Effect of non-specific reversal agents on anticoagulant activity
of dabigatran and rivaroxaban
Journal club:are vancomycin trough concentrations adequate for
optimal dosing?Michael N. Neely, Gilmer Youn, Brenda Jones, Roger
W. Jelliffe, George L. Drusano, Keith A. Rodvoid, Thomas P. Lodise
Antimicrobial Agents and Chemotherapy. 2014;58:309-16.Presented
by:Megan HandleyPharmD Candidate Class of 2015University of Arizona
College of PharmacyClinica BiblicaJanuary 29, 2015
ObjectivesBackground and overviewMethodsResults Authors
conclusionsStrengths and limitationsConclusions
-background and overview of vancomycin and how it is dosed
currently and why. Also introduce two studies that are referenced
in the journal club study being analyzed and how they relate-the
methods of the trial including the design, subject selection,
modeling and simulation methods, outcome measures, and statistical
analyses used-results of the article-conclusions proposed by the
authors based on the results-strengths and weaknesses that I found
to be true of the trial-and my set of conclusions based on my
analysis of the article2
Background: Current guidelines24hr area under the curve (AUC):
minimum inhibitory concentration (MIC) ratio has been established
as the most accurate parameter to measure the efficacy of
vancomycin against Staphylococcus aureus infections Target value
AUC:MIC 400Trough concentrations of are an adequate surrogate
marker for AUC Most adults with normal renal function (creatinine
clearance {CrCl} of 100mL/min) being treated for an infection by an
organism with a MIC 1mg/LA trough concentration in steady-state of
15-20mg/L would correlate with an AUC:MIC ratio 400 The most
accurate and practical method for monitoring efficacy
*******The vancomycin dosing guidelines published in 2009 by the
Infectious Diseases Society of America, the American Society of
Health-System Pharmacists, and the Society of Infectious Diseases
Pharmacists suggest the following points
3
BackgroundMonitoring peak concentrations has been abandoned as a
technique for dosingThere are no data connecting peak
concentrations with nephrotoxicity or efficacyTherefore, the only
rationale to support trough concentration monitoring is the
suggestion that trough concentrations are an adequate surrogate for
AUC.It is also suggested trough concentrations are simpler to
obtain than AUCIf this rationale is incorrect, the current
guidelines for dosing would not be justified.Recent data has
emerged revealing increased rates of nephrotoxicity with adherence
to the established guidelines
-nephrotoxicity is the primary adverse effect of vancomycin
treatment
-this is the main goal of the study, to determine whether trough
concentrations are an accurate surrogate marker for AUC4
Background: ToxicityExample: Lodise TP, et al. conducted a
retrospective study among 166 patients in Albany Medical Center
Hospital who were treated with vancomycin for over
48hrsExposure-toxicity response relationship existsTrough value was
the index that best described this association
-the occurrence of nephrotoxicity significantly increased as the
initial trough value increased-one study that supports the current
guidelines5
Bivariate analysis of the relationship between the vancomycin
exposure profile and nephrotoxicity.
-As you can see, initial vancomycin trough value was the only
statistically significant variable associated with the occurrence
of nephrotoxicity. P=0.001-76.2% of patients that experienced
nephrotoxicity had a trough value greater than or equal to 9.9. in
comparison, only 38.6% of patients without nephrotoxicity had a
trough value greater than 9.9. Therefore, if 100 patients with a
trough value greater than 9.9 were treated with vancomycin, the
absolute risk of acquiring nephrotoxicity is 37.6%.-Nephrotoxicity
was defined as an increase in SCr level of 0.5 mg/dL or 50%,
whichever was greater.6
Background: toxicity Conversely, Cataldo et al. conduced a
meta-analysis that associated continuous infusions of vancomycin
are associated with a significantly lower risk of nephrotoxicity
(RR 0.6, 95% CI 0.4-0.9, p=0.02) when compared with intermittent
dosing.
-Here is an example of a study that showed continuous infusions
of vanco are associated with a significantly lower risk of
nephrotoxicity compared to intermittent dosing. -since the relative
risks are all less than 1, the risk of nephrotoxicity is decreased
with continuous infusion compared to intermittent dosing. The
relative risk is 0.6, so the relative risk reduction of acquiring
nephrotoxicity with a continuous infusion is 40%. -Mortality,
however, did not differ significantly between the two groups.7
Study objectivesPrimary: to explore whether the assumption that
trough concentration is a good surrogate for AUC is true, and
whether or not it is simpler to obtain and measure than the
AUC.
Secondary: to explore the relationships among trough
concentrations, AUC, and the potential rate of nephrotoxicity.
