Back to the Top
The following message was posted to: PharmPK
Hi-
I apologize in advance if this question has already been posted, and for
being so basic. However, it has caused quite a stir among some of the
pharmacists.
For aminoglycoside target concentrations:
UTI/synergy 3-5
Bacteremia, soft tissue infection 6-8
Neutropenia, pneumonia 8-10
Do you consider these concs to be measured peaks (1 hr after the end of
a 30min infusion) or extrapolated peaks (concentrations at the end of a
30min infusion)?
Thanks for any help you can offer!
~Kara
Kara M. Bozik, Pharm.D.
Clinical Pediatric Specialist
UNC Children's Hospital
Chapel Hill, NC
[Sorry for the copy - I'm in the midst of changing hardware - and
listserv software version - db]
Back to the Top
The following message was posted to: PharmPK
Hi Kara,
Don't apologize. You would be surprised at the amount of controversy
regarding apparently "simple" concepts in clinical PK/PD.
Back to your question: I am in the camp that considers these
extrapolated
peaks.
Thanks,
Mike
Michael A. Jones, Pharm.D.
CPOE/e-CDS Clinical Specialist
Pharmacy Care Team
University of Colorado Hospital
Phone: 303-372-8214
michael.jones.aaa.uch.edu
www.uch.edu
Back to the Top
The following message was posted to: PharmPK
At our institution those levels are the drawn peaks at 30 minutes post
completion of the infusion.
Thanks..... Robert
Robert G. Aucoin, Pharm D.
Clinical Pharmacy Spec - Peds/PICU
Department of Pharmacy
Our Lady of the Lake RMC
Baton Rouge, La
Office: 225-765-7652
e-mail: raucoin.-a-.ololrmc.com
Back to the Top
The following message was posted to: PharmPK
Hi
I think a bigger question is...
Why are you using multiple daily dosing rather than extended interval
dosing eg every 24 hours?
Regards
Dr Carl Kirkpatrick
School of Pharmacy
University of Queensland
St Lucia, 4072
Brisbane, Australia.
Back to the Top
The following message was posted to: PharmPK
Kara
30 minutes after the end of infusion will be a better practice for
estimating the peak levels of aminoglycosides.
Krishna
Krishna Kumar, Ph. D. MPS
Professor of Biopharmaceutics & Pharmacokinetics
Howard University, School of Pharmacy,
2300 4th Street, N. W.
Washington DC 20059
Ph: 202 806 6540
Fax: 202 806 4636
Back to the Top
The following message was posted to: PharmPK
It is arguable that the aminoglycoside measured at 30 minutes post
infusion may reflect the drug still in the distribution phase, giving
possible erroneous elevated serum concentrations. What we customarily do
at our institution is draw blood at 2 and 6 hours post infusion, and
extrapolate the data to the time of the end of infusion, giving a true
Cmax.
Glenn
Glenn Whelan, Pharm.D.
Adjunct Clinical Pharmacologist/Clinical Coordinator
National Jewish Medical and Research Center
1400 Jackson Street
Office J329
Denver, CO 80206
phone: 303-398-1448
fax: 303-270-2189
Email: whelang.at.njc.org
Back to the Top
The following message was posted to: PharmPK
Dear Glenn,
You wrote:
> It is arguable that the aminoglycoside measured at 30 minutes post
> infusion may reflect the drug still in the distribution phase, giving
> possible erroneous elevated serum concentrations.
What's erroneous here? If you measure correctly, the serum concentration
cannot be 'erroneous elevated'. Probably you mean that a meausrement
in the
distribution phase may be higher than expected, and may be erroneously
interpreted as 'elevated'.
> What we customarily do
> at our institution is draw blood at 2 and 6 hours post infusion, and
> extrapolate the data to the time of the end of infusion, giving a
true
> Cmax.
This approach is probably not the worst thing to do for a good
interpretation of aminoglycosides levels, but literally this sentence is
strange. You cannot get a true Cmax by extrapolation, and certainly
not in
this case where measured values are higher than Cmax. Cmax is by
definition
the highest concentration reached. No discussion about that, I think.
And
Cmax is likely to be reached at or shortly after stopping the infusion.
Also from the point of interpretation I do not agree with this
approach. The
reason that one does not like measurements during the distribution
phase is
that it makes pharmacokinetic analysis more complicated, but not
impossible
using a two-compartment model and Bayesian estimation. For PK-PD the
'possible erroneous elevated serum concentrations' are likely to be
important. Bacteria do not have knowledge of pharmacokinetics; they
'feel'
only concentrations, and in case of aminoglycosides, in particular high
concentrations.
Best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
tel. 31-50 363 3292
fax 31-50 363 3247
Email: j.h.proost.-a-.rug.nl
Back to the Top
The following message was posted to: PharmPK
Hi Kara:
When you write "aminoglycosides" I think you want to say "gentamicin
or tobramycin",
no? In this sense your desired plasma concentrations are right.
You know that AMG are rapidly and widely distributed; even after IM
injections you
can obtain peaks after 30 - 120 minutes, with an average of 60 min
(erratic surely).
Normally you can take 30 minutes, although also you can take it after
60 min. Is very
important that you know (exactly) the time infusion, and the
extration time too. In
clinical practice these points not always are so easy to know it.
I agree about the controversial point of the subject, but there is
only limited evidence
about that you can take 2 or 3 or 6 or 8 hours post infusion (I am
not saying that is not
valid), mainly when you compare with classical 30 or 60 min.
extraction time
strategies, so, if you can, you are sure (with a lot of bibliography)
with peaks between
30 and 60 minutes "very well known".
Good luck.
Paulo.
Back to the Top
The following message was posted to: PharmPK
Hans and Glenn you both make excellent points and raise deeper
questions.
Glenn is obviously referring to a one-compartment model. This seems
appropriate to me (without doing a literature search) because that
was the
model used by most of the early clinical researchers to define
therapeutic
ranges. In addition, I also prefer to check two levels after a dose.
Hans' replay suggests other important questions.
1. Is the one-compartment model the best model for clinical use?
2. Are Bayesian methods mature enough for general use?
3. How sure are we of the therapeutic ranges for aminoglycosides?
Thanks,
Mike
Michael A. Jones, Pharm.D.
CPOE/e-CDS Clinical Specialist
Pharmacy Care Team
University of Colorado Hospital
Phone: 303-372-8214
michael.jones.-a-.uch.edu
www.uch.edu
Back to the Top
Dear Glenn:
What you say is most interesting. How is it that a measured
serum concentration is "falsely" elevated if it is obtained before
the distribution phase is complete? And how is it that a
concentration extrapolated back from measured 2 and 6 hr samples is
somehow "true"?
And in general, why is it that some people still feel that
serum samples must be obtained after distribution is complete, and
that the patient should be in a steady state?
I have never heard clearly just why a sample should not be
obtained before distribution is complete, for example. Can you help
me with this? And why should the patient be in a steady state? In
general, what are the best times to get serum samples, and why?
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Dr. Jelliffe:
The method by which I am writing about, is a one-compartment model,
not a multicompartment model. I may have mispoke before when Hans
pointed out that those serum concentrations are not 'erroneously
high', as that would refer to the assay method. However, obtaining
blood immediately after the end of the infusion would be erroneous
because you will get a serum concentration that is artificially high,
because of the drug still equlibrating/distributing into the
peripheral compartment. Therefore I would wait some time after then
end of the infusion, to grab two data points and back extrapolate,
and forward extrapolate the Cmax and Cmin. The Cmax, in this case
would reflect the central compartment's Cmax at the end of the infusion.
If I were to use the one-compartment model, and grab blood right
after the end of the infusion, and at another time point, and
performed calculations based on that (using the one compartment
model), I would arrive at a false elimination rate constant, and then
a falsely 'fast' elimination half life and clearance. I would also
suspect that if I looked at the serum concentration and decided that
it was within the desired range, I would then be under-treating the
patient.
I can see that you are encouraging the use of a population
pharmacokinetic model, which I fully agree with, that will allow us
to predict the patient's pharmacokinetic parameters by grabbing
sparse samples that are not necessarily in the 'post distribution'
phase, nor would the patient be at steady state.
Glenn
Back to the Top
Dear Glenn:
Thanks for your note. I would suggest that whether the 1
compartment model is correct or not is not the point. The point is,
with whatever imperfect model one has chosen to use, what are the
optimal times to get samples when they will let you make the most
precise parameter estimates for that model? Even prior to completion
of the distribution phase, the serum concentrations are neither
"erroneously" nor artificially" high. They simply are what they are.
When you are interested in getting the best rate constant for
elimination, did you ask what you had done to the perception of the
volume of distribution? And the relationship of that to clearance?
