Back to the Top
Can anyone please help me with vancomycin monitoring in peritoneal dialysis
patient?
Back to the Top
Hi S Ang
There are no easy answers to your questions and indeed any answers will
depend on the goals of therapy. But I might be able to help out a little
from a theoretical perspective - since I have no knowledge of the patient
etc...
>Can anyone please help me with vancomycin monitoring in peritoneal dialysis
>patient?
>>
>>1. A patient of ESRD on peritoneal dialysis every nite, is given a dose of
>>Vancomycin 1g intravenously every 7 days. When would be the appropriate
>>sampling time to monitor the Vancomycin level?
As we know vancomycin displays time dependent bactericidal activity.
Therefore the longer the concentrations are above the minimum bactericidal
concentration (or perhaps MIC if MBC isn't available) the better. Hence
monitoring at the trough would seem reasonable. Vancomycin is only
minimally eliminated via CAPD. The caviate however is that the commonly
used TDX assay cannot distinguish between the active vancomycin and its
degradation product (which has no/little activity) which accumulates in
patients with renal dysfunction. An HPLC assay would get around this
(although expensive due to set up costs etc...). There are various
publications on this. Routine monitoring, therefore, may be misleading - is
monitoring important to achieve the goals of therapy in this individual?
>>2. How would the sampling time change in a patient of ESRD on
haemodialysis
>>2 times a week?
I don't think vancomycin is significantly eliminated by HD and therefore the
same recommendation as above applies.
>>3. What would be the sampling time for a patient on peritoneal dialysis
>>every 7 days, with 40 cycles of dialysis with 16mg/2 litres of gentamicin
>>in the PD bag, and on gentamicn 40mg EOD? We would expect genta to
>>accumulate in the patient between the intervals of dialysis, how would the
>>drug level change after dialysis? I am calculating the patient's
>>pharmacokinetic parameters manually, should I use the same calculation as
>>in normal patient?
Clearly this is a complex situation. If the IV dose (40 mg) is for cover of
some procedure (eg changing the dialysis catheter) then there doesn't seem
to be any need to monitor (even if the patient remains on gentamicin
administered into the bag for a few days). But since it is being
administered q48h I assume that you are treating a g-ve sepsis secondary to
a g-ve peritonitis. In this case the usual goals of treating sepsis would
seem appropriate (ie high peaks and low troughs). The regimen you describe
would not, I expect, produce this. Calculating the parameter values will be
difficult since you have two input sites (intravenous and from the PD bag).
To pose a question to Roger Jelliffe:
Could you use USC-PACK to allow you to model this patient [assuming the
goals of therapy are adequately defined]?
(From my memory I don't think ABBOTTBASE could do this.)
I hope this may be some help.
Steve
===========================
Stephen Duffull
School of Pharmacy
University of Manchester
M13 9PL, Manchester, UK
Ph +44 161 2752355
Fax +44 161 275 2396
Email: sduffull.-at-.fs1.pa.man.ac.uk
Back to the Top
Dear Steve:
Thanks for your note. No, USC*PACK cannot do this either, as it
also uses
analytic solutions to the differential equations, and is limited to a 3
compartment model with 1 input and 1 output. However, we have other
population modeling software for nonlinear models. It uses the USC*PACK
clinical patient data files. It implements both the iterative Bayesian
(IT2B) and the NPEM approaches to population PK/PD modeling. It currently
has multiple outputs but still only 1 input. Multiple inputs should be
available this fall. Then, all that you mention can be done, at least from
the point of view of modeling. Designing dosage regimens for these models
will come later.
At the risk of being slightly provocative, let us discuss the ideas of
concentration dependent and "time dependent" (or better, I think,
non-concentration dependent) drugs. I believe the types of behavior which
we describe in this way are probably not a property of the drugs, but of
the ways we choose to use them. I have preferred to think of concentration
dependent and non-concentration dependent scenarios of therapy.
As suggested by France Mentre a number of years ago, there appears
to be a
general overall concentration - effect relationship, which can be
characterized by a nonlinear Michaaelis-Menten or Hill model. This has been
incorporated into the Zhi model, for example, of bacterial growth and kill.
If gentamicin, for example, were not so toxic, we could achieve much
higher concentrations, for example, peaks of 100, and troughs of 50, and we
would see "time dependent" behavior, since we would be able to achieve
concentrations well over 5 times the MIC, for example. We would easily be
in a region of the effect relationship where concentration is so high that
it is not very important any more, up on its saturated or flat part of the
relationship. It is specifically because we must choose, because of its
toxicity, to have low troughs that are at or below the MIC, that the
therapeutic scenario takes place on the steep portion of the concentration
- effect relationship, and is, therefore, "concentration - dependent".
