Leonid,
Thanks for pointing out the confusion in my response. I intended to write:
"Sequential methods are preferred over
simultaneous methods whenever there is a possibility of
mis-specification of the model linking concentration to effect. This
is nearly always a real risk so the simultaneous method is rarely a
sensible choice (see Zhang et al part II). "
The VIOA (very important old abbreviations
<http://www.cognigencorp.com/nonmem/nm/99dec192005.html>) IPP, PPP&D are
defined in the Zhang papers. Anybody who wants to do serious PKPD work
should be familiar with these papers.
The simulations performed by Zhang et al. show that simultaneous can be
worse than sequential (see Part II paper). That is why I encourage a
sequential approach using the PPP&D method.
Best wishes,
Nick
On 19/02/2013 1:57 p.m., Leonid Gibiansky wrote:
Hi Nick,
I am afraid, many users are not very familiar with all the important
abbreviations - PPP&D, IPP, VINA, etc. :)
I am not sure that I follow your points, looks contradictory:
> Sequential methods are preferred ....
> ...sequential method is rarely a sensible choice
In any case, I would not agree that PPP&D (use of both PK and PD data in
the simultaneous fit ?) is the only good method (or even that this
method is the best in all situations). In many cases, sequential method
gives nearly identical results to the simultaneous fit, and was easier
to implement numerically. I would be curious to see a good real life
example where sequential method provided the wrong answer while PP&D was
correct (that is, the example where these two methods resulted in
different clinically relevant conclusions).
Leonid
VINA = Very Important New Abbreviation
--------------------------------------
Leonid Gibiansky, Ph.D.
President, QuantPharm LLC
web: www.quantpharm.com
e-mail: LGibiansky at quantpharm.com
tel: (301) 767 5566
On 2/18/2013 3:23 PM, Nick Holford wrote:
Jakob, Leonid,
I think it should be pointed out more clearly that PKPD analyses can
be performed most simply using the PPP&D method. This method has
better properties compared with IPP and is easier to set up and run
(see Zhang et al Part I). Sequential methods are preferred over
simultaneous methods whenever there is a possibility of
mis-specification of the model linking concentration to effect. This
is nearly always a real risk so the sequential method is rarely a
sensible choice (see Zhang et al part II).
The IPP+SE method has properties similar to PPP&D but it technically
more challenging to use and cannot be used without a
variance-covariance matrix of the estimates. This can usually only be
obtained with oversimplified models (asymptotic $COV) or by time
consuming bootstraps.
The bottom line is to use PPP&D.
Leonid makes a good point about using VPC for model evaluation. The
VPC is capable of detecting structural model misspecification for
PKPD analyses but it is not foolproof. A misspecified random effects
model can compensate for a misspecified structural model. See
references below for an example.
Best wishes,
Nick
Karlsson MO, Holford NHG 2008. A Tutorial on Visual Predictive
Checks. PAGE 17 (2008) Abstr 1434 [wwwpage-meetingorg/?abstract=1434]
(last accessed 11 February 2012).
http://holford.fmhs.auckland.ac.nz/docs/vpc-tutorial-and-datatop.pdf
On 19/02/2013 7:55 a.m., Leonid Gibiansky wrote:
Hi Matts,
One method to investigate the problem would be to conduct VPC. If
VPC with model-estimated variances provides good (not inflated)
range of PK profiles then one can argue that the PK model provides
good description of the data and can be used for simulations
(including PK-PD).
Another test could be to do VPCs for the PK-PD model: one with
fixed PK parameters (as was used in the sequential PK-PD modeling
procedure) and the other one with model-simulated ETAs for both PK
and PD parts. Again, if both provide good coverage of observed
PK-PD data then combination of PK and PD models can be trusted, and
any of the approaches can be applied. If one of the VPCs is
inadequate, than it should be noticeable in the too narrow or too
wide prediction intervals.
Leonid
On 19/02/2013 5:20 a.m., Ribbing, Jakob wrote:
Resending, since my posting from this morning (below) has not yet
appeared on nmusers.
Apologies for any duplicate postings!