Extrapolation from linear regression on the logs of the
concentrations is a poor way to get parameter values. This is well
known. It assumes a weighting scheme that is usually erroneous, it
throws away all other that you may have had from previous doses, and
it is limited to only seeing a 1 compartment model. And it often gets
erroneous parameter values. Nonlinear regression on the data
themselves is much better, and MAP Bayesian methods are still better
when appropriate.
Whatever the model, it is much better to use some sort of
optimal design strategy. These have been around for over 30 years
now. A good article to read is by Dave D'Argenio - Optimal Sampling
Times for Pharmacokinetic Experiments. J Pharmacokin. Biopharm. 9:
739-756, 1981.
Yes, I am very much encouraging the use of population PK
approaches, especially nonparametric ones, and Bayesian adaptive
control, to study drug behavior and develop and adjust dosage
regimens. Optimal sampling strategies are an important part of that.
So is "Multiple Model" dosage design, and MM or IMM Bayesian adaptive
control (see the last reference). Some other references to consider
for practical clinical application are:
Jelliffe R, Schumitzky A, and Van Guilder M: Population
Pharmacokinetic / Pharmacodynamic Modeling: Parametric and
Nonparametric Methods. Therap. Drug Monit. 22: 354-365, 2000.
Jelliffe R, Bayard D, Milman M, Van Guilder M, and Schumitzky A:
Achieving Target Goals most Precisely using Nonparametric
Compartmental Models and "Multiple Model" Design of Dosage Regimens.
Therap. Drug Monit. 22: 346-353, 2000.
Jelliffe R: Goal-Oriented, Model-Based Drug Regimens: Setting
Individualized Goals for each Patient. Therap. Drug Monit. 22:
325-320, 2000.
Jelliffe R: Estimation of Creatinine Clearance in Patients with
Unstable Renal Function, without a Urine Specimen. Am. J. Nephrology,
22: 3200-324, 2002.
Bayard D, and Jelliffe R: A Bayesian Approach to Tracking Patients
having Changing Pharmacokinetic Parameters. J. Pharmacokin.
Pharmacodyn. 31 (1): 75-107, 2004.
Really, I believe linear regression on logs of the data is
DEAD as a PK method. I really do not understand why it is still used
at all, or still taught at all, for that matter. Any responses?
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Kara, aminoglycoside target concentrations defined in the handbooks
should represent measured peaks (a "time" after the end of the
infusion also defined in the handbook). I dont think the question is
so basic, because it is an important source of error.
Dr Guillermo Bramuglia
Profesor Adjunto
Catedra de Farmacologia,
Facultad de Farmacia y Bioquimica,
Universidad de Buenos Aires
Back to the Top
The following message was posted to: PharmPK
Guillermo Bramuglia wrote:
> PharmPK - Discussions about Pharmacokinetics
> Pharmacodynamics and related topics
>
> Kara, aminoglycoside target concentrations defined in the handbooks
> should represent measured peaks (a "time" after the end of the
> infusion also defined in the handbook). I dont think the question is
> so basic, because it is an important source of error.
I think the use of unidentified handbooks as a source of definitions
is 'an important source of error'. My own definition of target
concentration is certainly NOT the measured concentration e.g.
Holford NHG. Target Concentration Intervention: Beyond Y2K. British
Journal of Clinical Pharmacology 1999;48:9-13.
Holford NHG. The target concentration approach to clinical drug
development. Clinical Pharmacokinetics 1995;29(5):287-91
The measured concentrations are used to estimate individual PK
parameters (e.g. clearance and volume). These individualized
parameters can then be used to modify dosing to achieve the target
concentration. The timing of the measured concentrations is optimized
for estimation of the parameters. For aminoglycosides this typically
involves a measurement about 30 minutes after the end of the infusion
of the first dose and another about 8 hours later. Two measurements
are necessary in order to estimate 2 parameters.
My own suggestion for aminoglycoside concentrations is to achieve a
peak of 8 mg/L (8 hour dosing) or 24 mg/L (24 hour dosing) after the
first dose and an average of 3 mg/L for all subsequent dosing intervals.
There is no reason to measure concentrations at the same time as the
target concentrations are desired. However, for drugs which only need
an estimate of clearance (e.g. digoxin) then a measuring a
concentration in the middle of the dosing interval will often be
close to the average steady state target concentration.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-a-.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
> Nick Holford wrote:
I think the use of unidentified handbooks as a
> source of definitions
> is 'an important source of error'. My own definition
> of target
> concentration is certainly NOT the measured
> concentration
The unidentified handbooks or books are:
- Drug Information Handbook 9th Ed pag 1330, or
-Michael Winter, Basic Clinical Pharmacokinetics, Part
Two Aminoglycosides, 2nd Ed.
> My own suggestion for aminoglycoside concentrations
> is to achieve a
> peak of 8 mg/L (8 hour dosing) or 24 mg/L (24 hour
> dosing) after the
> first dose and an average of 3 mg/L for all
> subsequent dosing intervals.
>
> There is no reason to measure concentrations at the
> same time as the
> target concentrations are desired. However, for
> drugs which only need
> an estimate of clearance (e.g. digoxin) then a
> measuring a
> concentration in the middle of the dosing interval
> will often be
> close to the average steady state target
> concentration.
Ok, The problem is how to defined a "target
concentration" without measuring this concentration.
Guillermo
Dr Guillermo Bramuglia
Profesor Adjunto
Catedra de Farmacologia,
Facultad de Farmacia y Bioquimica,
Universidad de Buenos Aires
Back to the Top
The following message was posted to: PharmPK
Hi
I'm not sure what the confusion is here.
There are several methods for monitoring Ag's when given by extended
interval dosing. Nick provided one reasonable method (although I
don't know
that anyone has used it), other methods include the AUC method of
Begg and
the Hartford nomogram.
The problem with all methods is that probably none are optimal for
patient
care and indeed the optimal monitoring strategy is not known.
Certainly it
seems, anecdotally, that the act of monitoring your patient regularly
with
clearly set clinical goals (and PK goals) helps to reduce untoward
toxicity.
> > Nick Holford wrote:
>
> > There is no reason to measure concentrations at the >
> same time as the > target concentrations are desired.
If you use the any nomogram method (e.g. the Hartford nomogram) then you
have to measure concentrations at the time required by the monitoring
method. If you are doing modelling (e.g. Bayesian dose
individualisation)
then you can pretty much take your concentration at any time
(preferably an
informative time is good).
Dr Guillermo Bramuglia wrote:
> Ok, The problem is how to defined a "target concentration"
> without measuring this concentration.
To define the target you need to have a fully integrated PKPD model to
describe the exposure-response relationship. However I think you
mean to
achieve the target concentration - in which case yes you do have to
measure
a concentration but as Nick says you don't have to do it at the time
that
the monitoring strategy says if you use a modelling based approach.
Regards
Steve
==================================================================Stephen Duffull
School of Pharmacy, University of Queensland, Brisbane 4072, Australia
Tel +61 7 3365 8808, Fax +61 7 3365 1688, Email:
sduffull.-at-.pharmacy.uq.edu.au
www http://www.uq.edu.au/pharmacy/index.html?page=31309
Design: http://www.uq.edu.au/pharmacy/sduffull/POPT.htm
MCMC: http://www.uq.edu.au/pharmacy/sduffull/MCMC_eg.htm
University Provider Number: 00025B
Back to the Top
Guillermo,
Thanks for identifying the sources of your definition of a TC.
However, I do not agree with the idea that the TC is defined by a
measured concentration. I hope you will read the papers I cited which
explain this.
The important idea is to understand that the TC is the conc that is
expected to produce the target effect (TE). The TE is defined by the
patient and prescriber. Application of PKPK principles can be used to
predict the TC needed to get the TC. Development of a PKPD model
relating concs to effects needs measurements of concs and effects but
this is not done as part of TDM. It is done as part of a formal
clinical pharmacology study with sufficient conc and effect
meausurements to define the PKPD. TDM can then take advantage of the
PKPD study results to individulize patient treatment.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.aaa.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
OK, now I'm a bit confused. "achieve a peak of 8 mg/L (8 hour dosing)
or 24 mg/L (24 hour dosing) after the first dose and an average of 3
mg/L for all subsequent dosing intervals"? What happened to the PD
panacea that outdated PK? I very much agree with Nick ('any time',
optimal if possible, should do to take a good blood sample) and with
Roger too (there are no erroneous concentrations unless the assay kit
is bad). But in my view, the primary surrogate goal of antibiotherapy
at large is to achieve or sustain blood concentrations, or better
said, biophase levels, above the MIC characteristic for the pathogen.
