**8. Quality assurance and an uncertainty of result**

Each laboratory should possess a program of quality assurance of its analyses within good laboratory practice. In case of chromatographic methods steering the quality may be implemented through performing one or more of the activities listed below:


180 Chromatography – The Most Versatile Method of Chemical Analysis

concentration of the standard used for calibration.

Analyte Matrix LOQ CVr

Feedingstuff

Vitamin K3

Vitamin D3

Vitamin B1

Vitamin B2

Canthaxanthin

analyte. With chromatographic methods, the bottom limit of the method's application may be also regarded as the content of the analyzed component, which is equal to the lowest

%

Premixture 1.4 2.0 95.2 r=0.999

Premixture 1.0 2.0 96.4 r=0.999

r=0.999 Premixture 5.7 - 99.4

Premixture 200 IU/g 1.4 1.7 99.3 1.06-10.68 μg/ml; Preparation 1.3 1.3 98.4 r=0.999

Feedingstuff 1.0 mg/kg 5.6 98.9 0.2-1.0 μg/ml; Premixture 3.7 102.3 r=0.999

Feedingstuff 1.0 mg/kg 3.4 5.1 98.0 0.17-0.67 μg/ml; Premixture 2.3 6.2 98.3 r=0.999

%

Tryptophan Feedingstuff 10.0 mg/kg 4.1 4.0 94.9 12.5-100 nmol/l;

Ethoxyquin Feedingstuff 0.5 mg/kg 2.0 6.0 99.0 0.01-0.07 μg/ml;

MHA Feedingstuff 0.05% 2.8 - 96.7 0.05-0.45 mg/ml;

During the validation process in a laboratory the precision of a method is determined through examining such parameters as repeatability and within-laboratory reproducibility (intermediate precision). Within-laboratory reproducibility may be calculated on the basis of control charts or from the range between parallel results of an analysis (replications) of a feed additive, in compliance with the Nordtest [23] handbook. For two or more replications for the analyses of an analyte in each sample it is necessary to calculate the mean value, the difference between measurements (range), relative difference in % and next mean relative difference (%) for all samples of a particular type of feed. The mean range divided by the

Feedingstuff 1.0 mg/kg 4.7 7.9 97.3 0.7-8.5 μg/ml; Premixture 3.3 6.0 98.2 r=0.999

Preparation 1.0 1.4 99.7 r=0.999

Vitamin A Feedingstuff 1000 IU/kg 1.6 4.0 96.0 7.0-70 IU/ml;

Vitamin E Feedingstuff 6.0 mg/kg 2.0 2.0 96.7 0.05-0.3 mg/ml;

1.0 mg/kg

CVr – coefficient of variation; CVip – intermediate precision; rec. - recovery

Analyte Matrix LOQ CVr

**Table 8.** Validation parameters obtained for other feeds

Preparation 1.9 - 101.2

**Table 7.** Validation parameters obtained for selected feed additives – vitamins in feeds

CVip

6.4 - 100.9

CVip %

Rec.

% Linear range

r=0.999

r=0.997

% Rec. % Linear range

0.046-4.62 μg/ml;


Control material may be provided by certified reference material, CRM (matrix + analyzed substance), material from proficiency testing with a value assigned, enriched material prepared in the laboratory (fortified sample) and control material with recognized content of the tested and stable in time component, previously determined in the laboratory.

In compliance with the recommendations of the EN ISO/IEC 17025:2005 [24] standard and requirements defined in some regulations, in order to assess and interpret the result of a test, it is necessary to use the uncertainty of measurement. We hardly ever know the real content of the analyte and the result of the test is biased with an error. Hence, the idea of "uncertainty of measurement" has been introduced which is defined as "a parameter associated with the result of a measurement, that characterises the dispersion of the values that could reasonably be attributed to the measurand" [26]. The EN ISO/IEC 17025:2005 standard recommends at point 5.4.6.2 that testing laboratories should possess and make use of procedures for assessing the uncertainty of measurement. To assess the uncertainty of methods used to analyze feeds the most frequently used approach is the model, consistent with the GUM [26] guidebook, which consists of finding the components of uncertainty and uses the law of error propagation. Using this particular approach to assess uncertainty, it is

possible to obtain an underestimated value in case when we do not consider all the components. Other reasons for underestimating uncertainty during validation include a situation when while assessing uncertainty repeatability instead of within-laboratory reproducibility is taken into account or we often forget to consider the bias. Moreover, uncertainty assessment is done during validation when well-ground typical samples are analyzed in a short time, new standards are prepared, the apparatus is controlled (standard conditions). During routine activities we analyze various matrices and obtaining homogeneity is not easy in case of some samples. The conditions mentioned above may affect underestimation of uncertainty. That is why it is important to have a possibility to verify uncertainty with the help of other approaches.

New opportunities concerning verification and assessment of uncertainty can be found in the Eurolab [27] technical report, the Nordtest [23] handbook and in the paper [28] which recommend the use of experimental approaches in assessing uncertainty of laboratory methods, in particular:


Using the within-laboratory experimental approach in assessing uncertainty of measurement (u) within -laboratory reproducibility (s) is considered as a measure of precision, as well as the bias (b), in accordance with the equation below following the Eurolab [27] technical report.

$$
\sqrt{\mathbf{s}\_{Rw}^{\cdot^2} + \mathbf{b}^2} = \boldsymbol{\mu} \tag{1}
$$

Using High Performance Liquid Chromatography (HPLC) for Analyzing Feed Additives 183

= Δ (3)

<sup>2</sup> ( )*<sup>i</sup> bias n*

When certified reference materials are lacking (a frequent situation) and when no other analyses of bias have been performed in the laboratory (e.g. prior to applying the reference

A laboratory participating in PT may use the results of such tests in order to assess uncertainty of measurement for the testing method/procedure used. Similarly to determining uncertainty in within-laboratory experimental approach, the uncertainty of measurement (u) is equal to the root of the sum of squared values of standard deviation for within-laboratory reproducibility sRw and the bias (b), which can be calculated from the formulas 1,2 and 3.

With this approach two components of uncertainty are obtained from different sources. Precision is determined on the basis of the authors' own validation data (within-laboratory reproducibility), from the range or on the basis of measuring control charts (in-house). The bias is determined on the basis of PT results. Estimating the bias on the basis of a single participation in PT may have a limited range and should be treated as preliminary. If the data from several PTs are available (a wider range of matrices and concentrations) the

The results of analysing uncertainty on the basis of experimental approaches using the results of the authors' own results are presented below along with, for comparison, expanded uncertainties estimated with the help of Horwitz formula RSDR (%) = 2C-0.15 .

Additive Feed *sw* (%) bias (%) u (%) U = 2 ∙u (%) U (%) \* Vitamin A Feedingstuffs 4.0 12.4 13.1 26.2 23.8

Vitamin E Feedingstuffs 1.0 9.0 9.1 18.2 16.1

\*Expanded uncertainty for the HorRat value H=1 calculated from the Horwitz' formula RSDR=2 C-0,15; U (%) = 2 RSDR **Table 9.** Results of uncertainty evaluation for some feed additives in compound feeds and premixes

The chapter presents a brief review of the methods used for determining feed additives by means of high proficiency liquid chromatography, HPLC. The authors presented their own research procedures and special attention was given to the preparation of samples for testing, extraction, extract purification, chromatographic separation and the basic elements

Premixes 2.0 7.2 7.5 15.0 11.8

Premixes 2.0 6.1 6.4 12.8 8.2

Feedingstuffs 5.6 6.7 8.7 17.4 26.8 Premixes 3.7 6.7 7.6 15.2 18.6

Feedingstuffs 6.52 3.16 7.2 14.4 24.0 Premixes 5.09 3.16 6.0 12.0 10.7

method) bias can be estimated on the basis of proficiency testing, PT.

assessment of the bias may be referred to the complete measurement range.

Vitamin B1

Vitamin B2

**9. Conclusion** 

of method validation and quality control.

Precision of a research procedure in a laboratory is determined during validating the method or on the basis of measurements noted in control charts. Validation usually includes determining standard deviation of within-laboratory reproducibility srw or standard deviation of within -laboratory reproducibility sRw which is sometimes called intermediate precision. Bias is determined by means of standard deviation of measurement in relation to the reference value, e.g. while examining certified reference materials, using reference methods. The share of bias (b) in uncertainty of the measurement is determined with the help of mean deviation of measurements (Δ), uncertainty of reference material (uref ) and precision of measurements during examining that bias (s). The value of the expression s2 / n with a bigger number of measurements is low and can be omitted:

$$
\sqrt{\Delta^2 + \mu^2 \frac{\left(s\right)^2}{ref} + \frac{s^2}{n}} = b
\tag{2}
$$

In practice, different measurements result in different values of bias. In such a case the data may be combined and common assessment of a value of bias (uw) may be determined as a function of the measured value or, for typical data of matrices and levels, according to the formula below:

Using High Performance Liquid Chromatography (HPLC) for Analyzing Feed Additives 183

$$\sqrt{\frac{\sum (bias\_i)^2}{n}} = \Delta \tag{3}$$

When certified reference materials are lacking (a frequent situation) and when no other analyses of bias have been performed in the laboratory (e.g. prior to applying the reference method) bias can be estimated on the basis of proficiency testing, PT.

A laboratory participating in PT may use the results of such tests in order to assess uncertainty of measurement for the testing method/procedure used. Similarly to determining uncertainty in within-laboratory experimental approach, the uncertainty of measurement (u) is equal to the root of the sum of squared values of standard deviation for within-laboratory reproducibility sRw and the bias (b), which can be calculated from the formulas 1,2 and 3.

With this approach two components of uncertainty are obtained from different sources. Precision is determined on the basis of the authors' own validation data (within-laboratory reproducibility), from the range or on the basis of measuring control charts (in-house). The bias is determined on the basis of PT results. Estimating the bias on the basis of a single participation in PT may have a limited range and should be treated as preliminary. If the data from several PTs are available (a wider range of matrices and concentrations) the assessment of the bias may be referred to the complete measurement range.

The results of analysing uncertainty on the basis of experimental approaches using the results of the authors' own results are presented below along with, for comparison, expanded uncertainties estimated with the help of Horwitz formula RSDR (%) = 2C-0.15 .


\*Expanded uncertainty for the HorRat value H=1 calculated from the Horwitz' formula RSDR=2 C-0,15; U (%) = 2 RSDR **Table 9.** Results of uncertainty evaluation for some feed additives in compound feeds and premixes

## **9. Conclusion**

182 Chromatography – The Most Versatile Method of Chemical Analysis

verify uncertainty with the help of other approaches.

and the assessment of the method's bias, following CRM,

with a bigger number of measurements is low and can be omitted:

methods, in particular:

Eurolab [27] technical report.

formula below:

possible to obtain an underestimated value in case when we do not consider all the components. Other reasons for underestimating uncertainty during validation include a situation when while assessing uncertainty repeatability instead of within-laboratory reproducibility is taken into account or we often forget to consider the bias. Moreover, uncertainty assessment is done during validation when well-ground typical samples are analyzed in a short time, new standards are prepared, the apparatus is controlled (standard conditions). During routine activities we analyze various matrices and obtaining homogeneity is not easy in case of some samples. The conditions mentioned above may affect underestimation of uncertainty. That is why it is important to have a possibility to

New opportunities concerning verification and assessment of uncertainty can be found in the Eurolab [27] technical report, the Nordtest [23] handbook and in the paper [28] which recommend the use of experimental approaches in assessing uncertainty of laboratory



Using the within-laboratory experimental approach in assessing uncertainty of measurement (u) within -laboratory reproducibility (s) is considered as a measure of precision, as well as the bias (b), in accordance with the equation below following the

2 2

Precision of a research procedure in a laboratory is determined during validating the method or on the basis of measurements noted in control charts. Validation usually includes determining standard deviation of within-laboratory reproducibility srw or standard deviation of within -laboratory reproducibility sRw which is sometimes called intermediate precision. Bias is determined by means of standard deviation of measurement in relation to the reference value, e.g. while examining certified reference materials, using reference methods. The share of bias (b) in uncertainty of the measurement is determined with the help of mean deviation of measurements (Δ), uncertainty of reference material (uref ) and precision of measurements during examining that bias (s). The value of the expression s2 / n

2

2 2 *ref <sup>s</sup> u b n*

In practice, different measurements result in different values of bias. In such a case the data may be combined and common assessment of a value of bias (uw) may be determined as a function of the measured value or, for typical data of matrices and levels, according to the

*Rw s bu* + = (1)

Δ+ + = (2)

and the assessment of the method's bias, following participation in PT/ILC.

The chapter presents a brief review of the methods used for determining feed additives by means of high proficiency liquid chromatography, HPLC. The authors presented their own research procedures and special attention was given to the preparation of samples for testing, extraction, extract purification, chromatographic separation and the basic elements of method validation and quality control.

Using HPLC for testing fat-soluble vitamins in feed materials, mixtures and premixes enabled us to replace colorimetric methods and to eliminate bias, such as the positive error of vitamin A determination related to the presence of carotenoids in the analyzed feed. The problem of low precision of examining certain vitamins, e.g. vitamin A, in feed mixtures is often unrelated to the method of determination, but rather to non-homogenous distribution of vitamin A in the feed related to its being secured against losing activity, due to protective coating. This problem may be solved by preparing the analytical weighed amount of sufficiently high mass and grinding the sample immediately prior to determination procedure to particles sized 1 mm.

Using High Performance Liquid Chromatography (HPLC) for Analyzing Feed Additives 185

**Author details** 

**10. References** 

1095-1109.

43. http://eur-lex.europa.eu

http://eur-lex.europa.eu

(No.6):1579-1582.

 \*

Corresponding Author

Chromatography A, 1158:138-157.

Roche Publication Index, Basle, No. 1841: 8-11.

http://gazette.comune.jesi.an.it/2006/50/11.htm

feeds and premixes. Analyst, Vol. 120: 2175-2180.

in medicated animal feed. Analyst, Vol. 108: 1252-1256.

