Helmut
★★★

Vienna, Austria,
2021-12-01 13:09
(51 d 09:20 ago)

Posting: # 22680
Views: 439

## Manual integration [Bioanalytics]

Dear all,

recently I saw a deficiency of a European agency:

The Applicant should specify the integration method (automatic or manual) adopted to calculate chromatograms area for all submitted studies. If manual integration was adopted, the Applicant should specify the percentage of the total chromatograms and specify which samples (QC/calibrator or unknown) have been analysed.

Sure, that’s according to the 2011 guideline stating:

5.5. Integration
Chromatogram integration and re-integration should be described in a SOP. Any deviation from this SOP should be discussed in the analytical report. Chromatogram integration parameters and in case of re-inte­gration, initial and the final integration data should be documented at the laboratory and should be available upon request.

Instead of simply answering the question, i.e., that only automatic integration was used, the CRO additionally quoted from its SOP:

6.6 MANUAL INTEGRATION
Manual integration is PROHIBITED in any circumstance.

Oh dear, that’s bad scientific practice. Trust a silicon-based life-form more than a human brain?
An experienced analyst outperformed the software easily (esp. at low concentrations). Even ‘normal’ ones did so on the average.*
$$\small{CV\;(n=10)}$$$$\small{\begin{array}{lcc} \text{integration method} & \text{1 ng/mL} & \text{0.1 ng/mL}\\\hline \text{automated (smoothing 1, bunching 2)} & 6.5\% & 15.1\%\\ \text{manual correction (experienced analyst)} & 6.3\% & 11.1\%\\ \text{manual correction (ten analysts)} &5.2\% & 12.8\%\\ & (3.8\%-6.8\%) & (6.9\%-16.0\%)\\\hline \end{array}}$$ See also here (slides 26–29) and there (slides 24–31).

• Kirchherr H. Data Evaluation in LC-MS. In: Kuss H-J, Kromidas S, editors. Quantification in LC and GC. Wiley 2009; p 243–59.

Dif-tor heh smusma 🖖
Helmut Schütz

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

Denmark,
2021-12-01 14:40
(51 d 07:49 ago)

@ Helmut
Posting: # 22681
Views: 382

## Manual integration

Hi Helmut,

As I see it, FDA have been completely adverse to manual integrations for a decade at least. Manual integrations not accepted. At all. In any form.
And therefore many CROs have adopted similar policies - they have to run the validation and sample analysis without ever touching the MI function. Some CROs allow for MI if the intended submission is EU/ROW/WHO etc but do not open that option if the submission targets US.
The approach is, as I see it, completely safe: If you can run the val. and the sample analysis without MI then you have a great setup and you are not disturbing anyone. But still runs need to be approved and still someone should somehow sign off for the peak identification and correct integration. Which also means that it is a bit demanding from the equipment perspective. You need something which has a noise-less baseline, which means equipment sensitivity, colums integrity and method development is paramount.

I am aware of the wording in M10 which other suggests there may be a way for MI. I have not, hoever, heard of FDA taking any another stance in practice.

Finally, everything evolves, including integration algorithms. Perhaps what was pretty bad in 2009 is pretty good in 2021? There is no good way to evaluate it. I don't have a way to import WIFF files from 2009 into Analyst 1.6.5 or whatever today's version is. But if someone has a way to it, let us see some results

So, I am fine with CROs trying to avoid MI as long as what it is meant is the aim to have a method that is so sensitive that even at low concs the integration is good. That also means the deficiency letter is a walk in the park.

Pass or fail!
ElMaestro
Helmut
★★★

Vienna, Austria,
2021-12-01 17:38
(51 d 04:52 ago)

@ ElMaestro
Posting: # 22682
Views: 373

## Manual integration

Hi ElMaestro,

» As I see it, FDA have been completely adverse to manual integrations for a decade at least. Manual integrations not accepted. At all. In any form.

That’s bad. Well, it saves the assessor time.

» The approach is, as I see it, completely safe: If you can run the val. and the sample analysis without MI then you have a great setup and you are not disturbing anyone. But still runs need to be approved and still someone should somehow sign off for the peak identification and correct integration. Which also means that it is a bit demanding from the equipment perspective. You need something which has a noise-less baseline, which means equipment sensitivity, colums integrity and method development is paramount.

Fully ACK. However, let’s face it: If you have to measure really (‼) low levels, ‘a noise-less baseline’ belongs still to the realm of science fiction.

» I am aware of the wording in M10 which other suggests there may be a way for MI.

I’m absolutely fine with Section 3.6.3 of the GL:

Chromatogram integration and reintegration should be described in a study plan, protocol or SOP. Any deviation from the procedures described a priori should be discussed in the Bio­ana­lytical Report. The list of chro­matograms that required reintegration, including any manual integrations, and the reasons for reintegration should be included in the Bioanalytical Report. Original and reintegrated chroma­to­grams and initial and repeat integration results should be kept for future reference and submitted in the Bioanalytical Report for comparative BA/BE studies.

» I have not, hoever, heard of FDA taking any another stance in practice.

I believe it.

» Finally, everything evolves, …

All too true.

» … including integration algorithms. Perhaps what was pretty bad in 2009 is pretty good in 2021? There is no good way to evaluate it.

No idea. A CDS is a pitch-black box with Schrödinger’s cat lurking inside. In my CRO I had access to all raw data (peak slices). Though the vendor claimed which Al Gore Rhythms were implemented (without giving great detail), I never was able to reproduce it. Tried a lot. Gave up.

» So, I am fine with CROs trying to avoid MI as long as what it is meant is the aim to have a method that is so sensitive that even at low concs the integration is good.

Sure. I was only concerned about the sentence ‘Manual integration is PROHIBITED in any circumstance.’

Reminded me on:

Medical statistician: One who will not accept that Columbus discovered America…
because he said he was looking for India in the trial plan.
Stephen Senn

» That also means the deficiency letter is a walk in the park.

It was. (1) for the EMA and (2) only automatic anyhow.

Dif-tor heh smusma 🖖
Helmut Schütz

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