And this shouldn't even be generalized in terms of the Species or
variety, since a common intestinal disarrangement by E.coli is
certainly different from a nosocomial infection by the same bacteria,
taxonomically. "To define the target you need to have a fully
integrated PKPD model to describe the exposure-response
relationship."? Don't you just need a significant MIC and then plain
old PK to calculate dosing rate (models are from a different Pandora
box). In the past I've found the use of nomograms sometimes terribly
misleading, especially in particular clinical situations such as
pediatrics or ICU infections. But nowadays the references are just so
many about not generalizing antibiotherapy (unidentified handbooks,
yes, but it's just too late for me; maybe next time).
Luis
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.aaa.bos.mcphs.edu
Back to the Top
The following message was posted to: PharmPK
Luis,
The MIC and AUC/MIC and time above MIC approaches are all empirical
and have no reasonable theoretical basis e.g. they give no guidance
on how long treatment should continue or what the optimal dosing
interval should be. It is time for the anti-infective research
community to start thinking and not just continue with trial and
error approaches.
My suggested targets are based on my own interpretation of clinical
trials that have demonstrated effective treatment using empirical
dosing (q8h or q24h). There is unfortunately no adequate PKPD model
to provide any better guidance.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.at.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
Nick Holford wrote:
"The measured concentrations are used to estimate individual PK
parameters (e.g. clearance and volume). These individualized
parameters can then be used to modify dosing to achieve the target
concentration. The timing of the measured concentrations is optimized
for estimation of the parameters. For aminoglycosides this typically
involves a measurement about 30 minutes after the end of the infusion
of the first dose and another about 8 hours later. Two measurements
are necessary in order to estimate 2 parameters." [Not to nit pick, but
actually, we need three SDCs - the first one is the pre-infusion SDC
of 0
mcg/mL, and it is available at no cost. The advantages of obtaining
SDCs
after the first dose include one SDC at no cost, no assumptions about
any
SDC, and early optimization of dosage.]
I, for one fully agree with Nick's statement that the purpose of
obtaining
serum drug concentrations (SDC) is to estimate PK parameters. The
estimation of PK parameters also allows the clinician an opportunity to
uncover errors in the attributes of the SDC (e.g. recorded time vs
actual
SDC sampling time, and recorded time vs actual dosing time), and
errors in
the initial estimates of PK parameters (e.g. unrecognized fluid shifts,
overestimating renal function, etc.).
There seems to be two major philosophies of PK/PD practice.
1. Use measured SDCs to adjust dosage directly; and
2. Use measured SDCs to estimate PK parameters and then adjust dosage.
The problem with #1 is that there appears to be no appreciation or
knowledge
of the errors mentioned above, and therefore no investigation. There
are
too many assumptions about the patient's clinical circumstances with
this
approach.
I am interested in how the listserv feels about these two
philosophies of PK
practice, as well as any other modes of PK practice associated with the
aminoglycosides.
Thanks,
Mike
Michael A. Jones, Pharm.D.
CPOE/e-CDS Clinical Specialist
Pharmacy Care Team
University of Colorado Hospital
Phone: 303-372-8214
Pager: 303-266-6893
michael.jones.at.uch.edu
www.uch.edu
Back to the Top
The following message was posted to: PharmPK
Michael,
You wrote:
"[Not to nit pick, but actually, we need three SDCs - the first one
is the pre-infusion SDC of 0 mcg/mL, and it is available at no cost."
Thanks for your reply. I dont understand at all why you want a pre-
dose conc before the first dose? This conc is zero so why waste any
time thinking about it? It cannot add anything to the PK parameter
estimation obtained from the 2 concs obtained after the first dose.
Nick
Back to the Top
The following message was posted to: PharmPK
Hi Nick,
This statement is probably a result of my pre-Bayesian thinking. Vd
(one
compartment) is determined by the change in serum drug concentration
resulting from the first dose (0 --> Cmax).
Many of my contemporaries like to wait until "steady state" and draw two
SDCs, either a peak and trough, or a trough and peak. In both cases,
the
trough must do double duty, representing both a pre-infusion trough
and a
post-infusion SDC. This approach requires assumptions about "steady
state"
and timing of previous doses, etc.
By obtaining SDCs after the first dose, we eliminate the need for such
assumptions.
Thanks,
Mike
Michael A. Jones, Pharm.D.
CPOE/e-CDS Clinical Specialist
Pharmacy Care Team
University of Colorado Hospital
Phone: 303-372-8214
Pager: 303-266-6893
michael.jones.aaa.uch.edu
www.uch.edu
Back to the Top
The following message was posted to: PharmPK
Michael,
I don't think a pre-first dose conc of zero is every helpful --
Bayesian or Frequentist. It can of course be helpful if (as often
happens) the patients and clinicians have no idea what doses were
given and when. Measuring a conc prior to the first dose may reveal
that the patient has indeed been treated within the recent past with
an aminoglycoside.
I am glad you appreciate that it is utterly foolish to wait until
'steady state' to measure aminoglycoside concs. The people who
advocate this practice should have their licenses cancelled.
As a pragmatic matter it may be more practical to get a trough (at 8
hours) before the second dose and a peak after the second dose rather
than getting the peak after the first dose. But this is a matter of
local co-ordination with the lab that measures the concs.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-at-.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
Hi Nick,
I am sorry for the confusion, my earlier e-mail must not have been
clear.
What I was trying to suggest is drawing two concentrations after the
first
dose would actually provide three known concentrations at the price
of two.
The pre-first dose concentration is already known and does not need
to be
assayed.
Thanks,
Mike
Back to the Top
The following message was posted to: PharmPK
Dear Michael and Nick,
I would like to make some comments on the following
statement:
> The timing of the measured concentrations is optimized
> for estimation of the parameters. For aminoglycosides
> this typically involves a measurement about 30 minutes
> after the end of the infusion of the first dose and
> another about 8 hours later.
(a) I know the argument of distribution to take the
first('peak', confusing term here) sample about 30 minutes
after the end of the infusion. But where can we find the
evidence for this practice? Perhaps I missed something,
but I heard this argument for about 20 years without a
reference.
(b) The second sample about 8 hours later is not likely to
be the optimal sampling point to get the most precise
estimates of the PK parameters. Depending on the (assumed)
error in the measurements, the optimal sampling point is
at 1/k, or somewhat later. I guess that the 8 hours point
is related to the 8 hours dosing interval, and not to an
optimal sampling times analysis.
> There seems to be two major philosophies of PK/PD
> practice.
> 1. Use measured SDCs to adjust dosage directly; and
> 2. Use measured SDCs to estimate PK parameters and
> then adjust dosage.
As you will understand, I choose for option #2, extended
with a Bayesian approach.
Best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
tel. 31-50 363 3292
fax 31-50 363 3247
Email: j.h.proost.at.rug.nl
Back to the Top
The following message was posted to: PharmPK
Hans,
> I would like to make some comments on the following
> statement:
>
> > The timing of the measured concentrations is optimized
> > for estimation of the parameters. For aminoglycosides
> > this typically involves a measurement about 30 minutes
> > after the end of the infusion of the first dose and
> > another about 8 hours later.
>
> (a) I know the argument of distribution to take the
> first('peak', confusing term here) sample about 30 minutes
> after the end of the infusion. But where can we find the
> evidence for this practice? Perhaps I missed something,
> but I heard this argument for about 20 years without a
> reference.
I am afraid I cannot provide a reference for this. However, I am sure
you agree that in the real world it must take at least a few minutes
for mixing to occur within the central blood circulation. Taking a
sample exactly at the end of an infusion will certainly not be a good
time even with a 2 or 3 compartment model. The 30 minute suggestion
is a pragmatic clinical rule.
> (b) The second sample about 8 hours later is not likely to
> be the optimal sampling point to get the most precise
> estimates of the PK parameters. Depending on the (assumed)
> error in the measurements, the optimal sampling point is
> at 1/k, or somewhat later. I guess that the 8 hours point
> is related to the 8 hours dosing interval, and not to an
> optimal sampling times analysis.
The optimal sampling time will be later than 1/k because the residual
error is not homoscedastic. A typical additive error is 0.14 mg/L
(Matthews et al. 2004). Samples taken at 8 hours even with 8 hour
dosing can usually be expected to be well above this value and
therefore the proportional error will be dominant. Taking a sample as
late as possible is the optimal sampling time for a proportional
error. For 24 hour dosing samples taken at 24 hour will often be less
than the limit of detection. A pragmatic clinical rule is therefore
to advise an 8 hour sampling time for both 8 and 24 h dosing intervals.
