Introduction

Food safety is an essential element for the consumer. The process of globalization and the export of regional products has increased the trade flow between countries, making it increasingly necessary to control the production chains and product tracking carefully. In Europe, progress has been made with the issue of the Green Paper on the general principles of food law [1] and the publication of the White Paper on food safety [2]. Guidelines and procedures have been defined to assess and manage the food’s quality based on chemical and microbiological analyses that can define the possible presence of toxic chemicals or microbial agents [3]. Traditional analysis methods, recognized by the ISO international certifications, are universally accepted for their high accuracy and precision [4]. The new analytical measures must have sensitivity equivalent to the official methods, applicability to the matrix of interest, simplicity, speed, cost-effectiveness, and reliability [4,5,6,7]. The validation of the analysis methods guarantees the quality of the analytical data and the starting point for using new technologies for analytical and food certification purposes. The method of analysis applied by the laboratory must have stable performance over time to guarantee the comparability of the results, therefore, their “quality”. The UNI CEI EN ISO/IEC 17025: 2017 standard and related sections specify the measurement validation and reliability criterion [8,9,10]. Validation is a complex procedure, classifiable as primary and secondary validation. Primary validation is usually performed by the manufacturer or the laboratory that developed the method, as specified in the ISO 16140: 2003 standard. It is an exploratory-type experimental process whose purpose is to establish a new method’s operational limits and performance, a standardized method not adequately characterized, a standardized method modified by the laboratory, or a method developed by the laboratory (internal method [11, 12]).

The laboratories that intend to acquire an alternative method developed elsewhere must follow the procedure for secondary validation [13]. The secondary validation demonstrates the ability to execute the method with performances that are not lower than those declared by the validation protocol. The performance characteristics, as indicated by the ISO 17025 standard and international guidelines [14], to be evaluated are the measuring range; selectivity/specificity; accuracy (exactness); accuracy; limit of detection (LOD); limit of quantitation (LOQ); linearity range; sensitivity; robustness; measurement uncertainty (associated with the analytical data) and recovery. The process for the implementation of an analytical and experimental design involves the consideration of some variables (i.e., appropriate identification; scope; description of the type of object to be tested or calibrated; parameters or quantities and measuring ranges to be determined; equipment, including technical performance requirements; reference samples and required reference materials; environmental conditions and required stabilization period; description of the procedure, verification of the proper functioning of the equipment and, if required, calibration and fine-tuning before use; methods of recording observations and results; all measures of standard deviation; data to be recorded and the methods of analysis and presentation; uncertainty or uncertainty estimation procedures).

This work validated a rapid method for determining sodium chloride levels in canned tomatoes based on a colorimetric test.

Tomato (Solanum Lycopersicum L.) is the second horticultural crop cultivated worldwide after potatoes [15]. It is a climacteric fruit, highly delicate in the ripening process, and collectible at different stages of maturity (i.e., green mature, turning, breaker, red ripen, and pink or light red stage) [16]. The over-mature tomato fruits are susceptible during storage to biochemical activity, physical damage, disease, and insect infestation [17, 18]. NaCl and CaCl2 are added to canned tomatoes as dehydrating agents [19,20,21]. The NaCl levels must not exceed 3% in “peeled” and “unpeeled” tomatoes, 15% in tomato concentrate having a dry residue higher than 20%, and 3% for other concentrates and juice of tomato of the net weight (GU L 153 del 7.6.1986). The Volhard test is the official method used to determine chloride levels [22]. It involves a back titration with potassium thiocyanate to control the chloride ions level in a solution. Experienced personnel is required to perform the test. The colorimetric test subject to validation gives the chloride dosage in a few seconds and does not require staff training because it is carried out using a reagent kit and an automated apparatus. The colorimetric test validation was carried out to allow its use in routine food certifications, permitting considerable savings in time, costs, and laboratory personnel training.

The validation process can be developed only in laboratories where expert personnel can dedicate themselves to these problems. Therefore, it would be desirable that the scientific community assumes the burden of validating food analytical methods that employ new technologies to allow their routine use in analytical laboratories.

