Back to the Top
I wonder if anyone has pondered the question of drug disposition during
Extra Corporeal Membrane Oxygention (ECMO). I am a clinical pharmacist
working in Europe's largest ECMO centre. We treat patients with severe
cardiorespiratory failure, of all ages, on ECMO. For those of you who have
never come across ECMO, it is a prolonged form of cardiopulmonary (CPB)
bypass, i.e. whereas CPB lasts for 2-3 hours, ECMO may continue for weeks or
months. The technique involves draining venous blood out from the right side
of the heart, oxygenating it outside the body using artificial (silicone
membrane) oxygenators, and returning it back into the venous or arterial
circulation, thereby allowing the diseased native lung/heart to rest and
recover. This is akin to dialysis, but the volume of blood drained, the flow
rate and the size of the circuit is much larger in ECMO.
I have recently collected data on plasma concentrations of midazolam infused
continuously in neonates sedated during ECMO. I would like to model this
data using a population approach, however being a PK novice, I am not sure
as to the exact approach. Previous workers have fitted a 2 compartment
model, but the situation here is a bit more complex:-
1. The circulating volume in a neonate on ECMO is expanded (normal neonate =
250mls, neonate on ECMO=750mls).
2. It has been shown through in vitro studies that midazolam adsorbs to the
circuit tubing and oxygenator. In an in vitro investigation that we
conducted, infusing midazolam into the ECMO circuit at normal therapeutic
doses, there was significant attenuation of expected plasma concentrations,
at least for the first 120 mins, after which a steady state concentration
seemed to have been acheived. Thus, it appears that loss of drug to the
circuit may be time dependent.
Initial attempts at modelling with a 2 compartments has not produced a very
good fit. Do you think a 3 compartment approach is necessary? If so, how
would you incorporate loss in the circuit through the drug adsorption? How
would expect initial estimates of parameters to be changed by ECMO?
I would appreciate any comments or words of wisdom that you may have. I am
also appealing to the PK 'gurus' that have a much better understanding of
this clinical problem than I have.
Snr Clinical Pharmacist / Research Associate
Back to the Top
Dear Hussain Mulla:
Interesting, ECMO! We looked at the behavior of gentamicin in
newborns on and off ECMO, and found a different volume of
distribution on and off ECMO, as you might expect, in
Dodge W, Jelliffe RW, Zwischenberger J, Bellanger R, Hokanson J, and
Snodgrass W: Population Pharmacokinetic Models: Effect of Explicit
Versus Assumed Constant Serum Concentration Assay Error Patterns upon
Parameter Values of Gentamicin in Infants on and off Extracorporeal
Membrane Oxygenation. Therapeutic Drug Monit. 16: 552-559,1994.
One of the main points in that paper was about what assay
error pattern to use. When unity weighting was used, and findings
were different from those when the correct assay error pattern was
used. That was then, 1994. Today, we would recommend first, starting
with the true assay error polynomial determined over the working
range of the assay, and, second, using a parametric program such as
the iterative Bayesian 2 stage one (IT2B) to determine the remaining
environmental error (gamma), which is used as a scaling factor in
connection with the assay error polynomial. Then, having gamma, and
incorporating it into the error polynomial, use a nonparametric
program such as NPEM or the newer NPAG (adaptive grid) to get the
entire most likely parameter distributions. We think this stands the
best chance of getting the most likely parameter distributions given
the raw data and the error structure you have determined. This model
then can be used to develop the most precise achievement of desired
target goals (serum concentrations at desired times, etc.). For
gentamicin, Vd was greater, clearance less, and half time longer on
ECMO than off. Use of the carefully determined assay error pattern
was very helpful in this study.
You might also look at our web site (www.lapk.org) for more
info, especially under teaching topics, section 7, dealing with
parametric and nonparametric population modeling. The likelihood of
the results, given the raw data and the weighting or error scheme
employed, is a very important result, and is often not reported in
many studies of population modeling. I look forward to discussing
this with you more.
Very best regards,
Roger W. Jelliffe, M.D. Professor of Medicine, USC
USC Laboratory of Applied Pharmacokinetics
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.-a-.hsc.usc.edu
Our web site= http://www.usc.edu/hsc/lab_apk
Back to the Top
The following message was posted to: PharmPK
(by way of David Bourne) wrote:
>Initial attempts at modelling with a 2 compartments has not produced a very
>good fit. Do you think a 3 compartment approach is necessary? If so, how
>would you incorporate loss in the circuit through the drug adsorption? How
>would expect initial estimates of parameters to be changed by ECMO?
The best modeling approach would be to develop a complete
Physiologically Based Pharmacokinetic (PBPK) model that would account
also for the realistically measured partitioning (surface adsorption)
of the drug to the tubing (probably you will be able to describe it
as a saturable process, analogous to Michaelis-Menten kinetics).
You may find some additional ideas/references about PBPK modeling,
software, etc. in:
Byczkowski,J.Z.: Pharmacokinetic/Dynamic Modeling. In: The Standard
Handbook of Environmental Science, Health and Technology, Chapter 8.6
(Lehr,J., Ed), McGraw-Hill, New York, 2000.
Janusz Z. Byczkowski, Ph.D.,D.Sc.,D.A.B.T.
212 N. Central Ave.
Fairborn, OH 45324
JZB Consulting web site: http://members.aol.com/JanuszB/consult.htm
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (firstname.lastname@example.org)