Back to the Top
Ah, Dr. Jelliffe,
You are after my soul. I have been saying for a very long time that
clinical Pharmacokinetics and dynamics do not correlate with standard 1
compartment models. When deal with actual, live patients we see all types
of variation in distribution and clearance. Very few drugs act in straight
lines when mixed with inotropes, edematous fluid, reduced albumin and
exotoxins. The whole mess gets way out of control.
We have a small patient who is suffering from post scalding burn to 30% of
his body (2nd and 3rd degree). He developed seizure secondary to
electrolyte imbalance, secondary to initial burn and subsequent debridement.
His seizures compromised his airway and he was sedated and intubated. Four
days later he developed ARDS and a grew Pseudomonas and Staph Epi in his
blood and tracheal culture.
Now we have a real challenge. Electrolyte fluctuation continues (although
much more stable in the past 36 hours), ARDS and Sepsis.
What think you of standard kinetics in this patient.
I have been dosing his Vancomycin and Tobramycin very carefully. He is on
Inotropes (Dopamine .-at-. 5mcg/kg/min and Dobutamine at 10mcg/kg/min)for almost
60 hours. During this time his Albumin has dropped to 1.8mg/dl. We
corrected the Albumin with 0.5gm/kg 25% Albumin in his TPN (yes TPN, we
tried trophic feeds but he developed a terrible illus and we had to deal
with that). We corrected his electrolytes using the TPN plus Potassium and
Calcium and Mg riders. We dosed his Tobramycin at 3.8mg/kg q 12 hours (Pk
6.8, Tr 1.4) and his Vancomycin 12mg/kg q 6 hours (Tr 7.2).
Now I am very unsure of how true the drug levels are. With the reduced
Albumin and electrolyte sloughing and increased edema, it is hard to tell
how much drug was in the central vascular compartment and how much was in
the peripheral compartment.
I am going to get repeat levels of Tobramycin in the morning. The last dose
will be at 21:00 today. I postulate that some of the Tobramycin that seeped
into the peripheral compartment will leak back into the central compartment.
We'll see.
BTW: this child has 2 chest tubes and is on a high frequency oscillator.
Both tubes are draining. These two confounding variables will definitely
affect distribution.
We have had a horrible 10 days. This child and 4 others all on ventilators,
multiple organ involvement and many, many drugs etc. We lost one child this
morning. The real world is a terrible place to live.
Next time, I'm going to Barber School.
Take care all and continue to work at getting us good data to use in our
little patients.
robert
Robert Aucoin, RPh
Senior Clinical Pharmacist
The Children's Center at
Our Lady of the Lake RMC
Baton Rouge, La 70809
e-mail: raucoin.aaa.ololrmc.com
Back to the Top
Robert,
You may wish to consider using a constant rate vancomycin infusion
for your difficult patients. We find it easier when managing ITU
patients with variable kinetics. Here are some refs.
Wysocki et al, Continuous versus intermittent infusion of vancomycin
in severe staphylococcal infections: prospective multicentre
randomised study. Antimicrobial Agents Chemother 45: 2460-2467, 2001.
James et al, Comparison of conventional dosing versus continuous
infusion vancomycin therapy for patients with suspected or documented
gram-positive infections. Antimicrob Agents Chemother 40: 696 700,
1996.
Matthews, Vancomycin continuous infusions: a cohort of 23 intensive
care unit patients, Aus J Hos Pharm 31:108-110, 2001
Pawlostsky et al, Continuous rate infusion of vancomycin in premature
neonates: a new dosage schedule. Br J Clin Pharmacol 46: 163-167,
1998.
Alison
Back to the Top
Dear Dr. Aucoin:
Thanks so very much for your thoughtful answer, and for your
statement of your many problems with your highly unstable patients. Such
patients are extremely difficult to manage with conventional software. We
have been working with new software, and the bottom line here for you is
that it is designed, among other things, to track the behavior of drugs in
unstable patients, when their parameter values are changing in unsuspected
ways DURING the period of data analysis. We use an interacting multiple
model (IMM) approach to this problem, which comes from techniques of
tracking hostile targets when you want to shoot them down. With unstable
patients we want to do exactly the same thing - to track the behavior of
the drugs in them so we can do the best job of hitting our desired target
goals (serum concentrations, targets in peripheral compartments, etc), and
to do it all with maximum predicted precision (minimum expected weighted
squared error).
This approach is based first on nonparametric population PK/PD
models. This approach has been around since Alain Mallet first introduced
it about 1983. Go to our web site www.lapk.org. click on New advances in
population modeling, and see a comparison between parametric population
methods using the current FOCE approach and our new NPAG, for example. The
nonparametric methods are consistent. FOCE is not. The NP methods are also
more efficient than FOCE. The FOCE approximation greatly impairs both
statistical consistency (the more you sample from a population, the more
the answer approaches the true answer) and also efficiency. This work was
presented by Bob Leary at the PAGE meeting in Paris in June.
Based on the multiple models (the multiple support points) which
you get in an NP pop model, you now have a tool to estimate the probable
precision with which any dosage regimen will fail to hit a desired target,
and you can then find the regimen which specifically minimized that
expected error. That is multiple model (MM) dosage design. You cannot do
that with parametric models, which give you only one value for each
parameter (mean, median, etc). Then, when you get serum levels, you can use
the IMM sequential Bayesian posterior models to track the changes in the
drug in the patient. Then you can use your most recent values to develop
your next regimen to best hit your desired targets. Call me at
323-442-1300, and let's talk. Also, if anyone else is interested out there,
please do the same, and we can arrange for you to use the new MM software.
Our web site also has info on this. Click on teaching topics.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.at.hsc.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Robert and Alison:
I would support most strongly the idea of continuous vanco
infusions. The study by Wysocki et al looks very convincing, and the
patient's exposure to the drug is a lot less. You can set a target goal of
12 or 15, or even 10, for example. You can use our USCPACK software, and
specify that you want 3 infusion steps of 2 hours each, to bring the system
to equilibrium, followed by 3 infusion steps of 6 hours each, to finish out
the first day, individualized to the patient's weight and renal
function. Usually the first 3 steps bring everything to equilibrium, and
the last 3 are all about the same, and you can simply keep on at that rate
after that. Get a serum level the next morning, for example, and then all
you have to do is make a linear adjustment with the dose to go from the
level your present infusion rate is giving you to get your target level. I
have seen 2 patients with aplastic anemia probably from vanco. They were
conventionally dosed q 12 h or q 24 h, but their episodes might have been
less severe if they had not had the very high peaks that they probably had.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.aaa.hsc.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dr. Jelliffe,
Does it matter if a patient has meningitis, osteomyelitis or
endocarditis? Do you think the continuous infusion will do the same job
or the intermittent doses will bring the CSF or cardiac tissue/
vegetative tissue to a better level through passive diffusion and
maintained there? I am curious but don't have an answer yet.
Regards,
Thao Nguyen
Back to the Top
Dear Thao:
This is a good question. The main reason for what I say is the
general recognition that vanco is a "time dependent" drug.
While the concentrations that affect the degree of kill with
aminoglycosides (AG's) are very concentration dependent, those with vanco
are not. That main reason for this, it seems to me, is that with AG's we
choose high peaks to kill, and then low troughs (often well below the MIC)
to minimize toxicity. Because of this, we deliberately choose to play out
the therapeutic scenario on the very steep part of the concentration -
effect relationship with AG's.
In contrast, with vanco it is easy to be over 5 times the MIC,
even at the trough, without very much significant toxicity. Because of
this, once you obtain concentrations about 5 times the MIC, more doesn't
matter much, when you are talking about serum concentrations. This is the
general reason for advocating continuous infusion vanco.
Your point about concentrations in other compartments is very well
taken, and to my limited knowledge there is not much data. If anyone out
there has such data, it would greatly enhance this discussion to know about
penetration of vanco into these other compartments. Also, such data would
be very valuable in helping to make improved multicompartment population
models to permit us all to do the job better.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.-a-.hsc.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Dr. Jelliffe,
Thanks for your wonderful input.
In his presentation at ICAAC 2001, Dr. Ostergaard indicated that
the vancomycin concentration in CSF is approximately about 15% of serum
concentration at steady state in an experimental meningitis animal model
(http://www.isap.org/2001/post-ICAAC-Chicago/dias/Ostergaard/index.htm).
Compared to the serum peak, the CSF peak is somewhere there, though it
is not well-defined, delaying for 2-3 hours and then waning down. If
the continuous infusion level is kept approximately at 15 mcg/dl, the
extrapolated CSF level is probably about 2-3 mcg/dl (15% of 15) which is
possibly less than the goal level of 5X MIC (if the bug's MIC is 1 as
often seen with Staph aureus or S. pneumonia, 2 common bugs in
meningitis). To achieve constant level > 20 mcg/dl is probably not
desired for possible increased nephrotoxicity or ototoxicity. However,
whether it is appropriate to do the above extrapolation is a potential
problem since vancomycin CSF level depends on a number of factors, one
of which is the degree of inflammation.
Best regards,
Thao
Back to the Top
Dear Thao:
Thanks for your input, and for your comment on the relationship
between serum and CSF exchange and inflammation. Wouldn't it be really nice
if we could really make models of the exchange between serum and CSF, for
example, and correlate the rate constants with things like CSF protein or
cells or the WBC, etc. Then we can REALLY see what the relationship is
between inflammation and exchange with CSF, for example. All then can
generalize to evaluating exchanges between serum and other compartments.
What we need to do is to make good population PK models with both serum and
CSF data, for example, and the other CSF data.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
email= jelliffe.at.hsc.usc.edu
Our web site= http://www.lapk.org
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)