NONMEM Users Network Archive

Hosted by Cognigen

RE: Honest measurements

From: Mark Sale - Next Level Solutions <mark>
Date: Mon, 24 Aug 2009 07:08:12 -0700
Joachim et al.
  Moreover, assays are typically done= in duplicate or triplicate, with only the mean reported.  We had many= discussions at GSK with the chemists about getting "all" the data (BQL, al= l replicates) promising everything imaginable (first, the data would never = end up in a regulatory document, then that there would never be a publicati= on that would have anything other than model results, and finally that no i= nformation derived from these data would ever leave the company in any form= ), to no avail.  I hope others have better luck.  But, you are co= rrect, all forms of error in the assay (between replicate, issues of BQL) e= tc should ideally be addressed in the model.  Imagine how useful dupli= cate or triplicate data would be in defining residual error vs model misspe= cification.
I hadn't thought about including the calibration curve in th= e model, so that all that would be reported would be the absorption?


Mark Sale MD
Next Level Solutions, LLC

-------- Original Message --------
Subject: RE: [NMusers] Honest measurements
From: Grevel, Joachim <Joachim.Grevel Date: Mon, August 24, 2009 9:18 am
To: "nmusers" <nmusers
Dear all,

I am thinking back to early days of pharmacodynamics. The clinical people w= ould report to us "early" modellers response data in the form of percent ch= ange from baseline. Very soon we asked for the raw data rejecting their "mo= del" of generating response data as being subjective and biased (how was th= at "baseline" established?).

It was Nick among others who rejected any delivered baseline. He insisted t= hat the model "will tell us the baseline and its error". I somehow hear Nic= k telling us something similar about bioanalytical data: Do not truncate th= e data by some criterion established months (years) before the actual measu= rement. Give us the raw data and let us find the error in it.

Perhaps it is indeed time to rethink the ideas of validated methods and sta= ndard curves. Couldn't the standard curve be part of our modelling, and cou= ldn't the error in the standard curve be integrated into the error structur= e for our PK (PKPD) model?

I have a feeling that our current practice may have all the flaws we found = years ago in the "two-stage" PK modelling approach. We gave up on the two-s= tage approach because it did not describe variability correctly. (Mats, the= re were "millions" of two-stage analyses published and in submissions... an= d in the end we gave it up for something we had learned was better).

It will take a lot of theoretical work to make the case for an integrated b= ioanalytical-pharmacokinetic analysis. Nick, are you already working along = these lines?

Good luck to all,


--------------------------------------------------------------------------<= br> AstraZeneca UK Limited is a company incorporated in England and Wales with = registered number: 03674842 and a registered office at 15 Stanhope Gate, Lo= ndon W1K 1LN.
Confidentiality Notice: This message is private and may contain confidentia= l, proprietary and legally privileged information. If you have received thi= s message in error, please notify us and remove it from your system and not= e that you must not copy, distribute or take any action in reliance on it. = Any unauthorised use or disclosure of the contents of this message is not p= ermitted and may be unlawful.
Disclaimer: Email messages may be subject to delays, interception, non-deli= very and unauthorised alterations. Therefore, information expressed in this= message is not given or endorsed by AstraZeneca UK Limited unless otherwis= e notified by an authorised representative independent of this message. No = contractual relationship is created by this message by any person unless sp= ecifically indicated by agreement in writing other than email.
Monitoring: AstraZeneca UK Limited may monitor email traffic data and conte= nt for the purposes of the prevention and detection of crime, ensuring the = security of our computer systems and checking Compliance with our Code of C= onduct and Policies.
-----Original Message-----
From: owner-nmusers [mai= lto:owner-nmusers Sent: 24 August 2009 13:36
To: 'Stephen Duffull'; 'Nick Holford'; 'nmusers'
Subject: RE: [NMusers] Honest measurements

Sense and nonsense.

Analysts are constrained certainly but good analytical science is good
analytical science. During the method development stage several parameters=
are examined.

One parameter for example is recovery of analyte from a set of spiked
individual (matrices). Another exercise is the recovery of material from<= br> spiked matrix after exposure to a number of stability challenging
conditions, e.g., freeze thaw. During these tests it is certainly possible=
to "find less or, even "none" of the response(s) to the analyte(s).

Analysts would then "improve" the assay using a number of techniques
including the addition of preservatives or specifying the handling and
storage conditions. Too often however, the clinical collection parameters=
may have been defined before the conclusion or event the start of analytica= l
method development and perhaps "if pushed too far and incessantly" the
analyst will release "nonsense data" or instrument responses free of

For people at the next rung to use data this in any rational way suggests that those people should perhaps be using a seer rather than an analyst. When they are stung by the results of their data analysis, they will
immediately revert and "blame" the analyst (or if they do not, regulatory agencies will) for releasing "bad" data.

PK/PD and Analytical scientists usually work together during the analytical=
process. PK/PD driving with LLOQ, suggesting metabolites, Analytical
providing the best approach to meet that requirement including meeting all<= br> current method validation and reporting guidelines.

Edward F. O'Connor, PhD
78 Marbern Drive
Suffield, CT 06078

Tel 860-668=6201
Cel 860-324-6780
Edward F. O'Connor, PhD
78 Marbern Drive
Suffield, CT 06078

Tel 860-668=6201
Cel 860-324-6780
-----Original Message-----
From: owner-nmusers ation==top.location){Popup.composeWindow('pcompose.php?sendto=owner-n=');}else{top.Popup.composeWindow('pcompose.php?sendt=');} return false;" href="http://email0=" mce_href="http://email01.secures=">mailto:owner-nmusers /a>] On
Behalf Of Stephen Duffull
Sent: Monday, August 24, 2009 4:18 AM
To: Nick Holford; nmusers
Subject: RE: [NMusers] Honest measurements


I agree with Nick. Negative "observed" concns do occur for assays, even in=
my limited time working with HPLC I have seen them, however due to LOD/LOQ<= br> they are never really looked for and certainly never reported...


> -----Original Message-----
> From: owner-nmusers > nmusers > Sent: Monday, 24 August 2009 6:25 p.m.
> To: nmusers
> Subject: [NMusers] Honest measurements
> Mats Karlsson wrote:
> << Chemists, however pushed, would never report negative
> concentrations, not
> for past studies, not for future studies. The methods they use don't > even
> report them.>>
> I am working with a chemist using LC/MS who has been persuaded to look=
> honestly at his data without preconceived ideas of limits of
> quantitation and detection. Indeed when he opened his eyes he found > that his system was indeed giving negative concentration measurements<= br> > (at times when concentrations were expected to be very low).
> Of course we must do other things when the data is censored by bad
> scientific practice in the chemist's lab but with honest measurments a= n
> additive residual error model is required.
> Nick
> --
> Nick Holford, Professor Clinical Pharmacology
> Dept Pharmacology & Clinical Pharmacology
> University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New > Zealand
> n.holford > mobile: +64 21 46 23 53

No virus found in this incoming message.
Checked by AVG -
Version: 8.5.409 / Virus Database: 270.13.65/2322 - Release Date: 08/23/09<= br> 18:03:00

(image/png attachment: left.letterhead)

Received on Mon Aug 24 2009 - 10:08:12 EDT

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: