Accepted Manuscript

Analytical methods

Development and Validation of an Ionic Chromatography Method for the De-

termination of Nitrate, Nitrite and Chloride in Meat

Cristina López-Moreno, Isabel Viera Pérez, Ana M. Urbano

PII: S0308-8146(15)01200-5
DOI: [Link]
Reference: FOCH 17962

To appear in: Food Chemistry

Received Date: 28 March 2015

Revised Date: 28 July 2015
Accepted Date: 6 August 2015

Please cite this article as: López-Moreno, C., Viera Pérez, I., Urbano, A.M., Development and Validation of an Ionic
Chromatography Method for the Determination of Nitrate, Nitrite and Chloride in Meat, Food Chemistry (2015),
doi: [Link]

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
DEVELOPMENT AND VALIDATION OF AN IONIC CHROMATOGRAPHY METHOD FOR

THE DETERMINATION OF NITRATE, NITRITE AND CHLORIDE IN MEAT

*
Cristina López-Moreno , Isabel Viera Pérez y Ana M. Urbano

Environmental Services Laboratory of Malaga. Physical Chemistry Department. Plaza de la

Merced. Ed. Mercado de la Merced, s/n. 29012, Málaga, Spain.

ABSTRACT

The purpose of this study is to develop the validation of a method for the analysis of certain

preservatives in meat and to obtain a suitable Certified Reference Material (CRM) to achieve

this task. The preservatives studied were NO3-, NO2- and Cl - as they serve as important

antimicrobial agents in meat to inhibit the growth of bacteria spoilage. The meat samples were

prepared using a treatment that allowed the production of a known CRM concentration that is

highly homogeneous and stable in time. The matrix effects were also studied to evaluate the

influence on the analytical signal for the ions of interest, showing that the matrix influence does

not affect the final result. An assessment of the signal variation in time was carried out for the

ions. In this regard, although the chloride and nitrate signal remained stable for the duration of

the study, the nitrite signal decreased appreciably with time. A mathematical treatment of the

data gave a stable nitrite signal, obtaining a method suitable for the validation of these anions in

meat. A statistical study was needed for the validation of the method, where the precision,

accuracy, uncertainty and other mathematical parameters were evaluated obtaining satisfactory

results.

Keywords: Ionic Chromatography; method validation; meat preservatives determination;

Certified Reference Materials; matrix effects; mathematical treatment; signal stability;

uncertainty calculation; statistical study.

1
1. INTRODUCTION

It is becoming the norm that more laboratories are interested in the accreditation that

guarantees the technical competence of certain analyses. The relevant standard, UNE-EN

ISO/IEC 17025 (2005), describes the general conditions needed to obtain the technical

competence of assay and calibration laboratories. As this standard describes, the validation of a

method is a process which is essential in order to obtain the accreditation of the analysis to

ensure the suitability of the method for this determination and to determine the uncertainty value

associated with the analytical result (Eurachem/CITAC Guide CG4, 2004; ISO/IEC guide 98

Part 1, 2009; ISO/IEC guide 98 Part 2 and Part 3, 2008).

More than 3000 additives are available in the market to be used as antioxidants and anti-

microbial agents. The preservatives most used and controlled in meat industries are

antimicrobials and antioxidants (Gotterup et al., 2008; Ruiz-Capillas & Jiménez-Colmenero,

2008; Jastrzebska, 2010, Campos et al., 2010). Among these, nitrate and nitrite used in

combination with sodium chloride serve as important antimicrobial agents in meat to inhibit the

growth of bacterial spoilage. In this regard, nitrite prevents the development of extremely

dangerous bacteria, such as clostridium botulinum, which generates the botulin toxin,

responsible for muscular paralysis and neuronal complications (Cammack et al., 1999).

Additionally, nitrite is widely used in the alimentary industries because of its ability to react with

meat myoglobin generating mononitrosylhaemochrome which gives the characteristic red colour

to cured meat (Cammack et al., 1999). However, it has been demonstrated that high intakes of

nitrite can present risks to human health (Honikel, 2008; Zeilmaker et al., 2010; Li et al., 2011).

The Nitrite ion has the ability to change the oxihemoglobin to ferrohemoglobin, which causes

methemoglobinemia (Fan & Steinberg, 1996). This disease could mainly affect children under

one year of age, causing death by asphyxia (blue-baby syndrome) (Cammack et al., 1999).

The nitrate is used as a nitrite precursor through reduction by microbial enzymes (Honikel K.O.;

2008). In the seventies, nitrate was proved to be a nitrosamines precursor, which is a cause of

cancer in some animal species (Schweinsberg & Burkle, 1985; Cammack et al., 1999).

However, subsequent studies in humans proved that there was no clear relation between the

2
abusive intake of nitrate and cancer development, but on the contrary, nitrate has a strong

antimicrobial reaction that is beneficial for the organism (Lundberg et al., 2004).

Sodium chloride is another additive used in the production of meat because it has the ability to

prevent germ growth (Russell, Hugo & Ayliffe, 1999). Furthermore, the salt induced to decrease

pH helps the conversion from nitrate to nitrite, affecting the balance of these anions. However,

the adverse effects that abusive intake of salt can produce in human organism are well known

(Oliver et al., 1975; Beard et al. 1982; Charnley & Tannenbaum, 1985; Joosens & Geboers,

1987; He & Mac Gregor, 2015).

Due to the possible nitrate and nitrite adverse health effects, there are legal limits to control the

maximum concentration of these ions allowed in different meat products (Regulation (CE) Nº

853/2004). For this reason, the analytical control of these meat products is quite important.

Chromatographic techniques allow the separation, identification and quantification of the

species of interest in the same analytical batch, which makes this technique highlighted among

other colorimetric, turbimetric, titrimetric, etc., techniques. Thus, many studies are focused on

the determination of toxic, mutagenic and carcinogenic compounds in meat by chromatographic

methods (Siu & Henshall, 1998; Saccani et al., 2005; Calbiani et al., 2007; Zeleny et al., 2009).

Several types of chromatographic techniques, such as GC-NPD or FID (Wang, 2001), HPLC-

UV (Eggers & Cattle, 1986; Jobgen et al., 2007; Sun et al., 2007), IC-UV/VIS or conductivity

detector (Siu & Henshall, 1998), and GC-MS (Calbiani et al., 2007; Zeleny et al., 2009; Di

Stefano et al., 2012) are currently used for the analysis of the studied compounds. Among all,

Ion Chromatography is a fast and automatic method used to obtain trustworthy results,. For this

work, an ionic chromatograph equipped with a conductivity detector was optimized and the

analysis method was validated using statistical studies.

Eurachem/CITAC guide CG4 (2012) describes the process to obtain the uncertainty associated

to every result in an analytical determination. The uncertainty study is based on the evaluation

of all the steps implicated in the analytical process that could contribute to the total uncertainty

of the analytical determination. Thus, a deep knowledge of the technique, instruments, samples

and operators are needed in order to identify the uncertainty sources in the validation of the

method.

3
In this paper, the validation of a method for the determination of some preservatives in meat

was performed, studying the reliability of the analytical process, from the sample preparation to

the results acquisition. For this purpose, statistical parameters were evaluated, such as

repeatability, reproducibility, precision, accuracy, limits of detection and quantification, analytical

range and uncertainty of the analytical method.

2. MATERIAL AND METHOD

2.1. Instrument Setup

The instrument setup was composed by a compact ion chromatograph equipped with an

autosampler and conductivity detector. Fig. 1 illustrates the configuration of the experimental

setup and the chromatogram window with the main peaks of interest. The system was a

Metrohm 861 Advanced Compact IC, with chemical suppression system for the analysis of

anions. The autosampler was an 838 Advanced Sample Processor equipped with a dialysis

module that filtered the reagents through a 0.2 µm-pore-diameter membrane.

All parts of the 861 advanced compact IC, such as injection valve, high-pressure pump,

conductivity detector, suppressor module, peristaltic pump and autosampler were fully

controlled by the IC Net 2.3 program.

The column used for separation was a Metrosep A Sup 5 250/4, from Metrohm. The injection

volume, the flow rate and the temperature were set at 20 µl, 0.7 ml/min and 25 °C, respectively.

Using these features, the window chromatogram obtained was shown in Fig. 1 where the

approximated retention times for the peaks of interest are indicated.

2.2. Reactives

A solution of Na2CO3 3.2 mM / HNaCO3 1.0 mM was used as eluent for the chromatographic

analysis. This solution was prepared adding 0.34±0.005 g of Na2CO3 Reagent Grade (Merck

Millipore) and 0,084±0,0005 g of HNaCO3 (Sigma Aldrich) in a 1000 ml volumetric flask and

brought up to the mark with ultrapure water. The masses were measured in a balance (Sartorius

CP224S, maximum precision of ±0,0001 g). Ultrapure water and H2SO4 50 mM were used for

4
the chemical suppression module. The H2SO4 solution was prepared adding 2.8±0.01 ml of

H2SO4 96% Suprapur (Merck Millipore) in a 1000 ml volumetric flask and brought up to the mark

with ultrapure water. The solutions were prepared using ultrapure MilliQ water (MilliQ gradient

system, Millipore, Italia, resistivity 18.2 MΩ/cm).

Commercial Certified Reference Material (CRM) of NaCl (Merck Millipore, certipur quality) at

99.93% of richness and uncertainty of ±0.05% was used for the fortification of the samples,

together with certified KNO3 (CRM) (Merck Millipore, suprapur quality) at 98% and uncertainty of

±1.15% and commercial solution certified standard of 1000 mg/l nitrite in water from NaNO2

(CRM) (Merck Millipore, certipur quality) and uncertainty of ±5 mg/l.

2.3. Sample Preparation

In this study, three different types of fresh meat with equal mass (500 g. pork, 500 g. chicken

and 500 g. beef) including fat and nerves were used for the validation.

In any validation easy handling reference materials are recommended, free of moisture and as

chemically stable as possible. In order to obtain a homogeneous and stable sample, the

reference was prepared as follows:

a) The different types of meat were mashed (IKA A-11 basic) in 6 batches.

b) The different mashed samples were homogenized using a crushing machine (Selecta

RGL-500), obtaining a unique matrix sample.

c) The sample was boiled in 1500 ml ultrapure water in a beaker for 10 minutes and then

dried in a heater (Binder FD53) at 105ºC for 48 hours.

d) Finally each sample was mashed (IKA A-11 basic) again for 8 seconds to obtain a fine

powder with a sample weight of around 400 g.

2.4. Extract preparation

The extracts were each prepared with 3.00 g of the sample. For the preparation, 3 ± 0.01 g of

the processed sample were put in a 250 ml volume Erlenmeyer with 50 ml of ultrapure water

and the mixture was homogenized by shaking (Selecta Rotaterm Orbital shaker) at 300 rpm.

5
Then, 50 ml of ultrapure boiling water was added to ensure the ion extraction from the meat and

the mixture was set on an orbital shaker-heater (Selecta Rotaterm) at 40º C for 30 minutes.