Materials and methods

Apparatus

The following apparatuses were used for the experimental purposes: Thermo Fisher Scientific centrifuge (Thermo Fisher Scientific, Waltham, USA); ICubio imagic M7 (Shenzhen ICubio Biomedical Technology Co., Ltd; Shenzhen, China); Genius analytical balance (Genius Electronic Company, Mumbai, India); falcon tubes (50 mL) (Thermo Fisher Scientific, Waltham, USA), and micropipettes (P1000) (Gilson, Middleton, USA).

Reagents

Distilled water was purchased by Sigma-Aldrich (Milan, Italy); KCl min purity 99.0% by Carlo-Erba (Milan, Italy); paper filters by Whatman (Maidstone, UK); R-Biopharm reactive K.I.T. Enzytech™ Chloride (E2520) by R-Biopharm AG (Darmstadt, Germany); silver nitrate, ammonium sulfocyanate, concentrated nitric acid and saturated solution of ferric alum by Fisher Scientific (Hampton, USA).

Methods

Sample

The test was performed on 15 packs of canned tomatoes Sammarzano DOP (Protected Designation of Origin) bought in the Souther Italy market. 25 g of sample were diluted in 25 g of distilled water, centrifuged, filtered on filter paper (Whatman No. 4, Brentford, UK), and 1 mL put into each colorimeter cuvette.

Sample storage and environmental parameters

The laboratory area’s relative humidity (RH) was 60–70%. The temperature of the laboratory areas was between 18 and 28 °C. The canned tomatoes, once opened, were stored in a refrigerator (T = 4 °C, RH 65%) and processed in an hour.

Sodium chloride extraction

The canned tomatoes were homogenized with an immersion blender. 25 g of sample were taken in a falcon test tube, brought to a volume of 50 mL with H2O, stirred, centrifuged (4000 rpm, 5 min, 28 °C), and filtered.

Quantitative determination with a colorimetric method

The quantitative determination of sodium chloride was performed by ICubio imagic M7 apparatus.

Calibration curve

The calibration curve was obtained by using NaCl. Three repetitions were carried out for each point (1, 0.5, 0.25, 0.125, 0.0625 g/L).

Apparatus method

The instrument starts washing the cuvettes and dispensing reagents and samples. The sample positions (rack F or G, positions 1–10) and item chlorides were set for all the samples. The sequence of operations performed by the machine (Clean, Test1, Test2, etc.) was reported in the “Test Process box”. The “Test section” described the layout relating to reagents/samples/calibrators loaded on the machine in “RS Disk Status”, and the layout of the reaction disc with the progress of the tests in progress in “Incu Disk Status”. The results list was displayed in “Result box”.

Enzymatic method

The method is based on the extraction of chlorides through homogenization and extraction with water. The principle of the method is based on a direct colorimetric test with mercury thiocyanate. The chloride ions react with the mercury ions and release an equivalent quantity of thiocyanate ions, which form a red complex in the presence of iron ions. The color intensity of the complex is proportional to the concentration of chlorides present in the sample. The chloride dosage is carried out with an automated “ICUBIO IMAGIC M7” analyzer.

The samples were prepared as described in the kit leaflet (Table 1).

Table 1 Samples used in the automatized apparatus
Full size table

After 5 min, the absorbance was read.

$${\text{C}}\,{\text{sample}}\,\left[ {{\text{meq}}/{\text{L}}} \right] = \frac{{{\text{A}}\,{\text{sample}}}}{{{\text{A}}\,{\text{standard}}}} \times {\text{C}}\,{\text{standard}}\left[ {{\text{meq}}/{\text{L}}} \right]$$

A = absorbance, C = concentration.

As the concentration of the standard is fixed at 113 meq/L, the calculation becomes the following:

$${\text{C}}\,{\text{sample}}\,\left[ {{\text{meq}}/{\text{L}}} \right] = 113 \times \frac{{\Delta {\text{A}}\,{\text{sample}}}}{{\Delta {\text{A}}\,{\text{standard}}}}$$

Quantitative determination with the Vohlard test

The chlorides were precipitated by a silver nitrate excess and titrated with ammonium thiocyanate. The concentrated nitric acid (d20 = 1,40; 1 mL) was added to an aliquot of the sample, and silver nitrate (0.1N) over the chlorides present.

