NONMEM Users Network Archive

Hosted by Cognigen

RE: Simultaneous vs sequential for modeling parent AND metabolites in pop PK

From: James G Wright <james>
Date: Thu, 11 Dec 2008 11:48:08 -0000

In my experience the F-induced covariance among parameters is typically
identifiable for any drug with sub-100% bioavailability, and expressing
this as a variance on F is typically a parsimonious option. As this
term partially accounts for differences in the amount of drug entering
the system, it can reduce covariance between the parent and metabolite
When dealing with the metabolite parameters, I prefer the simple
notation of CL/F where clearance and bioavailability both refer to the
metabolite. If you work on prodrugs or multiple metabolites in
sequence, the fraction formed notation can become very cumbersome - and
sometimes the relationships between the chemical species are not clear
cut, so additional assumptions would be required for that notation.
If it is known that there is an appreciable first-pass metabolism, this
should be included in the model and usually produces a dramatically
improved fit in my experience. I always test this model unless it is
obviously not necessary. F and Fm remain unidentifiable of course
(there is no magic unfortunately) and the additional parameter is the
fraction of metabolite formation attributable to first-pass formation
(which I usually abbreviate to FFFP - fraction formed first-pass).
Best regards, James
James G Wright PhD
Wright Dose Ltd
Tel: 44 (0) 772 5636914
-----Original Message-----
From: owner-nmusers
On Behalf Of andreas lindauer
Sent: 11 December 2008 10:17
To: nmusers
Subject: Re: [NMusers] Simultaneous vs sequential for modeling parent
AND metabolites in pop PK
Nick, James and all,
I would like to add to the present discussion that when one estimates
CL/fm or V/fm of the metabolite, one should be aware that if the parent
drug is given orally (bioavailability unknown) this also affects the
metabolite parameters. In this case only CL/(fm*F) and V/(fm*F) can be
estimated. As has been already stated, this is not a problem for a
descriptive model. However, especially for simulations it is important
to account for the correlation of these parameters because CL/F and V/F
of the parent as well as CL/(fm*F) and V/(fm*F) of the metabolite are
all correlated because of F and fm. Thus my questions to the group: are
the between subject variabilities of fm and F identifiable? I know that
one can estimate them by fixing F and fm to 1 and estimate their omegas,
but will they really be identifiable and give meaningful results? I'm
wondering because also the parent drug's clearance and the fraction
metabolised are related. Shouldn't this also be taken into account?
Things get even more complicated when a significant first-past effect
occurs. Then (assuming 100% absorption) the remainder of the unknown F,
i.e. the fraction of the parent drug that is converted to the metabolite
via first-pass (1-F) must now also be considered for the metabolite
parameters and one estimates CL/(fm*F*(1-F)) and V/(fm*F*(1-F)). Or is F
(assuming complete absorption) identifiable in this case?
I'm looking forward to your comments.
Regards, Andreas.
Andreas Lindauer
Department of Clinical Pharmacy
Institute of Pharmacy
University of Bonn
An der Immenburg 4
D-53121 Bonn
phone: + 49 228 73 5781
fax: + 49 228 73 9757

Received on Thu Dec 11 2008 - 06:48:08 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: