Sample preparation: Analytical standards in iDQuant™ Standards Kit for Pesticide Analysis (SCIEX)4 represent a well-characterized mixture of 206 pesticides with various chemical properties and spanning different chemical classes.
Three different baby foods were selected, which will be referred to as A, B, and C. Each brand was analyzed as an unspiked matrix, spiked at 1 ppb of pesticide mix, and spiked at 10 ppb pesticide mix. Each sample and treatment combination was prepared in triplicate (Table 1).
The baby foods were extracted using acidic ACN (1% acetic acid in acetonitrile). For each, 15 g of homogenous food blend was added to a tube with 15 mL of the acidic ACN solvent. These were shaken vigorously for 1 minute, then centrifuged. The supernatant was then sampled for 1:10 dilution in mobile phase and LC-MS/MS analysis. This procedure is a “dilute-and-shoot” approach, adapted from the AOAC QuEChERS method but intended to produce a sample which has gone through very little matrix cleanup, in order to assess method performance in challenging matrix conditions.
Chromatography: Chromatography was performed using a Phenomenex Luna Omega Polar C18 (2.6 µm x 100 mm) at a flow rate of 0.4 mL/min, and an injection volume of 1 µL (Table 2) using the ExionLC™ System.
Mass spectrometry: The SCIEX Triple Quad 7500 LC-MS/MS System – QTRAP Ready was used to analyze the samples, to assess trace level quantitative performance of the analyte panel in both neat standards and the baby food blends.
Multiple reaction monitoring (MRM), employing two precursor to product ion transitions for each analyte in the panel, represents the most important analytical technique for trace level pesticide quantification. For each transition, optimized voltages are defined for compound-specific parameters such as Collision Energy (CE), but ion source parameters are also set which apply to all analytes in the method. The OptiFlow Pro Ion Source parameters were retuned as there are some key differences in design to the earlier Turbo V™ Ion Source. For example, the source temperature can be set lower than precedent methods, which can benefit the performance for thermally sensitive analytes. The source parameters can be found below (Table 3).
Data processing: All data were processed using the SCIEX OS Software. SCIEX OS Software, now available on SCIEX triple quadrupole platforms, integrates acquisition and processing into a single software platform to be used from sample analysis through to results reporting.
Most pesticides in the 209-compound panel were detected at levels below 1 ppb in the vial. Roughly half of them were measured at a lower limit of 0.1 ppb or lower (Figure 1). This is particularly notable as these values were achieved using a dilute-and-shoot method with a small (1 µL) injection volume.
Figures 2 and 3 illustrate the sensitivity and performance of this method on the SCIEX Triple Quad 7500 LC-MS/MS System – QTRAP Ready for some example pesticides. The chromatograms shown represent increasing concentrations of standard calibrators in neat solvent solution.
Analysis in baby food matrix
Each baby food blend was measured without fortification, and with a low- and high-level fortification. Figure 4 illustrates three example pesticides in the three matrix treatments. The EU Commission Directives outlines limits for pesticides in baby foods specifically.1 The measures, intended to protect vulnerable populations such as infants, set generic tolerance limits of 0.01 mg/kg in the food sample. Table 4 shows the direct correlation between the LOQs determined with this method and the mass on column and LOQ in sample. The mass on column value is critical for comparing LOQs across acquisition methods that differ by platform, injection volume, or separation strategy. The LOQ in sample is the value on which most regulatory limits are based, and its relationship to the LOQ in vial is dependent on sample extraction and preparation as well as injection volume.
With the 1:10 dilution, 1 µL injection volume, and absence of matrix cleanup steps, the achievement of LOQs as low as 0.01 represents exemplary instrument performance in challenging matrix conditions, and suggest that sample preparation adjustments to meet the 0.001 ng mass on column required to assess food products for EU compliance is readily achievable.
Qualitative confirmation of pesticide detection in complex matrix is achieved using ion ratios of two MRM transitions per compound and matching these to the compound-dependent ratios observed in calibration standards. This ion ratio confirmation can be automatically visualized, flagged, and filtered in the SCIEX OS Software results table (Figure 5).
For some compounds increased instrument sensitivity provides lower detection limits, however this added sensitivity can result in the upper range of the calibration curve becoming limited by detector saturation. In Figure 6, two examples of calibration curves are shown. For chlorfluazuron, the response across the concentration range remains linear. For desmedipham, a very good detection limit was observed (0.05 ppb) however, the signal saturation is apparent at the highest concentrations, as seen by a plateau in the calibration curve. The linear range extends from 0.05 ppb to 100 ppb.
Identifying the true linear dynamic range for calibration curves generated on a large panel of pesticides can be time consuming, as variable analyte responses are common, leading to different LLOQs and ULOQs across the analyte panel. One SCIEX OS Software feature that makes this aspect of data processing more streamlined is the Automatic Outlier feature. Here, the user defines the criteria the calibration standards must meet in order to achieve optimal quantitative method performance. In Figure 7, a screenshot from SCIEX OS Software shows the functionality of the outlier removal tool, along with a calibration curve where the high-end points which have been automatically excluded from the curve and model fit.
Reproducibility and robustness
It is readily apparent that method reproducibility as well as hardware robustness are critical for routine analysis of low-level residues in food. Two evaluations were performed in order to demonstrate reproducibility and robustness. First, triplicate analyses of the pesticide panel at a 10 ppb concentration in the baby food matrix were used to assess and report %CV values across the panel of compounds. Figure 8 shows a breakdown of where the majority of these %CV values fall.
Second, a test was executed in order to demonstrate extreme robustness and reproducibility across a great number of injections in a complex matrix. This test injected over 2000 samples of pesticide mixture in a black tea matrix used exclusively for this purpose. Figure 9 shows exemplary consistency from the first to the last injections with no cleaning or maintenance having taken place over the course of the test.
The SCIEX Triple Quad 7500 LC-MS/MS System - QTRAP Ready was used to analyze a panel of 209 pesticides in both neat solvent and baby food matrices. The primary objective was to evaluate the absolute sensitivity of the system and investigate how the method would perform in a baby food matrix composed of fruits and vegetables. To this end, it was found that sub-ppb levels were easily achievable with high-quality quantitative metrics (reproducibility, peak quality, linear response) even with a dilute sample and small injection volumes.
This promising new generation of analytical technology allows for the continued pursuit of low-level residues for food safety testing while addressing historical challenges of food analysis by LC-MS/MS. Larger dilutions and smaller injection volumes will help laboratories maintain maximum uptime and reduce the need for cleaning and decontamination.
The SCIEX OS Software platform for acquisition and processing makes the process seamless from sample to report, and advanced features accelerate and streamline the quantification process. Having access to greater sensitivity than what might be currently required is a growing trend in laboratory preparedness for future changes in the regulatory and analytical landscape.