Imran
☆

Mumbai,
2007-06-27 06:02
(5273 d 12:02 ago)

Posting: # 835
Views: 23,777

## Steady state calculation using WinNonlin [Software]

Dear PK Colleagues

I have come across a problem while runing PK of steady state using WinNonline software. My queries are:
1. should trough concentrations and concentration of profiling day should be run separately or in the same set of data to generate PK parameters?
2. Should only trough values should be used for calculation of % fluctuation?
Please suggest, if anyone having the insight in running the PK of Steady state using WinNonlin.

Thanking you in Anticipation

Regards
Dr.Imran Khan

Dr.Imran Khan
Helmut
★★★

Vienna, Austria,
2007-06-27 11:56
(5273 d 06:09 ago)

@ Imran
Posting: # 840
Views: 17,233

## Steady state calculation using WinNonlin

Dear Imran!

» 1. should trough concentrations and concentration of profiling day should
» be run separately or in the same set of data to generate PK parameters?
• If you are talking about PK modeling, all values should be included in the data set. But beware: in most (all?) situations WinNonlin's stripping procedure does not provide intial estimates; they have to be provided by the user.
• If you are talking about NCA, beware of 'PK/PD/NCA Modeling Questions FAQ #2' form Pharsight's support site:
Q: When WinNonlin gives an AUC following multiple doses, is this the total AUC or the AUC for the first dose?
A: When using Noncompartmental Modeling, the AUC given in the Final Parameters table is the AUC from 0 to infinity that would occur if only the first dose had been given.
Also beware of (quoting the manual):

When using steady-state data, WinNonlin computes AUC from dose time to dose time + tau, based on the tau value set above. However, in most studies, there will be sampling time deviations. That is, if dose time = 0 and tau = 24, the last sample might be at 23.975 or 24.083 hours. In this instance, the pro­gram will estimate the AUC based on the estimated concentration at 24 hours, and not the concentration at the actual observation time. For steady state data, Cmax, Tmax, Cmin and Tmin are found using observations taken at or after the dose time, but no later than dose time+tau.

» 2. Should only trough values should be used for calculation of % fluctuation?

%fluctutation (or %Peak-to-Trough Fluctuation, %PTF) is defined as:
$$\small{\%PTF=100\frac{C_{\textrm{max}}-C_{\textrm{min}}}{C_{\textrm{av}}}}$$
where $$\small{C_{\textrm{av}}=AUC_{\tau}/\tau}$$, and $$\small{C_{\textrm{max}}}$$, $$\small{C_{\textrm{min}}}$$ refer to extremes of the profile in steady state.

In WinNonlin you have to calculate %PTF manually (it's not given in the result workbook).

Dif-tor heh smusma 🖖
Helmut Schütz

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

Mumbai,
2007-06-29 05:13
(5271 d 12:52 ago)

(edited by Jaime_R on 2007-06-29 10:01)
@ Helmut
Posting: # 844
Views: 17,215

## Steady state calculation using WinNonlin

Dear HS
Thanks for ur mail and valuable suggestion on the issue. I data I want to estimate for SS study is given below.
 Subject Period Trough sample      Profiling till 72.00 hrs                   Day 1   Day-5   Day-6   Day-7  Day-7                   0.0 hr  0.0 hr  0.0 hr  0.0 hr  1.0 hr   2        2      BLQ    13.344  13.689  12.272   8.427   3        1      BLQ     2.408   2.88    3.299   2.944   4        1      BLQ     6.068   7.562   5.267   4.734   5        1      BLQ     4.711   6.01    6.407   5.082   6        2      BLQ     5.594   5.134   4.733   3.691   7        2      BLQ     4.043   5.055   3.842   3.867   8        1      BLQ     6.104   6.033   6.144   5.251   9        2      BLQ     7.272   8.825  10.937   6.892  10        2      BLQ     2.671   3.53    3.655   2.744  11        1      BLQ     2.238   6.671   3.635   3.092  12        2      BLQ     8.961   4.591   5.078   4.127  13        1      BLQ     9.957  11.607   8.85    7.571  14        2      BLQ     2.596   3.972   2.756   2.696  15        1      BLQ     5.578   5.959   4.593   3.244  16        2      BLQ     4.135   5.602   6.656   5.397  17        1      BLQ     3.339   3.985   2.957   3.241  18        2      BLQ     7.984   6.268   4.776   3.843  19        1      BLQ     5.478   5.781   5.357   4.043  20        2      BLQ    17.672  21.912  24.53   22.566

The step I followed calculation of PK Parameters is as:
PK/PD/NCA Analysis Wizard , noncompartment modeling, select model 200, data variables, dosage regimen and selecting steady state followed by modeling, generation of PK output.

