Population pharmacokinetics of daptomycin.
ABSTRACT Data from subjects in nine phase 1 (n = 153) and six phase 2/3 (n = 129) clinical trials were combined to identify factors contributing to interindividual variability in daptomycin pharmacokinetics (PK). Over 30 covariates were considered. A two-compartment model with first-order elimination provided the best fit for data on daptomycin concentrations in plasma over time. In the final population PK model, daptomycin plasma clearance (CL) was a function of renal function, body temperature, and sex. Of these factors, renal function contributed most significantly to interindividual variability. CL varied linearly with the estimated creatinine clearance. CL among dialysis subjects was approximately one-third that of healthy subjects (0.27 versus 0.81 liter/h). CL in females was 80% that in males; however, in clinical trials, the outcome was not affected by sex and therefore this effect is not considered clinically meaningful. The relationship with body temperature should be interpreted cautiously since the analysis included only a limited number of subjects who were hyperthermic. The volume of distribution of the peripheral compartment (V2) and intercompartmental clearance (Q) were linearly related to body weight. V2 increased approximately twofold in the presence of an acute infection. No factors were identified that significantly impacted V1. This analysis supports the dosing of daptomycin on a milligram-per-kilogram-of-body-weight basis and suggests that modified dosing regimens are indicated for patients with severe renal disease and for those undergoing dialysis.
- SourceAvailable from: Bruce A Mueller[Show abstract] [Hide abstract]
ABSTRACT: Continuous renal replacement therapy (CRRT) has given clinicians an important option in the care of critically ill patients. The slow and continuous dialysate and ultrafiltrate flow rates that are employed with CRRT can yield drug clearances similar to an analogous glomerular filtration rate of the native kidneys. Advantages such as superior volume control, excellent metabolic control, and hemodynamic tolerance by critically ill patients are well documented, but an understanding of drug dosing for CRRT is still a bit of a mystery. Although some pharmaceutical companies have dedicated postmarket research in this direction, many pharmaceutical companies have chosen not to pursue this information as it is not mandated and represents a relatively small part of their market. This lack of valuable information has created many challenges in the care of the critically ill patient as intermittent hemodialysis drug dosing recommendations cannot be extrapolated to CRRT. This drug dosing review will highlight factors that clinicians should consider when determining a pharmacotherapy regimen for a patient receiving CRRT.Seminars in Dialysis 03/2009; 22(2). · 2.25 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Increasing evidence suggests that antibiotic dosing in critically ill patients with acute kidney injury (AKI) often does not achieve pharmacodynamic goals, and the continued high mortality rate due to infectious causes appears to confirm these findings. Although there are compelling reasons why clinicians should use more aggressive antibiotic dosing, particularly in patients receiving aggressive renal replacement therapies, concerns for toxicity associated with higher doses are real. The presence of multisystem organ failure and polypharmacy predispose these patients to drug toxicity. This article examines the pharmacokinetic and pharmacodynamic consequences of critical illness, AKI, and renal replacement therapy and describes potential solutions to help clinicians give "enough but not too much" in these very complicated patients.Journal of Intensive Care Medicine 10/2014;
- [Show abstract] [Hide abstract]
ABSTRACT: Infections in critically ill patients are associated with persistently poor clinical outcomes. These patients have severely altered and variable antibiotic pharmacokinetics and are infected by less susceptible pathogens. Antibiotic dosing that does not account for these features is likely to result in suboptimum outcomes. In this Review, we explore the challenges related to patients and pathogens that contribute to inadequate antibiotic dosing and discuss how to implement a process for individualised antibiotic therapy that increases the accuracy of dosing and optimises care for critically ill patients. To improve antibiotic dosing, any physiological changes in patients that could alter antibiotic concentrations should first be established; such changes include altered fluid status, changes in serum albumin concentrations and renal and hepatic function, and microvascular failure. Second, antibiotic susceptibility of pathogens should be confirmed with microbiological techniques. Data for bacterial susceptibility could then be combined with measured data for antibiotic concentrations (when available) in clinical dosing software, which uses pharmacokinetic/pharmacodynamic derived models from critically ill patients to predict accurately the dosing needs for individual patients. Individualisation of dosing could optimise antibiotic exposure and maximise effectiveness.The Lancet Infectious Diseases 04/2014; · 19.45 Impact Factor
ANTIMICROBIAL AGENTS AND CHEMOTHERAPY, Aug. 2004, p. 2799–2807
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Vol. 48, No. 8
Population Pharmacokinetics of Daptomycin
Barry Dvorchik,1* Robert D. Arbeit,1† Julia Chung,2Susan Liu,2
William Knebel,2and Helen Kastrissios2
Cubist Pharmaceuticals, Inc., Lexington, Massachusetts 02421,1and GloboMax
Holdings LLC, Hanover, Maryland 210762
Received 9 November 2003/Returned for modification 26 January 2004/Accepted 4 April 2004
Data from subjects in nine phase 1 (n ? 153) and six phase 2/3 (n ? 129) clinical trials were combined to
identify factors contributing to interindividual variability in daptomycin pharmacokinetics (PK). Over 30
covariates were considered. A two-compartment model with first-order elimination provided the best fit for
data on daptomycin concentrations in plasma over time. In the final population PK model, daptomycin plasma
clearance (CL) was a function of renal function, body temperature, and sex. Of these factors, renal function
contributed most significantly to interindividual variability. CL varied linearly with the estimated creatinine
clearance. CL among dialysis subjects was approximately one-third that of healthy subjects (0.27 versus 0.81
liter/h). CL in females was 80% that in males; however, in clinical trials, the outcome was not affected by sex
and therefore this effect is not considered clinically meaningful. The relationship with body temperature should
be interpreted cautiously since the analysis included only a limited number of subjects who were hyperthermic.
The volume of distribution of the peripheral compartment (V2) and intercompartmental clearance (Q) were
linearly related to body weight. V2increased approximately twofold in the presence of an acute infection. No
factors were identified that significantly impacted V1. This analysis supports the dosing of daptomycin on a
milligram-per-kilogram-of-body-weight basis and suggests that modified dosing regimens are indicated for
patients with severe renal disease and for those undergoing dialysis.
tone) is a novel cyclic lipopeptide antibiotic derived from the
fermentation of Streptomyces roseosporus. Daptomycin was re-
cently approved for the treatment of complicated skin and skin
structure infections (cSSSI) caused by aerobic gram-positive bac-
teria, including those caused by methicillin-resistant Staphylococ-
cus aureus and methicillin-susceptible S. aureus. In vitro, dapto-
activity against most clinically relevant gram-positive pathogenic
bacteria, including bacterial isolates that are resistant to methi-
cillin, vancomycin, and linezolid (9). The MICs for daptomycin at
which 90% of isolates tested are inhibited are typically ?1 ?g/ml
for staphylococci and streptococci and 2 to 4 ?g/ml for entero-
cocci, including vancomycin-resistant isolates (7). Although the
mechanism of action has not been fully defined, it is distinct from
those of other antibiotics and appears to be mediated by the
disruption of multiple aspects of membrane function (5, 17). In
phase 3 trials for the treatment of cSSSI caused by susceptible
gram-positive bacteria, clinical and microbiological outcomes of
patients treated with daptomycin were comparable to those for
patients receiving conventional antibiotic therapy, such as pencil-
linase-resistant penicillins or vancomycin (1, 18).
Studies of healthy human subjects have demonstrated linear
pharmacokinetics after single (20) and multiple (8) intrave-
nous daptomycin doses up to 6 mg/kg of body weight over 14
days. After once-daily doses of 4 mg/kg, the average steady-
state trough daptomycin concentration in plasma was 5.89
?g/ml and varied by 27% among individuals (8). Daptomycin is
?90% bound to plasma proteins and has a low steady-state
volume of distribution, averaging 0.06 to 0.15 liter/kg (8, 19,
20), consistent with distribution into extracellular fluid. Elim-
ination is primarily achieved by renal excretion of unchanged
drug. In healthy adult subjects, the mean urinary recovery over
a 24-h period is 50 to 60% of the administered dose (8, 19). In
phase 1 studies involving healthy adult subjects and subjects
with graded renal insufficiencies, including end-stage renal dis-
ease, daptomycin plasma clearance (CL) was significantly re-
duced among subjects with creatinine clearances (CLCR) of
?40 ml/min or who were on dialysis (D. A. Sica, T. Gehr, and
B. Dvorchik, Abstr. 42nd Intersci. Conf. Antimicrob. Agents
Chemother., abstr. 2257, 2002).
For the present analysis, data from 15 clinical trials were
combined and a population analysis approach was used to
evaluate possible sources of interindividual variability in dap-
tomycin pharmacokinetics. The specific objectives were to de-
velop a model to describe the pharmacokinetics of daptomycin
in both healthy volunteers and subjects with acute bacterial
infections who were representative of the target patient pop-
ulation and to identify clinical characteristics that impact dap-
tomycin pharmacokinetics. The factors examined included age,
sex, weight, and the presence of bacterial infection, end-organ
dysfunction, comorbidities, and concomitant medications.
MATERIALS AND METHODS
Study design. The analysis included data collected from 282 subjects enrolled
in 15 phase 1, 2, and 3 clinical trials of daptomycin (Table 1). Frequent blood
samples were collected from each of 153 subjects in nine phase 1 clinical trials.
The subjects received single or multiple doses of 4 to 8 mg of daptomycin/kg
administered by a 30-min intravenous infusion every 24 h. Blood samples for
* Corresponding author. Present address: Barry Dvorchik & Asso-
ciates, Inc., 5809 Piney Lane Dr., Suite 105, Tampa, FL 33625. Phone:
(813) 951-2789. Fax: (832) 213-2008. E-mail: email@example.com
† Present address: Paratek Pharmaceuticals, Boston, MA 02111.
pharmacokinetic analysis were collected from a subset of 129 subjects with
gram-positive bacterial infections from six phase 2 and/or 3 clinical trials who
received various regimens of intravenous daptomycin. All subjects included in
the population pharmacokinetic analysis received at least one dose of daptomy-
cin, had at least one measured daptomycin concentration from plasma, and had
accurate documentation of the dates and times of the dose and concentration
measurements. Subject and study design details are shown in Table 1.
A validated high-performance liquid chromatography (HPLC) method was
used for an analysis of daptomycin concentrations in plasma for 13 studies (8).
The lower limit of quantitation was 3 ?g/ml and the interassay coefficient of
variation was 3.51%. A validated LC/MS/MS method (Cubist Pharmaceutical,
internal report) was used for the quantification of daptomycin concentrations in
plasma for a study of the pharmacokinetics of daptomycin in healthy and renally
impaired subjects. There was a close linear relationship (correlation coefficient,
0.965) between concentrations in plasma obtained by the LC/MS/MS and HPLC
methods. However, the LC/MS/MS assay was more sensitive than the HPLC
assay (lower limits of quantitation, 0.1 and 3 ?g/ml, respectively). For one study
of the use of daptomycin in healthy male volunteers, a microbiological assay was
used to quantify daptomycin concentrations in plasma (19). The interassay co-
efficient of variation and the limit of quantitation were 6.3% and 2 ?g/ml,
respectively. The results of the assay correlated well with those of the HPLC
assay, as evidenced by the fact that the pharmacokinetic parameters for dapto-
mycin for this group of subjects were in excellent agreement with data obtained
in other studies with healthy volunteers (8).
The covariates explored as possible sources of interindividual variability in
daptomycin pharmacokinetics are listed in Table 2. Body surface area was cal-
culated by the following formula (R. D. Mosteller, Letter, N. Engl. J. Med.
Body surface area (m2) ??
weight (kg) ? height (cm)
CLCRwas estimated for nondialysis subjects by use of the Cockcroft-Gault
CLCR(ml/min) ?[140 ? age (years)] ? weight (kg)
in which x ? 1 for males and 0.85 for females. For single-dose studies with
healthy subjects, hepatically impaired subjects, obese subjects, and healthy geri-
TABLE 1. Study designs
No. of samples/
Sample collection period (h)b
11Healthy4 mg/kg every 24 h for 7 days (n ? 6);
6 mg/kg every 24 h for 7 days (n ?
6), or 8 mg/kg every 24 h for 7 days
(n ? 6)
4 mg/kg ? i.v. aztreonam (crossover)
1827–3124 (1st dose), 72 (last dose),
Healthy, 18–30 yr old (n ? 12); or
healthy, ?75 yr old (n ? 12)
Healthy (BMI, ?25 kg/m2) (n ?