Later, the mixture was left to cool at room temperature for 90 minutes and the extract was

transferred to a 200 ml volumetric flask and brought up to the mark with ultrapure water before

filtering. A pleated paper filter was then used to filter the extract to get an aliquot higher than

100 ml. Finally, it was filtered again using a 0.2 µm pore diameter membrane filter until an

aliquot of 15 ml was obtained. The sample extracts were analyzed without further delay.

3. Results and discussion

3.1. Reference Materials

The validation of an analytical method can be understood as a diagnosis procedure of the

capability of a method to determinate an analyte under certain analytical conditions (Eurachem,

2014). There are several types of validation processes according to the method used, but,

standard concentration materials are always needed as reference values in quantitative

methods. These reference materials could be Certified Reference Materials (CRM), samples

previously analyzed by other validated methods, or routine samples fortified with CRM, leading

in this case to Fortified Reference Samples (FRS). The latter was used as a standard

concentration in this study, fortifying the meat sample with known concentrations of chloride,

nitrate and nitrite. This FRS is used first in the validation process and later in the control and

verification of the analytical method. Most analysis laboratories which seek accreditation for

analysis of meat samples, find a drawback due to the evolution with time of the components of

the meat sample used as reference material. This effect can cause differences in the

composition of the reference sample. A possible solution is the use of freeze-dried meat (Toribio

et al., 1999; Zeleny et al., 2009) whose main characteristics are temporal stability and simplicity

of use. Although this solution is effective, it is also time consuming and relatively expensive. In

this work a treatment process for the obtaining of a white meat sample that is relatively stable

and with options to be fortified with the desired analyte concentrations was developed. The use

6
of this meat sample together with the development of a mathematical correction of the signal in

function of time, allowed the use of this reference material to obtain satisfactory results.

In order to get a wide range of concentrations in the validation of the method, the meat sample

was fortified in three different concentration levels (high, medium and low) to guarantee the

reliability of the outcome across the whole analytical range.

Thus, a portion of 216.88 g of processed sample prepared as explained in section 2.3 was

fortified with commercial CRM of nitrite, nitrate and chloride in order to get the FRS with a high

concentration level. For this task, 4.0125 g of solid NaCl, 0.1481 g of solid KNO3 and a volume

of 28 ml certified standard of 1000 mg/l nitrite in water from NaNO2, was used.

Other portions of this fortified sample were mixed and homogenised in proportions of 50:50 and

25:75 with the blank sample(without fortification), to get FRS of medium and low concentration

levels, respectively.
-
The FRS of lower ion concentration level had 2805 mg of Cl,103 mg of NO3,and 32 mg of NO2

per Kg of FRS meat. The medium level concentration FRS had 5610 mg of Cl, 205 mg of NO3-
-
and 65 mg of NO2 per Kg of FRS meat. The higher level concentration FRS had 11220 mg of
- - -
Cl , 410 mg of NO3 and 129 mg of NO2 per Kg of FRS meat.

The meat sample obtained using this procedure was used to validate the method. The method

was also tested for fresh and cured meats samples using interlaboratory comparisons and

satisfactory results were obtained.

3.2. Calibration

Six standards and a blank were used for the calibrations. Multianion solutions of chloride, nitrate

and nitrite CRM were prepared and used as standards for the calibration curves. The analysis

ranges were set up to 250 mg/l (16700 mg/Kg) for chloride, up to 10 mg/l (670 mg/Kg) for nitrate

and up to 5 mg/l (330 mg/Kg) for nitrite, being mg/l the mass per litre of solution and mg/Kg the

mass per Kg of processed meat.

Ten calibration curves were performed for each analyte over ten different days and the average

result can be observed in Fig. 2. The linear values obtained (in percentage), quantification limit

(QL) and detection limit (DL) for each analyte were 98.4%, 881 mg/Kg and 473 mg/Kg,

7
respectively for chloride, 99.5%, 54 mg/Kg and 17 mg/Kg, respectively for nitrate and 98.9%, 10

mg/Kg and 5 mg/Kg, respectively for nitrite. In order to get the results of QL and DL, the

standard deviation of the lower concentration standard was multiplied by 10 and 3, respectively.

3.3. Matrix effects

If one analyte is analysed in different matrices, referring the matrix to the components of a

sample other than the analyte of interest, the result could be quite different due to the matrix

effect. This effect is based on the interference that components in the sample can cause on the

determination.

Usually, the determination of the presence and concentration of additives in meat is obtained by

comparison of the analytical signal obtained for the meat samples with the analytical signal

obtained from the calibration curve created with different concentration standards. However, the

extracted matrix of the samples is quite different than the matrix used in the standards, which

were prepared diluting chloride, nitrate and nitrite CRM with ultrapure water. Thus, the study of

the matrix effect is essential in order to assess the influence of the matrix sample in the

analytical signal.

For this study, two simultaneous calibration curves were created:

a) One was prepared by dilution of different CRM concentration of the analytes in ultrapure

water.

b) The other one was prepared by dilution of different CRM concentrations, but adding 70

ml of the blank meat extract before the flasks were made up to 100 ml with ultrapure

water, with each standard containing the same portion of diluted extract. Thus, each

standard contained a known amount of nitrite, nitrate and chloride plus 70 ml of extract

plus water up to 100 ml.

