kjeldahl method validation

[ydna1977]

Hi

Anybody know how to go about validating a kjedahl method

ie ICH guidelines , accuracy linearity etc or not applicable here

simple intermediate precision with 2 analyst perhaps ?

[HPLC chemist]

Why are you trying to validate the Kjeldahl method? As I recall it is used for feeds, foods, and water (TKN or Total Kjeldahl Nitrogen)!

[dblux_]

Why are you trying to validate the Kjeldahl method? As I recall it is used for feeds, foods, and water (TKN or Total Kjeldahl Nitrogen)!

No matter what it is used for. It may be ISO 17025 accredited and in this case has to be validated.

[HPLC chemist]

Dblux,

I disagree. EPA regulations superseed ISO 17205 for the TKN in water and wastewater. Similarly, foods and feeds are regulated by AOAC methodology.

However, the TKN is used in pharmaceutical raw materials (drug substance). In fact, it is required in USP monographs and may be 'verified (not validated)' by sections <1224>, <1225>, or <1226>.

[dblux_]

But EPA regulations have no power in Europe.
Here analytical methods are accredited for compliance with ISO 17025.

[ydna1977]

Hi

Thanks for you responses
This validation or verification is in a GMP environment
Quantitative results and determinations are obtained so following USP <1225> I assume , though where and to start
My background in GC/HPLC/dissolution development and validation

[Blazer]

I've had this method validated for determining percent nitrogen in peptide APIs. We validated following ICH Q2(R1) for specificity, linearity, repeatability, intermediate precision, accuracy, and robustness.

[ydna1977]

Hi
Thanks for your feedback
Any chance you could expand on the ICH element and procedures

ie
How did you carry out specificity , linearity

Acceptance criteria for this

Much appreciated

[Blazer]

Specificity: I've done this in two difference ways. One is to spike a sample with a non-nitrogen containing compound and show that the percent nitrogen results were not different. The other is to analyze a nitrogen containing compound and a non-nitrogen containing compound and demonstrate the difference in results.

Linearity: Analyze different weights of sample and measure the observed amount of nitrogen. Plot the linear regression. Correlation coefficient needed to be ≥ 0.99.

Repeatability: Analyze six samples, calculate % RSD. Our acceptance criterion is typically ≤ 5%, but depending on your instrument you may be able to set this tighter.

Intermediate Precision: 2nd analyst analyzes six samples, calculate % RSD of these six, overall % RSD (all 12 results from the two analysts), and an absolute difference between the two analysts of ≤ 0.5%.

Accuracy: Spike sample with another nitrogen containing compound at three different levels. We used acetanilide to spike. Calculate measured nitrogen versus expected nitrogen for % recovery. Our acceptance criterion was 97-103%.

Robustness: We studied the variability in the results when the sample was weighed out in environments of different humidity since our sample was known to be slightly hygroscopic.

[ydna1977]

Hi .
Thanks for your reply , much appreciated
This is making a bit more sense with respect to ICH Q r1 guidelines

to confirm (my understanding)

Specificity - OK
Linearity - weigh 30-120% of typical sample as detailed in method
prepare in duplicate ?
Repeatability - OK
Intermediate precision - OK
Accuracy - 80,100, 120% level ? amounts of acetanilide ?
in triplicate ?
Robustness - understood (depends on sample material I guess)


Thanks

[Blazer]

Linearity - weigh 30-120% of typical sample as detailed in method
prepare in duplicate ?

It's not an impurity method, so I don't think you'd need to have your linearity cover that far. You could probably get away with the typical 80-120%. If you want an expanded range, 50-150%. Duplicate, triplicate, or a single curve is up to you and your internal SOPs.

Accuracy - 80,100, 120% level ? amounts of acetanilide ?
in triplicate ?

Yes, that would be how we did it. Alternatively per ICH Q2(R1), you can determine accuracy using a known reference standard so if you can find one where the percent nitrogen has been determined, you can show that you get the same result (or reasonably close based on analytical variability) using this method.

[ydna1977]

Hi ,
Thanks

Making a bit more sense now , much appreciated

No doubt I will have more questions in the future on this

[DR]

Linearity - weigh 30-120% of typical sample as detailed in method
prepare in duplicate ?

It's not an impurity method, so I don't think you'd need to have your linearity cover that far. You could probably get away with the typical 80-120%. If you want an expanded range, 50-150%. Duplicate, triplicate, or a single curve is up to you and your internal SOPs.

Accuracy - 80,100, 120% level ? amounts of acetanilide ?
in triplicate ?

Yes, that would be how we did it. Alternatively per ICH Q2(R1), you can determine accuracy using a known reference standard so if you can find one where the percent nitrogen has been determined, you can show that you get the same result (or reasonably close based on analytical variability) using this method.

I do believe that Leucine is popular for shaking down a Kj analysis procedure and associated equipment (steam distillation/titration combination device). It is neither the easiest nor the hardest of the naturally occurring AAs to shake a nitrogen atom out of, so it is a decent test. Also, it is available in highly pure, well characterized gram quantities. The manual that accompanies your Kj apparatus should provide a useful range over which they consider the thing to be accurate, so I recommend using that as your guide for setting your ± range for a given sample.
Approach this the same way you would a Karl Fischer method (with a lengthy, hazardous sample prep)... limited linear range, pretty accurate once you get over a certain minimum titrant amount.

[ydna1977]

Hi ,

Thanks for you response
Treating like a KF validation does make sense

Do you think the below elements and acceptance criteria are suitable :

Linearity - will be assessed by preparing five solutions over the range 50 to 150% of the nominal working concentration /sample weight
Acceptance criteria - > 0.99
Accuracy - will be assessed by preparing three sets of sample solutions/ref std at 80%, 100% and 120% of the specification limit/sample weight.
OR
spiked solutions( with known amount of acetanilide) will be prepared in triplicate giving a total of nine spiked solutions (three solutions at each level; 80%, 100% and 120% nom weight
Acceptance criteria (nitrogen) - 95-105% with % RSD (n=3) < 5%
Precision – 6 preps of sample at 100% method weight
Acceptance criteria (nitrogen content) - % RSD (n=6) < 5%
Intermediate precision – as Precision by different analyst on a different day and acceptance criteria (n=6 preparations)
Acceptance criteria - Each analyst’s individual %RSD must be ≤20% and the mean of each analyst’s results must be within ±20% (relative) of the overall mean.

Specificity - Spike a sample with a non-nitrogen containing compound and show that the percent nitrogen results were not different.
Analyse a nitrogen containing compound and a non-nitrogen containing compound and demonstrate the difference in results
Robustness – Depends on sample material (hygroscopic – look at humidity, different environments) or solution stability testing over time ?


Thanks

[Nikita]

Hello!
We are doing the same validation on amino acid complex (API), so I have several questions:
1) What solvent did you use to dissolve acetanilide?
2) What concentrations and volumes of acetanilide solution did you add to sample object?

Thanks for your reply.