The XL file I have attached, the data processed is excluding trough sample (day1, 5, 6 and 7), should I process trough data and profiling data separately.

As u wrote in ur mail that in WinNonlin %PTF to be calculate manually, but here in my case I calculated all of the primary and secondary parameters (Excel file attached).

I have troubleshooting how the data to be presented in the WinNonline, I tried in all the possible way, but I am not sure about the output.

Primary Efficacy Parameters: Cmin, Cmax, AUC0 - &#61556;

Secondary Efficacy Parameters: Tmax, % Fluctuation

Regards
Dr.Imran Khan

Dr.Imran Khan
Jaime_R
★★

Barcelona,
2007-06-29 10:29
(5271 d 07:36 ago)

@ Imran
Posting: # 845
Views: 16,955

Dear Imran,

I edited your post (took me half an hour), because it came out as a mess.
Don't use <tabs>, use [ code]with blanks between columns[/code] instead.

» The XL file I have attached [...]

What's 'XL'? Do you mean Micro\$oft's Excel?
Although the Forum's software supports Uploads, this function is deactivated.
Attachments are not possible anyhow (this is not an e-mail list).

» [...] (Excel file attached).

OK, I was right, but just out of couriosity: what did you actually try, to 'attach a file'?

» I have troubleshooting how the data to be presented in the WinNonline, I tried in all the possible way, but I am not sure about the output.

Why don't you contact Pharsight's Support?

» Primary Efficacy Parameters: [...] AUC0 - &#61556;

'#61556' is in the private code range of UTF-8. No greek letters, Unicode, or ISO-8859 above DEC 255 are supported. You would have noticed this during Preview. Maybe you mean the greek letter 'tau' denoting a dosage interval (Unicode #964).

Don't take it personally, but I think the Forum should not become a Fastfood Restaurant (quickanddirty).

Regards, Jaime
Imran
☆

Mumbai,
2007-06-30 12:50
(5270 d 05:15 ago)

(edited by Jaime_R on 2007-06-30 13:22)
@ Jaime_R
Posting: # 848
Views: 16,937

Dear Jaime

Extremely sorry! I tried to send the data in best possible way, even checked the preview, but I could not trace attachement option in the mail

I have definately wrote a mail to Pharsight, but as a matter of discussion I wrote to Forum.

Even I am agree that the Forum should not become a Fastfood Restaurant

Best Regards,
Imran

--
Edit: Full quote removed. [Jaime]
Jaime_R
★★

Barcelona,
2007-06-30 13:33
(5270 d 04:31 ago)

@ Imran
Posting: # 849
Views: 16,854

Dear Imran!

» Extremely sorry!

OK, my reaction was also a little bit too harsh...

» [...] but I could not trace attachement option in the mail

Yes, because this is not a Mailing List, but a Forum.
File uploads may be possible in the new version.

» I have definately wrote a mail to Pharsight,[...]

Expect an 'RTFM'...

» [...] but as a matter of discussion I wrote to Forum.

Yes, that's fine!

Regards, Jaime
Helmut
★★★

Vienna, Austria,
2007-06-29 21:08
(5270 d 20:56 ago)

@ Imran
Posting: # 847
Views: 17,168

## Steady state calculation using WinNonlin

Dear Imran!

» The step I followed calculation of PK Parameters is as: PK/PD/NCA Analysis Wizard , noncompartment modeling, select model 200, data variables, dosage regimen and selecting steady state followed by modeling, generation of PK output.

Yes fine, but the order in Dosage Regimen is:
Dose
Time of Last Dose
Tau

» The XL file I have attached, the data processed is excluding trough sample (day1, 5, 6 and 7), [...]
Sure, you are interested in the profile (WinNonlin simply ignores the saturation phase, which is fine).

» should I process trough data and profiling data separately.

