Emrah Soner Özdeş
☆    

Turkey,
2014-11-21 08:25
(3415 d 08:13 ago)

Posting: # 13897
Views: 13,152
 

 Handling "Not Reportable" Data [Bioanalytics]

Hi All;

I have got a message from the analytical Lab I work with regarding a "not reportable" value as follows:

"This was a repeat value where the duplicate repeats were more than 20 percent different from one another and also differed from the original. Therefore according to the SOP it becomes Not Reportable."

Do you think is it safe to simply ignore that value during the PK analysis ?

Regards

Emrah Soner Özdeş


Edit: Category changed. [Helmut]
ElMaestro
★★★

Denmark,
2014-11-21 08:34
(3415 d 08:03 ago)

@ Emrah Soner Özdeş
Posting: # 13899
Views: 12,303
 

 Handling "Not Reportable" Data

Hi Emrah,

❝ "This was a repeat value where the duplicate repeats were more than 20 percent different from one another and also differed from the original. Therefore according to the SOP it becomes Not Reportable."


Why was it re-analysed ?

❝ Do you think is it safe to simply ignore that value during the PK analysis ?


Yes; it is likely safest, I think, to report the original value unless there is some other reason to consider the original value wrong.

Pass or fail!
ElMaestro
Emrah Soner Özdeş
☆    

Turkey,
2014-11-21 08:54
(3415 d 07:43 ago)

@ ElMaestro
Posting: # 13900
Views: 12,126
 

 Handling "Not Reportable" Data

Thanks for the prompt reply ElMaestro

Here is the situation:

Conc: 418.78
Re-analysis 1: 8.76
Re-analysis 2: 7.24
Means of Reanalyses : 8.00
Reported Conc: NR

NR = Not Reportable
Reason for re-analysis is Fault in sample processing
nobody
nothing

2014-11-21 09:07
(3415 d 07:30 ago)

@ Emrah Soner Özdeş
Posting: # 13901
Views: 12,191
 

 Handling "Not Reportable" Data

Educated guess, considering the values reported: The first concentration did not fit considering the PK of the compound ;-)

Maybe the lab staff screwed a little up with pipetting or there was something else leading to a contamination of the first analysis. What does the SOP of the CRO say for such a case? Completely ignore the plausible 2 repeats?

Kindest regards, nobody
Emrah Soner Özdeş
☆    

Turkey,
2014-11-21 09:18
(3415 d 07:19 ago)

@ nobody
Posting: # 13902
Views: 12,230
 

 Handling "Not Reportable" Data

Thanks for reply nobody:

It is apparent that there is a sample processing fault. Here is the related part in the study protocol:

"Individual samples will be re-assayed if at least one of the following criteria is met:
  1. Obtained study sample concentration is above the ULOQ. Such sample will be diluted with drugfree plasma and re-assayed.
  2. Internal standard and/or analyte response is not sufficient
  3. Fault in the sample processing
  4. Improper sample injection or malfunction of equipment
  5. Poor chromatography
  6. Identification of quantifiable analyte levels in pre-dose samples.
The individual samples selected for reanalysis that did not give a valid value (criteria a. – f.) will be reassayed in duplicate if sufficient plasma is available. Initial values, re-assayed results, finally accepted values and a justification for the acceptance will be reported. The individual samples selected for reassay will be assayed in a batch including calibration and QC samples.

I think neglecting that value in PK analysis will be the best practice as ElMaestro advised.
ElMaestro
★★★

Denmark,
2014-11-21 10:53
(3415 d 05:44 ago)

@ Emrah Soner Özdeş
Posting: # 13903
Views: 12,222
 

 Handling "Not Reportable" Data

Hi Emrah,

❝ I think neglecting that value in PK analysis will be the best practice as ElMaestro advised.


Careful, please. If you are submitting in a territory where outlier detection is the norm then perhaps you can neglect the original value after some number crunching gymnastics. But if you are not submitting in a territory where outliers testing is popular then you would perhaps need to either report the value however wrong it sounds, or, audit and self-inspect the study to see if you can verify that something really went wrong (and you need more than just an unexpected value for that); a verification of a problem would then allow you to report the re-analysed value.

Pass or fail!
ElMaestro
nobody
nothing

2014-11-21 12:11
(3415 d 04:26 ago)

@ Emrah Soner Özdeş
Posting: # 13904
Views: 12,151
 

 Handling "Not Reportable" Data

Which of those (a to f) did apply in this case to justify re-analysis of the sample?

Kindest regards, nobody
Emrah Soner Özdeş
☆    

Turkey,
2014-11-21 17:14
(3414 d 23:23 ago)

@ nobody
Posting: # 13908
Views: 12,044
 

 Handling "Not Reportable" Data

❝ Which of those (a to f) did apply in this case to justify re-analysis of the sample?