*From:*Ribbing, Jakob *Sent:* 18 February 2013 09:59 *To:*
"Kågedal, Matts"; [email protected] *Cc:* Ribbing, Jakob
*Subject:* RE: Simulation settgin in the precence of Shrinkage in
PK when doing PK-PD analysis
Hi Matts,
I think you are correct; the problem you describe has not had much
(public) discussion.
It is also correct like you say that this is mostly a problem in
case of all of the below
·sequential PK-PD analysis is applied (IPP approach, Zang et al)
·non-ignorable degree of shrinkage in PK parameters _of relevance_
oOf relevance: the PK parameters effectively driving PD for the
mechanism, e.g. CL/F if AUC is driving. In addition, if PD response
develops over several weeks/months then shrinkage in IOV may be
ignored even for relevant PK parameters
·I would also like to add that for this to be an issue individual
PK parameters must explain a fair degree of the variability in PD,
which is not always the case
oIf driving PD with typical PK parameters (along with dose and
other PD covariates) does not increase PD omegas, compared to IPP,
then either PK shrinkage is already massive, or else it is not an
issue for the IPP-PD model
If only the IPP approach is possible/practical a simplistic
approach to simulate PD data is as follows:
·sample (with replacement) the individual PK parameters along with
any potential covariates (maintaining correlation between IIP and
covariates, i.e. whole subject vectors for these entities, but
generally not for dose since generally should only have only random
association with IPP or PK/PD-covariates)
·then use the re-sampled datasets for simulating PD according to
the PD model (driven by IPP, covariates, dose, etc). The degree of
shrinkage is then the same for PD estimation and simulation.
This approach may for example allow to simulate realistic PD
response at multiple dosing, based on only single dose PD. When the
MD data becomes available then one may find that variabilities
shift between PK and PD due to different PK shrinkage, but I would
argue the simulated PD responses still were realistic. This
approach is useful for predictions into the same population
(especially if sufficient number of subjects available for
re-sampling), but may not allow extrapolation into other
populations where PK is projected to be different.
When possible the obvious solution is to apply one of the
alternative approaches to simultaneous PK-PD fit; after you have
arrived at a final-IPP model.
If a simultaneous fit is obtainable/practical this is the best
option, but notice that e.g. if you have rich PK data in healthy
and no PK data in patients (plus PD data in both populations): You
can estimate separate omegas for PD parameters in healthy vs.
patients, but it may be difficult to tell whether patients higher
PD variability is due to PK shrinkage, or due to the actual PD
variability being higher in this population (or both). PD
variability may be confounded by a number of other factors that are
actually variability in PK (fu, active metabolites and bio phase
distribution, just to mention a few where information may be absent
on the individual level). Depending on the purpose of the modelling
this often not an issue, however.
As you suggest there may be rare situations with IPP where a more
complicated approach is needed, with a) simulation and
re-estimation of PK model, to obtain Empirical-Bayes Estimates
based on simulated data, and then feed these into the subsequent PD
model. I would see this as a last resort. There are pitfalls in
that if PD parameters have been estimated under one degree of PK
shrinkage, then applying these estimates to a simulated example
with different PK shrinkage requires adjustment of PD variability.
I am not sure anyone has had to go down that route before and if
not I hope you do not have to either. Maybe others can advice on
this?
Best regards
Jakob
Two methodological references:
Simultaneous vs. sequential analysis for population PK/PD data II:
robustness of methods.
Zhang L, Beal SL, Sheiner LB.
J Pharmacokinet Pharmacodyn. 2003 Dec;30(6):405-16.
Simultaneous vs. sequential analysis for population PK/PD data I:
best-case performance.
Zhang L, Beal SL, Sheiner LB.
J Pharmacokinet Pharmacodyn. 2003 Dec;30(6):387-404.
e
--
Nick Holford, Professor Clinical Pharmacology
Dept Pharmacology & Clinical Pharmacology, Bldg 503 Room 302A
University of Auckland,85 Park Rd,Private Bag 92019,Auckland,New Zealand
tel:+64(9)923-6730 fax:+64(9)373-7090 mobile:+64(21)46 23 53
email: [email protected]
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford
