3.1.1.  Protective measures

Avoid breathing of dust, vapours, mist or gas. Wear full-face particulate filtering respirator type N100 (US) or type P3 (EN 143) respirator cartridges when working with the solid material. Wear protective gloves, goggles and clothing. Take special care to avoid skin exposure if handling solutions and work in adequately ventilated areas. Wash hands thoroughly after handling.

3.1.2.  Emergency procedures

General advice

Immediately call a POISON CENTER or doctor/physician. Show this safety data sheet to the doctor in attendance. Move out of dangerous area.

If inhaled

Move person into fresh air. If not breathing give artificial respiration. Consult a physician.

In case of skin contact

Wash off with soap and plenty of water. Consult a physician.

In case of eye contact

Rinse thoroughly with plenty of water for at least 15 minutes and consult a physician.

If swallowed

Immediately call a POISON CENTRE or doctor/physician. Never give anything by mouth to an unconscious person. Rinse mouth with water.

3.1.3.  Spillage / Projections

Contain spillage and then collect by wet-brushing and place in container for disposal. Keep in suitable, closed containers for disposal according to local regulations.

4.1.1.  Materials

• Acetonitrile. HPLC gradient grade (HiPerSolv Chromanorm, VWR)

• Methanol. HPLC gradient grade (HiPerSolv Chromanorm, VWR)

• Ultrapure water (Milli-Q)

• Aflatoxin B₁ stock (BIPM ref. OGP.030) and calibration (BIPM ref. OGP.029) solutions.

• Impurity standards: AfB₂, AfB_2a, AfP₁, AfQ₁ and DIOL (First Standard via NIM China).

4.1.2.  Sample preparation

Ampoules of the stock or calibration solution were opened and $1\mathrm{mL}$ of solution was transferred to glass injection vials and placed in the autosampler at $4\mathrm{°C}$ for immediate analysis.

4.1.3.  Instrumentation

Liquid chromatography system Agilent 1100 HPLC equipped with a diode array detector (DAD) and coupled to a Sciex 4000 Qtrap mass spectrometry detector.

4.1.4.  Liquid chromatography parameters

To avoid contamination of the sensitive MS instrument by the high content of the main AfB₁ compound, the mobile phase was diverted to waste during the elution window of AfB₁ so it was measured in the DAD detector but it did not reach the MS.

4.1.5.  DAD detection parameters

The absorption wavelengths used for the detection of the main component AfB₁ were $362\mathrm{nm}$ and $263\mathrm{nm}$ (reference $450\mathrm{nm}$).

4.1.6.  MS/MS detection parameters

The 4000 QTRAP was operated in a negative-positive switching electrospray ionization (ESI) mode. The capillary voltage was set at $\mathrm{5 500}\mathrm{V}$ and the source temperature at $600\mathrm{°C}$ for the positive ESI. For the negative ESI mode, the capillary voltage was $-\mathrm{4 500}\mathrm{V}$ and the source temperature $550\mathrm{°C}$. Nitrogen was used as the ion source gas, curtain gas and collision gas. The Gas 1 and Gas 2 of the ion source were set at 55 psi and 60 psi, respectively. The curtain gas (CUR) was set at 15 psi. The Collision Gas (CAD) was set at “Mid”. Table 1 lists the optimized transitions and conditions for multiple reaction monitoring (MRM) detection of AfB₁ and its most frequent, structurally related impurities depicted in Figure 2.

Table 1.  Transition ions and MS/MS parameters for the detection of AfB₁ and its impurities in MRM mode. Transitions marked with an asterisk were used for quantification purposes.

Figure 2-1 — Aflatoxin B₁, C₁₇H₁₂O₆; 312.06 Da; ${\text{P}}_{\text{kow}}$: -1.23

Figure 2-2 — Aflatoxin B₂, C₁₇H₁₄O₆; 314.06 Da; ${\text{P}}_{\text{kow}}$: 0.3

Figure 2-3 — Aflatoxin B₁ 8,9-dihydrodiol (DIOL); C₁₇H₁₄O₈, MW:346.07 Da;

Figure 2-4 — Aflatoxin B_2a; C₁₇H₁₄O₇, MW:330.07 Da; ${\text{P}}_{\text{kow}}$: 0.09

