Dear all,
I have a very unusual situation and wanted to see about getting the
collective opinion on the group regarding the best way to handle this
modeling problem. I have a drug that is given by 30 minute infusion.
Samples were collected at predose, mid-infusion, end of infusion, and
serial thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to
50%)than the end of infusion concentration. This phenomenon occurred
across multiple studies, on multiple days (although not always in the
same subject twice), and across multiple analytical runs. I have ruled
out switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have been able
to come up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the
same phenomenon and in that paper the authors discarded the midinfusion
data. I have tried using time-dependent volumes using continuous and
change-point functions. I get modest improvements in goodness of fit
compared to completely ignoring the phenomenon which has a residual
variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug
so it seems to me that modeling the iv data to the point of completely
capturing the phenomenon may be a modeling exercise and not of any real
value any longer.
Any opinions on the validity of throwing out the data, just running with
the model that ignores the phenomenon and has high residual variability,
or something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
mid-infusion higher than end of infusion
Title: Paul R
Peter:
The low hanging fruit answer might be that your drug is binding to
neutrophils, and that over the course of the infusion there is enhanced
margination of these PMNs. Marginated to the endothelium, they and
their adsorbed drug would not be measured.
PS. Your book is great.
Paul
Bonate, Peter wrote:
Dear all,
I have a
very unusual situation and wanted to see about getting the collective
opinion on the group regarding the best way to handle this modeling
problem. I have a drug that is given by 30 minute infusion. Samples
were collected at predose, mid-infusion, end of infusion, and serial
thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to
50%)than the end of infusion concentration. This phenomenon occurred
across multiple studies, on multiple days (although not always in the
same subject twice), and across multiple analytical runs. I have ruled
out switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how
best to model it or even explain it. The best I have been able to come
up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the
same phenomenon and in that paper the authors discarded the midinfusion
data. I have tried using time-dependent volumes using continuous and
change-point functions. I get modest improvements in goodness of
fit compared to completely ignoring the phenomenon which has a residual
variability of about 30% using a 3-C model.
As a company
we have decided to pursue an oral formulation of this drug so it seems
to me that modeling the iv data to the point of completely capturing
the phenomenon may be a modeling exercise and not of any real value any
longer.
Any opinions
on the validity of throwing out the data, just running with the model
that ignores the phenomenon and has high residual variability, or
something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD,
FCP
Genzyme Corporation
Senior Director,
Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX
78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell:
210-315-2713
--
Paul R.
Hutson, Pharm.D.
Associate
Professor
UW School
of Pharmacy
777
Highland Avenue
Madison
WI 53705-2222
Tel 608.263.2496
Fax
608.265.5421
Pager
608.265.7000, p7856
Peter, you do not specify what model you use. With any linear model in
PREDPP, or with the usual linear models in $DES, the drug in the depot
is governed by differential equations equivalent to
DADT(1)=R1-K10*A(1) ;where R1 is the infusion rate, K10 is elimination
This produces a monotonically rising A(1) which, if the infusion time is
great enough, reaches its steady state value DADT(1)=0, A(1)=R1/K10.
Sounds like this is what you expect to see.
But what if DADT is not linear? What if K10 increases with time, or
increases with rising A(1)? If for some reason the effective K10
increases with time, then A(1) will peak and then start dropping, even
as the constant rate infusion continues.
Or, F1 may in effect change with time. That is, effective R1 may
decrease with time, even though the infusion is constant.
Again, A(1) decreases from its peak.
Paul's suggestion that the drug is binding to something may well be
modelled as a change in F1 or effective K10.
If your drug goes into a depot and then to central compartment, then the
same remarks apply:
DADT(2)=K12*A(1)-K20*A(2)
may not apply. Maybe K20 increases with time, or increases with larger
A(2). Or maybe K12 falls with time, or falls with increasing A(2) or
decreasing A(1).
My suggestion is that you consider a non-linear differential equation,
using one of these ideas. You are all more familar with emax and
tolerance models than I am, so I will not say anything about the exact
form of such a model, or what the physiological explanation might be.
(Probably you know all this already, and it is only the physiology that
is in question.)
Quoted reply history
On Tue, 10 Jul 2007 14:11:26 -0500, "Paul Hutson"
<[EMAIL PROTECTED]> said:
> Peter: The low hanging fruit answer might be that your drug is binding
> to neutrophils, and that over the course of the infusion there is
> enhanced margination of these PMNs. Marginated to the endothelium,
> they and their adsorbed drug would not be measured.
> PS. Your book is great. Paul Bonate, Peter wrote:
>
> Dear all,
>
>
>
> I have a very unusual situation and wanted to see about getting the
> collective opinion on the group regarding the best way to handle this
> modeling problem. I have a drug that is given by 30 minute infusion.
> Samples were collected at predose, mid-infusion, end of infusion, and
> serial thereafter for 8 halflives. In about a third of the samples
> the mid-infusion sample had a considerably higher concentration (25 to
> 50%)than the end of infusion concentration. This phenomenon occurred
> across multiple studies, on multiple days (although not always in the
> same subject twice), and across multiple analytical runs. I have
> ruled out switched tubes and analytical error. For a variety of
> reasons this appears to be a valid phenomenon.
>
>
>
> Now, how best to model it or even explain it. The best I have been
> able to come up with is it is a distribution phenomenon. In
> discussions with another modeler I was informed that he just
> reviewed a paper having the same phenomenon and in that paper the
> authors discarded the midinfusion data. I have tried using time-
> dependent volumes using continuous and change-point functions. I
> get modest improvements in goodness of fit compared to completely
> ignoring the phenomenon which has a residual variability of about
> 30% using a 3-C model.
>
>
>
> As a company we have decided to pursue an oral formulation of this
> drug so it seems to me that modeling the iv data to the point of
> completely capturing the phenomenon may be a modeling exercise and not
> of any real value any longer.
>
>
>
> Any opinions on the validity of throwing out the data, just running
> with the model that ignores the phenomenon and has high residual
> variability, or something else I haven't been able to think of would
> be appreciated.
>
>
>
> Thanks,
>
>
>
> pete bonate
>
>
>
> Peter L. Bonate, PhD, FCP
>
> Genzyme Corporation
>
> Senior Director, Pharmacokinetics
>
> 4545 Horizon Hill Blvd
>
> San Antonio, TX 78229 USA
>
> [EMAIL PROTECTED]
>
> phone: 210-949-8662
>
> fax: 210-949-8219
>
> blackberry cell: 210-315-2713
>
>
>
> --
>
> Paul R. Hutson, Pharm.D.
>
> Associate Professor
>
> UW School of Pharmacy
>
> 777 Highland Avenue
>
> Madison WI 53705-2222
>
> Tel 608.263.2496
>
> Fax 608.265.5421
>
> Pager 608.265.7000, p7856
>
> References
>
> 1. mailto:[EMAIL PROTECTED]
--
Alison Boeckmann
[EMAIL PROTECTED]
We saw this behavior with doxorubicin infused over 20 min to parrots and
it seems to be quite reproducible (Gilbert et al. Aust Vet J 2004;
82:769-772) . We put it down to mid-infusion fluctuations in serum
levels as a result of altered cardiac output from the concentrated
infused drug, based on the papers by Richard Upton (Br J Anaesth 2004;
92:475-484 ; Intensive Care Med 2001; 27: 276-282).
Cheers
BC
Bruce CHARLES, PhD
Associate Professor
School of Pharmacy
The University of Queensland, 4072 Australia
[University Provider Number: 00025B]
TEL: +61 7 336 53194
FAX: +61 7 336 51688
[EMAIL PROTECTED]
http://www.uq.edu.au/pharmacy/brucecharles/charles.html
Quoted reply history
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Bonate, Peter
Sent: Wednesday, July 11, 2007 3:01 AM
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting the
collective opinion on the group regarding the best way to handle this
modeling problem. I have a drug that is given by 30 minute infusion.
Samples were collected at predose, mid-infusion, end of infusion, and
serial thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to
50%)than the end of infusion concentration. This phenomenon occurred
across multiple studies, on multiple days (although not always in the
same subject twice), and across multiple analytical runs. I have ruled
out switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have been able
to come up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the
same phenomenon and in that paper the authors discarded the midinfusion
data. I have tried using time-dependent volumes using continuous and
change-point functions. I get modest improvements in goodness of fit
compared to completely ignoring the phenomenon which has a residual
variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug
so it seems to me that modeling the iv data to the point of completely
capturing the phenomenon may be a modeling exercise and not of any real
value any longer.