12), moderately obese (BMI,
25–39.9 kg/m2) (n ? 6), or
extremely obese (BMI, ?40
kg/m2) (n ? 7)
Healthy (n ? 10), or hepatic
impairment, child pugh B (n ?
Healthy (n ? 5), mild renal
failure (n ? 6), moderate to
severe renal failure (n ? 7), on
hemodialysis (n ? 6), or on
peritoneal dialysis (n ? 5)
Moderate renal impairment
16 6 mg/kg19 15 48
17Healthy subjects, 4 mg/kg ? p.o.
probenecid (crossover design); all
other subjects, 4 mg/kg
184 mg/kg every 24 h for 14 days (n ?
4) or 6 mg/kg every 24 h for 14
days (n ? 4)
4 mg/kg, then six doses of 3 mg/kg
every 48 h (n ? 6), or 6 mg/kg,
then six doses of 4 mg/kg every
48 h (n ? 1)
6 mg/kg every 24 h for 7 to 14 days (n
? 13), 4 mg/kg every 24 h for 7 to
14 days (n ? 14), 6 mg/kg followed
by 3 mg/kg every 12 h for 7 to 14
days (n ? 13), or other regimen (n
6 mg/kg every 24 h for 7 to 14 days
8 2624 (1st and last doses),
19 On hemodialysis72636 (1st dose), 48 (last dose),
210Subjects with bacteremic
infections due to gram-positive
415 6 (on 5th day of dosing)
211Subjects with infections due to
gram-positive bacteria resistant
to vancomycin or otherwise
refractory to or contraindicated
for currently available therapy
Subjects with moderate to severe
bacterial pneumonia due to
Streptococcus pneumoniae or
other gram-positive cocci
Subjects with moderate to severe
bacterial pneumonia due to S.
pneumoniae or other gram-
Subjects with cSSSI due to gram-
Subjects with cSSSI due to gram-
2756 (on 5th to 7th day of
3 124 mg/kg every 24 h for 5 to 10 days6412 (on 5th day of dosing)
3 13 4 mg/kg every 24 h for 5 to 10 days12412 (on 5th day of dosing)
3 144 mg/kg every 24 h for 7 to 14 days 1656 (on at least 3rd day of
6 (on at least 3rd day of
315 4 mg/kg every 24 h for 7 to 14 days275
aAdministered by intravenous infusion for 30 min.
bPeriod following dose administration.
cBMI, body mass index.
2800DVORCHIK ET AL.ANTIMICROB. AGENTS CHEMOTHER.
atric subjects, creatinine levels in serum and body weights were determined at
screening, on the morning of dose administration, and prior to discharge from
the clinical research unit. For each study, intraindividual differences in these
parameters were within 10% and estimated CLCRvalues for these subjects were
considered stable. For subjects in the single-dose graded renal study (excluding
those undergoing dialysis), entry criteria required that CLCRbe determined from
two separately measured 24-hour CLCRvalues performed within 21 days of
dosing. Subjects with CLCRvalues within 30% of each other were considered to
have a stable CLCRand were enrolled in the study. Creatinine levels in serum
and body weights were also obtained prior to dosing and prior to discharge from
the clinical research unit. Intraindividual estimates of CLCRwere within the
protocol criteria for a stable CLCR. For subjects in multiple-dose renal studies
(excluding those undergoing dialysis), creatinine levels in serum were deter-
mined at screening, predose, and every other day from day 3 to discharge (day
15) from the clinical research unit. No clinically significant change was observed
in the intraindividual values over time. For phase 2 and/or 3 clinical studies,
creatinine levels in serum and body weights were determined at various times
throughout the duration of the study. Estimated CLCRvalues were calculated by
using creatinine levels in serum obtained on the day of the pharmacokinetic
blood draw and the day closest to the pharmacokinetic day and the mean of the
highest and lowest creatinine levels in serum obtained over the course of the
entire study. No clinically significant intraindividual differences were observed,
regardless of the value chosen. To maintain consistency with single-dose studies,
we calculated the estimated CLCRvalues used in the final analysis by using the
creatinine level in serum and the actual body weight measured closest to the
pharmacokinetic day. The weight used was the actual body weight on the day of
the pharmacokinetic draw or the day closest to the pharmacokinetic day. Indi-
vidual CLCRvalues were estimated to be ?150 ml/min for 38 subjects and were
all set to 150 ml/min. These subjects represented multiple different studies; their
median body weight was 85.3 kg (range, 52 to 153 kg). Sixty-three percent had
normal creatinine levels in serum, with actual body weights ranging from 72.6 to
152.8 kg; the remaining 37% had creatinine levels in serum ranging from 0.2 to
0.6 mg/dl, with actual body weights ranging from 56.3 to 130.6 kg. The lowest
estimated CLCRfor subjects was 14 ml/min. Body temperature was evaluated
only for subjects in phase 2 and/or 3 studies. Concomitant medications examined
for possible pharmacokinetic interactions with daptomycin included acidic drugs
that are actively secreted in the renal tubule and drugs that are highly (?95%)
bound to albumin (3, 13, 15). Comorbidities included diseases producing fluid
TABLE 2. Subject characteristics
Median (continuous variables)
or no. (categorical variables)
Range (continuous variables)
or % (categorical variable)
Body weight (kg)
Body surface area (m2)
Body temp on day of pharmacokinetic study (°C)a
Baseline serum albumin (g/dl)b
Baseline alkaline phosphatase (IU/liter)
Baseline ALT (IU/liter)
Baseline AST (IU/liter)
Baseline total bilirubin (mg/dl)
Baseline BUN (mg/dl)
Baseline blood glucose (mg/dl)
Baseline serum creatinine (mg/dl)c
Baseline creatinine clearance (ml/min)d,e
Average serum creatinine (mg/dl)c,f
Average creatinine clearance (ml/min)d,g
Study (phase 1, phase 2/3)
Sex (male, female)
Race (Caucasian, African-American, other)
Dialysis (yes, no)
Renal function 5 categories (?80 ml/min, ?50 to ?80 ml/min,
?30 to ?50 ml/min, ?30 ml/min, on dialysis)
Renal function 4 categories (?80 mL/min, ?40 to ?80 ml/min,
?40 ml/min, on dialysis)
Elevated baseline BUN (?ULN ? 25 mg/dl)
Elevated baseline serum creatinine (?ULN ? 1.4 mg/dl)
Elevated baseline blood glucose (?ULN ? 109 mg/dl)
Has congestive heart failure
Has fluid accumulation (edema or ascites)
Has gram-positive bacterial infection
Taking concomitant aztreonam
Taking concomitant metronidazole
Taking concomitant medication that is secreted in the renal
tubule (including probenecid)
Taking concomitant medication that is ?95% bound to albumin
163, 51, 68
165, 64, 24, 8, 21
58, 18, 24
59, 23, 9, 3, 7
165, 80, 16, 2159, 28, 6, 7
aOn day of blood samples obtained for pharmacokinetics (n ? 100). Data are missing for all phase 1 studies and for 29 subjects in phase 2/3 studies.