Figure 2-5 — Aflatoxin Q₁, C₁₇H₁₂O₇; 328.06 Da; ${\text{P}}_{\text{kow}}$: 0.3

Figure 2-6 — Aflatoxin P₁, C₁₆H₁₀O₆; 298.04 Da; ${\text{P}}_{\text{kow}}$: 1.77

Figure 2-7 — Aflatoxin M₁, C₁₇H₁₂O₇; 328.06 Da; ${\text{P}}_{\text{kow}}$: -0.5

Figure 2-8 — Aflatoxin G₁ , C₁₇H₁₂O₇; 328.07 Da; ${\text{P}}_{\text{kow}}$: 0.83

Figure 2-9 — Aflatoxin G₂, C₁₇H₁₄O₇; 330.07 Da; ${\text{P}}_{\text{kow}}$: 0.09

Figure 2-10 — Aflatoxicol, C₁₇H₁₄O₆; 314.08 Da; ${\text{P}}_{\text{kow}}$: -1.46

Figure 2 — Chemical structure of Aflatoxin B₁ related impurities optimised for detection by LC-MS/MS.

4.1.7.  Data analysis

Data was evaluated using Analyst 1.6.3 software (SCIEX). Peak integration was verified manually for all samples and standards. Peak areas were extracted for quantification and uncertainty evaluation.

4.2.1.  Materials

• Acetonitrile. HPLC gradient grade (HiPerSolv Chromanorm, VWR)

• Ultra-Micro Cell Quartz Cuvette, $10\mathrm{mm}$ light path (Perkin Elmer).

• Aflatoxin B₁ stock (OGP.030) and calibration (OGP.029) solutions.

4.2.2.  Sample preparation

Ampoules of the stock or calibration solution were opened and an aliquot of the material was transferred to the cuvette for analysis without further manipulation. Acetonitrile was used in a reference cuvette to perform the instrument auto-zero (blank subtraction).

4.2.3.  Instrumentation

Measurements were performed in a PerkinElmer Lambda 650 UV/VIS spectrometer.

4.2.4.  UV-spectrophotometry parameters

A wavelength scan measurement method was used for qualitative analysis (i.e. identification of absorption maxima) and a fixed wavelength method to determine the absorbance value of solutions for quantitative analysis.

Wavelength scan method parameters:

• Deuterium lamp: on

• Tungsten lamp: on

• Scan from $370.00\mathrm{nm}$ to $190.00\mathrm{nm}$

• Data interval: $1.00\mathrm{nm}$, scan speed: $266.75\mathrm{nm}/\min$

• Ordinate mode: A (Absorbance)

• Cycle: 1

• Slit: $2\mathrm{nm}$

• No cell changer

Fixed wavelength method parameters:

• Deuterium lamp: on

• Tungsten lamp: on

• Wavelengths: $223\mathrm{nm}$, $263\mathrm{nm}$ and $360\mathrm{nm}$

• Ordinate mode: A (Absorbance)

• Cycle: 3

• Slit: $1\mathrm{nm}$

• Gain: Auto

• Response $0.2\mathrm{s}$

• No cell changer

4.2.5.  Data analysis

The typical wavelength spectrum of AfB₁ is represented in Figure 3. The three observed absorption maxima at $223\mathrm{nm}$, $263\mathrm{nm}$ and $360\mathrm{nm}$ were selected as the fixed wavelengths for the quantitative analysis of AfB₁.

Figure 3 — UV-vis absorption spectrum of AfB₁.

Data were acquired using the Perkin Elmer UV WinLab software and absorbance measurements were extracted for data evaluation.

where

$m_p$

weighed mass of AfB₁ powder

$b_p$

buoyancy correction of powder weighing

$w_p$

purity of AfB₁ powder

$m_{\text{sol}}$

weighed mass of stock solution

$b_{\text{sol}}$

buoyancy correction of solution weighing

6.2.1.  Study design

Short-term stability studies consider the impact of temperature and time to simulate potential transport conditions and/or storage conditions. Any significant influence of light, UV-radiation, moisture, etc. is excluded provided that the storage facilities and transport/packaging conditions are appropriate.

The stability study of OGP.029 followed an isochronous design [9] with a reference temperature of $-20\mathrm{°C}$ and study temperatures of $4\mathrm{°C}$, $22\mathrm{°C}$ and $40\mathrm{°C}$ and storage in the dark. Selected sample units were transferred from study temperatures to the reference temperature every two weeks until the end of the eight-week study.

The sample units were selected using a random stratified sampling scheme from each of the quartiles of the approximately 200-unit batch. The study was composed of three units at the reference temperature and twelve units at each of the study temperatures, requiring 39 samples in total (Table 6).

Table 6.  Temperatures, time points and sample units selected for the stability study of OGP.029. Units between parentheses were kept as back-up.

6.2.2.  Stability study measurements

Two samples of each time point and temperature conditions were measured under repeatability conditions (same day and run) in a randomised manner using the LC-DAD method for AfB₁ and the LC-MS/MS method for the related structure impurities. Ampoules were vortexed before opening and two aliquots of $0.5\mathrm{mL}$ were transferred into separate injection vials to have duplicate measurements of each sample (4 measurements for each condition). Representative TIC and DAD chromatograms of OGP.029 samples are shown in Figure 5.

Figure 5 — Total ion chromatogram (top) and DAD 362nm chromatogram (bottom) of a representative sample of OGP.029.

The only two structure-related impurities found in OGP.029 above the limit of detection of the LC-MS/MS method were AfB₂ and AfB_2a. Four to five standard calibration solutions containing AfB₂ and AfB_2a standards in the ranges $4.3-21.4\mathrm{ng}/\mathrm{g}$ and $9.5-652\mathrm{ng}/\mathrm{g}$, respectively, were prepared to quantify these impurities by external calibration (Figure 6). Triplicate injections per standard level were spread over the analytical sequence. For the main component AfB₁, no calibration was performed but instead the peak areas from the LC-DAD chromatograms were directly evaluated.

Figure 6 — External calibration functions for the quantification of structure-related impurities AfB₂ (top) and AfB_2a (bottom) in OGP.029.

6.2.3.  Stability data evaluation

Calculated mass fraction values of impurities AfB₂ and AfB_2a and peak area values of the main component AfB₁ were normalized to the average values of the reference samples (stored at $-20\mathrm{°C}$) to render results comparable. Statistical outliers were only removed in case of known technical reasons. As a first evaluation step, normalized data were plotted according to the injection sequence to discard any potential analytical drift. The slopes of the fitted regression lines were not significant (t-test) at the 95% confidence level (Figure 7).

Figure 7 — OGP.029 stability data analysis to identify potential trends in the analytical sequence. Data correspond to normalized mass fractions of AfB₂ (a) and AfB_2a (b) impurities and normalized peak areas of the main compound AfB₁ as detected by LC-DAD at $263\mathrm{nm}$ © and $362\mathrm{nm}$ (d).

For each temperature, regression lines of the normalized values versus storage time were calculated. The fitted regression model was tested for overall significance (loss/increase due to storage) using an F-test (95% confidence level). The stability results of the main component and the impurities at each of the studied temperatures are shown in Figure 8.

Figure 8 — Stability results of OGP.029 for the main component AfB₁ and impurities at the three studied temperatures. Data correspond to normalized mass fractions of AfB₂ (top) and AfB_2a (middle) impurities and normalized peak areas of the main compound AfB₁ as detected by LC-DAD at $263\mathrm{nm}$ (bottom). AfB₁ data acquired at $362\mathrm{nm}$ were similar to those at $263\mathrm{nm}$ (not shown). Dotted lines represent stability-associated uncertainty intervals of the normalized values as a function of the storage time. Single lines are fitted where the stability trend was found to be significant at the 95% confidence level.

At $4\mathrm{°C}$, all studied compounds in OGP.029 were stable for 8 weeks, as evidenced by the absence of a significant trend (F-test, 95% confidence level). At $22\mathrm{°C}$, a small declining trend was observed for the main component AfB₁ over the studied period although the changes were considered negligible at 2 weeks. At $40\mathrm{°C}$, both the main component and the AfB_2a impurity degraded significantly over the 8-week period.

In conclusion, the OGP.029 calibration solution can be shipped safely in the dark at $22\mathrm{°C}$ if the transport time does not exceed 2 weeks. In addition, storage at $4\mathrm{°C}$ for a period of up to 8 weeks does not result in significant changes in composition.

6.3.1.  Study design

Homogeneity between ampoules is evaluated to ensure that the assigned value of the calibration solution is valid for all units of the material, within the stated uncertainty. It is therefore necessary to determine this between-unit variation and incorporate it in a combined uncertainty estimate.

Ten ampoules were selected from the OGP.029 batch following a randomly stratified sampling scheme. They were measured under repeatability conditions using UV-spectrophotometry for total aflatoxin content and LC-DAD-MS/MS for AfB₁ and the detected structure-related impurities AfB₂ and AfB_2a.

6.3.2.  Homogeneity study measurements

The selected ampoules were allowed to equilibrate at room temperature were vortexed before opening. They were analyzed in a random order to ensure that any trends in the ampouling process could be distinguished from potential trends in the analytical sequence.

Three aliquots per ampoule were measured consecutively by UV-spectrophotometry using the fixed wavelength method at $223$, $265$ and $360\mathrm{nm}$. Triplicate measurements of each aliquot gave rise to a total of 9 measurements per ampoule and wavelength. Another three aliquots ($>500\mathrm{\mu L}$) were transferred into glass injection vials and stored at $-20\mathrm{°C}$ until LC-DAD-MS/MS analysis.

6.3.3.  Homogeneity data evaluation

Absorbance and peak area values were normalized with respect to the average result for each of the studied compounds. Statistical outliers were only removed in case of known technical reasons. Linear regression functions were calculated for the normalized values arranged in ampouling and analysis order. The slopes of the lines were tested for significance at a 95 % confidence level to discard the presence of trends. Figure 9 shows the UV-$360\mathrm{nm}$ aflatoxin measurements displayed according to the order of analysis and of ampouling. The dispersion of measurement results across the analytical sequence implied no need for trend correction and this variability is encompassed by the uncertainty associated to homogeneity.

Figure 9 — Homogeneity results of OGP.029 as determined by UV-spectrophotometry at $360\mathrm{nm}$ plotted according to the analysis (top) or ampouling (bottom) order.

Homogeneity samples results for the main compound AfB₁ and related impurities AfB₂ and AfB_2a obtained by LC-DAD-MS/MS are shown in Figure 10 as normalized peak areas.

Figure 10 — Homogeneity results of OGP.029 calibration solution as determined by LC-DAD-MS/MS for AfB₁ ($362\mathrm{nm}$ and $263\mathrm{nm}$ detection, top) and related impurities AfB₂ and AfB_2a (MS/MS detection, bottom).

Quantification of between-unit heterogeneity was done by analysis of variance (ANOVA), which allows for the separation of the variation between ampoules ($s_{\text{bb}}$) from that associated with the method repeatability ($s_{\text{wb}}$). These variances are calculated as follows:

where $\mathrm{M}{\mathrm{S}}_{\text{btw}}$ and $\mathrm{M}{\mathrm{S}}_{\text{with}}$ are the mean sums of squares between- and within-units obtained by the ANOVA evaluation and $n$ is the number of replicates per ampoule ($n=3$).

The standard deviation between the sample units is used as the estimator for the between-units variability. The measurement variation sets a lower limit to this estimator reflected in $\mathrm{M}{\mathrm{S}}_{\text{btw}}$ being smaller than $\mathrm{M}{\mathrm{S}}_{\text{with}}$. This does not imply that the material is perfectly homogeneous, but only shows that the study set-up was not adequate to detect evidence of heterogeneity. In this case, the maximum heterogeneity that could be hidden by the intrinsic variability of the method, $u{*}_{\text{bb}}$, is calculated according to the equation below: [10]

where $p$ is the number of measured ampoules ($p=10$) and $n$ is the number of measurement replicates per ampoule ($n=3$).

The final uncertainty from homogeneity ($u_{\text{bb}}$) is estimated as $s_{\text{bb}}$ or $u{*}_{\text{bb}}$, depending on which of these is larger. This uncertainty is presented in Table 7 for every measured compound using the LC-DAD-MS/MS method. The F-test at the 95% confidence level did not detect significant differences between ampoules for any of the studied compounds.

Table 7.  Homogeneity uncertainty results of OGP.029 from data generated by LC-DAD-MS/MS

Homogeneity results obtained by UV-spectrophotometry at the 3 measured wavelengths are shown in Table 8. They confirm the findings of the chromatographic method and therefore the AfB₁ calibration solution can be regarded as homogeneous.

Table 8.  Homogeneity uncertainty results of OGP.029 from data generated by UV-spectrophotometry at the three measured wavelengths $223$, $265$ and $360\mathrm{nm}$.

where

$m_{\text{stock}}$

weighed mass of AfB₁ stock solution

$b_{\text{stock}}$

buoyancy correction of stock solution weighing

$w_{\text{stock}}$

AfB₁ mass fraction of the stock solution

$m_{\text{sol}}$

weighed mass of calibration solution OGP.029

$b_{\text{sol}}$

buoyancy correction of calibration solution weighing