Any opinions on the validity of throwing out the data, just running with
the model that ignores the phenomenon and has high residual variability,
or something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
Hi Pete
I assume that this compound is NOT a biologic, owing to the plans for oral
formulation, but can you give any information on the general class of agent
that you are dealing with? That can provide some information on possible
causes of this sort of behavior such as what Bruce suggests below. You
could also get extravasation or some local inflammation that is causing
altered PK in the case of anti-neoplastic agents
Diane
_____
Quoted reply history
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On
Behalf Of Bruce Charles
Sent: Tuesday, July 10, 2007 8:06 PM
To: Bonate, Peter; [email protected]; [EMAIL PROTECTED]
Subject: RE: [NMusers] mid-infusion higher than end of infusion
We saw this behavior with doxorubicin infused over 20 min to parrots and it
seems to be quite reproducible (Gilbert et al. Aust Vet J 2004; 82:769-772)
. We put it down to mid-infusion fluctuations in serum levels as a result of
altered cardiac output from the concentrated infused drug, based on the
papers by Richard Upton (Br J Anaesth 2004; 92:475-484 ; Intensive Care Med
2001; 27: 276-282).
Cheers
BC
Bruce CHARLES, PhD
Associate Professor
School of Pharmacy
The University of Queensland, 4072 Australia
[University Provider Number: 00025B]
TEL: +61 7 336 53194
FAX: +61 7 336 51688
[EMAIL PROTECTED]
http://www.uq.edu.au/pharmacy/brucecharles/charles.html
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On
Behalf Of Bonate, Peter
Sent: Wednesday, July 11, 2007 3:01 AM
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting the
collective opinion on the group regarding the best way to handle this
modeling problem. I have a drug that is given by 30 minute infusion.
Samples were collected at predose, mid-infusion, end of infusion, and serial
thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to 50%)than
the end of infusion concentration. This phenomenon occurred across multiple
studies, on multiple days (although not always in the same subject twice),
and across multiple analytical runs. I have ruled out switched tubes and
analytical error. For a variety of reasons this appears to be a valid
phenomenon.
Now, how best to model it or even explain it. The best I have been able to
come up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the same
phenomenon and in that paper the authors discarded the midinfusion data. I
have tried using time-dependent volumes using continuous and change-point
functions. I get modest improvements in goodness of fit compared to
completely ignoring the phenomenon which has a residual variability of about
30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug so
it seems to me that modeling the iv data to the point of completely
capturing the phenomenon may be a modeling exercise and not of any real
value any longer.
Any opinions on the validity of throwing out the data, just running with the
model that ignores the phenomenon and has high residual variability, or
something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
Hi Peter,
A difference of substantial magnitude in a 30-minute infusion would
suggest to me that the rate of input or time of sampling is not
precisely controlled. Sorry to be boring, but every time I have seen
this so far it has been explained by the administration or sampling
procedure. That the phenomenon is not necessarily reproducible in the
same subject partially supports this interpretation, as an explanation
requires substantial within-subject variation (which is actually why the
netrophil-binding idea is particularly clever).
It is common in studies for the end-of-infusion sample to in fact be
just after the end of the infusion (when concentration is falling,
usually rapidly) even when your protocol specifically disallows this.
Best regards, James
James G Wright PhD
Scientist
Wright Dose Ltd
Tel: 44 (0) 772 5636914
www.wright-dose.com http://www.wright-dose.com/
Quoted reply history
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Bonate, Peter
Sent: 10 July 2007 18:01
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting the
collective opinion on the group regarding the best way to handle this
modeling problem. I have a drug that is given by 30 minute infusion.
Samples were collected at predose, mid-infusion, end of infusion, and
serial thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to
50%)than the end of infusion concentration. This phenomenon occurred
across multiple studies, on multiple days (although not always in the
same subject twice), and across multiple analytical runs. I have ruled
out switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have been able
to come up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the
same phenomenon and in that paper the authors discarded the midinfusion
data. I have tried using time-dependent volumes using continuous and
change-point functions. I get modest improvements in goodness of fit
compared to completely ignoring the phenomenon which has a residual
variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug
so it seems to me that modeling the iv data to the point of completely
capturing the phenomenon may be a modeling exercise and not of any real
value any longer.
