant reductions in renal function in SCI and lead to appreciably higher doses and potential adverse effect.37 Although these limitations of creatinine-based equations are well known in SCI, the authors thought it prudent to include them because of their common reference in drug dosing recommendations, automatic appearance in laboratory result
TABLE 7. Accuracy of the PK Model Using eCrCl and eGFR Methods to Predict Serum Vancomycin Concentration Within the 30% Measured (P30) P30
P value
67.6% versus 21.6% 67.6% versus 40.5%
,0.001 0.013
40.5% versus 21.6%
0.039
Equation CKD-EPI cystatin C versus CG-IBW CKD-EPI cystatin C versus Mirahmadi et al Mirahmadi et al versus CG-IBW
2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
reporting, and an absence of well-known alternatives. The more recent MDRD and CKD-EPI creatinine were also included because of an absence of data examining their use in the SCI population. This report provides new data regarding 2 alternative methods to assess renal function for drug dosing purposes in the SCI population. Both account for the SCI-related decreased creatinine production by providing a creatinine correction factor for substitution into CG. Mirahmadi et al use a correction based on their measurements of decreased endogenous creatinine production in paraplegia (20%) and quadriplegia (40%) compared with control subjects. The method of Lee–Wang relies on adjusting all serum creatinine concentrations less than 1.0 mg/dL upward to that value. This corrected creatinine is then substituted into a modified CG-IBW equation derived using linear regression from their SCI population to estimate vancomycin clearance and dose. The predictive ability of the PK model using the Mirahmadi et al CG-SCI creatinine correction was found more accurate. The authors presume the comparatively decreased accuracy of the Lee–Wang method to be due to the practice of adjusting varied serum creatinine concentrations to an identical value. For example, patients with a measured serum creatinine from 0.2 mg/dL to 1.0 mg/dL will use an identical corrected creatinine value of 1.0 mg/dL to estimate vancomycin clearance. In contrast, no correction is provided for a serum creatinine of 1.0 mg/dL or larger implying no change in the creatinine production despite the well-known reduction in patients with SCI. The authors note the Lee–Wang method seemed to provide better results with baseline serum creatinine values closer to 1.0 mg/dL but lost predictive abilities when lower values necessitated more substantial adjustment. Although the PK model predictions using Mirahmadi et al were superior to all other creatinine-based equations, its P30 accuracy was still less than 50%. Limitations of this study include its observational and retrospective nature. The sample size of 37 is small and consisted of only male patients. A definition of unstable renal function specifically for the SCI population is not available, and the one used could be questioned. In the absence of literature guidance, the authors adopted a strict definition of unstable renal function as any change of serum creatinine of 0.3 mg/dL over a 28-day period compared with the definition of the Kidney Disease-Improving Global Outcomes group of the same change over 48 hours in the general population.38 It must be emphasized that these results are not applicable when renal function is unstable. It is possible that a systematic bias of the population-based estimates used in the PK model may affect results; however, this would be expected to affect all equations to a similar degree. The MDRD equation is known to be less accurate when GFR is greater than 60 mL/min per 1.73 m2. The authors included MDRD because of its widespread availability in routine laboratory reports39 and the lack of data in SCI. Including a measured CrCl through a timed 24-hours urine collection would have been desirable but was not available for this analysis. However, measured CrCl has been shown to be unreliable in the SCI population.21,40,41 It also requires
637
Ther Drug Monit Volume 36, Number 5, October 2014
DeCarolis et al
24 hours for collection and additional time for processing further limiting usefulness for initial dose design. A full PK analysis with multiple vancomycin samplings would provide more robust data, however, our methods reflect current clinical practice and the latest guidelines for monitoring vancomycin trough concentration at steady-state. Of importance, the use of the population-based PK model is not intended to replace the individualized method of assessing serum vancomycin concentration at steady state, but rather as an adjunct to provide predictions to assist with dosing before its availability. There are limitations to the use of serum cystatin C measurement. There are non-GFR determinants of cystatin C such as inflammation, diabetes, and fat mass of unknown significance, which we did not control for.42 Despite its increasing use, its availability may be limited and costs may be higher compared with creatinine. At our institution, serum cystatin C measurement is available daily with a short turnaround time at an acquisition cost less than $10.00. Medicare reimbursement costs in 2013 are $18.69 for serum cystatin C and $7.04 for serum creatinine.43 It should be noted that a cystatin C reagent calibrated against the international certified reference material was not available at the time of the study. Actual results reflect cystatin C as measured by particle-enhanced immunonephelometry on the Siemens BNII analyzer in the United States. However, applying the conversion factor provided by Siemens Healthcare Diagnostics to reflect International Federation of Clinical Chemistry standardization did not significantly alter the results.
CONCLUSIONS In patients with SCI, the use of the CKD-EPI cystatin C equation in a population-based PK model provided significantly greater accuracy with less bias and greater precision for predicting steady-state serum vancomycin concentration compared with other equations that estimate renal function. This superior predictive ability may provide necessary guidance for improved initial dose selection in the SCI population before, or in the absence of a measured serum vancomycin concentration. If a serum cystatin C measurement is unavailable, the CG-SCI creatinine correction equation of Mirahmadi et al improved predictive ability compared with other creatinine-based equations. Both methods would require close monitoring of serum vancomycin concentration to assure optimal dosing. We encourage further study to validate this finding in additional SCI populations. REFERENCES 1. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41. 2. Levey AS, Coresh J, Greene T, et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Ann Intern Med. 2006;145:247–254. 3. Levey AS, Stevens LA, Schmid CH, et al; for the Chronic Kidney Disease Epidemiology Collaboration. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150:604–612. 4. Perrone RD, Madias NE, Levey AS. S.Cr. as an index of renal function: new insights into old concepts. Clin Chem. 1992;38:1933–1953.
638
5. National Kidney Disease Education Program. Chronic kidney disease and drug dosing: information for providers (Revised January 2010). Available at: http://www.nkdep.nih.gov/resources/ckd-drug-dosing-508. pdf. Accessed August 30, 2013. 6. Mirahmadi MK, Byrne C, Barton C, et al. Prediction of creatinine clearance from serum creatinine in spinal cord injury patients. Paraplegia. 1983;21:23–29. 7. Lee JP, Wang YJ. Testing the predictive ability of the “spinal cord injury equation” in estimating vancomycin clearance. Am J Health Syst Pharm. 2013;70:669–674. 8. Simonsen O, Grubb A, Thysell H. The blood serum concentration of cystatin C (gamma-trace) as a measure of the glomerular filtration rate. Scand J Clin Lab Invest. 1985;45:97–101. 9. Jacobsson B, Lignelid H, Bergerheim US. Transthyretin and CysC are catabolized in proximal tubular epithelial cells and the proteins are not useful as markers for renal cell carcinomas. Histopathology. 1995;26: 559–564. 10. Vinge E, Lindergard B, Nilsson-Ehle P, et al. Relationships among serum cystatin C, serum creatinine, lean tissue mass and glomerular filtration rate in healthy adults. Scand J Clin Lab Invest. 1999;59: 587–592. 11. Baxmann AC, Ahmed MS, Marques NC, et al. Influence of muscle mass and physical activity on serum and urinary creatinine and serum cystatin C. Clin J Am Soc Nephrol. 2008;3:348–354. 12. Dharnidharka VR, Kwon C, Stevens G. Cystatin C is superior to S.Cr. as a marker of kidney function: a meta-analysis. Am J Kidney Dis. 2002;40: 221–226. 13. Inker LA, Schmid CH, Tighiouart H, et al. Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med. 2012;367: 20–29. 14. Cheung RPF, DiPiro JT. Vancomycin: an update. Pharmacotherapy. 1986;6:153–169. 15. Matzke GR, McGory RW, Halstenson CE, et al. Pharmacokinetics of vancomycin in patients with various degrees of renal function. Antimicrob Agents Chemother. 1984;25:433–437. 16. Rybak M, Lomaestro B, Rotschafer JC, et al. Therapeutic monitoring of vancomycin in adult patients: a consensus review of the American Society of Health-System Pharmacists, the Infectious Diseases Society of America, and the Society of Infectious Diseases Pharmacists. Am J Health Syst Pharm. 2009;66:82–98. 17. Rowland M, Tozer TN. Clinical Pharmacokinetics, Concepts and Applications. 2nd ed. Philadephia, PA: Lea & Febiger; 1989: 238–254. 18. Stevens LA, Nolin T, Levey AS. In reply to ‘estimated GFR for drug dosing: a bedside formula,’ ‘drug dose adjustments in patients with renal impairment,’ ‘use of the MDRD study equation for drug dosing,’ and ‘estimated GFR vs creatinine clearance for drug dosing’. Am J Kidney Dis. 2009;54:985–986. 19. Sawchuk RJ, Zaske DE. Pharmacokinetics of dosing regimens which utilize multiple intravenous infusions: gentamicin in burn patients. J Pharmacokinet Biopharm. 1976;4:183–195. 20. Grubb A, Blirup-Jensen S, Lindstrom V, et al. First certified reference material for cystatin C in human serum ERM-DA471/IFCC. Clin Chem Lab Med. 2010;48:1619–1621. 21. National Kidney Foundation. KDOQI clinical practice guidelines for chronic kidney disease: evaluation, classification and stratification. part 5. Evaluation of laboratory measurements for clinical assessment of kidney disease. Am J Kidney Dis. 2002;39:S76–S110. 22. Earley A, Miskulin D, Lamb EJ, et al. Estimating equations for glomerular filtration rate in the era of creatinine standardization: a systematic review. Ann Intern Med. 2012;156:785–795. 23. Sheiner LB, Beal SL. Some suggestions for measuring predictive performance. J Pharmacokinet Biopharm. 1981;9:503–512. 24. Thomassen SA, Johannesen IL, Erlandsen EJ, et al. Serum cystatin C as a marker of the renal function in patients with spinal cord injury. Spinal Cord. 2002;40:524–528. 25. Jenkins MA, Brown DJ, Ierino FL, et al. Cystatin C for estimation of glomerular filtration rate in patients with spinal cord injury. Ann Clin Biochem. 2003;40:364–368. 26. Beringer PM, Hidayat L, Heed A, et al. GFR estimates using cystatin C are superior to serum creatinine in cystic fibrosis. J Cyst Fibros. 2009;8: 19–25.