Matthews I, Kirkpatrick C, Holford NHG. Quantitative justification
for target concentration intervention - Parameter variability and
predictive performance using population pharmacokinetic models for
aminoglycosides. British Journal of Clinical Pharmacology 2004;58(1):
8-19
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-a-.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
[Nick, thanks for pointing out the interaction between optimal
sampling and 'sample' error - it has been available in ADAPT Sample
module for some time but I haven't seen much discussion or use of
this interaction in practical (clinical) setting. I had explored it a
little without conclusion (lacking clinical experience) See page 7-8
of http://www.boomer.org/c/p2/Ch33W.pdf but this information is no
longer considered core curriculum material ;-) - db]
Back to the Top
The following message was posted to: PharmPK
Dear forum,
I just would like to add to this discussion about the sampling time for
amoniglycosides.
In patients with adequate renal function (CLcr >60 mL/min), a trough
aminoglycoside level (taken during the last hour of the dosing interval)
should be near zero i.e. <<1 mg/L. If the levels are found so, the
initial
dosing interval may be maintained and no further drug levels are
necessary
as long as CLcr remains unchanged.
If the trough level concentration is > 1 mg/L, an extension of the
dosing
interval will be necessary. In such a case, an accurately timed random
level may be obtained 7 - 14 hrs after the end of infusion especially
for
gentamicin, netilmicin, and tobramycin.
It is also said that the choice of a 24-hrs dosing interval (pulse
dosing) for patients
with CLcr > 60 mL/min has been prescribed primarily by convenience
rather
by pharmacokinetic considerations.
Secondly, the peak-to-MIC ratio for amoniglycosides should be > 10,
which
is easily achieved in most patients. Therefore, there is usually no need
to measure the peak level (30 minutes sample) except in patients with
markedly expanded volume of distribution.
Especially in case of pulse doing of amoniglycosides, it is said that
every dose is "first dose", so steady state will not occure. Drug
monitoring should begun from
first dose.
One of the study also suggested a need to monitor two serum
concentrations at least one half-life apart to identify patients with
atypical clearance (Ref.: van der Anwera P. Pharmacokinetic
evaluation of
single daily dose amikacin. J Antimicrob Chemother 1991;27(Suppl C):
63-67.)
Prashant
--
Kole Prashant, M.Pharm (Ph.D)
Faculty,
Pharmacy Group,
Birla Institute of Technology and Science, Pilani
Rajasthan, 333031
Phone: (O) 01596-245074-Ext-458
Fax: 01596-244183
Alt.Email: prashantkole.at.rediffmail.com,plkole.at.yahoo.com
Back to the Top
The following message was posted to: PharmPK
Dear Nick, Proost, Prashant and All
I partly agree with Nick's discussion about AUC, and the part I
disagree pertains to the understanding that AUC is in essence a
measure of exposure and that's why time is "averaged out" in the
integration process. With antibiotics for which therapeutic effect
relates to time of exposure AUC is a good PK estimator. Similar cases
occur with TDM of cytotoxics and immunosuppressants. However, being
Gram- bacteria susceptible to aminoglycosides in a concentration-
dependent fashion that's the scientific rationale for considering
'peak' levels rather than the notion of exposure.
But maybe I wasn't clear in my earlier posting. What I find crucial
to consider is not the AUC, but rather a measure of the pathogen
susceptibility. MIC and MBC are perhaps the oldest, but growth rate
versus kill rate, time to kill, and most of the other metrics I
alluded to (not to repeat), incorporate the notion of
chronopharmacodynamics most relevant to microorganism's population
dynamics (I strongly recommend a close reading of Derendorf, Forrest
and Dalhoff's work).
Nick> ...MIC...is dependendent on the duration of incubation.
I don't understand how MIC dependence on incubation time should cloud
its usefulness. It also depends on temperature, pH and culturing
medium, but all these variables are standardized, as incubation time,
to infer about in vivo infection. Our entire scientific edifice is
based on assumptions that precede any conclusion (no magic). The fact
of the matter is that if drug concentrations are above the MIC for a
time that compensates the duplication rate of bacteria, the
likelihood for wiping the infection and preventing resistance is the
greatest.
Proost> ... take the first ('peak', confusing term here) sample about
30 minutes after the end of the infusion.
Nick> The 30 minute suggestion is a pragmatic clinical rule.
Proost> ... where can we find the evidence for this practice?
This is just another example of Occam's razor, aka the parsimony
principle or law of ecconomy. The explanation is certainly related
with distribution kinetics (many references, from early Schentag J.
et al., Antimicr.Agents Chemother. 1981,19:5,859 all the way to
Andras V. et al. TDM, 2000,22:6,676). Gentamicin (which by the way is
not a single compound but rather a mixture of three with slightly
different PK characteristics) sustains in vivo a bi-exponential
disposition, akin to saying that a 2-compt. open model most
adequately describes rich data type conc.vs.time. Its distribution
half-life has been reported to be between 21.8 and 41.6 min. So a
quick calculation about the fraction remaining to be distributed
after let's say 3 half-lives, clearly shows that a significant amount
of drug is still being distributed to peripheral spaces. Now, in
clinical terms and for TDM purposes this can be perceived as
irrelevant, focusing on peaks and troughs, and thus the single
exponential approximation (one compt. model). But to be consistent
drug concentrations can only be dealt with in the post-distribution
phase, which by convention has been established to be sufficiently
reached 30min. after the end of the infusion.
Prashant> ... a trough aminoglycoside level (taken during the last
hour of the dosing interval) should be near zero i.e. <<1 mg/L.
Prashant is absolutely right and this is a tremendously important
remark. Both from a kinetics stand point and from a clinical
perspective (post-antibiotic effect, prevention of resistance,
nefrotoxicity) pre-dose gentamicin levels should be essentially zero.
The accumulation factor, 1/(1-e^-ke.tau), is approximately 1.5 for
tid and 1.0 for qd, when the notion of steady-state doesn't apply, or
if preferred, ss is reached right after dose one. It has been
established for a long time now that generally the benefits of
gentamicin qd dosing are very significant (e.g. "...once daily
therapy: 8/8 meta-analyses support the use" in Ann Intern Med
1996;124:717).
So I must say again, trashing other people's work while waiting for a
'golden age' doesn't serve the advancement of science. Anti-infective
drugs are a very serious business. Let me suggest just two more:
"Novel Pharmacokinetic-Pharmacodynamic Model..." by Meagher A.K. et
al., Antimicr.Agents Chemother. 2004,48:6,2061 , and "Pharmacokinetic/
pharmacodynamic evaluation of anti-infective agents" by Schuck E. et
al., Expert Rev. of Anti-infective Ther. 2005,3:3,361
Yours truly,
Luis
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.-a-.bos.mcphs.edu
Back to the Top
The following message was posted to: PharmPK
Interesting discussion on aminoglycoside monitoring issues.
There is another interesting and economic issue with the idea of getting
first dose or even second dose aminoglycoside levels (and hopefully I
won't
have my "license cancelled" for this commentary :-).
It is quite common that aminoglycoside administration is part of an
initial
empiric regimen. It is also very common to see the aminoglycoside
discontinued within 48 to 72 hours of initiating therapy (because of
culture
& sensitivity results, change in patient status, change in empiric
therapy,
change to oral therapy, etc).
I was involved in an aminoglycoside DUE that showed an impressive
number of
aminoglycoside levels that were obtained and never utilized because the
patient's aminoglycoside had been discontinued by the time the
results were
back (sorry, I don't have the data at hand). In 24-hour dosing
schemes, it
seems to be common that therapy is discontinued/changed between the
second
and third dose of aminoglycoside. This is one reason that I am not a
big
advocate of getting levels (at least reflexively) surrounding the second
dose (certainly not for q8h dosing or q12 dosing). Patient/disease
circumstances of course prevail.
I do think it is prudent to not be too anxious to see the numbers
with some
exceptions for those who are "more sick" or who have less than very good
renal and/or other organ function (vague statement I know, but obviously
this needs to be patient specific).
Has anyone else looked at how many levels they get that are essentially
wasted?
Regards,
Peter
Someone wrote:
>I am glad you appreciate that it is utterly foolish to wait until
>'steady state' to measure aminoglycoside concs. The people who
>advocate this practice should have their licenses cancelled.
>As a pragmatic matter it may be more practical to get a trough (at 8
>hours) before the second dose and a peak after the second dose rather
>than getting the peak after the first dose. But this is a matter of
>local co-ordination with the lab that measures the concs.
*****
Peter Underwood, Pharm.D.
Clinical Pharmacist
Jarosz Regulatory Services, Inc.
1634 W. Wildwood Rd.
Whitewater, WI 53190 USA
Back to the Top
The following message was posted to: PharmPK
Peter
You raise a good point. There is (of course) a corollary argument - how
many people are started on what was thought to be a 2 or 3 day course
and
then stay on it a whole week? We have come across several cases of
this -
and these are usually the most dangerous as everyday is (incorrectly)
thought to be the last day and hence no "levels" are ever taken.
So, I have a contrasting view: if it is only a single dose (e.g. some
sort
of prophylaxis before a procedure) then take no levels. If it is for a
treatment (even if for only 1 or 2 days) then take levels on the
FIRST dose
(not second or third). If dosing using an extended interval take
(preferably) 2 levels, one at about 30 minutes post-infusion and one at
about 6-14 hours later.
The reason for the timing of the samples is 2 fold:
1) The later sample is taken sufficiently before the next dose that
the lab
can measure the Ag and if necessary the second dose can be changed. The
thought of taking the 2nd level so late in the dose interval that it
could
not influence the choice of the second dose does not seem sensible.
2) The times are reasonably informative from a design of experiments
view
point. For a 1-cpt model and additive error the locally optimal
sampling
times are at time=0 and time=1/k (as mentioned by others before). As
also
pointed out it is likely that Ags follow more complex 2 or 3 cpt
characteristics and therefore if you are using a 1-cpt model (most
common)
then you need to take a sample at a time close to zero but one that
won't
introduce bias due to model misspecification. With respect to the
second
sample it has been pointed out that the residual error model is most
likely
proportional with a slight additive component. If it were only
proportional
then Nick correctly indicates that the second sampling time would be
at the
trough - so i) a sample between 1/k and the dose interval, ii) a sample
early enough to allow the lab to assay the sample *before* the next
dose is
due and iii) at a time when the sample will be greater than the limit of
detection. That is: approx between 6 and 14 hours post-dose.
As far as wasted samples. In a time of limited resources then
perhaps the
question: "Is it better to waste x samples or risk y cases of
nephrotoxicity?" should be answered by the budget holders after an
appropriate safety-economic analysis?
Steve
==================================================================Stephen Duffull
School of Pharmacy, University of Queensland, Brisbane 4072, Australia
Tel +61 7 3365 8808, Fax +61 7 3365 1688, Email:
sduffull.-at-.pharmacy.uq.edu.au
www http://www.uq.edu.au/pharmacy/index.html?page=31309
Design: http://www.uq.edu.au/pharmacy/sduffull/POPT.htm
MCMC: http://www.uq.edu.au/pharmacy/sduffull/MCMC_eg.htm
University Provider Number: 00025B
Back to the Top
The following message was posted to: PharmPK
Stephen
Sorry for intruding in your reply to Peter, but this is just too
important. No antibiotic (particularly an aminoglycoside) should be
administered as a single dose EVER. Even if a patient becomes
asymptomatic or is discharged, including surgical prophylaxis,
antibiotherapy (even without isolating the pathogen) OUGHT to be
continued for at least 3-5 days (given anticipated bacterial growth)
or more based on susceptibility data. Opportunistic infections occur,
patients get superinfected, common strains become resistant, and next
time the patient is admitted the same antibiotic won't work and there
may be just no other to choose from. I subscribe your
pharmacoeconomic analysis, but let me rephrase it in this way: Each
dollar saved on skipping a dose for a full treatment, many times
represents hundreds more spent later on readmitting the same patient.
Luis
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.-a-.bos.mcphs.edu
Back to the Top
The following message was posted to: PharmPK
***It is incorrect to suggest that one should never use a single dose
of an
antibiotic and that one should always treat for at least 3-5 days in all
cases. Here are a few exceptions (taken from the British National
Formulary):
Prophylaxis of meningococcal meningitis in contacts: oral
ciprofloxacin 500 mg
or i.m. ceftriaxone 250 mg, each as a single dose.
Prophylaxis of infective endocarditis in susceptible individuals
before dental
procedures: a single dose of amoxicillin, clindamycin, or azithromycin.
Treatment of dental abscesses: amoxicillin 3 g repeated after 8 hours.
Treatment of urinary tract infections: amoxicillin 3 g repeated after
10-12
hours.
In discussing recommendations for antimicrobial drug therapy, the
results of
clinical trials of efficacy must be taken into consideration. There
is a large
literature on the use of single doses of antibiotics, much of it
conflicting;
one cannot be dogmatic. It is not yet clear whether single-dose
treatment, in
cases in which it is currently recommended, causes more or less
bacterial
resistance.
I haven't followed all of this discussion, but has anyone raised the
question
of the postantibiotic effect of aminoglycosides? The relation between
concentration and beneficial effect is poor because of it. Harmful and
beneficial effects need to be discussed separately.
Jeff
--
J K Aronson, MA MBChB DPhil FRCP FBPharmacolS.
University Department of Clinical Pharmacology,
Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE.
Tel: (01865) 224626. Fax: (01865) 791712.
Back to the Top
The following message was posted to: PharmPK
Jeff,
I agree with your assertion that one should not be dogmatic about the
merits of single dose antibiotic treatment.
But I would caution you about the dogma you imply with your statement
about the 'poor relation' of aminglycoside concentration to
beneficial effect.
> I haven't followed all of this discussion, but has anyone raised
the question
> of the postantibiotic effect of aminoglycosides? The relation between
> concentration and beneficial effect is poor because of it.
As far as I am aware it is true for almost* all drugs that the
relationship between plasma concentration and effect is delayed. The
naive interpretation of an immediate plasma concentration to effect
relationship leads to strange conclusions. E.g. after a rapid
injection of digoxin the plasma concentration falls while the
inotropic effect increases. The simple conclusion would be that
digoxin effects are negatively correlated with concentration.
Fortunately the elegant work of Weiss & Kang (2004) has elucidated a
plausible mechanism for this observation.
The clinical effects of aminoglycosides (bacterial clearance from
blood, recovery from infection, nephro and ototoxicity) all take days
to achieve yet aminoglycoside concentrations go up very rapidly (0.5
h infusion) and fall quickly (typical half-life of a few hours). So
why would one expect an immediate relationship between (measured)
concentrations and effects?
The in vitro description of the post-antibiotic effect has been known
for a long time but the mechanistic basis for it and a robust model
to predict it's analogue in vivo effects still seems to elude us.
Dosing schedules for aminoglycosides have been developed by trial and
error. Daily dosing compared with 8 hourly dosing seems at least to
be no less effective and possibly less toxic.
"For the pooled efficacy outcomes, the risk ratio for bacteriologic
cure is 1.02 (95% CI, 0.99 to 1.05), and the risk ratio for mortality
is 0.91 (CI, 0.63 to 1.31). For the pooled toxicity outcomes, the
risk ratio for nephrotoxicity is 0.87 (CI, 0.60 to 1.26), and the
risk ratio for ototoxicity is 0.67 (CI, 0.35 to 1.28). " (Hatala et
al. 1996)
[I have to disagree with the interpretation of Luis Pereira who said
recently on this thread:
'It has been established for a long time now that generally the
benefits of gentamicin qd dosing are very significant (e.g. "...once
daily therapy: 8/8 meta-analyses support the use" in Ann Intern Med
1996;124:717).']
Note that conclusions about dosing schedules are based on clinical
trials comparing dosing schedules and not directly comparing
concentration profiles. A target concentration controlled trial (see
Kraiczi et al. 2003) is long overdue in this area to test the long
debated hypotheses about the importance of peak, trough, average
steady state, AUC, etc as targets for dose individualization.
Nick
*=Very few drugs have their action in blood/plasma. Possible
exceptions are subtances like heparin. It must always take time for a
drug to bind to its receptor (digoxin [e.g. Weiss & Kang 2004]). For
drugs which do not act on blood components directly we must consider
the time for movement by perfusion and diffusion to the the target
organ and cell. After that the cell must respond and the eventual
clinical benefit/toxicity revealed which can take hours (levodopa
[e.g. Chan et al. 2004]), days (warfarin [e.g. Sheiner 1969]) or
months/years (bisphosphonates [e.g. Reid et al. 2002).
Chan PLS, Nutt JG, Holford NHG. Modeling the short and long duration
responses to exogenous levodopa and to endogenous levodopa production
in Parkinson's disease. Journal of Pharmacokinetics &
Pharmacodynamics 2004;31(3):243-268
Hatala R, Dinh T, Cook DJ. Once-Daily Aminoglycoside Dosing in
Immunocompetent Adults: A Meta-Analysis. Ann Intern Med 1996;124(8):
717-725.