In versions 5.1+ you may run a linear regression of e.g., the last three trough values to check whether you have reached steady state (95% CI of the slope should include zero).
Model 502 (Linear) of example's data below give:
Last 4 troughs (24 h - 96 h): slope 0.063, CI -0.049, +0.176
Last 3 troughs (48 h - 96 h): slope 0.019, CI -0.120, +0.158
...so you may assume steady state conditions.

» As u wrote in ur mail that in WinNonlin %PTF to be calculate manually, but here in my case I calculated all of the primary and secondary parameters [...]

Oh, you are right!

» I have troubleshooting how the data to be presented in the WinNonline, I tried in all the possible way, but I am not sure about the output.

Why don't you keep it simple and check the results manually?
OK, I've done that for you...

D=100, V/F=1, k01=0.6931, k10=0.06931, tau=24h for five days gives (rounded to 1 decimal place):
time conc  0    0.0  24  21.1  48  25.0  72  25.8  96  25.9  98  91.5  99  97.4 100  96.9 102  88.7 104  78.3 106  68.4 109  55.7 112  45.2 116  34.3 120  26.0

Now let's have a look at the output (Non-Transposed Final Parameters Tab), and a 'manual' calculation
          WinNonlin5.2  'manual' AUClast       1425.4     1425.4 Cmax            97.4      97.4 Cmin            25.9      25.9 Tmin            96        96 Cav             59.4      59.4 %Fluctuation   120.3873  120.3873 AUCtau        1425.4    1425.4 AUCall        1425.4    1425.4

Fine.
Please note, that AUClast = AUCtau = AUCall.
Generally nobody is interested in the sampling beyond the last administration + tau, but in your case obviously this was the case (+72 hours).

If we add two more samples (144 h, 168 h), WinNonlin's output might get a little bit confusing for a novice, since:
time conc   0   0.0 ...   ... 120  26.0 144   4.9 168   0.9

          WinNonlin5.2  'manual' AUClast       1865.8    1865.8 Cmax            97.4      97.4 Cmin            25.9      25.9 Tmin            96        96 Cav             59.4      59.4 %Fluctuation   120.3873  120.3873 AUCtau        1425.4    1425.4 AUCall        1865.8    1865.8

WinNonlin is 'clever' enough to keep AUCtau (and therefore Cav, and %Fluctuation correct within tau), but please note AUClast # AUCtau and AUCall!

If we sample for another 48 hours until we get a BLQ:
time conc   0   0.0 ...   ... 168   0.9 192   0.2 216   BLQ

          WinNonlin5.2  'manual' AUClast       1879.0    1879.0 ...              ...       .. AUCtau        1425.4    1425.4 AUCall        1881.4    1881.4

Now finally we ran into AUClast # AUCtau # AUCall!
AUCall adds a triangle from the Tlast to the next sampling time point, assuming C=0. If you know of any serious reference using AUCall, please tell me. It's Pharsight's invention and IMHO useless - I removed it from my templates...
It's also funny that Pharsight's business rival ThermoScientific added this 'metric' to Kinetica as well.

In short:
• Sample only until last administration + tau,
• WinNonlin gives the correct result for %Fluctuation,
• Don't get confused; it's only AUCtau you are interested in.

» Primary Efficacy Parameters: Cmin, Cmax, AUC0 - &#61556;
» Secondary Efficacy Parameters: Tmax, % Fluctuation

In BA/BE we should not talk about Efficacy...

Dif-tor heh smusma 🖖
Helmut Schütz

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

Mumbai,
2007-08-02 13:52
(5237 d 04:12 ago)

@ Helmut
Posting: # 957
Views: 16,823

## Steady state calculation using WinNonlin

Dear HS

Thanks for the solutions. I worked with WinNonlin and I am through with my results.

One more Issue I would like to discuss with u.
How can we calculate or prove that SS is achieved.
I had used SAS for the same by repeated measures of ANOVA for Cpd calculation, and withdrawn the conclusion based on time* formulation interaction (p value.)

What procedure to be followed by using Winnonlin 5.0.1. or any other suitable method can be used to prove SS is achived.

Pls comment.

Best Regards
Dr.Imran

--
Edit: Full quote removed. Please see this post! [HS]

Dr.Imran Khan
Helmut
★★★

Vienna, Austria,
2007-08-02 15:07
(5237 d 02:58 ago)

@ Imran
Posting: # 958
Views: 16,892

## Steady state calculation using WinNonlin

Dear Imran!