2014 Lippincott Williams & Wilkins
Ther Drug Monit Volume 36, Number 5, October 2014
27. Delanaye P, Cavalier E, Radermecker RP, et al. Cystatin C or creatinine for detection of stage 3 chronic kidney disease in anorexia nervosa. Nephron Clin Pract. 2008;110:c158–c163. 28. Tanaka A, Suemaru K, Otsuka T, et al. Estimation of the initial dose setting of vancomycin therapy with use of cystatin C as a new marker of renal function. Ther Drug Monit. 2007;29: 261–264. 29. Tanaka A, Aiba T, Otsuka T, et al. Population pharmacokinetic analysis of vancomycin using serum cystatin C as a marker of renal function. Antimicrob Agents Chemother. 2010;54: 778–782. 30. Suzukia A, Imanishi Y, Nakanob S, et al. Usefulness of serum cystatin C to determine the dose of vancomycin in critically ill patients. J Pharm Pharmacol. 2010;62:901–907. 31. Okamoto G, Sakamoto T, Kimura M, et al. Serum cystatin C as a better marker of vancomycin clearance than serum creatinine in elderly patients. Clin Biochem. 2007;40:485–490. 32. Mohler JL, Barton SD, Blouin RA, et al. The evaluation of creatinine clearance in spinal cord injury patients. J Urol. 1986; 136:366–369. 33. Sawyer WT, Hutchins K. Assessment and predictability of renal function in spinal cord injury patients. Urology. 1982;19:377–380. 34. Mohler JL, Ellison MF, Flanigan RC. Creatinine clearance prediction in spinal cord injury patients: comparison of 6 prediction equations. J Urol. 1988;139:706–709. 35. Kaji D, Strauss I, Kahn T. Serum creatinine in patients with spinal cord injury. Mt Sinai J Med. 1990;57:160–164. 36. MacDiarmid SA, McIntyre WJ, Anthony A, et al. Monitoring of renal function in patients with spinal cord injury. BJU Int. 2000;85: 1014–1018. 37. Vaidyanathan S, Watt JWH, Singh G, et al. Dosage of once-daily gentamicin in spinal cord injury patients. Spinal Cord. 2000;38: 197–198. 38. Kidney Disease: Improving Global Outcomes (KDIGO) Acute Kidney Injury Work Group. KDIGO clinical practice guideline for acute kidney injury. Kidney Inter Suppl. 2012;2:1–138. 39. Accetta NA, Gladstone EH, DiSogra C, et al. Prevalence of estimated GFR reporting among US clinical laboratories. Am J Kidney Dis. 2008; 52:778–787. 40. Walser M. Assessing renal function from creatinine measurements in adults with chronic renal failure. Am J Kidney Dis. 1998;32: 23–31. 41. Sepahpanah F, Burns SP, McKnight B, et al. Role of creatinine clearance as a screening test in persons with spinal cord injury. Arch Phys Med Rehabil. 2006;87:524–528. 42. Stevens LA, Schmid CH, Greene T, et al. Factors other than glomerular filtration rate affect serum cystatin C levels. Kidney Int. 2009;75: 652–660. 43. Centers for Medicare and Medicaid Services. Clinical laboratory fee schedule. Available at: http://www.cms.gov/Medicare/MedicareFee-for-Service-Payment/ClinicalLabFeeSched/clinlab.html. Accessed August 30, 2013.
2014 Lippincott Williams & Wilkins
Cystatin C to Predict Vancomycin Concentration
APPENDIX 1. One-Compartment, PopulationBased PK Model for the Prediction of SteadyState Serum Vancomycin Concentration Step 1. Estimate half-life and convert to elimination rate constant (k) a. Calculate eCrCl or eGFR in mL/min and place the result in equation, Estimated t1=217 ¼
vancomcyin half -life in normal renal functionð7:9 hoursÞ15 ; 1 2 ffe · ½1 2 ðeGFR=100Þg
fe = fraction of drug excreted unchanged in urine (0.9). b. Determine k from the estimated half-life. k¼
0:693 : estimated half -life
Step 2. Estimate Volume of Distribution (Vd) MVAMC population average Vd = 0.84 L/kg.
Step 3. Prediction of steady-state vancomycin serum concentration19: a. Predict the peak concentration (Cpmax)(mcg/mL)
Cpmax ¼
0 dose=infusion timeðmg=hrÞ 1 2 e 2 kt · ðk · Vd Þ ð1 2 e 2 kT Þ
T = dose interval, t0 = infusion time. b. Predict the concentration at trough concentration (for study, this time was the same as the time of the actual measured vancomycin concentration): 0
Cpmin ¼ Cpmax · e 2 kðT 2 t Þ
639