On the other hand, if you want to see a "time dependent" drug such as
Vancomycin develop "concentration dependent" behavior, simply stop giving
it, and see the bugs grow out again. We usually are able to be well above
the MIC with Vancomycin concentrations, and because of this, the most
useful thing to do is simply to keep it up for as long as it takes to do
the job. If Vancomycin is stopped prematurely, the concentrations fall and
the bugs grow out again, with "concentration - dependent" behavior in that
scenario.
This behavior, and the above scenarios, can be well seen if one
uses a Zhi
model, as described by Corvaisier et al, in Antimicrob. Agents +
Chemother., 42: 1731-1737, 1998. The Zhi model uses Hill model of the
killing effect of the drug to yield the rate constant for kill, which is
balanced against the rate constant for logarithmic growth without the drug.
It is a good "worst case" model of the growth and kill relationship, as
described in that paper. Vinks and others have made other more complex
models as well, in which the number of bugs reaches a maximum as substrates
get low, and the MIC increases with time as resistance increases. However,
all these models reflect the basic saturable relationship between
concentration and effect, and they show fairly convincingly that it is the
scenario of the concentration profile over time, not the drug itself, that
determines whether the effect relationship is concentration dependent or not.
Making combined PK/PD models of this type, including combinations
of drugs
and their shared effects, is one of the main reasons why we have developed
the present IT2B and NPEM population modeling resource on the Cray T3E at
the San Diego Supercomputer Center, supported by a grant from the NCRR. We
look forward to being useful to people who want to study relationships of
this type. If anyone is interested, please let us know.
As to the times to get Vancomycin levels, you can get one at any
time. It
doesn't have to be a trough. Most software packages can fit a level
obtained at any time. Usually, to my knowledge, one wants a trough level of
about 10 ug/ml, to be about 5 times above the usual MIC of 1, which is
really 2, because Vanco is about half bound to protein. Because of this, a
trough of 10 is usually effective, and all other levels are higher. The
idea, I believe, is to make an individual model of the behavior of Vanco in
that patient, and you usually need, optimally, at least 2 levels, something
high near the peak, and something low near the trough. The other problem is
that Vanco has such clear 2 compartment behavior, with the levels at 2-3
hours after the end of the infusion being about half the true peak at the
end of the infusion. There are many ad hoc ways to try to avoid this
problem, usually involving not getting the true peak, and ignoring
everything until 2-3 hours after the end of the infusion, and then using
linear regression on logs of the levels. I think this is a poor method. I
think you can obviously get only a trough level, since that is what we are
interested in, but in order to be able to achoeve it with optimal
precision, you usually need some other level, or levels, as well, to get
better parameter estimates, perhaps a true peak and another 2-3 hours
later. Sander Vinks has done computations of the D-optimal times for this,
and can contribute a lot here, I think, especially with regard to his
cystic fibrosis patients, but his work has general relevance. Sander, can
you help us further?
Another thought is that since Vancomycin is poorly removed by
hemodialysis, that may be true as well for peritoneal dialysis. If you
assay what comes out of the bottles from the peritoneum, one may be able to
compute the net uptake of vanco from that route. One could treat this
peritoneal lavage as a prolonged IV infusion, and use the currently
available USC*PACK or Abbott software to model the combined IP and IV
process as a series of short and long IV infusions. To evaluate baseline
renal function, one can use the pair of serum creatinines from a
postdialysis sample to the next predialysis sample to get an estimate of
the patient's baseline creatinine clearance. We have been using a feature
in the USC*PACK programs to do this for many years, and it has worked well.
Does anyone else have any other ideas?
Sincerely,
Roger Jelliffe
************************************************
Roger W. Jelliffe, M.D.
USC Lab of Applied Pharmacokinetics
CSC 134-B, 2250 Alcazar St, Los Angeles CA 90033
**Note our new area codes below, since 6/15/98!**
Phone **(NOTE NEW AREA CODE AND PREFIX)** (323)442-1300, Fax (323)442-1302
email=jelliffe.-a-.hsc.usc.edu
************************************************
You might also look at our Web page for announcements of
new software and upcoming workshops and events. It is
http://www.usc.edu/hsc/lab_apk/
************************************************
Back to the Top
[A few replies - db]
Date: Thu, 27 Aug 1998 23:10:39 -0700
From: Brennan
MIME-Version: 1.0
To: PharmPK.-at-.pharm.cpb.uokhsc.edu
CC: Multiple recipients of PharmPK - Sent by
Subject: Re: PharmPK Vanco monitoring in IPD
What you are contemplating is elegant but not practical. You have to get
back
to basics here. First, in a dialysis patient are you concerned with
nephrotoxicity? No you are not, the focus for both aminoglycosides and
vancomycin shifts. The concern is sustained high drug levels and the
toxicities
associated with them.
