NONMEM Users Network Archive

Hosted by Cognigen

Re: cyst size modeling

From: Nick Holford <n.holford>
Date: Sun, 20 Jul 2008 14:06:33 +1200

Nele,

You give quite a lot of information about your problem but no details of
what kind of cysts (origin, size, composition) or what biological
process you think might be involved in making them get smaller or what
pharmacological process might modify the natural history. Without this
kind of information it is really hard to propose anything that is not
simply empirical e.g. exponential natural history ('baseline') plus
effect compartment for drug treatment.

You say the objective is to "find the dose that would result in cyst
eradication if the baseline effect would not be there". The best dose is
obviously a dose of zero because the cysts will be eradicated anyway
without any risk of adverse effects <grin>. So I suspect you are also
interested in eradicating the cysts more quickly.

There does seem to be a more practical problem however. With only
baseline and 4 week observations it is impossible to distinguish a
natural history model from the time course of drug effect. One
additional measurement at 2 weeks for the highest dose might give you a
small clue about the time course of drug effect but it will be
confounded with the exposure response model. I think you will agree that
the design of the experiment is pretty bad for learning anything useful
about the time course of response (with or without treatment). Can you
persuade your experimental pharmacologists to work a bit harder?

Best wishes,

Nick


nele.plock
> Dear nmusers,
>
> I would like to seek your advice on a PKPD modeling problem to maybe get
> some further ideas on how to address this problem.
> A pharmacological experiment in rats was performed, where cysts are placed
> by surgery. 3 weeks later, the size of all cysts is assessed (this time
> point is defined as zero). In addition, I know the cyst size without
> treatment 2 weeks and 4 weeks after. Overall, the size remains the same
> for the first two weeks, but decreases afterwards. This is not unusual, as
> it is also known that the cysts will all eventually go away after a while,
> even without treatment. So I know that I would need a baseline model of
> some sort in order to derive an additional compound effect.
> The compound was tested in 3 different doses. Cyst size after treatment
> with the two lower doses was only assessed after 4 weeks of treatment,
> whereas for the highest dose the size is available after 2 and 4 weeks of
> treatment. Thus, I have some information about effect over time (i.e. time
> zero, after 2 weeks and after 4 weeks).
> My question is: What kind of baseline model would be a good choice? Does
> anybody have experience with cyst modeling and can share what would be
> pharmacologically plausible? What should be the natural course of cyst
> shrinkage (linear, turnover)?
> Moreover I am wondering how to best describe the influence of the drug. I
> know that it will not directly kill the cysts, so I thought about
> including an effect compartment. I have written a Berkeley Madonna code to
> play around with the model. What are your thoughts on a model like this
> (code below)? Would the available data be able to support the parameters?
> The model is intended to find the dose that would result in cyst
> eradication if the baseline effect would not be there.
>
> Thank you and best regards
> Nele
>
> PS: If you send any answers, I would be very grateful if you could also
> send a copy to neleplock
> Thank you!
> ________________________________________________________________________________
> Berkeley Madonna code (parameters were chosen arbitrarily as no analysis
> has been performed yet):
> METHOD RK4
>
> STARTTIME = 0
> STOPTIME=696
> DT = 0.2
>
> INIT (gut) = X
> INIT (central) = 0
> INIT(effect)=0
> INIT(PD)=1
>
> D/DT(gut)=-KA*gut+DOSING
> D/DT(central)=KA*gut -K20*central
> D/DT(effect)= KEO*central-KEO*effect
> D/DT(PD) = KIN - KOUT*FAC*PD- KOUT*effect*PD
>
> cyst=100*PD ;100 as real cyst size at the start of the experiment
>
> Ccentral=central/V2*1000
>
> CL=0.387
> V2=1.83
> KA=0.724
> KEO=0.01
> KIN=0.0001
> KOUT=KIN
> FAC=IF TIME < 336 THEN 0 ELSE 10
>
> X=0.1
> K20=CL/V2
>
> ;--------MULTIPLE DOSING--------
> DOSING=PULSE(DOS,START,INTERVAL)
> DOS= IF TIME < 100000 THEN X ELSE 0
> START= 24
> INTERVAL=24
> ________________________________________________________________________________
> Dr. Nele Plock
> Bayer Schering Pharma AG
> Drug Metabolism & Pharmacokinetics
> Development Pharmacokinetics
> Scientific Expert Development Pharmacokinetics
> D- 13342 Berlin
>
> Phone : +49-30-468 15146
> Fax: +49-30-468 95146
> nele.plock
> http://www.bayerscheringpharma.de
>
> Vorstand: Arthur J. Higgins, Vorsitzender | Werner Baumann, Andreas Busch,
> Ulrich Köstlin, Kemal Malik, Gunnar Riemann
> Vorsitzender des Aufsichtsrats: Werner Wenning
> Sitz der Gesellschaft: Berlin | Eintragung: Amtsgericht Charlottenburg 93
> HRB 283
>

--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand
n.holford
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford

Received on Sat Jul 19 2008 - 22:06:33 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request@iconplc.com.

Once subscribed, you may contribute to the discussion by emailing: nmusers@globomaxnm.com.