, Waldemar Korol and GrażynaBielecka

*National Research Institute of Animal Production, National Feed Laboratory, Poland* 

[1] Regulation (EC) No 1831/2003 of the European Parliament and of the Council of 22 September 2003 on additives for use in animal nutrition. OJ L 268 from 18.10.2003, 29-

[2] Dejaegher B, Heyden Y (2007) Ruggedness and robustness testing. Journal of

[3] Horwitz, W. & Albert, R. (2006). The Horwitz Ratio (HorRat): A useful index of method performance with respect to precision. Journal of AOAC International, Vol. 89(No. 4):

[4] Commission Regulation (EC) No 152/2009 of 27 January 2009 laying down the methods of sampling and analysis for the official control of feed. OJ L 54 from 26.02.2009, 1-130.

[5] Manz, U. & Philipp, K. (1988). Determination of Vitamin D3 in Complete Feeds and Premixes with HPLC, in Analitycal Methods for Vitamins and Carotenoids in Feed,

[6] Manz, U. & Maurer, R. (1982). Method for the determination of vitamin K3 in premixes and animal feedstuffs with the aid of high performance liquid chromatography.

[7] Fedder, R. & Plöger, A.(2005). Solid-phase extraction of vitamins A and E from animal feeds: A substitute for liquid-liquid extraction. Journal of AOAC International, Vol. 88

[8] VDLUFA-Methodenbuch. Band III - 5 Erg. (2004) - 6 Erg. (2006). Die chemische

[9] Decreto 20 febbraio 2006. Approvazione del metodo ufficiale di analisi per la deteminazione della vitamina B1 negli alimenti per animali – Supplemento n.19

[10] Analytical Methods Committee. (1983). Determination of halofuginone hydrobromide

[11] Analytical Methods Committee. (1995). Determination of lasalocid sodium in poultry

[12] Dusi, G. & Gamba, V. (1999). Liquid chromatography with ultraviolet detection of lasalocid, monensin, salinomycin and narasin in poultry feeds using pre-column

[13] Horwitz, W. & Latimer, Jr., G.W. (2011). Official Methods of Analysis of AOAC International, 18th Edition, Revision 4 , AOAC International, Gaithersburg, Maryland, USA. [14] Regulation (EC) No 882/2004 of the European Parliament and of the Council of 29 April 2004 on official controls performed to ensure the verification of compliance with feed

International Journal for Vitamin and Nutrition Research, Vol.52: 248-252.

Untersuchun von Futtermitteln. VDLUFA – Verlag, Darmstadt.

derivatisation. Journal of Chromatography A, Vol. 835: 243-246.

JolantaRubaj\*

Progress in the area of examining the content of water-soluble vitamins is also related to introducing the methods of liquid chromatography. The authors included their own procedures of analyzing vitamins B1 and B2, thiamine and riboflavin, with the use of HPLC methods and gave their characteristic parameters which meet the current requirements regarding the assessment of content and interpretation of results. These methods may be used especially to examine low content of thiamine and riboflavin, endogenic and added, in feed materials and mixtures.

HPLC methods have been widely used for testing cocciodiostats in feed preparations, premixes and mixtures. They contributed to improving the safety of using these additives, controlling concordance with manufacturer's declaration and not exceeding the maximum content in feed mixtures, as well as controlling the withdrawal period. Without liquid chromatography with mass spectrometry (LCMS) it wouldn't be possible to analyze effectively the remains of coccidiostats in the tissues and food products of animal origin. To reduce the risk of cross contamination in non-target feeds maximum content values for coccidiostats were determined recently at the level from 0.01 mg/kg (diclazuril) to 1.25 mg/kg (narasin, monensin), [29]. This created a need to develop some test methods adequate for the level of acceptable cross contamination and verifying them in interlaboratory tests. Future research will focus on checking the LCMS method for this particular purpose.

The official methods of separating and determining amino acids in feedingstuffs [13,8] are based mainly on ion exchange chromatography. However, in examining free amino acids (amino acids used as additives: lysine, methionine, threonine, tryptophan, valine, arginine and cysteine) HPLC methods are becoming increasingly more popular as they make the analyses shorter in time. In some cases a HCLP method is the only solution, e.g. while determining methionine hydroxy analog, verified in the authors' own studies. The need to perform a large number of analyses in a shorter time determines the direction of future studies of amino acids in feedingstuffs and using ultra-performance liquid chromatography, UPLC, for this purpose.

In the testings of feed colorants the most frequently used means were spectrophotometric methods [13,8]. The diversity of feed products and the resulting changeability of matrix, as well as determining the maximum content of colorants in feed mixtures, were the reasons for searching for new methods of examining colorants, including HPLC. An example of such a method in reference to canthaxanthin and a procedure based on the authors' own research is quoted in the present work. Future research in this respect will use the LCMS method to a higher degree as it enables detecting and determining several feed colorants in a single sample in view of cis-trans steroisomers.