bn ? 164 (data are missing for five phase 2/3 studies).
cn ? 282.
dn ? 261 (excludes subjects on dialysis).
eIncludes 38 subjects (14.6%) for whom the estimated creatinine clearance was set to 150 ml/min.
fAverage of minimum and maximum values recorded over the period of the pharmacokinetic study.
gIncludes 34 subjects (13.0%) for whom the estimated creatinine clearance was set to 150 ml/min.
VOL. 48, 2004 POPULATION PHARMACOKINETICS OF DAPTOMYCIN2801
accumulation (e.g., ascites and edema), the presence of infection, diabetes,
hypertension, and congestive heart failure. All covariates were values recorded at
baseline, with the exception of body temperature, which was the value recorded
on the day of pharmacokinetic sampling. Missing continuous covariates were
replaced with the median value of the covariate for subjects of the same sex in the
Pharmacokinetic analysis. Population pharmacokinetic models were built by a
nonlinear mixed-effects modeling approach and first-order conditional maximum
likelihood estimation with eta-epsilon interaction in the NONMEM program
(double precision, version V, level 1.1) (2).
(i) Base model selection. One-, two-, and three-compartment structural mod-
els were fit to the data for concentrations in plasma over time; graphical displays
of the data were also evaluated. Hypothesis testing to discriminate among alter-
native hierarchical structural models was performed by using the likelihood ratio
test (16). For comparisons of alternative models, the difference in the NONMEM
objective function was approximately chi-square distributed, with n degrees of
freedom, where n was the difference in the number of parameters between the
hierarchical models. A decrease of ?3.84 in the value of the NONMEM objec-
tive function, which is less than twice the maximum logarithm of the likelihood
of the data, is significant in the likelihood ratio test (n ? 1; P ? 0.05). The
goodness of fit was evaluated by using diagnostic scatter plots (not shown).
The duration of infusion (D1) was estimated for a subset of 108 subjects
included in the phase 3 clinical trials for whom the date and time of the start of
daptomycin infusion and the time of the first pharmacokinetic blood draw for
concentration measurements, but not the time of cessation of the infusion, were
recorded. All estimates were consistent with the protocols, which specified an
infusion time of 30 min. Interindividual variability in D1could not be estimated
and was not included in the model.
All pharmacokinetic parameters were assumed to be logarithmically normally
distributed, and exponential interindividual variability terms were included in the
pharmacokinetic parameters in the model. Various residual error models were
tested, including an evaluation of possible systematic differences between phase
1 and phase 2/3 studies and among studies that used different assay methods.
(ii) Population pharmacokinetic model building. Exploratory analyses were
used to guide the model building process. Relationships between individual
covariates and Bayesian estimates of the pharmacokinetic parameters were ex-
plored graphically. Generalized additive models were used to evaluate both
linear and nonlinear relationships between parameters and covariates (14). In
addition, measures of body size and renal function markers were tested as
possible sources of interindividual variability for each pharmacokinetic param-
eter. All possible covariate-parameter relationships thus selected were tested,
with the exception that possible drug interactions and the effect of the dapto-
mycin dose were examined only for CL and the volume of the central compart-
ment (V1), as appropriate.
Continuous covariates were entered into the population pharmacokinetic
model according to the following equation:
P ? ?1? ?2? (COV ? COV)
where P is the individual estimate of the parameter, COV is the value of the
covariate, and COV is the median value of the covariate in the study population.
?1is the typical value of the parameter (when COV ? COV) and ?2is the slope
of the effect of the covariate on the parameter.
Categorical covariates were included in the model by using indicator variables,
as shown in the following equation:
P ? ?1? ?2
where P is the individual estimate of the parameter, ?1is the typical value of the
parameter when the covariate is not present (IND ? 0), and ?2is the fractional
change in the value of P when the covariate is present (IND ? 1).
The statistical significance of each covariate-parameter relationship was
screened individually in NONMEM, and the model was built by stepwise addi-
tions to obtain a full model. Stepwise deletions were used to obtain the final
(reduced) model. The likelihood ratio test was used for hypothesis testing to
discriminate among alternative hierarchical models. A strict inclusion criterion
(P ? 0.001) corresponding to a change in the value of the NONMEM objective
function of 10.83 (n ? 1 degrees of freedom) was used to account for multiple
hypothesis testing. At each stage of the analysis, the goodness of fit was evaluated
by using diagnostic scatterplots.
(iii) Pharmacokinetic parameter calculations. The terminal half-life (t1/2),
volume of distribution at a steady state (Vss), and area under the curve from time
zero to infinity [AUC(0-?)] were calculated from individual pharmacokinetic
parameter estimates obtained by Bayesian estimation from the final population
pharmacokinetic model. For a two-compartment model (10), the following equa-
tions were used:
Vss? V1? V2
where ? is the terminal phase rate constant (per hour), CL is the daptomycin
plasma clearance (liters per hour), Q is the intercompartmental clearance (liters
per hour), V1is the volume of the central compartment (liters), V2is the volume
of the peripheral compartment (liters), and t1/2is the terminal phase half-life
(hours). All calculations were performed in NONMEM.
(iv) Statistics. Individual Bayes estimates and calculated pharmacokinetic
parameter values were grouped according to four renal function categories.
Three categories were defined by using the estimated CLCRvalues: the groups
were values of ?80 ml/min, ?40 to ?80 ml/min, and ?40 ml/min. These ranges
were chosen based on an analysis of phase 1 studies with renally impaired
subjects (Sica et al., 42nd ICAAC). Subjects on dialysis comprised a fourth
category (thus, the categories were CLCRvalues of ?80 ml/min, ?80 to ?40
ml/min, and ?40 ml/min and subjects on dialysis). Differences between groups
were evaluated by analysis of variance with Scheffe’s test (S-PLUS Professional;
Insightful Corp., Seattle, Wash.). P values of ?0.05 were considered significant.