The sample was heated until the silver chloride coagulated and then left to cool in the dark. A saturated ferric alum solution was used as an indicator. The titration with ammonium sulphocyanate (0.1N) was carried out until red [22].

Results’ calculation

$${\text{NaCl}}\left( {\text{g}} \right) = \left( {{\text{N}} - {\text{n}}} \right) \times 0.005845$$

N = silver nitrate 0.1N (mL), n = Saturated solution of ammonium sulfocyanate.

Spectrophotometric method validation parameters

The determination of the method validation parameters was evaluated as reported by [5]. The regression line, detection (LOD, LOD = 3.3σS) and quantification (LOQ; LOQ = 10σS) limits were found from the calibration line.

Type A and B uncertainties (U) were determined following the EURACHEM/CITAC guide [23].

$${\text{U}}\,{\text{Type}}\,{\text{A}} = \sqrt {\frac{{{\text{Variance}}}}{{{\text{Degrees}}\,{\text{of}}\,{\text{freedom}}}}}$$
(1)

Type B was defined with a metrology approach.

  • U(a) (it was associated with analytical balance) was obtained considering a certificate of calibration (0.00060 g), stability (0.000032 g), and repeatability (0.000029 g);

  • U(p) (it was associated with 0.2 mL pipette) was obtained considering a certificate of repeatability (0.00020 mL) and calibration (0.096 mL);

  • U(mr) (it was associated with KCl);

  • U(ct) (associated with the calibration curve) was obtained by measuring the standard in triplicate at three concentrations.

    $${\text{U}}\left( {{\text{ct}}} \right){\text{S}}\frac{x/y}{b} \times \sqrt {1/{\text{n}} + 1/{\text{m}}}$$
    (2)

S = standard deviation of the residual, \(\mathrm{n}\) = points used for the calibration line, m = readings taken for each sample.

Recovery was obtained by dosing NaCl content ten times in the tomato sample.

The repeatability was calculated at two concentrations of sodium chloride, 0.04%, and 0.1%. Shapiro–Wilk, Uber, and Dixon tests evaluated the data’s reliability.

The method reproducibility wasobtained by performing ten analyses of a same sample.

$${\text{Reproducibility}}=\frac{\text{The standard deviation of samples obtained by the tested method}}{\text{The standard deviation of samples tested by the Voholard method}}$$
(3)

The accuracy was obtained as following:

$$\mathrm{Accuracy}=\frac{|{\overline{\mathrm{ X}} }_{\mathrm{ Voholard }test}-\overline{\mathrm{ X} }|}{\sqrt{{\mathrm{S}}_{\mathrm{r}}^{2}+{\mathrm{U}}_{\mathrm{Voholard test}}^{2}}}\le {t}_{p\nu }$$

\({\overline{\mathrm{X}} }_{\mathrm{ Voholard }test}\) = concentrazioni medie ottenute con il test Voholard, \(\overline{\mathrm{X} }\) = concentrazioni medie ottenute con il metodo testato, S = deviazione standard, U = incertezza, t = t student.

Statistical analysis

Statistica software version 7.0 (StatSoft, Hamburg, Germany) was used to obtain statistical significance.

Results

This work validated a colorimetric method to determine NaCl levels in canned tomatoes. The validation plan compared the results obtained by the automatized colorimetric method to those obtained by the Volhard test, considered the reference method [22]. The linearity range, LOQ, LOD, measuring range, uncertainty, and accuracy were appraised.

Linearity range

Linearity defines the range over which laboratory tests are accurate. The measurement interval was established based on the data obtained from the calibration curve. R2 = 0.99 confirmed the linearity of the curve (Fig. 1, Table S1).

Fig. 1
figure 1

Calibration curve

Full size image

The ANOVA test proved the normal residual distribution (Fig. 2, Table S2). The hypothesis was considered valid since was less than a 10% variance between observed and expected values at each level.

Fig. 2
figure 2

Residual distribution determined by the Anova test. ei = absolute residuals; eNi = normalized residuals); eSi = studentized residuals; eji = standardized residuals

Full size image

Detection limit, quantification limit, and measuring range

The detection (LOD) and quantification (LOQ) limits were obtained from the calibration line (Fig. 2). Three repetitions were carried out for each point.