» How can we calculate or prove that SS is achieved.
» I had used SAS for the same by repeated measures of ANOVA for Cpd calculation, and withdrawn the conclusion based on time* formulation interaction (p value.)

In never understood this procedure. OK, statistically yes, but would you really throw away an entire study based on such a result? I always preferred it to look at the time course of the subjects and eventually exclude only ones who were not in steady state on the profile day. Of course you have to state this procedure in the protocol.

» What procedure to be followed by using Winnonlin 5.0.1. or any other suitable method can be used to prove SS is achived.

For my personal suggestions see this tread.

If you want to perform the analogue of SAS' repeated measures ANOVA, you will have to fight yourself through WinNonlin's Linear Mixed Effects Modeling feature - and if you succeed, come back and tell us how you did it!

Dif-tor heh smusma 🖖
Helmut Schütz

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

Mumbai,
2007-08-06 11:48
(5233 d 06:17 ago)

(edited by Jaime_R on 2007-08-06 12:24)
@ Helmut
Posting: # 965
Views: 16,916

## Steady state calculation using WinNonlin

Dear HS and Colleagues
Hi!

In SS study Cpd Analysis should we calculate SS acheived for each subject, or it should be calulated as a mean of all observation.

What if a subject 'A' has not acheived SS, sholud the subject be included in the PK analysis.

I understand we calculate ABE, but would be it questioned if any subject is not in SS state.

Best regards

Dr.Imran

--
Edit: Full quote removed. Please see this post! [Jaime]

Dr.Imran Khan
Helmut
★★★

Vienna, Austria,
2007-08-06 22:16
(5232 d 19:49 ago)

@ Imran
Posting: # 966
Views: 17,051

## Exclusion of subjects who are not in SS

Dear Imran!

» In SS study Cpd Analysis should we calculate SS acheived for each subject, or it should be calulated as a mean of all observation.

As already told in this thread, my personal preferences are always to assess achievement of SS for individual subjects.

» What if a subject 'A' has not acheived SS, sholud the subject be included in the PK analysis.

IMHO this subject should be excluded, because if the subject was not in steady state, the superposition principle of pharmacokinetics is not fulfilled.
You may only make valid statements (about accumulation, etc.) if
AUCinf (single dose) = AUCtau (steady state).

» I understand we calculate ABE, but would be it questioned if any subject is not in SS state.

At least I would do so!

Dif-tor heh smusma 🖖
Helmut Schütz

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

India,
2009-04-13 14:34
(4617 d 03:30 ago)

(edited by Ohlbe on 2009-04-13 22:30)
@ Helmut
Posting: # 3516
Views: 15,618

## Exclusion of subjects who are not in SS

Dear Helmut,

Adding to that, if dosing interval for reference product is 24 hours and that of test product is 12 hours i.e. tau=24 hrs. for reference and tau=12 hrs. for test product then how can we calculate AUCtau ? means if we put tau=12 hrs. for test product in the model then which AUC area will be estimated by WinNonlin i.e first 0-12 hrs or last 0-12 hrs ?

Moreover, if we calculate AUC0-24 for both test and reference in above case then how WinNonlin calculates ? Should i give common dosing interval i.e. tau=24 hrs. for both test and reference products and how much AUC0-24 calculation is reliable for test product?

Thanks and regards,

KR
--
Edit: Full quote removed. Please see this post! [Ohlbe]
vixen
☆

Czechia,
2020-04-27 10:10
(585 d 07:54 ago)

@ Helmut
Posting: # 21350
Views: 5,365

## Exclusion of subjects who are not in SS

Hello,

I like Helmut's suggestion to assess achievement of SS for individual subjects. I wonder though whether I should evaluate the Cpds on a normal or logarythmic scale. What is your approach?

Thanks,
vixen
Helmut
★★★

Vienna, Austria,
2020-04-27 11:45
(585 d 06:20 ago)

@ vixen
Posting: # 21351
Views: 5,350

## Common sense instead of statistics

Hi vixen,

» I like Helmut's suggestion to assess achievement of SS for individual subjects.

My suggestion dates back to 2007. A while ago…
I don’t use it since 2010 and don’t recommend it any more (see this presentation, slide 14).

» I wonder though whether I should evaluate the Cpds on a normal or logarythmic scale.

Normal – otherwise you cannot deal with BQLs and missings.

See the presentation.

Dif-tor heh smusma 🖖
Helmut Schütz

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