Fig. 2 compares detection of CRM in water versus matrix. For the latter, the concentrations

obtained for a blank standard in matrix were subtracted from the results obtained for the fortified

standards in matrix. Fig. 2 also shows the difference between the slopes of both calibrations for

each preservative. This difference was calculated by dividing the difference of the slopes by the

8
slope of the calibration in the matrix. As can be observed, the most significant difference is that

obtained for nitrate calibration, but even in this case, the difference is lower than 2.5%. The

conclusion is that there is no significant matrix effect that could affect the results when the

calibration standards are prepared in ultrapure water when the samples are extracted as

described in section 2.4.

3.4. Nitrite kinetic stability assessment

Quantification of the uncertainty for the reference material used in the validation process is

needed to obtain the total uncertainty of the analytical method. If we regularly change the

reference material or the fortified reference sample (FRS), then the calculation would be quite

complicated, due to the number of variables and the multiple repetitions necessary to quantify

the uncertainty. . Therefore, the use of the same reference material in the whole validation

process leads to a simplification of the uncertainty calculation.

Nitrite ion behaviour in meat samples a dynamic process since the nitrite maintains the

continuous reduction-oxidation reactions (Marco et al., 2006). The meat sample was observed

to have a nitrite concentration that varies in time, which obstructs the validation using the same

FRS. This effect was not observed in the chloride or nitrate samples

In order to study the variation of the nitrite concentration with time, the peak area signal of

different concentrations of the FRS was represented over 25 days, being the total length of the

whole validation exercise. Fig. 3A shows the behaviour of the nitrite analytical signal for the

three levels of FRS in function of time. As can be seen, the sample fortified with the lowest

concentration level presents higher stability with time, followed by the medium concentration

level and finally by the highest concentration level. Therefore, as the nitrite concentration in the

sample increases, the chromatographic signal reduces in proportion.

In order to get through this drawback, a mathematical approximation to correct the signal of

nitrite in time was performed, thus avoiding the need to change the FRS every day (Slickers,

1993). This mathematical approximation has been widely used and proved in other areas, for

example in medicine (Fan et al., 2008) or in industrial manufacturing (Lopez-Moreno et al.,

2005), where the change of every variable in time, such as sample temperature, pressure,

9
humidity, and the chemical composition, can make necessary a mathematical approximation to

obviate the variable contribution and turn the sample into a stable material. The proposed

method uses a transfer operation that transforms the obtained area signal in the expected area

to the value which would obtained if the sample had remained stable in time. For this, the slope

calculation in the linear approximation of the analytical signal versus time is needed, and is

indicated in fig. 3A. In our case, the experiments took 25 days and, in that period, the temporal

variation in the three levels is linear, and thus, this transfer method is perfectly suitable for this

purpose. For each analytical point, a correction factor “f” was determined, which is given by the

following expression:

f = m·d (1)

Being:

m = slope obtained in the linear temporal evolution

d = days passed from the sample preparation

As observed in expression 1, the factor depends on the slope but also on the days passed from

the sample preparation day, and thus, the factor increases with time at the same rate that the

nitrite signal decreases. The factor is indicative of the signal variation percentage related to that

obtained the first day. The factor was added to the area obtained for every point, as shown in

the following expression:

AT = AM + f (2)

Where:

AT =Transformed nitrite area

AM =Empirical nitrite area

10
In this way, the nitrite signal is transformed in a time stable function, as schematised in Fig. 3B.

In this validation, all the nitrite data obtained subsequent to the first day of analysis were

corrected. Satisfactory results were obtained using this mathematical method.

3.5. Validation of the analytical method

The first step in order to achieve validation of an analytical method is the establishment of the

analysis tolerances that ensure the reliability of the results obtained. Only if these tolerances

tests are passed, can the method be accepted as valid for the analysis. Table 1 details the

expressions used and the values established as tolerances, being CI, the compatibility index,

VRi, the reference value, Xi, the average value, Si, the standard deviation, Sr, the standard

deviation obtained in repeatability conditions, SR, the standard deviation obtained in

reproducibility conditions, n, the number of experiments, VC, the variation coefficient, U, the

uncertainty, DL, the detection limit, CT, the correction term, and finally, the UVRi, the reference

value uncertainty. The tolerances were established according to the exigencies of the

accreditation entity (ENAC) auditors, who determined the limits as necessary to consider the

analysis method validated.

The knowledge of the error contributions to the whole analytical process is essential to obtain a

proper validation of the method (Quintela et al., 2012). There are two models widely used in the

uncertainty calculation of an analytical method (Eurachem/CITAC Guide CG4, 2004; ISO/IEC

guide 98 Part, 1, 2009; ISO/IEC guide 98 Part 2 and Part 3, 2008): The first one is the

“uncertainty propagation” and is based on a theoretical calculation where the uncertainty in

every single step of the method is estimated and finally combined in an unique uncertainty. This

model was used only for the determination of the uncertainty obtained in the FRS preparation.

The second one is the “black box” model. This is the most used because of its simplicity and

also because it offers a more empirical and thus more realistic uncertainty value. This model

was used for the uncertainty calculation of the whole analytical process except the FRS

preparation, where, as mentioned previously, the uncertainty propagation method was used.

All contributions to the uncertainty calculation were considered, i.e., sample and reactive

dilutions, calibration curves, mass measurement, instrumental and analysts contributions, etc.

11
Following the black box model, the analysis must be developed in two different analytical

conditions: Repeatability and reproducibility conditions. The former one is based on the

repetition (three times per day) of the analysis in a short time, using the same procedure, the

same analyst, the same calibration plot and the same instrumental under the same work

conditions. On the other hand, the reproducibility conditions are based on the development of

the analysis in different days, using different calibration plots, and if possible, different analyst

and instrumental. In this way, the uncertainty obtained for the reproducibility conditions analysis

must be higher than that obtained in repeatability conditions. In this study, the reproducibility

conditions were developed using ten different days for the analysis and ten different calibration

plots.

Reference samples were made by fortification of the meat sample using the analytes of interest

as described in the subsection 3.1. The repeats used the same FRS, which was prepared only

once. For the uncertainty calculation in the preparation of the FRS with the uncertainty

propagation model, the theoretical uncertainty that takes place in each individual step of the

sample fortification procedure was calculated. In Table 2, the uncertainties that take part in the

preparation of the chloride FRS are shown, being U(xi)A, the certified uncertainty of the

standard used for the fortification or the estimated theoretical uncertainty for the Xi value; U(xi)B,

the value of uncertainty obtained by the use of the U(xi)A percentage to the value; U(xi)C, the

uncertainty obtained by dividing U(xi)B by the square root of 3 just in the required cases; C, the

concentration of chloride obtained in the FRS in mg/Kg and U(xi)Comb, the combined

uncertainty of the whole FRS preparation process.

The combined uncertainty is the only value that includes the individual uncertainties of each

single step. For the combined uncertainty calculation, the following expression is used:

2
 U(x i )
Ucomb. = C i ⋅ ∑  
 (3)
 xi 

Being:

Ci = Concentration of the level (high, medium or low)

Xi = Measurement of the parameter that takes part in the error contribution (weight, volume

dissolution, etc)

12
U(xi) = Uncertainty associated to each contribution.

For the uncertainty calculation of each step in the FRS preparation, a rectangular distribution of

the Xi value was assumed, and thus, the certified or theoretical uncertainty was divided by the

squared root of 3 (EURACHEM/CITAC, Guide CG4, 2012). However, in the case of medium

and low concentration levels, the uncertainty of the standard used for the fortification is

previously obtained for the FRS high level, since this preparation stage consists in the mixture

of this sample with the meat product matrix in blank. In these cases, the division by the root of 3

is not appropriated.

Table 2 shows the uncertainty obtained using the propagation model for the chloride FRS. As

observed the uncertainty increases for the preparation of the sample with higher concentration

level, being lower for the medium and low concentration levels.

In the nitrate and nitrite fortification case, the process was similar, and the uncertainties of the

FRS preparation are indicated as “uncertainty of reference material” in Table 3, where the

accuracy, precision and uncertainty estimation in the validation method are summarized.

In the determination of the Total Uncertainty of an analytical process, the Compatibility Index

(CI) (see Table 1) takes an important role. The CI is a parameter indicative of the deviation

between the theoretical (or reference) concentration value and the averaged value obtained

empirically with the method. As is denoted in Table 1, only if the CI is higher than 2, the

combined uncertainty is increased by a Correction Term (CT) (Lopez-Moreno et al., 2010,

Centrich et al, 2011, Vicente y Oliva et al., 2011), that is calculated using the following

expression:

2
 C 
CT =   (4)
 3
 

Where C is the difference between the reference standard concentration and the average value

obtained empirically.

In this way the uncertainty increases only when the variation between the theoretical value of

the reference material and the empirical result is higher than the established limit, but, even in

these cases, the expanded uncertainty results, indicated in Table 3, were under 17%.

13
This expanded uncertainty represents the uncertainty of every concentration value of the

samples used in the analysis and is the result of multiplying the combined uncertainty by a

constant factor. This factor must be elected considering the confidence level required, the error

distribution in every step, and the number of assays developed. In this work, the calculated

uncertainty follows a Gaussian distribution, more than 6 validation assays were carried out and

the confidence level was 95%. Using these data, the constant factor obtained is 2, which means

that the expanded uncertainty is twice the combined uncertainty

Following the black box model proposal, fortified reference samples and the blank sample

(without addition) were analysed in triplicate along ten different analysis sessions carried out on

ten different days and by different analysts. The concentration obtained for the blank sample

was subtracted from the results obtained for the fortified samples. In this way, only the added

concentrations were evaluated, without having into account for the validation of the initial

analyte content in the blank sample before fortifying.

As can be observed in Table 3, the statistical results obtained in the validation process of the

method for chloride, nitrite and nitrate satisfy the tolerance levels established in Table 1. Table 3

also shows the combined and expanded uncertainties of the method, the latter being the one

that must be always indicated in the analysis report together with the analytical result. As can be

observed, the uncertainty is higher for the low concentration levels in every analyte, and

decreases for the medium and high concentration levels, where the expanded uncertainty

values do not exceed 7%. The same occurs with the accuracy and precision in reproducibility

and repeatability conditions, where the values increase for the lower concentration levels. This

increase in the uncertainty for the lower concentration levels has been observed before in other

works of this group (Lopez-Moreno et al., 2010), concluding that the instrument resolution limit

is the main cause of this increase of the uncertainty.

4. Conclusions

In this study, the validation results obtained for the analysis of some additives in meat using a

chromatographic method are reported, showing that the method meets the tolerance levels

established by this laboratory to ensure the validation of the analytical process.

14
A blank meat matrix was prepared using a treatment that allowed the production of meat

reference samples quickly and easily. Part of this blank meat matrix was fortified in order to

obtain fortified reference samples (FRS) homogeneous and stable, with chloride, nitrite and

nitrate in the required concentration levels to cover the whole concentration limits of the

validation.

A matrix study was also carried out in order to assess the matrix effects and additives influence

on the analytical signal, showing that the matrix effects are not significant for this determination.

To avoid the temporal decay of the nitrite signal in the high level concentration, a signal study

was performed with time, and a transfer function was used to obtain a concentration value

stable with time. This transfer function gave satisfactory results, evidencing the validity of the

method and the mathematical arguments used.

To complete the validation of the method, a statistical study was carried out in order to prove the

reliability of the method, showing, among others, the precision, accuracy and uncertainty of the

analytical results, giving in all cases satisfactory values.

5. Acknowledgements

This research was supported by the Council of Malaga, which is gratefully acknowledged.

6. References

Beard T.C., Cooke H.M., Gary W.R. & Barge R. (1982). Randomised controlled trial of a no-added

sodium diet for mild hypertension, Lancet 2 455-458.

Calbiani F., Careri M., Elvira L., Mangia A. & Zagnoni I. (2007). Validation of an ion-pair liquid

chromatography-electrospray-tandem mass spectrometry method for the determination of

heterocyclic aromatic amines in meat-based infant foods. Foods Additives and Contaminants, 24(8),

833-841.

Cammack R., Joannou C.L., Cui X.Y., Torres Martinez C., Maraj S.R. & Hughes M.N.. (1999). Nitrite

and nitrosyl compounds in food preservation, Biochimica et Biophysica Act Bioenergetics, 1411,

475-488.

Campos I., Masot R., Alcaniz M., Gil L., Soto J., Vivancos J.L., García-Breijo E., Labrador R.H., Barat

J.M. & Martinez-Manez R. (2010). Accurate concentration determination of anions nitrate, nitrite and

15
 chloride in minced meat using a voltammetric electronic tongue, Sensors and Actuators B. 149 (1),

71-78.

Centrich F., Subirana T., Granados M., Companyó R., (2011), Practical Quality Control: the

Experiences of a Public Health Laboratory, Modern Approaches To Quality Control, Dr. Ahmed Badr

Eldin (Ed.), ISBN: 978-953-307-971-4, InTech. Page 425.

Charnley G. & Tannenbaum S.R. (1985). Flow cytometric analysis of the effect of sodium chloride on

gastric cancer risk in the rat, Cancer Res. 45, 5608-5616.

Di Stefano V., Avellone A., Bongiorno D., Cunsolo V., Muccilli V., Sforza S., Dossena A., Drahos L. &

Vékey K. (2012). Applications of liquid chromatography–mass spectrometry for food analysis, J.

Chromatogr. A, 1259, 74-85.

Eggers N.J. & Cattle D.L. (1986). High-performance liquid chromatographic method for the

determination of nitrate and nitrite in cured meat, J. Chromatogr. A, 354, 490-194.

EURACHEM/CITAC, Guide CG4. (2012). Quantifying Uncertainty in Analytical Measurement. URL

[Link]

EURACHEM guide (2014). The Fitness for Purpose of Analytical Methods A Laboratory Guide to

Method Validation and Related Topics.

Fan W., Liang Y., Yuan D & Wang J. (2008). Calibration model transfer for near-infrared spectra

based on canonical correlation analysis, Analytica Chimica Acta, 623, 1, 22-29.

Fan A.M. & Steinberg V.E. (1996). Health Implications of Nitrate and Nitrite in Drinking Water: An

Update on Methemoglobinemia Occurrence and Reproductive and Developmental Toxicity, Regul.

Toxicol. Pharmacol, 23, 35-43.

Gotterup J., Olsen K., Knochel S., Tjener K., Stahnke L.H., & Moller J.K.S. (2008). Colour formation in

fermented sausages by meat-associated staphylococci with different nitrite- and nitrate-reductase

activities, Meat Sci., 78, 492-501.

He F.J. & Mac Gregor G.A. (2015). Salt and sugar: Their effects on blood pressure. European Journal

of phycology, 465, 3, 577-586.

Honikel K.O. (2008). The use and control of nitrate and nitrite for the processing of meat products,

Meat Sci, 78, 68-76.

ISO/IEC Guide 98. (2009). Part 1: Introduction to the expression of uncertainty in measurement. URL

[Link].

ISO/IEC Guide 98. (2008). Part 2: Concepts and basic principles. URL [Link].

ISO/IEC Guide 98. (2008). Part 3: Guide to the expression of uncertainty in measurement. URL

[Link].

16
Jastrzebska A. (2010). Application of capillary isotachophoretic method to the determination of nitrate

and nitrite ions in meat products. J. Anal. Chem. 65 (11) 1170-1175.

Jobgen W.S., Jobgen S.C., Li H., Meininger C.J. & Wu G. (2007). Analysis of nitrite and nitrate in

biological samples using high-performance liquid chromatography, J. Chromatogr. B, 851, 71-82.

Joosens J.V. & Geboers J. (1987). Dietary salt and risks to health, American Journal of Clinical

Nutrition, 45 1277-1288.

Li Z.W., Zhang L., Ye R.W., Liu J.M., Pei L.J., Zheng X.Y. & Ren A.G. (2011). Maternal

periconceptional consumption of pickled vegetables and risk of neural tube defects in offspring.

Chinese Med J-Peking. 124, 1629–1633.

Lopez-Moreno C., Viera I. & Urbano A.M. (2010). Validation of an ion chromatographic method for the

quantification of anions in water, Desalination 261, 111-116.

Lopez-Moreno C., Palanco S. & Laserna J.J. (2005). Calibration transfer method for the quantitative

analysis of high-temperature materials with stand-off laser-induced breakdown spectroscopy, J.

Anal. Atom. Spectrom. 20 (11) 1275-1279.

Lundberg J.O., Weitzberg E., Cole J.A. & Benjamin N. (2004). Nitrate, bacteria and human health,

Nature Rev.: Microbiology 2 (7) 593-602.

Marco A., Navarro J.L. & Flores M. (2006). The influence of nitrite and nitrate on microbial, chemical

and sensory parameters of slow dry fermented sausage, Meat Sci. 73, 660-673.

Oliver W.J., Cohen E.L. & Neel J.V. (1975). Blood pressure, sodium intake, and sodium related

hormones in the Yanomamo Indians, a "no-salt" culture, Circulation, 52, 146-151.

Quintela M., Báguela J., Gotor G., Blanco M.J. & Broto F. (2012). Estimation of the uncertainty

associated with the results based on the validation of chromatographic analysis procedures:

Application to the determination of chlorides by high performance liquid chromatography and of fatty

acids by high resolution gas chromatography, J. Chromatogr. A, 1223, 107-117.

Regulation (EC) No 853/2004 of the European Parliament and Of the Council of 29 April 2004 laying

down specific hygiene rules for on the hygiene of foodstuffs.

Ruiz-Capillas C. & Jiménez-Colmenero F. (2008). Determination of preservatives in meat products by

flow injection analysis (FIA), Food Additives and Contaminants A, 25 (10) 1167-1178.

Russell A.D., Huga W.B. & Ayliffe G.A.J. (1999). Principles and Practice of disinfection preservation

and sterilization, Wiley-Blackwell. Page 102.

Saccani G., Tanzi E., Pastore P. & Cavalli S., Rey M. (2005). Determination of biogenic amines in

fresh and processed meat by suppressed ion chromatography-mass spectrometry using a cation-

exchange column, J. Chromatogr. A, 1082, 43-50.

17
Vicente y Oliva, J., Sanchez Perez A.M., Hernandez Perdomo, W. (2011), Virtual

Instruments in dimensional metrology, Research in Engineering Education Symposium 2011,

REES2011, Madrid, España. pp. 106-118.

Schweinsberg F. & Burkle V.J. (1985). Nitrite: A Co-carcinogen? J. Cancer Res. Clin. Oncol. 109, 200-

202.

Siu D.C. & Henshall A. (1998). Ion chromatographic determination of nitrate and nitrite in meat

products, J. Chromatogr. A, 804, 157-160.

nd
Slickers K. (1993). Automatic Atomic-Emission-Spectroscopy, (2 Ed.), Brühlsche

Universitätsdruckerei, Griessen, Germany. Page 328.

Sun H.W., Wang F.C. & Ai L.F. (2007). Simultaneous determination of seven nitroimidazole residues

in meat by using HPLC-UV detection with solid-phase extraction, J. Chromatogr. B, 857, 296-300.

Toribio F., Puignou L. & Galceran M.T. (1999). Evaluation of different clean-up procedures for the

analysis of heterocyclic aromatic amines in a lyophilized meat extract, J. Chromatogr. A, 836, 223-

233.

UNE-EN-ISO/IEC 17025. (2005). General Requirements for the Competence of Testing and

Calibration Laboratories, URL [Link].

Wang J.H. (2001). Determination of three nitroimidazole residues in poultry meat by gas

chromatography with nitrogen–phosphorus detection, J. Chromatogr. A, 918, 435-438.

Zeilmaker M.J., Bakker M.I., Schothorst R. & Slob W. (2010). Risk assessment of N-

nitrosodimethylamine formed endogenously after fish-with-vegetable meals. Toxicological Sci. 116,

323–335.

Zeleny R., Harbeck S. & Schimmel H. (2009). Validation of a liquid chromatography–tandem mass

spectrometry method for the identification and quantification of 5-nitroimidazole drugs and their

corresponding hydroxy metabolites in lyophilised pork meat, J. Chromatogr. A, 1216, 249-256.

18
FIGURE CAPTIONS

Fig. 1. Instrumental setup and chromatogram window with the main peaks of interest, elution order and

their respective retention times: 1, autosampler; 2, peristaltic pump; 3, dialysis filter; 4, ultrapure water; 5,

carbonate/bicarbonate tampon; 6, diluted sulphuric acid; 7, isocratic pump; 8, purge valve; 9, pulse shock-

absorber; 10, injection valve with 20 µl long loop; 11, conductivity detector; 12, chemical suppression

device; 13, separation column; 14, peristaltic pump; 15, control computer.

Fig. 2. Calibration curves for chloride, nitrate and nitrite using standards of CRM in meat matrix (□) and

CRM in water (■). Difference of the slopes and equation of corresponding curves.

Fig. 3. A) Nitrite peak area variation with time for low (■), medium (●) and high (▲) concentration levels

with the equation of corresponding curves; B) diagram of the effect obtained when the mathematical

transfer method was used to correct the empirical nitrite area (□) and transferred to the stable with time

area (■). “f” is de correction factor and “m” is the slope of the signal decay with time.

19
 18,0

17,5

Nitrate (16.57 min)

Chloride (9.28 min)
Conductivity (µS/cm)
17,0

Nitrite (11.10 min)

16,5

16,0

5
15,5

4 6

15,0
8 9 10 11 12 13 14 15 16 17 18 19 20
2
Time (minutes)
11

3
1
10 13

7 12

14
9

15

Fig. 1
 5000
Matrix
Water
4000
Slopes difference: 0.62%

Chloride Peak Area

3000
y = 17.096x - 93.667

2000
y = 17.203 x + 36.706

1000

0 50 100 150 200 250

Chloride Concentration (mg/l)

70

Matrix
60
Water

50 Slopes difference: 2.47%

Nitrate Peak Area
40 y = 5.9372x - 0.0133

30
y = 6.0843x - 0.3381
20

10

0 2 4 6 8 10
Nitrate Concentration (mg/l)

35
Matrix
30 Water
Slope difference: 2.10%
25
y = 6.6311 x - 0.2495
Nitrite Peak Area

20
y = 6.7734 x - 0.1872
15

10

0 Fig. 2
-5
0 1 2 3 4 5
Nitrite Concentration (mg/l)
 160
Low concentration level
Medium concentration level
140 High concentration level

120

Nitrite Peak Area

y=-1.9601x+149.95

A) 100

80 y=-0.5269x+79.76

60
y=-0.1462x+44.48

40

0 5 10 15 20 25
Time (days)
170
Transferred Area
160 Empirical Area

150
Nitrite Peak Area

140

B) 130

120 f

110

100

0 5 10 15 20 25
Fig. 3
Time (days)
 Parameter Expression Tolerance
VRi − xi ≤2
C.I. = For CI >2, the uncertainty will be
Compatibility Index  S 
2

U 2 P + i  incremented by a correction term

V
Ri  n  (CT)

VRi − xi ≤16% for low concentration levels

Accuracy A= ⋅100
VRi ≤ 4% for medium and high levels
Precision in reproducibility SR
VCR = ·100 ≤10%
conditions (Variation Coefficient) Xi
Precision in repeatability conditions Sr
VCr = ·100 ≤10%
(Variation Coefficient) Xi
≤ 500 mg/Kg for Chloride
Detection Limit (DL) D.L.= 3·SRi ≤ 20 mg/Kg for Nitrate
≤ 5 mg/Kg for nitrite
Other used expressions without established tolerances:
xi
Average Xi =
n
2
Standard Deviation in ∑ (x i − X)
SR =
reproducibility conditions (n −1)

Standard Deviation in repeatability 2

∑ si
conditions Sr =
n
2 2
 S   S 
Uncertainty
2
U = UVR + R  +  r  + CT a
 n   n 
 R   r 

a
CT term is used just in cases described in section 3.3

Table 1. Established tolerances for the validation of the method and

expressions used.

20
 Uncertainty U(xi)A U(xi)B U (xi)C C U(xi) Comb
Xi
source (%) (g) (g) (mg/Kg) (mg/Kg)
High Level
Sample Weight 216 g 0.5 1.1 0.6
Standard
4.01 g 0.1 0.004 0.002 11219.3 33.2
Weight
Standard
99.9 % 0.05 0.05 0.03
Concentration
Medium Level
Sample Weight 50 g 0.5 0.3 0.1
Standard
11219.3 g Combined 33.2 33.2 5609.6 28.3
Weight
Standard
50 % 0.5 0.3 0.1
Concentration
Low Level
Sample Weight 75 g 0.5 0.4 0.2
Standard
11219.3 g Combined 33.2 33.2 2804.8 11.6
Weight
Standard
25 % 0.5 0.1 0.07
Concentration

Table 2. Uncertainty obtained using the “Uncertainty Propagation” model for the chloride FRS

21
 Low Medium High
Chloride
Level Level Level
Accuracy (%) 8.2 3.8 1.0
Accuracy CI Value 8.3 3.9 2.3
CI Result >2 >2 >2
Repeatability (%) VCr 2.4 1.4 1.2
Precision
Reproducibility (%) VCR 3.1 2.7 1.1
Ref. Material (mg/Kg) 11.6 28.3 33.2
Correction (CI >2) (mg/Kg) 132.7 122.7 65.5
Uncertainty Combined U(x) (mg/Kg) 146.9 170.2 118.5
Expanded U(x) (mg/Kg) 288.9 340.4 241.4
U(x) exp. Relative (%) 11.2 6.3 2.2
Low Medium High
Nitrate
Level Level Level
Accuracy (%) 1.4 0.7 1.9
Accuracy CI Value 0.6 0.5 1.7
CI Result <2 <2 <2
Repeatability (%) VCr 5.3 2.8 1.3
Precision
Reproducibility (%) VCR 7.4 3.7 2.8
Ref. Material (mg/Kg) 0.8 1.7 3.0
Correction (CI >2) (mg/Kg) --- --- ---
Uncertainty Combined U(x) (mg/Kg) 5.9 6.0 9.6
Expanded U(x) (mg/Kg) 12.8 12.9 20.5
U(x) exp. Relative (%) 12.7 6.4 4.9
Low Medium High
Nitrite
Level Level Level
Accuracy (%) 15.0 3.1 3.1
Accuracy CI Value 9.0 3.3 4.2
CI Result >2 >2 >2
Repeatability (%) VCr 1.3 1.1 1.1
Precision
Reproducibility (%) VCR 4.3 2.1 1.8
Ref. Material (mg/Kg) 0.2 0.4 0.7
Correction (CI >2) (mg/Kg) 2.8 1.2 2.3
Uncertainty Combined U(x) (mg/Kg) 3.1 1.7 3.0
Expanded U(x) (mg/Kg) 6.0 3.3 5.9
U(x) exp. Relative (%) 16.2 4.9 4.7

Table 3. Parameters and uncertainty estimation in the validation method for chloride, nitrate and nitrite.

22
Highlights

- Validation of an ionic chromatography method for the analysis of meat is

proposed.
- Fortified reference meat samples containing Cl-, NO3- and NO2- were prepared.
- Nitrite signal showed a temporal decay for high concentration samples.
- Using a mathematical model, stable nitrite values were achieved for reference
meat.
- Satisfactory statistical results of the validation were obtained.

23