Any opinions on the validity of throwing out the data, just running with
the model that ignores the phenomenon and has high residual variability,
or something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
Although physiological explanations can not be completely discarded, by
the principles of Occam's Razor, one should look for simple explanations
first. I agree with James that infusion variability may be a likely
cause. In my experience infusion pumps are not nearly as precise as you'd
think they'd be.
----- Forwarded by Michael J Fossler/PharmRD/GSK on 07/11/2007 08:38 AM
-----
"James G Wright" <[EMAIL PROTECTED]>
Sent by: [EMAIL PROTECTED]
11-Jul-2007 05:24
To
[email protected]
cc
Subject
RE: [NMusers] mid-infusion higher than end of infusion
Hi Peter,
A difference of substantial magnitude in a 30-minute infusion would
suggest to me that the rate of input or time of sampling is not precisely
controlled. Sorry to be boring, but every time I have seen this so far it
has been explained by the administration or sampling procedure. That the
phenomenon is not necessarily reproducible in the same subject partially
supports this interpretation, as an explanation requires substantial
within-subject variation (which is actually why the netrophil-binding idea
is particularly clever).
It is common in studies for the end-of-infusion sample to in fact be just
after the end of the infusion (when concentration is falling, usually
rapidly) even when your protocol specifically disallows this.
Best regards, James
James G Wright PhD
Scientist
Wright Dose Ltd
Tel: 44 (0) 772 5636914
www.wright-dose.com
Quoted reply history
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of Bonate, Peter
Sent: 10 July 2007 18:01
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting the
collective opinion on the group regarding the best way to handle this
modeling problem. I have a drug that is given by 30 minute infusion.
Samples were collected at predose, mid-infusion, end of infusion, and
serial thereafter for 8 halflives. In about a third of the samples the
mid-infusion sample had a considerably higher concentration (25 to
50%)than the end of infusion concentration. This phenomenon occurred
across multiple studies, on multiple days (although not always in the same
subject twice), and across multiple analytical runs. I have ruled out
switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have been able
to come up with is it is a distribution phenomenon. In discussions with
another modeler I was informed that he just reviewed a paper having the
same phenomenon and in that paper the authors discarded the midinfusion
data. I have tried using time-dependent volumes using continuous and
change-point functions. I get modest improvements in goodness of fit
compared to completely ignoring the phenomenon which has a residual
variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug so
it seems to me that modeling the iv data to the point of completely
capturing the phenomenon may be a modeling exercise and not of any real
value any longer.
Any opinions on the validity of throwing out the data, just running with
the model that ignores the phenomenon and has high residual variability,
or something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
<<image/jpeg>>
Hi everybody,
Of course poor control of the infusion rate can be an issue and also
make sure that the end of infusion sample is performed just before the
end of infusion rather than just (or some time) after. If these issues
are clarified then binding of the drug to circulating cells or receptors
could be an explanation... for example, this phenomenon is pretty common
with monoclonal antibodies that bind to circulating receptors.
Rene
Quoted reply history
________________________________
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of James G Wright
Sent: Wednesday, July 11, 2007 11:25 AM
To: [email protected]
Subject: RE: [NMusers] mid-infusion higher than end of infusion
Hi Peter,
A difference of substantial magnitude in a 30-minute infusion would
suggest to me that the rate of input or time of sampling is not
precisely controlled. Sorry to be boring, but every time I have seen
this so far it has been explained by the administration or sampling
procedure. That the phenomenon is not necessarily reproducible in the
same subject partially supports this interpretation, as an explanation
requires substantial within-subject variation (which is actually why the
netrophil-binding idea is particularly clever).
It is common in studies for the end-of-infusion sample to in fact be
just after the end of the infusion (when concentration is falling,
usually rapidly) even when your protocol specifically disallows this.
Best regards, James
James G Wright PhD
Scientist
Wright Dose Ltd
Tel: 44 (0) 772 5636914
www.wright-dose.com http://www.wright-dose.com/
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Bonate, Peter
Sent: 10 July 2007 18:01
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting
the collective opinion on the group regarding the best way to handle
this modeling problem. I have a drug that is given by 30 minute
infusion. Samples were collected at predose, mid-infusion, end of
infusion, and serial thereafter for 8 halflives. In about a third of
the samples the mid-infusion sample had a considerably higher
concentration (25 to 50%)than the end of infusion concentration. This
phenomenon occurred across multiple studies, on multiple days (although
not always in the same subject twice), and across multiple analytical
runs. I have ruled out switched tubes and analytical error. For a
variety of reasons this appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have
been able to come up with is it is a distribution phenomenon. In
discussions with another modeler I was informed that he just reviewed a
paper having the same phenomenon and in that paper the authors discarded
the midinfusion data. I have tried using time-dependent volumes using
continuous and change-point functions. I get modest improvements in
goodness of fit compared to completely ignoring the phenomenon which has
a residual variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of
this drug so it seems to me that modeling the iv data to the point of
completely capturing the phenomenon may be a modeling exercise and not
of any real value any longer.