Kraiczi H, Jang T, Ludden T, Peck CC. Randomized concentration-
controlled trials: motivations, use, and limitations. Clin Pharmacol
Ther 2003;74(3):203-14
Reid IR, Brown JP, Burckhardt P, Horowitz Z, Richardson P, Trechsel
U, et al. Intravenous zoledronic acid in postmenopausal women with
low bone mineral density. New England Journal of Medicine 2002;346(9):
653-661
Sheiner LB. Computer-aided long-term anticoagulation therapy. Comput
Biomed Res 1969;2(6):507-18
Weiss M, Kang W. Inotropic effect of digoxin in humans: mechanistic
pharmacokinetic/pharmacodynamic model based on slow receptor binding.
Pharm Res 2004;21(2):231-6
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.aaa.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
Dear Lawrence and Jeffrey
Thanks for your reply. Unidentified handbooks, I remember, but
precisely because I'm against all dogmas in science, this issue is
certainly worth a Sunday afternoon. So pulling out my files on this,
many references are published recommending AMP (antimicrobial
prophylaxis) as: "the shortest effective course of prophylactic
antibiotics should be used", Page CP et al. Arch.Surg.
1993,128:79-88; "1 day postop ... as good as 3", Williams DN et al.,
Clin. Orthopaed. 1984, 190:83-8; ; "1 day postop ... as good as 7",
Nelson CL et al. Clin. Orthopaed. 1983, 176:258-63; "1 dose cefazolin
as good as 6", Olak J et al., Ann.Thorac.Surg. 1991,51:956-8; "the
administration of additional doses during the 24 hours after the
start of the operation appears unnecessary", Nichols RL,
Infect.Dis.Clin.Pract. 1993,2:149-57, and citing IDSA /SHEA /SIS /
PIDS /CDC/ APIC "it is likely that no further benefit is conferred by
the administration of additional doses after the patient has left the
operating room ... postoperative administration is not recommended",
Dillinger EP et al., Clin.Infect.Dis. 1994,18:422-7. (sorry for the
acronyms to save space, please google them).
However, as per ASHP Therapeutic Guidelines (ISBN: 1-58528-055-0),
2003 (p.426-73) (w/ 559 published references) it can also be
recommended per procedure (please notice that I'm skipping all the
cases pointing to <24h dosing, already reflected above):
Cardiothoracic- cefazolin 1g iv q8h for up to 72h, cefuroxime 1.5g iv
q12h for up to 72h; Transplantation/Heart - cefazolin 1g iv q8h for
48-72h, cefuroxime 1.5g iv q12h for 48-72h; Transpl./Lung and Heart-
lung - cefazolin 1g iv q8h for 48-72h, cefuroxime 1.5g iv q12h for
48-72h, cefamandole 1g iv q6h for 48-72h; Transpl./Liver - cefotaxime
1g iv plus ampicillin 1g iv q6h during procedure and for 48h beyond,
Cardiothoracic - cefazolin 20-30mg/kg iv q8h for up to 72h,
cefuroxime 50mg/kg iv q8h for up to 72h: Patients undergoing lung
transplantation for cystic fibrosis should receive 7-14 days of
prophylaxis with antimicrobial pretransplant culture and
susceptibility results. Strength of evidence that supports the use or
nonuse of prophylaxis is classified as A(levels I-III), B(levels IV-
VI), or C(level VII) from Level I-evidence from large, well-conducted
randomized clin.trials, to Level Vi-expert opinion (please refer to
the actual document).
Running the risk of infuriating db for the length of the post (break
in two?) I'll quote further:
"The shortest effective duration of antimicrobial administration for
preventing postoperative infection is not known ... for most
procedures the duration of antimicrobial prophylaxis should be 24
hours or less, with the exception of cardiothoracic procedures
(48-72h) and ophthalmic procedures (duration not clearly
established)", DiPiro JT, et al. Am.J.Surg. 1986,152:522-9
"three-day course of ampicillin was significantly more effective",
Elliot JP et al., Am.J.Obstet.Gynecol. 1982,143:740-4
"No well-controlled studies have evaluated the efficacy of
antimicrobial prophylaxis in pediatric patients undergoing
cardiovascular procedures ... predominant practice is to use
cefazolin for two days", Lee KR et al. Pediatr.Infect.Dis.J. 1995,
14:267-9
"the optimal duration of antimicrobial prophylaxis for radical
hysterectomy has not been established ... ranged from one dose to
four days", Marsden DE et al. Am.J.Obstet.Gynecol. 1985,152:817-21
"the available data do not specifically address duration and timing
.... ranged from one to five days", Christy NE et al. Ann.Ophthalmol.
1979,11:1261-5
"patients randomly assigned to receive a short course of
antimicrobial prophylaxis ... were more likely to develop surgical
wound infection than patients randomly assigned to receive a longer
course ... cefamandole 750mg every 6h for 48h", Richet S. et al.
Am.J.Med. 1991,91:171S-2S.
"studies involving total hip replacement have used antimicrobials
from 12 hours to 14 days postoperatively", Hill C et al. Lancet
1981,1:795-7
And the list goes on and on. The "Guideline for Prevention of
Surgical Site Infection, 1999" by Mangram AJ et al., Infect.Control
Hosp.Epidimiol. 20:4,247, contains another 497 reference list, and
reproducing the picture above, recommends: "Administer a prophylactic
antimicrobial agent only when indicated, and select it based on its
efficacy against the most common pathogens causing SSI (Surgical Site
Infection).
I have no problem apologizing the entire world and surroundings for
having used the capitalized word "EVER". But that way we got to talk
about ABX usage and that's far more important to me.
Nick's message just popped on my screen and I rejoice for agreeing
with him (although I think qd gentamicin dosing as been documented by
clinical trials). Timing is everything in life (not a dogma, just a
fact) and the term "bad bugs" bears a lot of evolutionary theory
(some food for thought). Bacteria are highly adaptable living
organisms. They've been around certainly longer than we. And their
metabolism is incredibly faster and simpler than ours.
The ongoing SIP project has been justified by findings such as:
"Patients who develop surgical site infections are sixty percent more
likely to spend time in an ICU, five times more likely to be
readmitted to the hospital and have twice the incidence of
mortality.", Kirklan KB, et al. Infect Control Hosp Epidemiol
1999,20:725-30
"27% to 54% of all cases did not receive ABX prophylaxis in a timely
fashion", Silver et al. Am.J.Surg. 1996,171:548-52
"Each postoperative infection is estimated to increase a hospital
stay by an average of 7 days and add over $3,000 in hospital
expenses." Martone WJ, et al. In: Bennett JV, Brachman PS, eds.
Hospital Infections. Little Brown Co. 1992, 577-96
Unfortunately plots cannot be posted, but about the emergence of MRSA
(methicillin resistant Staphylococcus aureus) Fukatsu et al. present
very interesting data (Arch.Surg. 1997,132:1320-5) with a significant
decrease after 1988-90, but upward trends from then on for E.coli,
enteroccocus and fungi resistance.
I can't go much longer but currently available PKPD population models
provide very good insights about all this!!
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.-at-.bos.mcphs.edu
Back to the Top
The following message was posted to: PharmPK
Dear Luis,
The thread on Aminoglycoside target concentrations become quite long and
diverse now, but I have still a few replies to your comments:
> But maybe I wasn't clear in my earlier posting. What I find crucial
> to consider is not the AUC, but rather a measure of the pathogen
> susceptibility.
Of course one should consider a measure of the pathogen
susceptibility. But
one should also consider a measure of drug exposure (e.g. AUC or peak
concentration) to evaluate (expected) efficacy of antibiotic therapy.
So I
don't understand that you can state that it is 'crucial not to consider
AUC'.
> The fact
> of the matter is that if drug concentrations are above the MIC for a
> time that compensates the duplication rate of bacteria, the
> likelihood for wiping the infection and preventing resistance is the
> greatest.
OK, but the higher the concentration, the more rapid the killing of
bacteria, until the effect levels off at high concentrations. This
implies
that AUC or peak concentration is important for efficacy of antibiotic
therapy. This has been demonstrated clearly in many references.
> But to be consistent
> drug concentrations can only be dealt with in the post-distribution
> phase, which by convention has been established to be sufficiently
> reached 30min. after the end of the infusion.
Thank you for your references on aminoglycoside distribution
kinetics. But
your above-quoted statement does not apply to TDM using Bayesian MAP
estimating and a two-compartment model.
> Prashant> ... a trough aminoglycoside level (taken during the last
> hour of the dosing interval) should be near zero i.e. <<1 mg/L.