Tertiary: to test the level of adherence to the
guideline-recommended timing of blood sampling (just prior to the
fourth dose in those with normal renal function) in routine
inpatient settings
8
methods
Study design: data setsObtained 3 independent data sets of
adults receiving vancomycin (n=47)15 records consisting of: dosing
history, concentrations at frequent intervals, and patient
covariates such as weight and CrClOutpatients: samples obtained
0.5hr, 1h, at two random times, and 24hrs after the start of the
doseInpatients: samples just before and 1, 2, 3, 8, and 12hrs after
a dose22 patient records, various levels of renal function. Purpose
was to measure influences of age, protein binding, and renal
function on pharmacokinetics (PK)Had 11-13 samples taken over a
12-24hr period. 10 adult volunteers with normal CrCl enrolled in a
study to measure PK in various fluids in the body. Received 9 doses
of 1000mg every 12 hours followed by 7 blood samplings up to 24hrs
after the last dose.
The data sets were used to draw conclusions regarding the first
objective: whether or not trough value is an adequate surrogate for
AUC, and the second objective, to explore the relationships between
trough, AUC, and nephrotoxicity.
-Jettliffe formula used in first data set, is similar to
C-G-second data set used C-GThe collected data sets were intended
to represent a wide variety of patients with various degrees of
renal function, enabling a relevant model of vancomycin
pharmacokinetics. 3. normal CrCl was not defined10
Subject selectionInclusion criteriaData set 1:Adults with
prosthetic cardiac valves each receiving single dose prior to
outpatient dental procedure (n=12) ORORAcutely ill adults in the
cardiac ICU with suspected or proven staphylococcal infections
(n=7)Data set 2: Adult patients (n=22)Data set 3: Healthy adult
volunteers (n=10)Normal CrCl (unspecified)Exclusion criteria: none
mentioned
-information on the cohort subjects was very limited in the
article. -The goal was to obtain a diverse group of patients with
varying renal function to build a relevent vancomycin PK
model11
Study protocolUsed Pmetrics (the nonparametric population
modeling and simulation package) to:Model the PK dataSimulate from
the modelGenerate plotsPerform standard data summaries and
statistical testsWithin Pmetrics, the nonparametric adaptive grid
(NPAG) algorithm was used to build a population PK model for the
pooled data sets (used two-compartment model)
-the authors relied heavily on this program to create a
pharmacokinetic model to determine AUCs and draw conclusions.
-these parameters were estimated based on the data set in full
and two subsets
Two compartment model: Linear elimination (Ke) from central
compartmentVolume (Vc)Linear transfer to peripheral compartment
(KCP)Linear transfer from peripheral compartment (KPC)
12
Modeling and simulation: trough vs. aucEstimated PK parameters
in population by using full data set (ModelF) with two subsets:
peak and trough concentrations (ModelPT) and trough concentrations
only (ModelT)Created concentration-time profiles at 12-minute
intervals using the median of the joint distribution (Bayesian
approach)Compared AUCs from ModelPT and ModelT to the gold standard
ModelF using a Wilcoxon signed-rank test and linear regression
-These modeling techniques were used to draw conclusions on
objective one: trough vs. AUC
-the two subsets were data-depleted since they only contained
peak and trough concentrations-Peak concentration is concentration
closest to 1 hour after end of infusion-Trough concentration
closest to 1 hour before subsequent dose
-these concentration-time profiles allowed the authors to
calculate the AUC for the various subsets (peak-trough, trough) and
the full data set-The Bayesian approach involved determining the
posterior parameter distribution which is the probability
distribution of an unknown quantity based on the already
established evidence on vanco PK
- AUC estimated from the full data was considered the gold
standard gold standard: sort of like the control, what the subset
results will be compared to -Wilcoxon signed rank test is
equivalent to a paired student t-test but for non-normally
distributed data (not in a symmetrical bell-shaped graph) It
determines whether two population mean ranks differ. Able to use in
this case since the same data is being utilizedOf note, evaluated
the utility of a Bayesian approach to estimate the AUC from the
limited sampling strategyUsed ModelPT and ModelF data sets as
Bayesian priors to estimate the Bayesian posterior AUCs in the
ModelT data set and justified their use of a bayesian approach
covariate=has an effect on the outcome13
-The first row, dataset, includes FULL (all 47 patient),
considered rich sampling because vancomycin concentration
measurements are taken at all times. -The trough data set consists
of the full data set, except all non-trough measurements were
eliminated-the peak/trough data set contained only the
concentrations that were peaks and troughs.
-in the second row, you can see each dataset was used to
construct a population model that differed only in the parameter
value distributions based on the data. (Pmetrics computer
program)
-From each population model, all 47 subjects AUCs were
calculated using Bayesian calculations.
-The full data set was what was used to simulate the population
pharmacokinetic model 14
Modeling and simulation: nephrotoxicityCombined the data sets
with available CrCl values to use CrCl as a linear covariate,
reestimated the population parameter value distributionsUsed
model-predicted vs. observed concentration values by linear
regression to calculate slope, intercepts, and R2 valuesUsed values
and Pmetrics to simulate 5000 concentration-time profiles each for
1000 and 1500mg doses administered over 1hr every 12hrs for 5 days
Used previous studies to define trough concentration intervals
of