It is reported as "Fault in the sample processing".
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-11-21 17:45
(3414 d 22:53 ago)

@ Emrah Soner Özdeş
Posting: # 13909
Views: 12,227
 

 IS carry-over?

Merhaba Emrah!

❝ It is reported as "Fault in the sample processing".


OK, but how can you know that? Was anything documented in the lab journal like “I’m not sure whether I added the IS to this sample”? Can you give us:
  • The LLOQ and ULOQ of the method.
  • The concentration of samples analyzed in the batch before and after the doubt­ful value.
  • The concentration of samples of the subject’s profile before and after the doubt­ful value + their sampling time points.
  • The ranges of Cmax-values in the study.

❝ ❝ Conc: 418.78

❝ ❝ Re-analysis 1: 8.76

❝ ❝ Re-analysis 2: 7.24

❝ ❝ Means of Reanalyses : 8.00

❝ ❝ Reported Conc: NR


NR because “duplicate repeats were more than 20 percent different from one another”. That’s a matter of perspective:

Re-analysis 1: 8.76 (+21.0% of 7.24, –9.5% of x)
Re-analysis 2: 7.24 (–19.9% of 8.76, +9.5% of x)

Obviously the decision to not report the value was based on the first replicate – why not on the second? Looks like an ambiguity in the SOP. :cherry picking:
If ever possible (i.e., suf­fi­cient sample) run triplicates and use the median for assess­ment.

In which country do you want to submit the study? Bad luck for the EU and seemingly for Turkey as well.
However, I would explore potential impact of IS carry-over. Inject a sample spiked to a similar concentration like the sample in the batch before the doubtful value followed by a spiked sample without IS.

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
nobody
nothing

2014-11-21 18:19
(3414 d 22:18 ago)

@ Helmut
Posting: # 13910
Views: 12,064
 

 IS carry-over?

...but that should end up in category b, huh?

Kindest regards, nobody
Emrah Soner Özdeş
☆    

Turkey,
2014-11-24 09:58
(3412 d 06:40 ago)

@ Helmut
Posting: # 13921
Views: 11,974
 

 IS carry-over?

Hi Helmut;

I guess I will be able to answer your questions after I obtain the Analytical Report.

Thanks for reply
Emrah Soner Özdeş
☆    

Turkey,
2014-11-27 11:28
(3409 d 05:10 ago)

@ Helmut
Posting: # 13927
Views: 11,853
 

 IS carry-over?

Hi again Helmut. The previous values were from another study. Here are the values from the current study and the thing you ask for.

Conc: 0.00
Re-analysis 1: 2.63
Re-analysis 2: 2.15
Means of Reanalyses: 2.39
Reported Conc: NR

❝ ❝ It is reported as "Fault in the sample processing".


❝ OK, but how can you know that? Was anything documented in the lab journal like “I’m not sure whether I added the IS to this sample”?


I guess I wont be able to access lab journals.

❝ NR because “duplicate repeats were more than 20 percent different from one another”. That’s a matter of perspective:

❝ Re-analysis 1: 8.76 (+21.0% of 7.24, –9.5% of x)

❝ Re-analysis 2: 7.24 (–19.9% of 8.76, +9.5% of x)

❝ Obviously the decision to not report the value was based on the first replicate – why not on the second? Looks like an ambiguity in the SOP. :cherry picking:


In case of these values original value is 0.00 which makes the situation more interesting i guess. Flow Chart states that If duplicates differ from each other more than 20%, the next step is checking whether difference between each duplicate and original is more than 20%. According to Lab, the difference is more than %20 and hence the value is "NR". But in my opinion it is "Not Applicable" to find the percent difference from a value of "0.00"

❝ Can you give us:

❝ – The LLOQ and ULOQ of the method.


1.31 - 193.21 ng/mL

❝ – The concentration of samples analyzed in the batch before and after the doubt­ful value.

❝ – The concentration of samples of the subject’s profile before and after the doubt­ful value + their sampling time points.


0.0-0.00
0.8-2.52
1.5-10.79
3.0-32.81
4.5-61.07
5.0-69.93
5.5-76.10
6.0-76.64
6.5-83.04
7.0-89.39
7.5-84.73
8.0-88.83
10.0-89.91
12.0-64.55
24.0-17.11
48.0-4.74
72.0-NR

❝ – The ranges of Cmax-values in the study.


43.41 - 170.70 ng/mL

Regards
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-11-27 15:21
(3409 d 01:17 ago)

@ Emrah Soner Özdeş
Posting: # 13934
Views: 11,748
 

 Another story!