Please note that one should discriminate between the 'true trough
level',
i.e. the pre-dose concentration (which indeed should be low to be
sure that
the drug does not accumulate) and the 'measured trough level', or better
'measured second level to estimate clearance'. The latter should not
be near
zero, since it becomes almost non-informative due to the influence of
measurement error. Applying TDM with (Bayesian) PK analysis there is
no need
to measure a trough level. One should measure at the (estimated) most
informative time points. The 'true trough level' can then be
calculated from
the model, and likely more accurately than by measuring.
Best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
tel. 31-50 363 3292
fax 31-50 363 3247
Email: j.h.proost.aaa.rug.nl
Back to the Top
The following message was posted to: PharmPK
Dear Hans
Obviously there was a misunderstanding here, but that's the price we
have to pay for talking by email and not around a couple of pints :-)
> So I don't understand that you can state that it is 'crucial not
to consider AUC'.
I didn't say this. Remember, I even stated that AUC is a good PK
estimator for drugs with exposure-related efficacy. And I gave
examples. What I meant to say was, along with what Nick Holford
stated earlier (as good as I understood him), for aminoglycosides,
that elicit a beneficial effect strongly related with concentration
levels with an adequate time course, AUC is not the primary PK
parameter of interest. As said before, you may have two numerically
equal AUC's with totally different Cvs.T profiles. Moving to the PD
hemisphere, I'll state again that I find crucial to pay a great deal
of attention to MIC or any other susceptibility metric.
> the higher the concentration, the more rapid the killing of bacteria.
Having said that monitoring AUC you won't be able to distinguish
between 'higher concentrations', I'll add to this comment essentially
what Holford and Wolf also said. If peripheral drug levels are
intended (most cases), time to reach distribution steady-state and
actual biophase levels are key. We cannot expect immediate effect at
tmax. Working with high MIC pathogens that may be located in
vegetations hard to reach, if we crank up the dose we may achieve
unbearable toxicity for the patient. Then we may choose a continuous
infusion, such as for vanco. So higher is not always better.
> statement does not apply to TDM using Bayesian MAP estimating and
a two-compartment model.
Again certainly because of typing constraints I don't follow your
point. What I tried to said was that with a t1/2,alphamin, 3 half-
lives=60 mins. and that's why we take a sample 30 mins. after a 0.5h
infusion to be described by a single exponential decay. The choice of
the structural model in a Bayesian approach is exactly the same.
Either consider earlier samples with a full '2compt.model', or just
post-distribution data with a '1compt.model' approximation.
> a trough aminoglycoside level (taken during the last > hour of
the dosing interval) should be near zero i.e. <<1 mg/L.
I understand, and kind of agree, with the trough issue. But like you
said with respect to the 'peak' it's again a poor choice of words (as
only pharmacokineticists are capable of).'True trough', 'measured
trough','measured second level' trough; why don't we just say conc.X
at time X (certainly the most informative possible).
But I'll just take this opportunity again to say that antibiotherapy
is perhaps one of the most significant examples where PK gains
tremendously with a PKPD approach (and thus a research interest of
mine). On the sensitive issue of single dosed antibiotics, how do we
distinguish proper patient preparation, shaving, skin disinfection,
surgical asepsis, from successful ABX prophylaxis? On the other hand,
the easiest way to isolate a resistant strain in the lab, is to seed
a series of Petri dishes with a small concentration of antibiotic in
the medium. That's why I say that single doses have the potential for
trouble.
Warmest regards,
Luis
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.aaa.bos.mcphs.edu
Back to the Top
The following message was posted to: PharmPK
Dear Luis,
Thank you for your reply.
> > So I don't understand that you can state that it is 'crucial not
> to consider AUC'.
>
> I didn't say this.
Indeed, I rephrased your comment, and sorry for that. You said:
> What I find crucial
> to consider is not the AUC, but rather a measure of the pathogen
> susceptibility.
>From this I understand that a measure of pathogen susceptibility,
e.g. MIC,
is more crucial than AUC. I don't agree. Effective antibiotic therapy is
equally dependent on antibiotic concentration at the site of infection
(related to AUC) and the susceptibility of the pathogen. It makes no
sense
to give more weight to one of these factors.
> What I meant to say was, along with what Nick Holford
> stated earlier (as good as I understood him), for aminoglycosides,
> that elicit a beneficial effect strongly related with concentration
> levels with an adequate time course, AUC is not the primary PK
> parameter of interest. As said before, you may have two numerically
> equal AUC's with totally different Cvs.T profiles. Moving to the PD
> hemisphere, I'll state again that I find crucial to pay a great deal
> of attention to MIC or any other susceptibility metric.
I agree, but in my opinion aminoglycosides are not a good example for
the
latter statement. In fact, using extended dosing intervals, the plasma
concentration of aminoglycosides are a prolonged time below MIC.
> > the higher the concentration, the more rapid the killing of
bacteria.
>
> Having said that monitoring AUC you won't be able to distinguish
> between 'higher concentrations',
A higher AUC means a higher average concentration. Of course the time
course
plays also a role, and equal AUC will not result in the same efficacy.
> Working with high MIC pathogens that may be located in
> vegetations hard to reach, if we crank up the dose we may achieve
> unbearable toxicity for the patient. Then we may choose a continuous
> infusion, such as for vanco. So higher is not always better.
I did not say that a higher peak is always better. My statement was
perhaps
not complete, and should be rephrased to:
" the higher the concentration at the site of infection, the more
rapid the
killing of bacteria."
> > statement does not apply to TDM using Bayesian MAP estimating and
> a two-compartment model.
>
> Again certainly because of typing constraints I don't follow your
> point. What I tried to said was that with a t1/2,alphamin, 3 half-
> lives=60 mins. and that's why we take a sample 30 mins. after a 0.5h
> infusion to be described by a single exponential decay. The choice of
> the structural model in a Bayesian approach is exactly the same.
> Either consider earlier samples with a full '2compt.model', or just
> post-distribution data with a '1compt.model' approximation.
I fully agree with the latter. But if one has good population data for a
2-comp model, I would prefer this over a 1-comp model. And the optimal
sampling point for estimating V is shortly after stopping the infusion
(allowing a few minutes for adequate mixing over the vascular space).
This
also allows to get the 'true' peak concentration. Please note that
the high
peak concentrations during the distribution phase are relevant for their
efficacy, at least in case of aminoglycosides. These high levels are
ignored
by sampling at 30 or 60 min after stopping the infusion.
> > a trough aminoglycoside level (taken during the last > hour of
> the dosing interval) should be near zero i.e. <<1 mg/L.
>
> I understand, and kind of agree, with the trough issue. But like you
> said with respect to the 'peak' it's again a poor choice of words (as
> only pharmacokineticists are capable of).'True trough', 'measured
> trough','measured second level' trough; why don't we just say conc.X
> at time X (certainly the most informative possible).
I agree with respect to trough levels at other time points than just
before
the next dose. A level at 8 hours in a 24 hour interval is not a trough
level. But levels just before the next dose are correctly called 'trough
levels'. If they are measured these are correctly called 'measured
trough
levels'. I don't see any problem here.
Best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
tel. 31-50 363 3292
fax 31-50 363 3247
Email: j.h.proost.-at-.rug.nl
Back to the Top
The following message was posted to: PharmPK
All this from Nick obscures the point. It is common knowledge that
there is
often a discrepancy between the time-course of the rise and fall of
plasma
concentrations after a dose and the time-course of the rise and fall
of the
resulting effect. So for most drugs one doesn't expect a direct relation
between plasma concentration and effect--hence Lew Sheiner's direct
PK-PD
model and later indirect models. But to mention digoxin in the same
breath as
aminoglycosides is illogical and unhelpful--they have radically
different
pharmacokinetic and pharmacodynamic characteristics.
The primary problems that I as a clinician perceive about the
aminoglycosides
are:
1. The slow tissue accumulation during repeated therapy over several
weeks
while average plasma concentrations hardly change. Even if there were
good
relations between plasma concentration and beneficial/adverse effects
after
the first dose the relations would change with time.
2. The fact that the aminoglycosides have their beneficial effects
via a hit-
and-run action (the so-called postantibiotic effect), rather than an
on-off
action, for which knowing the receptor on and off rates helps. [But the
adverse effects are on-off (at least I think that they are--is there any
information about hit-and-run adverse effects?).]
3. In vitro MICs do not always relate to in vivo effects (which is not
infrequently the case for other antimicrobial drugs too), partly
because of
the poor relation between plasma concentration and effect and partly
because
bacteria do not behave the same way in vitro and in vivo.
Secondary problems are that:
4. Other antimicrobials that are used concurrently may shift the
aminoglycoside concentration-effect curve to the left (and perhaps
change its
characteristics--slope, maximal efficacy??).
5. Aminoglycoside kinetics may change as the patient gets better,
because of
changing renal function (e.g. in the patient with septicaemic shock).