Good, sound analysis of these patients is important but from a practical
point of view, drug doses of gent and vanco should be based on pre-dialysis
levels vs post dialysis levels. Post dialysis dosing becomes like a loading
dose and should be treeated in the same manner.
Still, if you can record good dosing and drawings times, your model becomes
better and a population might be developed from that. It would be interesting.
Bob B.
---
From: Nick Holford
Sender: nhol004.-at-.auckland.ac.nz
Reply-To: n.holford.aaa.auckland.ac.nz
To: PharmPK.at.pharm.cpb.uokhsc.edu
Subject: PharmPK Re: Vanco monitoring in IPD
Date: Fri, 28 Aug 1998 14:13:18 -0700 (Pacific Daylight Time)
Priority: NORMAL
X-Authentication: none
MIME-Version: 1.0
Roger,
On Fri, 28 Aug 1998 11:36:54 -0500 Roger Jelliffewrote:
> At the risk of being slightly provocative, let us discuss the ideas of
> concentration dependent and "time dependent" (or better, I think,
> non-concentration dependent) drugs.
Your post clearly shows that you understand that all drug effects are
concentration dependent so I
want to quibble with you over your preference for the term
"non-concentration dependent". A literal
interpretation of this term would mean a conc of 0 would be effective which
is clearly absurd.
I think the terminology problem is akin to that used by some to describe
capacity limited elimination
e.g. ethanol. Some people refer to "zero-order elimination" but in fact
elimination *IS* dependent on
conc and it is not truly zero-order even though it may appear so when
studied over a narrow range of
concs.
So please lets saccept that drugs that appear to have little dependence on
conc because they are
studied at concs which produce effects approaching the maximum are in fact
truly concentration
dependent. It is the inadequate experimental design that produces the
misleading conclusion that the
effects are independent of concs. It is not a property of the drug.
--
Nick Holford, Dept Neurology, L226
OHSU,3181 SW Sam Jackson Park Rd,Portland,OR 97201,USA
n.holford.-at-.auckland.ac.nz tel:+1 (503) 494-7228 fax:494-7242
http://www.phm.auckland.ac.nz/Staff/NHolford/nholford.htm
---
Date: Sat, 29 Aug 1998 23:04:30 -0700
From: Brennan
MIME-Version: 1.0
To: PharmPK.at.pharm.cpb.uokhsc.edu
CC: Multiple recipients of PharmPK - Sent by
Subject: Re: PharmPK Re: Vanco monitoring in IPD
Regarding vancomycin monitoring
From my point of view, kinetic monitoring of aminoglycosides and vancomycin
are important for two different reasons.
1. Aminoglycosides are monitored to prevent toxicity while providing
appropriate therapy
2. Vancomycin is monitored to insure appropriate therapeutic levels
It seems to be generally agreed that vancomycin is not as toxic as was
thought
when it was first developed (at least not at therapeutic levels) What is of
greater concern is the time above MIC. Modeling of the patient easily provides
solutions to this (especially when vanco levels are available). The problem I
seem to run into the most is that the general emperic response to elevated SCr
values is to extend the dosing interval rather than reduce the dose. For
example,
a patient with a SCr of 1.5 - 2.0 is more likely to receive vancomycin at
1g q24h
than 500 mg q12h empirically. Yet the 500 mg q12h is often more appropriate.
The problem seems to be that vancomycin is all to often considered in
the same
way as gentamicin when in actuality, they should be considered opposites.
Back to the Top
Hi everyone
It is pleasing to see such debate on terminology of the schedule dependent
nature of the bactericidal effects of Vanc and Ags. It should be noted that
I had not intended to be controversial by using terms such as conc-dependent
or time-dependent in my original reply... and I think both Nick and Roger
have given good discussions. Having said that I thought that I might
clarify the choice of those terms, with respect to schedule dependence of
Ags/Vanc. The following interpretation may help:
Ags: Concentration-dependent is used to refer to the conc above the MIC (not
the AUC or time above the MIC). The higher the conc the better the
bacterial kill (and also {to a lesser extent} kill rate). Prolonged concs
above the MIC are of *no* value and merely promote adaptive resistance
(hence a trough of 50 mg/L - even if not toxic - would be worse than a
trough of 0.01 mg/L). Perhaps (I say this because we don't know) the best
Ag dose is to achieve a *very* high peak and then some 60 minutes or so
later to get rid of it entirely and don't dose again for the next 1-2
days???
Vanc: Time-dependent is used to refer to the duration that the conc is above
the MIC. The longer the time the longer the kill period. This is not the
same as non-concentration dependent kill. Obviously when the concentration
falls below the MIC (or MBC or whatever) the bugs will resume growing unless
the body has done its bit.
Clearly these definitions do not help us describe the underlying PD model
but do give us insight into their clinical usage.
Other points. We still have to worry about ototoxicity (which dialysis
patients may have a greater propensity for even without the use of ototoxic
drugs). The post-dialysis dose is not a loading dose since removal by
dialysis is more akin to CL than a change in Vd. Schedule dependency of
these drugs must still be considered even if it is difficult to do
clinically, and dosing of Ags in this setting is certainly that!
Cheers
Steve
===========================
Stephen Duffull
School of Pharmacy
University of Manchester
M13 9PL, UK
Ph +44 161 275 2355
Fax+44 161 275 2396
Email: sduffull.-at-.fs1.pa.man.ac.uk
Back to the Top
In regards to concentration dependence and the aminoglycosides....there
is little dispute as to the fact that higher concentrations result in
more rapid killing. Peak/MIC ratios are associated with improved
outcomes, however aminoglycosides (and quinolones) have often been
thought of as an AUC drug....and in fact, AUC/MIC ratio has been
demonstrated to be a better predictor of outcome than peak/mic most of
the time (although the difference is small). Bill Craig has published
most extensively on this topic.....
Also keep in mind that all of these parameters are inter-related. As
AUC/MIC increases, time above mic and peak/mic likewise increase. (For
example, an auc/mic ratio of 125 results in being above the mic for 80%
of the dosing interval).
Regards,
Patrick Smith
State University of NY .-a-. Buffalo
Clinical Pharmacokinetics Lab
Back to the Top
[Two replies - db]
From: Nick Holford
Sender: nhol004.-at-.auckland.ac.nz
Reply-To: Nick Holford
To: PharmPK.-a-.pharm.cpb.uokhsc.edu
Subject: PharmPK Re: Vanco monitoring in IPD
Date: Thu, 3 Sep 1998 14:08:56 +0100 (British Summer Time)
Priority: NORMAL
X-Authentication: none
MIME-Version: 1.0
One might argue with your choice of the word "parameter" in
this context. Parameters in the strict sense are by
definition not inter-related. Part of the problem for
predicting the effects of antibiotics is the lack of
established models which predict the relationship between
the time course of concs and bacterial killing. If such
models existed then their parameters would be definable and
would be seen not to be inter-related.
AUC and AUC/MIC are empirical targets with nothing but
trial and error to support their use. Trial and error is OK
- it has been the basis of essentially all medical practice
over the millenia and has changed little even in this
century. But a rational approach (in contrast to the
empirical approach) would at least be driven by a model
that "understood" how the time course of concs influenced
outcome.
--
Nick Holford, L226,Dept of Neurology,OHSU
3181 SW Sam Jackson Park Road,Portland,OR 97201-3098
n.holford.-at-.auckland.ac.nz,(503)494-4778,fax 494-7242
http://www.phm.auckland.ac.nz/Staff/NHolford/nholford.htm
---
X-Sender: jelliffe.-at-.hsc.usc.edu
Date: Thu, 03 Sep 1998 15:02:53 -0700
To: PharmPK.at.pharm.cpb.uokhsc.edu
From: Roger Jelliffe
Subject: Re: PharmPK Re: Vanco monitoring in IPD
Mime-Version: 1.0
Dear Patrick:
It is good to see your comments. It is quite true that when the dosage
regimen is given in a certain stated format, that a change in the dose
amount results in a change in everything. For example, peak/mic, AUC,
AIC/mic, and time above mic, for example, are all interrelated, and in fact
cannot be separated except as abstract entities. The most useful thing, I
think, is not to argue about which of these things, or others, has the
"best" empirical correletion with some outcome, but actually to sit down
and try to model the entire PK/PD relationship, as Zhi, and more recently,
Maire, Bouvier D'Yvoire, Vinks, Mouton, and others, are doing.
Sincerely,
Roger Jelliffe
************************************************
Roger W. Jelliffe, M.D.
USC Lab of Applied Pharmacokinetics
CSC 134-B, 2250 Alcazar St, Los Angeles CA 90033
**Note our new area codes below, since 6/15/98!**
Phone **(NOTE NEW AREA CODE AND PREFIX)** (323)442-1300, Fax (323)442-1302
email=jelliffe.at.hsc.usc.edu
************************************************
You might also look at our Web page for announcements of
new software and upcoming workshops and events. It is
http://www.usc.edu/hsc/lab_apk/
************************************************
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)