The measuring range was calculated considering the linearity range and the LOQ (Fig. 3, Table S3).

Fig. 3
figure 3

LOD, LOQ, and decision limit of the automatized colorimetric method

Full size image

The method’s limits were adequate to detect the current chloride contents in canned tomatoes.

Uncertainty

The total uncertainty is a parameter that evaluates the random error sources attributed to the measurement, including systematic error dispersion. It is calculated by estimating the errors associated with the various stages of the analysis (e.g. pre-analytical effects, homogenization, pipetting, weighing, extraction, injection, derivatization, recovery, and calibration curves). Type A uncertainties are associated with method repeatability. It was calculated by repeating the analysis ten times and evaluating the variations between the results (the value with a more significant relative difference was considered in calculating the type A uncertainty). The standard deviation (0.05) confirms the test’s repeatibility (Table 2).

Table 2 Type A uncertainty
Full size table

The uncertainties of category B are those associated with the reference materials. The type B uncertainty was calculated based on the minimum purity criterion. The type B uncertainty of the scales and the micropipette was obtained by doubling the last recorded uncertainty contribution, which can be obtained from the respective calibration certificates, to become independent of the minor variations between one calibration and another. The measured uncertainty (resulting uncertainty) was less than 10% (Table 3).

Table 3 Type B uncertainty
Full size table

Recovery

Recovery is the fraction of analyte present or added to the portion of material under test, extracted, and measured. The recovery depends on the analyte concentration. It can be assessed by analyzing a certified reference material, fortifying a white, fortifying a matrix containing the analyte, and comparing the results obtained with a standard method. The experimental procedure analyzed a tomato sample ten times to detect the NaCl content. The recovery was determined at one point as the measurement range was linear and not very extensive (theoretical recovery = 0.139; experimental recovery = 0.145).

The recovery was considered suitable for experimental purposes since the difference between the theoretical and experimental values (0.0007%) was less than the uncertainty of the method (Table S4).

Precision

Precision estimates the agreement between the results of subsequent measurements of the same measure.

Repeatability and reproducibility are two types of precision. Repeatability is the goodness of the agreement between the results of subsequent measurements conducted under the same measurement conditions. The repeatability of the automatized test was calculated at the two concentrations of sodium chloride considered values admissible in canned tomatoes by current legislation: 0.04% (Fig. 4) and 0.1% (Fig. 5). Reproducibility is the goodness of the agreement between the results of subsequent measurements of the same sample by measuring conducted under non-homogeneous measurement conditions. The reproducibility limit (R) is the maximum value, predictable at a certain confidence level, of the absolute difference between two results obtained under reproducibility conditions. The results are suspicious if the difference between the two results is more significant than R. The Shapiro–Wilk test can detect the normality of distribution; the Dixon and Huber test finds anomalous values [24].

Fig. 4
figure 4

The precision of the method for the determination of NaCl (0.04%) in canned tomatoes

Full size image
Fig. 5
figure 5

The precision of the method for the determination of NaCl (0.1%) in canned tomatoes

Full size image

The reproducibility of the automatized test was confirmed by the normality of distribution tested by Shapiro–Wilks test and zero anomalous values in Dixon and Huber tests.

Accuracy

Accuracy is the goodness of the agreement between the average value obtained from an adequately numerous series of results and the accepted reference value. It can be evaluated by comparing the results obtained by analyzing a series of samples (standard or real) with the method to be validated and with a reference method. [25].

The accuracy was calculated at two sodium chloride concentrations of 0.04% (Fig. 6) and 0.1% (Fig. 7). The statistical analysis (T-test) showed that the tested method had a lower variability than the Voholard method, verifying the hypothesis H0 for a significance level greater than 0.05 for a coverage of 95% or greater than 0.01 for 99% coverage.