So, what practical advice would you give me about monitoring therapy
in a
patient whose illness I have recently had to manage, a man with
Abiotrophia
defectiva endocarditis (left-sided), normal renal function, and no other
complications, for whom the microbiologists recommended six full
weeks of
treatment with intravenous benzylpenicillin plus gentamicin. Is once
daily
gentamicin therapy OK? What plasma concentrations should I aim for--
both for
benefit and to minimize the risk of toxicity? How often should I
measure them?
How does duration of therapy affect the target concentrations for
benefit and
toxicity?
As a small aside--has anyone developed a satisfactory PK-PD model for
the hit-
and-run type of effect for any drug, i.e. irreversible effects or
effects that
reverse only slowly even after the drug has completely dissociated?
Regards
Jeff
--
J K Aronson, MA MBChB DPhil FRCP FBPharmacolS.
University Department of Clinical Pharmacology,
Radcliffe Infirmary, Woodstock Road, Oxford OX2 6HE.
Tel: (01865) 224626. Fax: (01865) 791712.
Back to the Top
The following message was posted to: PharmPK
Jeff,
I'm disappointed that your white coat clinical perspective finds
clinical pharmacological science obscure.
You seem to be living in the 19th century with your Ehrlichian 'hit and
run' ideas about antibiotic action. Can you offer any serious
contemporary science (with references) to support your beliefs? Do you
really belief that aminoglycoside action is mediated by something other
than concentration driven drug binding to bacterial target sites? Do you
think that drug molecules literally hit the bacterium and knock it down
then run away? These unsupported assertions may satisfy medical students
but this thread has been trying to take a more serious approach to
understanding the target concentration concept.
The lack of formal clinical trials of even empirical target
concentration hypotheses means that I cannot offer you any scientific
advice on how to manage the patient you describe. You will just have to
rely on the age old trial and error approach that characterises most
clinical practice today.
As far as I am aware Lewis Sheiner did not propose the term 'direct
PK-PD model' and the term 'indirect effect model' was introduced by Bill
Jusko. In recent times even Bill uses the more appropriate 'turnover
model' to describe this class of delayed drug action.
Nick
Back to the Top
The following message was posted to: PharmPK
In message
PharmPK.-at-.boomer.org writes:
> PharmPK - Discussions about Pharmacokinetics
> Pharmacodynamics and related topics
>
> The following message was posted to: PharmPK
>
> Jeff,
>
> I'm disappointed that your white coat clinical perspective finds
> clinical pharmacological science obscure.
>
> You seem to be living in the 19th century with your Ehrlichian
'hit and
> run' ideas about antibiotic action. Can you offer any serious
> contemporary science (with references) to support your beliefs? Do
you
> really belief that aminoglycoside action is mediated by something
other
> than concentration driven drug binding to bacterial target sites?
Do you
> think that drug molecules literally hit the bacterium and knock it
down
> then run away? These unsupported assertions may satisfy medical
students
> but this thread has been trying to take a more serious approach to
> understanding the target concentration concept.
***No--of course I don't believe that. Indeed, I believe that pretty
much
everything at the site of action is concentration related (see, for
example,
Aronson & Ferner BMJ 2003; 327: 1222-5). However, we are talking
about plasma
concentration versus effect relations, and from that point of view
the action
is hit-and-run. [By the way, don't knock the 19th century--the law of
mass
action was a 19th century concept.]
> The lack of formal clinical trials of even empirical target
> concentration hypotheses means that I cannot offer you any scientific
> advice on how to manage the patient you describe. You will just
have to
> rely on the age old trial and error approach that characterises most
> clinical practice today.
***That's a pity, because I had hoped that some practical guidance might
emerge from all the theory that has been discussed. In the absence of
trial
evidence, one still ought to be able to formulate guidance based on
theoretical considerations. Can someone out there help answer my
questions?
[Incidentally, I invite you to join a (UK) ward round some day and
see the
extent to which we try to apply scientific principles to the practice of
medicine at the bedside.]
> As far as I am aware Lewis Sheiner did not propose the term 'direct
> PK-PD model' and the term 'indirect effect model' was introduced
by Bill
> Jusko. In recent times even Bill uses the more appropriate 'turnover
> model' to describe this class of delayed drug action.
***I didn't say that Lew proposed the term, but that is what his
model has
come to be called. It wasn't necessary to call it direct until there
was an
indirect model to compare it with.
Regards
Jeff
Back to the Top
The following message was posted to: PharmPK
Dear Jeff
I'll take your challenge. Although my white coat is a bit old now, I
still have one and I understand your reasoning. However, not agreeing
at all with the form of Nick's comment, I must agree with its content
from a scientific point of view, except for the hand-waving and trial
and error recommendation at the end.
> man with Abiotrophia defectiva endocarditis (left-sided), normal
renal function, and no other
>complications, for whom the microbiologists recommended six full
weeks of treatment with intravenous
>benzylpenicillin plus gentamicin
Great example. As any Bayesian would say, one should build up on
prior experience to anticipate some posterior observation, in spite
of not knowing the real underlying relationship. When more
information comes, a new prior is constructed to maybe infer about a
different posterior. No trial-and-error, please.
So, the risk of A.defectiva bacteriaemias are greatly dependent of
phenotypic characterization, many times very unusual (commensal flora
in mouth, high affinity to the endocardium). Identification can be
made by the ATB expression system, and susceptibility can be
determined by E-test for penicillin and Kirby-Bauer method for other
antibiotics (and this is not always done, or doable). Obviously your
microbiologists did a great job giving you the strain and the ABX to
kill it (growing reports of macrolide resistance, used for
prophylaxis)(my point in earlier postings). But the biggest task, as
you say, is to treat the patient, who by the way can die if just
empirically treated. Having done the above I imagine that a minimum
bactericidal concentration of the ABX could also be estimated, let's
say MBC for this strain is 0.12mg/L for benzylpen. and 0.5mg/mL for
genta (Johnson AP et al. BMJ, 2001,322:7283,395-6). Elimination half-
lives for benzylpenicillin and gentamicin may be estimated as 1h and
4h, respectively (Jack DB, Handbook of Clinical Pkarmacokinetic Data,
Macmillan Publishers Ltd.1994). Patient has normal renal function
now, but he most probably won't half-way through or at the end of
treatment. So creatininemia is due and since glomerular filtration is
major excretion route, I would start monitoring CLcr daily to adjust
genta dosing (Jelliffe's method). Finally, since this patient most
probably has a peripheral catheter available I would sample optimally
and I would model the data according to a biexponential disposition.
However, I'll do it here quick and dirty just to exemplify (and I
would do it also to immediately start treatment). Bacterial doubling
time could be 20min, although A.defective grows slowly (keep the
microbiology people close by), so kill rate must at least match this,
0.4-0.8 logCFU/mL/h would do. Betalactams efficacy appears to be
related with how long concentrations are sustained above the MIC, so
I would choose frequent dosing and I would monitor t>MIC of benzpen
(not AUC) shortening the dosing interval or increasing the dose to
have at least 3t1/2 above the MIC per dosing interval. Let's say
20millionU qid. Aminoglycosides efficacy appears to be related with
the biggest possible maximum conc above MIC (I also don't like the
term hit-and-run) so I would choose a 24h dosing interval with a dose
that elicits peak/MIC of about 10. Let's say 3mg/Kg qd. Note that
with Vd=0.2L/kg this would imply a Cmax of about 15mg/L and a MIC of
1.5mg/L to maintain the rule. So having a documented MIC value I
would increase the dose up to the maximum tolerated by the patient to
maximize the ratio, and with t1/2=4h I would assure that about 2/3 of
the dosing interval are above the MIC (resistance,PAE,toxicity). As I
learned from an old professor, first you treat the life threatening
infection, and later you treat the patient's kidneys and inner ears
because he will still be alive. So as CLcr starts to go down genta
dosing would have to be adjusted. I would maintain the treatment for
4-6 weeks evaluating periodically, again based on actual kill rate
and bacterial doubling time.
Of course this can all be done in a much more elegant way making use
of a PKPD model, but I think I stressed this before. A good
comforting paper for helping with the clinical decision along with
the above maybe Baddour LM et al. Circulation, June 2005,394-434.
I hope this helps.
Regards
Luis
--
Luis M. Pereira, Ph.D.
Assistant Professor, Biopharmaceutics and Pharmacokinetics
Massachusetts College of Pharmacy and Health Sciences
179 Longwood Ave, Boston, MA 02115
Phone: (617) 732-2905
Fax: (617) 732-2228
Luis.Pereira.at.bos.mcphs.edu
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)