Hi Emrah,

❝ The previous values were from another study.


Yes, but it would have been nice to understand what was going on there.

❝ Here are the values from the current study and the thing you ask for.

❝ Conc: 0.00

❝ Re-analysis 1: 2.63

❝ Re-analysis 2: 2.15

❝ Means of Reanalyses: 2.39

❝ Reported Conc: NR


OK, another story. Conc = 0 post dose? I don’t believe it. Rather BLQ at 72 hours. Did the CRO report other concentrations as zero?

[image]

Why was the sample at 72 hours re-analysed? Without the 72 h-value the esti­mate of the elimination is not perfect, but acceptable (or has the CRO another stupid rule like “R²adj ≥0.95” in place?). From it we could expect the concen­tra­tion at 72 hours to be <LLOQ (~0.80). On the other hand it looks like the drug follows a two-compartment model. It would have been nice to have a sampling time at ~16 h. Based on the 24/48-concentrations the repeated values at 72 h seem to be reasonable.
But actually it is not important. The extrapolated part is <5% of AUC. I don’t get why the CRO repeated the value at all.

❝ ❝ ❝ It is reported as "Fault in the sample processing".

❝ ❝ OK, but how can you know that? Was anything documented in the lab journal like “I’m not sure whether I added the IS to this sample”?

❝ I guess I wont be able to access lab journals.


If you are the sponsor of the study perform an audit. If the CRO is not willing to show you the lab journals never do a study with them again.

❝ Flow Chart states that If duplicates differ from each other more than 20%, the next step is checking whether difference between each duplicate and original is more than 20%. According to Lab, the difference is more than %20 and hence the value is "NR". But in my opinion it is "Not Applicable" to find the percent difference from a value of "0.00"


All these flow-charts are crap anyway, IMHO.

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
Emrah Soner Özdeş
☆    

Turkey,
2014-11-28 11:00
(3408 d 05:37 ago)

@ Helmut
Posting: # 13944
Views: 11,600
 

 Another story!

Hi Helmut;

I would like to thank you for all your help on the subject.

My guess is that the lab decided to renalayse the sample because of the "0.00" post dosing concentration. All other values are expressed BLOQ instead of "0.00" which makes me think that they really did a mistake in sample processing. (blank injection maybe). So they reanalysed the sample.

If this is the scenario, avarage of the reanalyses which was "2.39" should be the reported value, i guess.

I will question the lab with regard to those points.

Thanks again

Regards
nobody
nothing

2014-11-28 11:18
(3408 d 05:20 ago)

@ Helmut
Posting: # 13945
Views: 11,585
 

 Another story!

...but from visible inspection the concentrations found with re-analysis make more sense as the third data point for linear regression than the point close to Cmax, whih is more likely not really in the terminal phase ;-)

I still don't get why the concentration was reported as ZERO for the first analysis, makes absolutely no sense at all. Would torture the lab to hand out the chromatograms of the whole analytical run for inspection by the sponsor...

Kindest regards, nobody
AngusMcLean
★★  

USA,
2014-11-28 20:09
(3407 d 20:28 ago)

@ Emrah Soner Özdeş
Posting: # 13954
Views: 11,524
 

 Handling "Not Reportable" Data

it seems to me that the value > 400 WAS discordant with respect to the pharmacoknetic trend of the values prior to it. As such it can be repeated as a pharmacokinetic outlier according to an appropriate SOP. If the two new values for the duplicate repeats carried out are within 20% of each other then according to appropriate SOP then they would be averaged and report this value. The >400 values is rejected as a pharmcokinetic outlier. You need an SOP policy in place to follow and in the analytical report all repeats are Tabulated together with the reason for repeats and final values. This report will reference the Lab. SOPs for repeats

Angus
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2014-11-28 20:16
(3407 d 20:21 ago)

@ AngusMcLean
Posting: # 13955
Views: 11,631
 

 EMA and similar regulations…

Hi Angus,

agree with everything you wrote. On this side of the pond common sense is (no more) the driving force behind (some) regulations. Repeating samples for PK reasons is not acceptable over here. A single value can screw up an entire study. :crying:

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
jag009
★★★

NJ,
2014-12-01 20:54
(3404 d 19:43 ago)

(edited by jag009 on 2014-12-02 16:07)
@ Helmut
Posting: # 13962
Views: 11,446
 

 EMA and similar regulations…

Hi Helmut,

On a side note (sort of). What if you have data from a subject (1 treatment) that is all BLQs? I mean the entire profile... No, there is no clinical evidence of cheating, vomiting etc etc...

Thanks
John
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,638 registered users;
77 visitors (1 registered, 76 guests [including 7 identified bots]).
Forum time: 16:38 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