Data. Measurements of daptomycin concentrations over
time were available for 3,325 plasma specimens collected from
282 adult subjects. Two outlying concentrations, one for a
specimen reported as a trough plasma with a daptomycin con-
centration of 106 ?g/ml (sixfold higher than the average trough
value) and one for a specimen reported as obtained 0.58 h after
dose administration with a daptomycin concentration of 309
?g/ml (fivefold higher than the average plasma concentration
at 0.58 h), were excluded from the analysis.
Imputed covariate values were generated for a total of 37
subjects; laboratory tests of hepatic function were the most
frequently missing values. Height, used in the calculation of
body surface area, was imputed for 3 subjects; baseline creat-
inine values in serum were imputed for 13 subjects. Albumin
levels in serum were not recorded in one phase 1 study and
three phase 2/3 studies and were considered to be missing for
all subjects in those studies. Missing body temperature values
on the day of pharmacokinetic sampling were not imputed for
29 subjects from the phase 2/3 studies. The population phar-
macokinetic model was coded to remove the effect of missing
covariates in the model.
Descriptive statistics for all covariates are presented in Table
Pharmacokinetic analysis. (i) Base model. Plots of data for
concentrations in plasma versus time (not shown) showed a
biphasic disposition of daptomycin. A review of the minimum
objective function and diagnostic plots showed that the data
for daptomycin concentrations in plasma over time were best
described by using a two-compartment open model with first-
order elimination. The structural pharmacokinetic model used
the following parameters: clearance (CL), the volume of the
central compartment (V1), intercompartmental clearance (Q),
and the volume of the peripheral compartment (V2). In addi-
tion, the duration of infusion (D1) was estimated for several
2802DVORCHIK ET AL.ANTIMICROB. AGENTS CHEMOTHER.
phase 3 subjects. Estimates of D1were consistent with the
duration of infusion specified in the clinical protocols.
The median daptomycin clearance for the study population
was estimated to be 0.688 liter/h (11.5 ml/min) and the volume
of the central compartment was 4.8 liters. Median estimates for
the intercompartmental clearance and volume of distribution
of the peripheral compartment were 3.6 liters/h and 3.6 liters,
respectively. All pharmacokinetic parameters were precisely
estimated, with relative standard errors (RSEs) of ?3%. The
estimated median duration of infusion for 108 subjects en-
rolled in phase 2/3 trials was 0.402 h (24 min), with an RSE of
Interindividual variabilities were estimated to be 52.1% for
CL, 60.6% for the volume of the central compartment, 31.9%
for the volume of the peripheral compartment, and 74.4% for
intercompartmental clearance. A simple additive residual er-
ror model based on diagnostic plots provided the best fit for
the data. A further evaluation of diagnostic plots indicated that
there was a larger degree of misfit of model predictions for
observations collected in phase 2/3 studies; therefore, different
error structures were evaluated for phase 1 versus phase 2/3
studies. On the basis of the likelihood ratio test, the final
residual error model was described by a combination of addi-
tive errors, reflecting the different assay methods used to de-
termine daptomycin concentrations in plasma. The residual
error was slightly lower for the study in which daptomycin
concentrations in plasma were assayed by LC/MS/MS than for
studies assayed by HPLC (2.08 versus 4.72 ?g/ml, respectively),
consistent with the higher sensitivity of the former method.
(ii) Population model. Exploratory graphical analyses re-
vealed a direct correlation between daptomycin clearance and
various markers of renal function, including estimated creati-
nine clearance, renal function category, and laboratory mark-
ers of renal function. Intercompartmental clearance (in liters
per hour) and the volume of the peripheral compartment (in
liters) were correlated with body weight. There were no obvi-
ous relationships between V1and any of the tested covariates
and no significant association between daptomycin pharmaco-
kinetics and either the concomitant medications or the con-
comitant diseases evaluated in this patient population.
The final model for daptomycin clearance was determined to
be the following:
CL ? [CLR? 0.14 ? (TEMP ? 37.2)] ? y in which CLR
0.807 ? 0.00514 ? (CLCR? 91.2) in others
0.269 in dialysis subjects
where CL ? daptomycin clearance (liters per hour), CLR?
daptomycin clearance as a function of renal function only (li-
ters per hour), CLCR? estimated creatinine clearance (milli-
liters per minute), TEMP ? body temperature (oC), and y ?
0.8 for females and 1 for males.
Both intercompartmental clearance and the volume of the
peripheral compartment were determined to be functions of
body weight, as follows:
Q ? 3.46 ? 0.0593 ? (WT ? 75.1)
V2? ?3.13 ? 0.0458 ? (WT ? 75.1)] ? z
where WT ? body weight (kilograms) and z ? 1.93 for subjects
with a bacterial infection and 1 for noninfected subjects.
The median value for V1was estimated to be 4.80 liters.
Although the interindividual variability in V1was 57%, none of
the covariates investigated, including body weight, was identi-
fied as a significant source of variability in V1.
Additional exploratory analyses were performed to evaluate
whether any other covariate could explain the effect of sex on
daptomycin clearance or the effect of the presence of infection
on V2. A review of the covariate graphics indicated that body
weight, body surface area, and race differed by sex. Each of
these covariates was substituted into the clearance model to
determine if it could be substituted for sex in the model, but
none produced a significant change in the objective function
value. Consequently, the clearance model including sex repre-
sented the final model.
Infections were only present in subjects in the phase 2 and/or
3 clinical trials. Therefore, in the model the presence of infec-
tion could be a marker for an unmonitored covariate that
differed between the phase 1 and phase 2/3 clinical trials. In a
graphical evaluation, age and serum albumin were determined
to differ between the two groups. Each of these covariates, as
well as body temperature, was substituted into the V2model,
and none produced a significant change in the objective func-
tion value. The V2model including infection represented the
Parameter estimates for the final population pharmacoki-
netic model are presented in Table 3. Pharmacokinetic param-
eters were precisely estimated and diagnostic plots showed a
good fit of the final model to the observed daptomycin con-
centrations in plasma (Fig. 1).
Based on the final population pharmacokinetic model, the
apparent Vssfor a typical healthy subject with a median body
weight of 75 kg was estimated to be 7.9 liters. In comparison,
the Vsswas increased 37%, to 10.8 liters, if the subject had an
acute bacterial infection. The median terminal elimination t1/2
of daptomycin was determined to be 7.07 h in a typical nor-
mothermic male with normal renal function. The median ter-
minal elimination t1/2for a male subject with a creatinine
clearance of 40 ml/min was 10.36 h, and for a male subject
receiving dialysis, the median terminal elimination t1/2was
Individual estimates of CL and V1were obtained from the
final population pharmacokinetic model by Bayesian estima-
tion. These were used to calculate individual estimates of the
t1/2, Vss, and AUC(0-?)for a single 4-mg/kg intravenous dose
from the individual parameter estimates and were summarized
by renal function category. Summary statistics for these esti-
mates are presented for all subjects (Table 4) and separately
for phase 1 and phase 2/3 subjects (Table 5).
These analyses indicated that the daptomycin CL, t1/2, and
AUC(0-?)for a single 4-mg/kg intravenous dose were depen-
dent on renal function. An analysis of variance indicated
that compared with subjects with CLCRvalues of ?40 ml/
min, subjects whose CLCRvalues were ?40 ml/min or who
were on dialysis had significantly larger volumes of the cen-
tral compartment. This factor was not significant in the
population pharmacokinetic analysis, most likely because of
the large variation in V1and the relatively small number of
VOL. 48, 2004POPULATION PHARMACOKINETICS OF DAPTOMYCIN2803
subjects on dialysis. Vsswas not dependent on renal func-
Relative to that in subjects with normal renal function
(CLCRvalues of ?80 ml/min), the daptomycin half-life was
increased 2.3-fold in subjects with CLCRvalues of ?40 ml/min
and 3.5-fold in subjects who were on dialysis; changes in dose-
normalized AUC(0-?)values were 1.8-fold and 3-fold, respec-
tively (Table 5). In comparison, median half-life and dose-
normalized AUC(0-?)values in subjects with CLCRvalues of
?80 ml/min and in subjects with CLCRvalues of ?80 and ?40
ml/min differed ?10%. These differences, although statistically
significant, were not considered clinically meaningful.
FIG. 1. Diagnostic plots for daptomycin population pharmacokinetic model. Observed versus predicted daptomycin concentrations in plasma
(left), observed versus individual predicted daptomycin concentrations in plasma (middle), and weighted residuals versus predicted daptomycin
concentrations in plasma (right panel) are shown. Circles represent individual data points. Dashed lines represent regression lines. Solid lines
TABLE 3. Population pharmacokinetic parameter estimates for daptomycinb
Structural model parameter
CV (RSE [%])
CL (liter/h) for male subject with median creatinine clearance (91.2 ml/min)
Change in CL (liter/h) for each 10 ml/min that creatinine clearance differs from the median value
CL (liter/h) for subject on dialysis
Fractional change in CL for female subject
Change in CL (liter/h) for each °C that temp differs from the median value (37.2°)
Q (liter/h) for subject with median body weight (75 kg)
Change in Q (liter/h) for each 10 kg that body weight differs from the median value
V2(liters) for subject with median body weight (75 kg)
Change in V2(liters) for each 10 kg that body weight differs from the median value
Fractional change in V2for subject with infection
Duration of infusion (h)
aNE, not estimated.
bFor the residual error parameter ?2addfor studies that used the LC/MS/MS assay, the estimated value was 4.28, with an RSE of 23.1% and an intraindividual error
SD of 2.07 ?g/ml. For the parameter ?2addfor studies that used the HPLC assay, the estimated value was 22.4, with an RSE of 20.3% and an intraindividual error SD
of 4.73 ?g/ml.
2804 DVORCHIK ET AL.ANTIMICROB. AGENTS CHEMOTHER.
Daptomycin is a novel lipopeptide antibiotic which was re-
cently approved for the treatment of cSSSI caused by suscep-
tible gram-positive microorganisms. It has a unique mechanism
of action, and in vitro it demonstrates rapid, concentration-
dependent bactericidal activity against drug-resistant clinical
isolates of gram-positive microorganisms, a long postantibiotic
effect, and a low rate of spontaneous resistance (4, 9, 11, 17).
In vivo, daptomycin exhibits linear pharmacokinetics after both
single and multiple once-daily doses (8, 20). Pharmacodynamic
studies using a model of S. aureus thigh infections in neutro-
penic mice indicated that bacterial eradication best correlates
with the ratio of AUC24 hto the MIC (12).
This report represents the first population pharmacokinetic
TABLE 4. Summary of pharmacokinetic parameters sorted by estimated CLCRand obtained by Bayesian estimation from the final model
Value for subjects in CLCRgroup
(n ? 165)
?80 to ?40 ml/min
(n ? 80)
(n ? 16)
(n ? 21)
AUC(0??)(?g ? h/ml)a
aCalculated for a single 4-mg/kg dose.
bSignificantly different from ?80-ml/min group.
cSignificantly different from ?80- to ?40-ml/min group.
dSignificantly different from ?40-ml/min group.
TABLE 5. Summary of pharmacokinetic parameters by study phase and estimated CLCRof individual parameters obtained by Bayesian
estimation from the final model
Value for phase 1 subjects in CLCRgroup
Value for phase 2/3 subjects in CLCRgroup
(n ? 79)
?80 to ?40 ml/min
(n ? 48)
(n ? 8)
(n ? 18)
(n ? 86)
?80 to ?40 ml/min
(n ? 32)
(n ? 8)
(n ? 3)
AUC(0-?)(?g ? h/ml)a
aCalculated for a single 4-mg/kg i.v. dose.
VOL. 48, 2004POPULATION PHARMACOKINETICS OF DAPTOMYCIN2805
analysis of daptomycin and includes subjects from all three
phases of the clinical development program. The increased use
of population pharmacokinetic analysis has generated a num-
ber of newer software programs that, like NONMEM, have
advantages and disadvantages. One topic of discussion has
been the ability and ease of use of NONMEM to detect the
presence of a nonnormal distribution, especially within a sub-
population. The intelligent use of any pharmacokinetic pro-
gram is a prerequisite for a meaningful analysis. NONMEM,
when used by trained personnel, does allow one to detect a
distribution that is substantially nonnormal.
Among healthy subjects, the estimated pharmacokinetics
were consistent with those previously reported for a phase 1
study in which single doses of 0.5 to 6 mg of daptomycin/kg
were administered intravenously to healthy volunteers (20).
The population analysis defined quantitatively the decrease in
daptomycin CL associated with reduced renal function, a re-
lationship that was suggested by earlier phase 1 studies. New
findings included the increase in the volume of the peripheral
compartment (V2) in subjects with bacterial infections relative
to healthy subjects as well as the associations between weight
and both intercompartmental clearance (Q) and V2.
Renal function, sex, and body temperature accounted for
21.5% of the interindividual variability in daptomycin clear-
ance, with renal function being the single most significant ex-
planatory variable. During the screening of covariates, adding
just renal function (i.e., CLCRin nondialysis subjects and a flag
for subjects on dialysis) to the clearance model reduced the
interindividual variability by 18.9%, from 52.1% in the base
model (no covariates) to 33.2% (with the addition of renal
function markers) (data not shown). This finding is consistent
with the fact that daptomycin, like other hydrophilic antibiot-
ics, is cleared primarily by renal excretion (20).
The median estimated daptomycin clearance for a normo-
thermic male with an estimated CLCRof 91.2 ml/min was 0.807
liter/h (13.5 ml/min). Among subjects on dialysis, the median
daptomycin clearance was estimated to be 0.269 liter/h (4.5
ml/min), or approximately one-third that of nondialysis sub-
jects. Among subjects who were not on dialysis, daptomycin
clearance was a linear function of CLCR. For example, for an
increase or decrease in the estimated CLCRof 10 ml/min, the
daptomycin clearance increased or decreased by 0.05 liter/h
The dose of daptomycin recommended for the treatment of
cSSSI is 4 mg/kg administered by intravenous infusion once
every 24 h for subjects with CLCRof ?30 ml/min and once
every 48 h for subjects with lower CLCRvalues, including those
who are on dialysis. These recommendations are based on
several observations in addition to the data presented in this
report. Sica et al. (42nd ICAAC) determined mean Cmaxand
AUCssvalues among 44 subjects with graded renal impairment
or who were undergoing dialysis. The Cmaxwas consistent for
all subjects and the AUCsswas similar in all subjects who had
an estimated CLCRof ?40 ml/min. For subjects with an esti-
mated CLCRof ?40 ml/min, the AUC was increased 2.33-fold
compared to that for subjects with a CLCRof ?80 ml/min. The
two phase 3 trials for the treatment of cSSSI included a limited
number of subjects with estimated CLCRvalues between 30
and 40 ml/min. There was no increase in adverse events attrib-
uted to daptomycin among these subjects; none participated in
the pharmacokinetic studies reported here. Additional studies
of the pharmacokinetics and safety of daptomycin in renally
impaired subjects and in those undergoing dialysis are in
Daptomycin clearance was influenced to a lesser extent by
sex and body temperature. Clearance in females was estimated
to be approximately 80% that of male subjects with similar
renal function. Among subjects with cSSSI treated with dap-
tomycin in two recent large phase 3 trials, there were no clin-
ically or statistically significant differences between the success
rates for males (n ? 230) and females (n ? 192) (74.8 versus
77.1%, respectively; 95% confidence intervals, ?7.2 and 8.3)
(data on file, Cubist Pharmaceuticals). Thus, although the dif-
ference in daptomycin clearance related to sex was statistically
significant, it does not appear to be clinically meaningful.
The observation that daptomycin clearance increased with
elevated body temperatures (?37.2°C) should be interpreted
cautiously since the analysis was limited to data obtained from
100 subjects in the phase 2/3 clinical studies, of whom only 14%
were hyperthermic (body temperature of ?38°C).
Comorbidities, including diseases producing fluid accumu-
lation (e.g., ascites and edema), diabetes, hypertension, and
congestive heart failure, were not significantly correlated with
daptomycin clearance. Medications that were tested for possi-
ble pharmacokinetic interactions with daptomycin included
acidic drugs that are actively secreted in the renal tubule and
drugs that are highly (?95%) bound to albumin (3, 13, 15).
These had no effect on daptomycin pharmacokinetics.
The estimated increases in Q and V2for daptomycin with
increased body weights were consistent with the physicochem-
ical properties of daptomycin and the physiologic effects of
weight. Daptomycin appears to be restricted to the extracellu-
lar space which increases with body weight (15). Similarly, the
extravascular distribution of daptomycin occurs via diffusion
(15), which would also be facilitated by the increased fluid
(water) associated with an increased body weight.
V2was estimated to be approximately twofold larger in sub-
jects with acute bacterial infections than in uninfected subjects.
This is consistent with the pathophysiology of acute bacterial
infections, which is characterized by an inflammatory response
associated with increased vascular permeability and the collec-
tion of extracellular fluid at the site of infection. However,
since bacterial infections were only present among subjects in
the phase 2/3 clinical trials, it is also possible that this factor
was a surrogate for another, possibly unmonitored, covariate
or an unidentified systematic difference between the phase 1
and phase 2/3 clinical trials. Currently, there is no recommen-
dation regarding increased doses of daptomycin for patients
with exceptionally severe infections or impaired host defenses.
A trial of daptomycin at 6 mg/kg intravenously once a day for
the treatment of infective endocarditis due to S. aureus is in
Although there was an appreciable variability in the esti-
mates of V1, none of the covariates investigated was identified
as a significant source of this variability. In a recently reported
phase 1 study of daptomycin pharmacokinetics using subjects
with graded renal insufficiencies and end-stage renal disease
(Sica et al., 42nd ICAAC), the total volume of distribution for
daptomycin was increased among subjects on hemodialysis. In
the present larger study, this relationship was extended and
2806 DVORCHIK ET AL.ANTIMICROB. AGENTS CHEMOTHER.
further defined as an increase in V1in subjects with CLCR
values of ?40 ml/min as well as in subjects on dialysis, but only
in a supplemental analysis of variance (Table 5). This may be
because of the relatively small proportion of subjects who were
undergoing dialysis or perhaps because the effect represents
another, possibly unmonitored, covariate. Additional studies
of daptomycin pharmacokinetics in subjects on hemodialysis
are in progress.