Fig. 6
figure 6

Accuracy of the method for the determination of NaCl (0.04%) in canned tomatoes

Full size image
Fig. 7
figure 7

Accuracy of the method for the determination of NaCl (0.1%) in canned tomatoes

Full size image

Discussion

In the practice of commodity laboratories, innovative analysis methods that use automated equipment often replace the official methods carried out by specialized personnel to analyze laboratory samples. The automated analyzers decrease the analysis time, improve safety, and reduce the cost of analyses. All new analytical methods must be subjected to robust validation procedures to ensure the results’ reliability. Analytical method validation indicates if an analytical procedure is appropriate for its planned purpose. This work validated a colorimetric method, performed by an automated analyzer, to determine NaCl levels in canned tomatoes using the international guidelines [14, 26]. The objective was achieved by comparing the results obtained by a colorimetric method to those obtained by the Volhard test, considered a reference method (Ministerial Decree 03/02/1989—SO GU SG n 168 20/07/1989 Met 33). According to ISO/IEC, 2005, the parameters evaluated were linearity range, LOQ, LOD, measuring range, recovery, measurement uncertainty (associated with the analytical data), precision, and accuracy. Statistical analyses were used to estimate validation qualities against fixed acceptance criteria.

The range over which there is a proportional relationship between signal and analyte concentration is defined by the linearity of the calibration curve points (linearity range) [27]. This work verified linearity by running three replicates at five concentrations of NaCl (Standard) over the claimed measuring interval. The calibration curve’s linearity was acceptable since the correlation factor (R) was 0.999. Statistical tests assessed the acceptability of the estimated result (random residues were less than 10%) [5, 8, 28, 29].

The test lower range limits were obtained from LOD and LOQ. LOD, or minimum detectable quantity, is the concentration of analyte that produces a signal significantly different from the blank. LOQ, or minimum detectable quantity, is the minimum concentration of analyte that can be detected. The low value of LOD and LOQ and linear calibration in the studied range confirmed method sensitivity.

The minimum precision value was obtained by testing the method’s repeatability when the two sodium chloride concentrations admissible by law in canned tomatoes (0.04% and 0.1%) were checked. The results obtained with the two tests (Volhard and colorimetric) were compared. The Shapiro-Wilks test proved that it was verified at a probability level of p = 95% (a = 5%), and Dixon and Huber’s tests verified the absence of anomalous data. The Dixon test identifies a single anomalous value (too low or too high) placed at the beginning or end of a data sequence ordered in the direction of growth. Huber’s test is one of the more “robust” tests for detecting anomalous data. It uses a median-centric evaluation unaffected by the extreme values of an ordered data series. The test, also known as MAD (Median Absolute Deviation), allows for eliminating an unspecified number of anomalous data. In Dixon and Huber tests, the values are considered anomalous if the verification result is more significant than the critical value reported in the Unichim Manual 179/1: 2001 [30]. In this case, the anomalous data are eliminated, and the procedure is repeated on the remaining data.

In our validation process, when determining the repeatability of the determination of the NaCl concentration (0.1%) in canned tomatoes, the presence of two anomalous data was found when statistical tests [Grupps 5% (pair) and Huber 5%] were performed. Therefore the two values (0.63 and 0.64) were eliminated, and the tests were repeated.

The uncertainty was calculated using the metrological method, considering all the sources of significant uncertainty due to the measurement procedure (calibration curve) and the apparatus (technical balance, flask, micropipette, sample dilution) to calculate the composite standard uncertainty. The measured uncertainties were considered irrelevant since they were less than 10% of the results [31].

The recovery value was considered congruous since the difference obtained between the theoretical and experimental values was less than the uncertainty of the method.

The accuracy was evaluated by comparing the data series of the colorimetric method with those obtained by the Voholard test. The results showed that the accuracy between the two methods was equivalent since the significance level (P) greater than 0.05 for coverage of 95% was insignificant, and T experimental was < T critical.

Robustness studies of the colorimetric method were not needed because the method presented no deviations.

Conclusions

This work subjected an automated spectrophotometric method for the determination of NaCl in canned tomatoes to the validation process. The colorimetric method’s performance was equivalent to those obtained by the reference method (Volhard test) in terms of linearity, LOD, LOQ, measured range, precision, accuracy, and recovery. The automatized method improves workflow efficiency, standardizes processes, decreases errors due to intense staff concentration and fatigue caused by excessive experiments, and makes staff’s time more productive. Therefore, the validated method could be helpful in commodity laboratories where many samples need to be processed quickly. The scientific community should validate the automatic methods available on the market as soon as possible in the hope that, in a short time, they can be routinely used for food analysis.