Helmut
★★★
avatar
Homepage
Vienna, Austria,
2008-05-26 17:26
(5785 d 07:51 ago)

Posting: # 1872
Views: 11,004
 

 Lag time in WinNonlin - bug or feature? [Software]

Dear all,

I just discovered another nice one.

Quoting WinNonlin’s 5.2 online help (Noncompartmental analysis > Plasma or serum data > Table B-3.):
Tlag    Extravascular input (model 200) only. Tlag is the time prior
        to the first measur­able (non-zero) concentration.

Beware that non-zero concentration must be taken literally for all other values, i.e., if you have values prior to the first measurable coded as ‘Missing’ or ‘BQL’ or any other non-numeric value, WinNonlin will come up with a Tlag of zero!
Example:
+-------+---------+-------+-------+
| time  | code 1  | code 2| =zero |
+-------+---------+-------+-------+
|  0.00 | MissingBQL  |  0    |
|  0.25 | MissingBQL  |  0    |
|  0.50 | MissingBQL  |  0    |
|  0.75 | MissingBQL  |  0    |
|  1.00 | MissingBQL  |  0    |
|  1.5  |  10.5   | 10.5  | 10.5  |
|  2.0  |  15.9   | 15.9  | 15.9  |
|  2.5  |  18.4   | 18.4  | 18.4  |
|  3.0  |  19.4   | 19.4  | 19.4  |
|  4    |  19.4   | 19.4  | 19.4  |
|  6    |  17.0   | 17.0  | 17.0  |
|  9    |  13.4   | 13.4  | 13.4  |
| 12    |  10.6   | 10.6  | 10.6  |
| 16    |   7.68  |  7.68 |  7.68 |
| 24    |   4.05  |  4.05 |  4.05 |
| 36    |   1.55  |  1.55 |  1.55 |
+-------+---------+-------+-------+
| Tlag  |   0     |  0    |  1.5  |
(see Edit below)
+-------+---------+-------+-------+

If you want to get correct results you must recode your values first (Tools > Status Codes)!


Edit: See this post. WinNonlin's result for the third column is 1.0 (in agreement with the manual).

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
d_labes
★★★

Berlin, Germany,
2008-05-29 18:47
(5782 d 06:31 ago)

@ Helmut
Posting: # 1883
Views: 8,985
 

 Lag time in WinNonlin - bug or feature?

Dear HS,

IMHO it must be a feature. For what ever. ;-)

But I'm a little bit confused about the correct? entry. According to the definition (underlined by me)

... Tlag is the time prior to the first measur­able (non-zero) concentration ...


I would guess 1.0<Tlag<1.5 .
Or is this also a feature?

Regards,

Detlew
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2008-05-29 22:03
(5782 d 03:14 ago)

@ d_labes
Posting: # 1885
Views: 9,138
 

 WinNonlin <--> Kinetica... / Survey

Dear DLabes!

❝ IMHO it must be a feature. For what ever. ;-)


[image]Of course it’s a feature, bugs don’t exist for software vendors…

❝ But I'm a little bit confused about the correct? entry. According to the definition (underlined by me)

❝ ❝ … Tlag is the time prior to the first measur­able (non-zero) concentration …

❝ I would guess 1.0<Tlag<1.5 .


Yes, sure! Now a couple of points.
WinNonlin comes up with 1.5, which is nice – and the value pragmatically most people would also give in NCA, but contrary to the definition given in the manual (last value prior […], which is 1).
OK, I cheated; my data are artificial ones using a one-compartment model with a lag-time of 1 hour:

\(C(t)=25.5\cdot(\textrm{e}^{-0.08\cdot(t-1)}-\textrm{e}^{-1.2\cdot(t-1)})\)

Kinetica v4.1.1 gives no definition of tlag in the manual at all, but comes up with 1.

Little is published about NCA of tlag, except in Cawello (2003), where an interval (like “between 1.0 and 1.5”) is suggested.* In the Glossary (p. 151) of the same reference another definition is given:

Lag-time: refers to the time period between administration of a drug and the occurrence of the first measurable concentration of the active component in the blood stream.
(my emphases)


Unfortunately the interval definition is scientifically valid but unsuitable for common statistical comparisons (yes, I know, fuzzy logic would help…). Tlag may sound exotic, but it can be an important parameter (e.g., for gastric resistant formulations or drugs with presystemic metabolism).

Long ago I played around with linear back extrapolation and solving for C=0, but unsuccessfully. In our example we would end up with lag-times of 0.53 (linear, 2 points) or 0.89 (quadratic, 3 points), which are contradictory to <LLOQs ‘reported’ at 0.75 and/or 1.
I also worked a little bit on splines, but this went well with some subjects and terrible with others… All back-extrapolation methods I tried were very sensitive to ‘noise’. Such a behavior is in agreement with compartmental methods, where fitting of the absorption process is often quite demanding.

So my question to the all of the forum’s members:
  What method are you using?
  [X] time point of first concentration > LLOQ
  [ ] time point prior to first concentration > LLOQ
  [ ] mean of these two time points
  [ ] a verbatim statement (like: between x and y hours)
  [ ] some kind of back-extrapolation (which one?)
  [ ] I’m using software (which one?) giving me a correct answer (which one?)


  • Weimann H-J. Drug concentrations and directly derived parameters. In: Cawello W, editor. Parameters for Compartment-free Pharmacokinetics. Aachen: Shaker Verlag; 2003. p. 29-30.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
d_labes
★★★

Berlin, Germany,
2008-05-30 10:12
(5781 d 15:05 ago)

@ Helmut
Posting: # 1887
Views: 9,031
 

 Tlag

Dear HS,

up to now I have not come into the trouble of using Tlag. If I had to use this parameter, I would opt for:

  [ ] time point of first concentration > LLOQ
  [ ] time point prior to first concentration > LLOQ
  [X] mean of these two time points
  [ ] a verbatim statement (like: between x and y hours)
  [ ] some kind of back-extrapolation (which one?)
  [ ] I'm using software (which one?) giving me a correct answer (which one?)

Regards,

Detlew
d_labes
★★★

Berlin, Germany,
2009-02-02 13:48
(5533 d 10:29 ago)

@ Helmut
Posting: # 3171
Views: 8,944
 

 First time lag time

Dear Helmut, dear All,

❝ Little is published about NCA of tlag, except in Cawello (2003),


Meanwhile I came into touch with the lag time (first in my long, long career in BE studies) in a food BE study for one of the "(in)famous" ...prazoles.

Searching again the literature / I-net I found the following paper I'm not noticed so long (not so new if one sees the publication year :-( ):

Czismadia F and L Endrényi
Model-independent estimation of lag-times with first-order absorption and disposition
J Pharm Sci 87, 608–12 (1998)

It contains a comprehensive description of methods for NCA estimation of tlag. Interesting enough no one which defines the time of the first conc. >LLOQ.

After a first look their new (for me) method - named tE, with weighted regression of values between first time with conc. >LLOQ and tmax - seems promising. Although their premises in deriving this method seems a little bit restricting ("first-order absorption and disposition", no one knows really about this).

What do you think?

What should be done with this parameter? Reporting the values only (with summary statistics) or any statistical analysis of differences between formulations under study? Which statistical method, model?
There are questions upon questions :ponder: .

Regards,

Detlew
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2009-02-02 14:08
(5533 d 10:09 ago)

@ d_labes
Posting: # 3172
Views: 8,846
 

 First time lag time

Dear D. Labes!

❝ Meanwhile I came into touch with the lag time (first in my long, long career in BE studies) in a food BE study for one of the "(in)famous" ...prazoles.


Oh dear!

❝ […] I found the following paper […]



❝ What do you think?


Hhm, I didn’t know this paper also. The approach seems reasonably to me.
In PK modeling one would come up with exactly such a value.

❝ What should be done with this parameter? Reporting the values only (with summary statistics) or any statistical analysis of differences between formulations under study? Which statistical method, model?


I remember a paper by H. Blume dealing with gastric resistant diclofenac, where he used tmax-tlag as a parameter describing the rate (drug dependent) and evaluated tlag separately (formulation dependent).
The collection of Blume/Mutschler (in German, out of print) contains some examples.

Henning Blume, Ernst Mutschler
Bioäquivalenz
Qualitätsbewertung wirkstoffgleicher Fertigarzneimittel

GOVI-Verlag, Frankfurt am Main/Eschborn, 1989-1996


Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
d_labes
★★★

Berlin, Germany,
2009-02-02 14:42
(5533 d 09:35 ago)

@ Helmut
Posting: # 3173
Views: 8,624
 

 lag time paper

Dear Helmut,

❝ […] Hhm, I didn’t know this paper also. […]



I have written to l.endrenyi(at)utoronto.ca and got a copy.

Regards,

Detlew
SDavis
★★  
Homepage
UK,
2009-02-04 13:49
(5531 d 10:29 ago)

@ d_labes
Posting: # 3178
Views: 8,941
 

 Lag time in WinNonlin - bug or feature?

Dear Helmut, d_labes,

I'm also a little confused about the results that Helmut quoted above. Running that same data in 5.2.1 I get the result of 1 for Tlag in the third group which agrees with the definition in the help file. I don't recall this function being changed for some versions in WNL so is there a typo in your table, or can you send me the workspace for further investigation?

... Tlag is the time prior to the first measur­able (non-zero) concentration ...


I would agree that probably Tlag lies somewhere between 1.0 and <1.5 but with NCA being model-independent I guess Pharsight decided to go for the prior one but obviously this will be heavily dependent on the sampling schedule so compartmental modelling is probably going to be more useful if you need to define this more 'precisely'.

I concur it would be useful for WNL to 'intrinsically' recognise BLQs but Helmut correctly states that the appropriate process is currently to recode non-numeric values first using (Tools > Status Codes).

Simon
Senior Scientific Trainer, Certara™
[link=https://www.youtube.com/watch?v=xX-yCO5Rzag[/link]
https://www.certarauniversity.com/dashboard
https://support.certara.com/forums/
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2009-02-04 14:05
(5531 d 10:12 ago)

@ SDavis
Posting: # 3179
Views: 9,160
 

 Lag time in WinNonlin - bug or feature?

Dear Simon,

❝ I'm also a little confused about the results that Helmut quoted above.


:confused: is a nice state of mind, isn’t it?

❝ Running that same data in 5.2.1 I get the result of 1 for Tlag in the third group which agrees with the definition in the help file. I don't recall this function being changed for some versions in WNL so is there a typo in your table […]



Oops, checked it again: you are right!

❝ […] so compartmental modelling is probably going to be more useful if you need to define this more 'precisely'.


Yes, but
  • PK modeling is not acceptable in the field of BE from a regulatory point of view.
  • In real life datasets absorption regularily is difficult to model. Quite often k01 is very inprecise; you have to play around with constraints (WinNonlin’s terminology: bounds) to get a ‘resonable’ value of tlag

❝ I concur it would be useful for WNL to 'intrinsically' recognise BLQs […]



Yes, please!

❝ […] the appropriate process is currently to recode non-numeric values first using (Tools > Status Codes).


An often overlooked feature… :-D

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
82 visitors (0 registered, 82 guests [including 7 identified bots]).
Forum time: 00:18 CET (Europe/Vienna)

Nothing shows a lack of mathematical education more
than an overly precise calculation.    Carl Friedrich Gauß

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5