Any opinions on the validity of throwing out the data, just
running with the model that ignores the phenomenon and has high residual
variability, or something else I haven't been able to think of would be
appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
I saw similar problems because of operation procedures for compounds with rapid
decay after the end of infusion. Therefore it is essential to make sure the PK
sample at the end of infusion is collected right before the end of infusion
rather than post the end of infusion.
Just curious about the binding issue. Could anyone describe a little bit more
about this? Will this happen for a 30-minute infusion? How about for a 6-hour
infusion or 24-hour continuous infusion? That is, is there a time scale/limit
to observe this phenomenon (because of the binding variation with time)? Is
this also dependent on the infusion rate and concentration?
Alan
Quoted reply history
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Rene Bruno
Sent: Wednesday, July 11, 2007 8:50 AM
To: James G Wright; [email protected]
Subject: RE: [NMusers] mid-infusion higher than end of infusion
Hi everybody,
Of course poor control of the infusion rate can be an issue and also make sure
that the end of infusion sample is performed just before the end of infusion
rather than just (or some time) after. If these issues are clarified then
binding of the drug to circulating cells or receptors could be an
explanation... for example, this phenomenon is pretty common with monoclonal
antibodies that bind to circulating receptors.
Rene
_____
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of James G Wright
Sent: Wednesday, July 11, 2007 11:25 AM
To: [email protected]
Subject: RE: [NMusers] mid-infusion higher than end of infusion
Hi Peter,
A difference of substantial magnitude in a 30-minute infusion would suggest to
me that the rate of input or time of sampling is not precisely controlled.
Sorry to be boring, but every time I have seen this so far it has been
explained by the administration or sampling procedure. That the phenomenon is
not necessarily reproducible in the same subject partially supports this
interpretation, as an explanation requires substantial within-subject variation
(which is actually why the netrophil-binding idea is particularly clever).
It is common in studies for the end-of-infusion sample to in fact be just after
the end of the infusion (when concentration is falling, usually rapidly) even
when your protocol specifically disallows this.
Best regards, James
James G Wright PhD
Scientist
Wright Dose Ltd
Tel: 44 (0) 772 5636914
www.wright-dose.com http://www.wright-dose.com/
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Bonate, Peter
Sent: 10 July 2007 18:01
To: [email protected]; [EMAIL PROTECTED]
Subject: [NMusers] mid-infusion higher than end of infusion
Dear all,
I have a very unusual situation and wanted to see about getting the collective
opinion on the group regarding the best way to handle this modeling problem. I
have a drug that is given by 30 minute infusion. Samples were collected at
predose, mid-infusion, end of infusion, and serial thereafter for 8 halflives.
In about a third of the samples the mid-infusion sample had a considerably
higher concentration (25 to 50%)than the end of infusion concentration. This
phenomenon occurred across multiple studies, on multiple days (although not
always in the same subject twice), and across multiple analytical runs. I have
ruled out switched tubes and analytical error. For a variety of reasons this
appears to be a valid phenomenon.
Now, how best to model it or even explain it. The best I have been able to
come up with is it is a distribution phenomenon. In discussions with another
modeler I was informed that he just reviewed a paper having the same phenomenon
and in that paper the authors discarded the midinfusion data. I have tried
using time-dependent volumes using continuous and change-point functions. I
get modest improvements in goodness of fit compared to completely ignoring the
phenomenon which has a residual variability of about 30% using a 3-C model.
As a company we have decided to pursue an oral formulation of this drug so it
seems to me that modeling the iv data to the point of completely capturing the
phenomenon may be a modeling exercise and not of any real value any longer.
Any opinions on the validity of throwing out the data, just running with the
model that ignores the phenomenon and has high residual variability, or
something else I haven't been able to think of would be appreciated.
Thanks,
pete bonate
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713