In conclusion, this population analysis of daptomycin phar-
macokinetics indicates that renal function is the single most
significant factor contributing to interindividual variabilities in
daptomycin clearance. Because of their reduced daptomycin
clearance, patients on dialysis and those with severe renal
disease (CLCRof ?30 ml/min) will require adjusted dosage
regimens to achieve systemic exposures that are clinically and
pharmacologically comparable to those seen in subjects with
higher levels of renal function. Daptomycin clearance was also
impacted by sex and body temperature. However, an analysis
of clinical outcomes suggested that the variation associated
with sex is not clinically meaningful. The relationship with body
temperature should be interpreted cautiously since the analysis
was limited to the subset of subjects from phase 2/3 clinical
studies, of which only 14% were hyperthermic. The relation-
ships between body weight and the rate and extent of extravas-
cular distribution support the dosing of daptomycin on the
basis of milligrams per kilogram of body weight.
This work was supported by and conducted under the auspices of
Cubist Pharmaceuticals, Inc.
We acknowledge the cooperation and assistance of the subjects,
investigators, and study personnel who participated in these trials and
the support of our colleagues in the Cubist Clinical Department.
1. Arbeit, R. D., D. Maki, F. P. Tally, E. Campanaro, B. I. Eisenstein, and the
Daptomycin 98–01 and 99–01 Investigators. 2004. The safety and efficacy of
daptomycin in the treatment of complicated skin and skin structure infec-
tions. Clin. Infect. Dis. 1673–1681.
2. Beal, S. L., and L. B. Sheiner (ed.). 1988–1998. NONMEM users guides.
Parts I to VIII. GloboMax LLC, Hanover, Md.
3. Brater, D. C., and P. Chennavasin. 1984. Effects of renal disease: pharma-
cokinetic considerations, p. 119–148. In L. Z. Benet, N. Massoud, and J. G.
Gambertoglio (ed.), Pharmacokinetic basis for drug treatment. Raven Press,
New York, N.Y.
4. Bush, L. M., J. A. Boscia, M. Wendeler, P. G. Pitsakis, and D. Kaye. 1989.
In vitro postantibiotic effect of daptomycin (L146032) against Enterococcus
faecalis and methicillin-susceptible and methicillin-resistant Staphylococcus
aureus strains. Antimicrob. Agents Chemother. 33:1198–1200.
5. Canepari, P., M. Boaretti, M. Marlleo, and G. Satta. 1990. Lipoteichoic acid
as a new target for activity of antibiotics. Antimicrob. Agents Chemother.
6. Cockcroft, D. W., and M. H. Gault. 1976. Prediction of creatinine clearance
from serum creatinine. Nephron 16:31–41.
7. Critchley, I. A., D. C. Draghi, D. F. Sahm, C. Thornsberry, M. E. Jones, and
J. A. Karlowsky. 2003. Activity of daptomycin against susceptible and multi-
drug resistant gram-positive pathogens collected in the SECURE study (Eu-
rope) during 2000–2001. J. Antimicrob. Chemother. 51:639–649.
8. Dvorchik, B. H., D. Brazier, M. F. DeBruin, and R. D. Arbeit. 2003. Dap-
tomycin pharmacokinetics and safety following administration of escalating
doses once daily to healthy subjects. Antimicrob. Agents Chemother. 47:
9. Fuchs, P. C., A. L. Barry, and S. D. Brown. 2002. In vitro bactericidal activity
of daptomycin against staphylococci. J. Antimicrob. Chemother. 49:467–470.
10. Gibaldi, M., and D. Perrier. 1982. Pharmacokinetics, 2nd ed. Marcel Dek-
ker, Inc., New York, N.Y.
11. Hanbeger, H., L. E. Nilsson, R. Mailer, and B. Isaksson. 1991. Pharmaco-
dynamics of daptomycin and vancomycin on Enterococcus faecalis and Staph-
ylococcus aureus demonstrated by studies of initial killing and postantibiotic
effect and influence of Ca2?and albumin on these drugs. Antimicrob. Agents
12. Louie, A., P. Kaw, W. Liu, N. Jumbe, M. H. Miller, and G. L. Drusano. 2001.
Pharmacodynamics of daptomycin in a murine thigh model of Staphylococ-
cus aureus infection. Antimicrob. Agents Chemother. 45:845–851.
13. Mackichan, J. J. 1992. Influence of protein binding and use of unbound free
drug concentrations, p. 5–48. In W. E. Evans, J. J. Schentag, and W. J. Jusko
(ed.), Applied pharmacokinetics—principles of therapeutic drug monitoring,
3rd ed. Applied Therapeutics Inc., Vancouver, Canada.
14. Mandema, J. W., D. Verotta, and L. B. Sheiner. 1992. Building population
pharmacokinetic-pharmacodynamic models. I. Models for covariate effects.
J. Pharmacokinet. Biopharm. 20:511–528.
15. Rowland, M., and T. N. Tozer. 1995. Clinical pharmacokinetics: concepts and
applications, 3rd ed. Williams and Wilkins, Baltimore, Md.
16. Sheiner, L. B., B. Rosenberg, and V. V. Marathe. 1977. Estimation of pop-
ulation characteristics of pharmacokinetic parameters from routine clinical
data. J. Pharmacokinet. Biopharm. 5:445–479.
17. Silverman, J. A., N. Oliver, T. Andrew, and T. Li. 2001. Resistance studies
with daptomycin. Antimicrob. Agents Chemother. 45:1799–1802.
18. Wesson, K. M., D. S. Lerner, N. B. Silverberg, and J. M. Weinberg. 2003.
Linezolid, quinupristin/dalfopristin and daptomycin in dermatology. Clin.
19. Wise, R., T. Gee, J. M. Andrews, B. Dvorchik, and G. Marshall. 2002.
Pharmacokinetic and inflammatory fluid penetration of intravenous dapto-
mycin in volunteers. Antimicrob. Agents Chemother. 46:31–33.
20. Woodworth, J. R., E. H. Nyhart, G. L. Brier, J. D. Wolny, and H. R. Black.
1992. Single-dose pharmacokinetics and antibacterial activity of daptomycin,
a new lipopeptide antibiotic, in healthy subjects. Antimicrob. Agents Che-
VOL. 48, 2004 POPULATION PHARMACOKINETICS OF DAPTOMYCIN2807